sample_number int64 8 203k | abstract stringlengths 21 17.4k | article stringlengths 0 18.9k | compressed_article stringlengths 0 10.9k | compression_time float64 0 0.78 | original_tokens int64 0 3.79k | compressed_tokens int64 0 2.29k | token_saving stringclasses 2 values |
|---|---|---|---|---|---|---|---|
177,060 | we investigate the relation between trees of fragmenting granules ( tfgs ) and the locations of concentrated magnetic flux in internetwork areas .
the former have previously been identified with mesogranulation . while a relationship has been suggested to exist between these features , no direct evidence has yet been provided .
we present some preliminary results that show that concentrated magnetic flux indeed collects on the borders of tfgs . | magnetic field in the quiet solar photosphere , while ubiquitously present , is concentrated at the edges of convective cells .
we refer the interested reader to @xcite for a comprehensive review of small - scale magnetism in the lower solar atmosphere .
the convective flows expunge field from the interiors of granules and concentrate it as `` magnetic elements '' ( mes ) in the intergranular downdrafts .
supergranular flows advect field to supergranular boundaries where it forms the magnetic network .
gas motions are also expected to push field to the edges of cells of an intermediate , so - called mesogranular scale .
this scale has indeed been observed in the positions of photospheric magnetic elements in internetwork areas .
@xcite also observed `` voids '' in active network .
however , no evidence has been presented that the observed cells outlined by mes correspond to mesogranular cells .
mesogranules have been associated with so - called `` trees of fragmenting granules '' ( , called `` active granules '' by @xcite ) .
tfgs consist of repeatedly splitting granules that originate from a single granule .
they may live for several hours , much longer than the lifetime of an individual granule .
flow fields derived from granular motions also show convergence at the borders of tfgs .
one would thus expect mes to lie predominantly on tfg boundaries .
an image sequence spanning several hours of moderately quiet sun was recorded using the solar optical telescope on the hinode spacecraft @xcite on march 30 , 2007 .
the narrowband filter imager was used to record stokes i and v in the photospheric fe i line at 630.2 nm .
more details can be found in @xcite .
their study of magnetism in quiet - sun internetwork yielded the locations of strong concentrations of magnetic flux that we use in this analysis . to reduce noise , the fe i intensity data is initially filtered using an optimum filter , and averaged over 3 frames in time . to ensure proper segmentation and separation of granules ,
the data is scaled up spatially by a factor of two .
noise is further reduced through convolution with a 8-pixel gaussian .
we identify granules by computing the curvature @xmath0 in each pixel in 4 directions .
if the minimum curvature is positive , the pixel is labeled as part of a granule .
the granule mask is then extended to values of curvature up to @xmath1 , following the cst algorithm .
we apply erosion - dilation processing with a 3-pixel + -shaped kernel to remove very small features .
finally , granules are grouped into families by following them in space and time .
two granules are considered to be members of the same family if there is a path through the granule mask forward in time that connects them . at the start of the sequence ,
all families consist of a single granule .
as time progresses , some families die out , others grow , and new families appear .
if two granules of different families merge , we keep the oldest family .
the lifetime of tfgs is on the order of several hours , so one must expect to wait a similar amount of time before the tfg pattern is established . here , we will study a single frame taken about 5 hours after the start of the sequence .
the segmentation of the granules and the subsequent grouping into families has been performed for several variations of the number of frames over which is averaged in time , the width of the gaussian , and the level to which the masks are extended to negative curvature values .
the chosen parameters appear to be fairly robust , i.e. , the resulting pattern does not change much within some range of the chosen values for the parameters .
the process is most sensitive to the amount that the mask is extended .
too much extension may merge separate granules , causing larger tfgs , while too little extension results in very limited grouping .
frame number 524 in the fe i sequence , recorded at 05:25:33 , is shown in fig . [
fig : fig1 ] .
it shows the granular pattern , with the borders of tfgs overlaid in black , and the positions of mes overlaid as white diamonds .
the granular pattern is shaded in dark gray in network areas .
the segmentation there is not good , because of the confusion between granules and bright points . as a result ,
the tfg pattern there is not trustworthy .
also , the analysis by @xcite did not identify mes in the network .
this is not a problem for our analysis , because our interest is primarily in the internetwork . a closeup of the white box is shown in fig .
[ fig : fig2 ] .
many mes appear to lie preferentially near the borders of tfgs .
the interiors of large tfgs are mostly devoid of mes .
these preliminary results are encouraging .
it thus seems highly likely that the cells described by , e.g. , and the `` voids '' in maps made with the hinode sp instrument found by @xcite correspond to mesogranules .
further statistical study is required to quantify the relationship between the borders of tfgs and the positions of mes .
while these preliminary results are encouraging , it must be verified that mes lie statistically closer to the borders of tfgs than , e.g. , to the border of randomly placed cells of similar size .
mes that emerge in tfg interiors are expected to migrate to the borders in a matter of perhaps an hour , then along the borders of tfgs to the network . in this context
it is of interest to study the motions of mes in detail and to quantify the contribution of tfgs to the formation of the magnetic network and the diffusion of magnetic field .
_ hinode _ is a japanese mission developed and launched by isas / jaxa , with naoj as domestic partner and nasa and stfc ( uk ) as international partners .
it is operated by these agencies in co - operation with esa and nsc ( norway ) . | magnetic field in solar photosphere concentrated at edges convective cells to @xcite for review small - scale magnetism solar convective flows expunge field from granules concentrate magnetic elements in intergranular downdrafts supergranular flows advect field to boundaries magnetic network gas motions push field to edges cells mesogranular scale observed in photospheric magnetic elements @xcite observed voids in active network no evidence cells outlined by mes correspond to mesogranular cells mesogranules associated with trees of fragmenting granules active granules tfgs granules live several hours longer than lifetime flow fields from granular motions show convergence at borders tfgs mes on tfg boundaries image sequence quiet sun recorded solar optical telescope @xcite on march 30 2007 narrowband filter stokes i and v in photospheric fe i line at 630.2 nm details in @xcite study magnetism quiet - sun internetwork yielded locations of strong concentrations magnetic flux fe i intensity data filtered using optimum filter averaged over 3 framessegmentation separation of granules data scaled spatially by factor two noise reduced through convolution with 8-pixel gaussian identify granules computing curvature @xmath0 in each pixel 4 directions if minimum curvature positive pixel labeled part of granule granule mask extended to curvature up to @xmath1 cst algorithm erosion - dilation processing with 3-pixel + -shaped kernel remove small features granules grouped into families space time two granules members same family if path through granule mask connects them start families consist of single granule time some families die others grow new families appear if two granules families merge keep oldest family lifetime of tfgs several hours wait similar before tfg pattern established study single frame 5 hours after start sequence segmentation grouping into families performed for variations of number of frames averaged time width gaussian masks extended to negative curvature values chosen parameters robust resulting pattern change process sensitive to mask too much extension merge granules larger tfgs too little extension limited grouping frame number 524 in sequence recorded at 05:25:33 shown in fig .shows granular pattern borders tfgs overlaid black positions mes white diamonds granular pattern shaded in dark gray in network areas segmentation not good confusion between granules bright points tfg pattern not trustworthy analysis @xcite identify mes in network not problem for our analysis interest in internetwork closeup white box shown in fig . mes lie near borders tfgs interiors of large tfgs devoid of mes preliminary results encouraging likely cells described `` voids '' in maps hinode sp instrument @xcite correspond to mesogranules further statistical study required relationship between borders tfgs positions mes results must verified mes lie closer to borders tfgs than randomly placed cells similar size mes in tfg interiors migrate to borders hour to network interest to study motions of mes quantify contribution tfgs to magnetic network diffusion magnetic field hinode _ japanese mission developed launched by isas / jaxa naoj domestic partner nasa stfc international partners operated by agencies co - operation with esa nsc ( norway ) | 0.190696 | 1,281 | 626 | , Saving $0.0 in GPT-4. |
177,064 | within the realm of qcd sum rules , one of the most important areas of application of this nonperturbative approach is the prediction of the decay constants of heavy mesons .
however , in spite of the fact that , indisputably , the adopted techniques are , of course , very similar , we encounter rather dissimilar challenges , or obstacles , when extracting from two - point correlators of appropriate heavy - light currents interpolating the mesons , the characteristics of charmed mesons with different spin . in view of this
, it seems worthwhile to us to revisit this issue for the case of charmed pseudoscalar mesons @xmath0 and vector mesons @xmath1 address = institute for high energy physics , austrian academy of sciences , nikolsdorfergasse 18 , a-1050 vienna , austria address = institute for high energy physics , austrian academy of sciences , nikolsdorfergasse 18 , a-1050 vienna , austria , altaddress = d .
v. skobeltsyn institute of nuclear physics , m. v. lomonosov moscow state university , 119991 , moscow , russia address = infn , sezione di roma tre , via della vasca navale 84 , i-00146 , roma , italy | any description of physical systems bound by the strong interactions that deserves to be attributed as _ reliable _ should be based on quantum chromodynamics ( qcd , the quantum field theory that governs the strong interactions ) and should be of _ non_-perturbative nature .
one formalism that in contrast to , for instance , lattice gauge theory offers the prospect of providing analytical insights , namely , in form of relations between features of hadrons and the parameters of qcd is realized by the technique of _ qcd sum rules _
their formulation proceeds along a well - established sequence of steps : * define the correlation function of a nonlocal product of operators ( in particular , of appropriate quark currents ) that interpolate the hadron under study , i.e. , have nonvanishing matrix elements between vacuum and this hadron state . *
evaluate this correlation function at the hadron level , by inserting a complete set of states , and at the qcd level , by applying wilson s operator product expansion ( ope ) reshaping any nonlocal product to a series of local operators , to obtain perturbative contributions , represented by dispersion integrals of spectral densities , and non - perturbative ( np ) terms , labelled as `` power '' contributions , representing the `` vacuum condensates '' of the local ope operators .
* get rid of subtraction terms left behind by cauchy s integral formula and suppress the effects of hadron excitations and continuum , by performing a borel transformation from momentum to another variable , the borel parameter @xmath2 * hide your ignorance about higher states by postulating quark
hadron duality ; thus assume that all contributions of hadronic excited and continuum states cancel against those of perturbative qcd above effective thresholds @xmath3 .
taking advantage of the experimental knowledge @xcite of the masses @xmath5 of the mesons discussed , our goal is to perform advanced @xcite extractions of the decay constants , @xmath6 of both pseudoscalar ( p ) @xcite and vector ( v ) @xcite charmed mesons ( regarded as bound states of a charmed quark @xmath7 of mass @xmath8 and , in the non - strange case , of a light quark @xmath9 of mass @xmath10 or , in the strange case , of a light quark @xmath11 of mass @xmath12 ) from the two - point correlation functions of adequately chosen interpolating currents . as indicated , by way of construction the qcd sum rules derived along the lines sketched above are expressed , at qcd level , in terms of spectral densities @xmath13 and non - perturbative terms @xmath14 at appropriate renormalization scale @xmath15 terming the qcd side of such sum rule as the _ dual _ correlator , @xmath16 we refer to the characteristics predicted by this sum rule for a ground - state meson as its _ dual _ mass and _ dual _ decay constant : * for the charmed pseudoscalar mesons @xmath17 we select as their interpolating operator the pseudoscalar current @xmath18 to extract @xcite both @xmath19 and decay constants @xmath20 defined by @xmath21 : @xmath22 * for the charmed vector mesons @xmath23 we use as interpolating operator the vector current @xmath24 to obtain @xcite the masses @xmath25 and decay constants @xmath26 defined by @xmath27 from the sum rule@xmath28 99 m. a. shifman , a. i. vainshtein , and v. i. zakharov , _ nucl .
b _ * 147 * ( 1979 ) 385 .
k. a. olive et al .
( particle data group ) , _ chin .
c _ * 38 * ( 2014 ) 090001 .
w. lucha , d. melikhov , and s. simula , _ phys .
d _ * 76 * ( 2007 ) 036002 , arxiv:0705.0470 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ phys .
b _ * 657 * ( 2007 ) 148 , arxiv:0709.1584 [ hep - ph ] .
w. lucha , d. i. melikhov , and s. simula , _ phys . atom .
nucl . _ * 71 * ( 2008 ) 1461 .
w. lucha , d. melikhov , and s. simula , _ phys .
b _ * 671 * ( 2009 ) 445 , arxiv:0810.1920 [ hep - ph ] .
d. melikhov , _ phys .
b _ * 671 * ( 2009 ) 450 , arxiv:0810.4497 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ phys .
d _ * 79 * ( 2009 ) 096011 , arxiv:0902.4202 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ j. phys .
g _ * 37 * ( 2010 ) 035003 , arxiv:0905.0963 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ phys .
b _ * 687 * ( 2010 ) 48 , arxiv : 0912.5017 [ hep - ph ] .
w. lucha , d. i. melikhov , and s. simula , _ phys . atom .
_ * 73 * ( 2010 ) 1770 , arxiv:1003.1463 [ hep - ph ] .
w. lucha , d. melikhov , h. sazdjian , and s. simula , _ phys .
d _ * 80 * ( 2009 ) 114028 , arxiv:0910.3164 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _
j. phys .
g _ * 38 * ( 2011 ) 105002 , arxiv:1008.2698 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ phys .
b _ * 701 * ( 2011 ) 82 , arxiv:1101.5986 [ hep - ph ] .
w. lucha , d. melikhov , and s. simula , _ phys .
b _ * 735 * ( 2014 ) 12 , arxiv:1404.0293 [ hep - ph ] . k. g. chetyrkin and m. steinhauser , _ phys .
b _ * 502 * ( 2001 ) 104 , arxiv : hep - ph/0012002 .
k. g. chetyrkin and m. steinhauser , _ eur .
j. c _ * 21 * ( 2001 ) 319 , arxiv : hep - ph/0108017 .
p. gelhausen , a. khodjamirian , a. a. pivovarov , and d. rosenthal , _ phys .
d _ * 88 * ( 2013 ) 014015 ; * 89 * ( 2014 ) 099901(e ) .
m. jamin and b. o. lange , _ phys .
d _ * 65 * ( 2002 ) 056005 , arxiv : hep - ph/0108135 . | description of physical systems bound by strong interactions based on quantum chromodynamics non_-perturbative lattice gauge theory analytical insights relations between features hadrons parameters qcd by qcd sum rules formulation define correlation function of nonlocal product of operators appropriate quark currents interpolate hadron nonvanishing matrix elements between vacuum hadron state evaluate correlation function at hadron level inserting complete states qcd level applying wilson s operator product expansion reshaping nonlocal product to local operators obtain perturbative contributions by dispersion integrals of spectral densities non - perturbative terms as `` power '' contributions representing vacuum condensates of local ope operators get rid of subtraction terms by cauchy s integral formula suppress effects of hadron excitations continuum borel transformation from momentum to borel parameter hide ignorance about postulating quark hadron duality assume contributions of hadronic excited continuum states cancel against of perturbative qcd above thresholdsexperimental knowledge @xcite masses @xmath5 mesons goal perform advanced @xcite extractions decay constants pseudoscalar p vector v @xcite charmed mesons bound states charmed quark @xmath7 mass @xmath8 light quark @xmath9 mass @xmath10 light quark @xmath11 mass @xmath12 two - point correlation functions interpolating currents qcd sum rules expressed spectral densities @xmath13 non - perturbative terms @xmath14 renormalization scale @xmath15 qcd side sum rule dual _ correlator characteristics sum rule ground - state meson dual mass decay constant charmed pseudoscalar mesons @xmath17 select pseudoscalar current @xmath18 extract @xcite @xmath19 decay constants @xmath20 @xmath21 charmed vector mesons @xmath23 use interpolating vector current @xmath24 obtain @xcite masses @xmath25 decay constants @xmath26 @xmath27 sum rule@xmath28. shifman vainshtein zakharov nucl 147 ( 1979 ) 385 olive et al . particle data group chin 38 ( 2014 ) 090001 luchamelikhov s simula phys 76 2007 ) 036002 arxiv:0705.0470 hep - ph lucha melikhov s simula . 657 2007 ) 148 arxiv:0709.1584 hep - ph lucha melikhov s simula phys 71 2008 ) 1461 lucha melikhov s simula. 671 2009 ) 445 arxiv:0810.1920 hep - ph melikhov. 671 2009 ) 450 arxiv:0810.4497 hep - ph lucha melikhov s simula phys 79 2009 ) 096011 arxiv:0902.4202 hep - ph lucha melikhov s simula . 37 2010 ) 035003 arxiv:0905.0963 hep - ph lucha melikhov s simula . 687 2010 ) 48 arxiv 0912.5017 hep - ph lucha melikhov s simula phys atom 73 2010 ) 1770 arxiv:1003.1463 [ hep - ph lucha melikhov sazdjian s simula phys . 80 2009 ) 114028 arxiv:0910.3164 [ hep - ph lucha melikhov s simula j phys .38 2011 ) 105002 arxiv:1008.2698 hep - ph lucha d. melikhov s simula phys 701 2011 ) 82 arxiv:1101.5986 hep - ph lucha melikhov simula 735 2014 ) 12 arxiv:1404.0293 hep - ph chetyrkin m. 502 2001 ) 104 arxiv hep - ph/0012002 chetyrkin steinhauser 21 2001 ) 319 hep - ph/0108017 p. gelhausen. khodjamirian pivovarov d. rosenthal phys 88 2013 ) 014015 89 2014 ) 099901(e ) m. jamin b o. lange 65 2002 ) 056005 arxiv hep - ph/0108135 | 0.264119 | 1,754 | 911 | , Saving $0.1 in GPT-4. |
177,082 | we report on the accuracy of the measurement of the two photon decay width of a higgs boson with the mass of 120 gev at the tesla photon collider , assuming a @xmath0 integrated luminosity of 80 fb@xmath1 in the hard part of the spectrum .
the qcd radiative corrections for the quark pair background processes are taken into account by a reweighting procedure .
we found that the product @xmath2 can be measured with the statistical error of 1.8@xmath3 in one year of run .
_ desy zeuthen , platanenallee 6 , 15738 zeuthen , germany _ | the search for the higgs boson to explore the mechanism of symmetry breaking is one of the most important tasks of the experimental program at the future colliders .
while it can only be produced in association with another particle at an @xmath4 collider , the higgs boson can be produced singly in the s - channel of the colliding photons at a photon collider .
if the mass of the higgs boson is light , as indicated by the current electroweak measurements , then it will be found by the time a @xmath0 collider is constructed .
the aim of such collider will be then a precise measurement of higgs properties . among these
the measurement of the two photon decay width of the higgs boson is particularly important .
the higgs boson can be produced in @xmath5 collisions via a one - loop diagram , in which all charged particles whose masses derive from the higgs mechanism can appear inside the loop . in the standard model ( sm ) ,
the dominant contributions come from the top and w boson loops .
a deviation of the two photon width from the sm prediction indicates additional contribution from unknown particles , and is a signature of physics beyond the sm .
for example , the minimal extension of the sm predicts the ratio of the two photon width @xmath6/ @xmath7 1.2 @xcite for a higgs boson with a mass of 120 gev , assuming a supersymmetry scale of 1 tev and the chargino mass parameters @xmath8 and @xmath9 of 300 and 100 gev , respectively . at a @xmath0 collider ,
the two photon decay width is obtained by measuring the higgs formation cross section in the reaction @xmath10 , where x is the detected final state .
the number of detected events is proportional to the product @xmath11@xmath12br@xmath13 .
measuring the formation cross section will determine this product .
an independent measurement of the branching ratio br(@xmath14 ) at @xmath4 collider allows a determination of the @xmath0 partial width . in this paper
we study the accuracy of the measurement of the two photon decay width for a higgs boson with the mass of 120 gev .
the study is performed for the photon collider option of the tesla linear collider operated at centre - of - mass energy of 210 gev .
the assumed data volume corresponds to a total @xmath0 integrated luminosity of 400 fb@xmath1 @xcite , which amounts to about @xmath15 = 80 fb@xmath1 in the hard part of the spectrum .
the feasibility of the measurement of the two photon decay width of the higgs boson in this mass region has also been reported by @xcite .
the cross section for the higgs boson formation is given by a breit - wigner approximation @xmath16 where @xmath17 is the higgs boson mass , @xmath11 and @xmath18 are the two photon and total decay width of the higgs boson , @xmath19 and @xmath20 are the initial photon helicities and @xmath21 is the @xmath0 centre - of - mass energy .
the initial photons should have equal helicities , so that @xmath22 = 0 , in order to make a spin-0 resonance as it is the case of the higgs boson .
if polarised photon beams are used , the signal cross section is increased up to a factor of 2 .
the experimentally observed cross section is obtained by folding this basic cross section with the @xmath0 collider luminosity distribution .
a higgs boson with the mass of 120 gev is detected in the @xmath23 final state .
the event rate is given by the formula : @xmath24 where the conversion factor @xmath25 is 3.8937966@xmath26@xmath27 fb gev@xmath28 .
it depends strongly on the differential luminosity , @xmath29 .
the shape of the luminosity distribution depends on the electron and laser beam parameters .
the electron and laser beam energy considered for this study are 105 gev and 1 ev , respectively , resulting in the maximum photon energy of about 75 gev , suitable to study a higgs boson with the mass of 120 gev .
the combination of the polarisation of the laser and the electron are chosen such to make the generated photon spectrum peak at its maximum energy .
the helicity combination of the two high energy photons is arranged such that @xmath22 = 0 state is dominant .
the luminosity distributions for the @xmath22=0 and @xmath30 = 2 are provided by the circe2 @xcite program and shown in figure [ fig_lumi ] .
beamstrahlung , secondary collisions between electrons and laser beam photons , and other non - linear effects are taken into account when calculating the @xmath0 luminosity .
the resulting value of @xmath29 is 1.6 fb@xmath1/gev . the branching ratios br(h @xmath31 ) , br(h @xmath32 ) and the total width are taken to be 0.22@xmath3 , 68@xmath3 and 4 mev , respectively , and are calculated with the hdecay @xcite program including qcd radiative corrections .
a signal rate of about 20000 events per year can be achieved under these conditions .
the main background processes are the direct continuum @xmath33 and @xmath34 production . due to helicity conservation
, the continuum background production proceeds mainly through the states of opposite photon helicities , making the states @xmath35 .
choosing equal helicity photon polarisations the cross section of the continuum background is suppressed by a factor @xmath36 , with @xmath37 being the quark mass .
unfortunately , this suppression does not apply to the process @xmath38 , because after the gluon radiation the @xmath39 system is not necessarily in a @xmath40 state .
the surviving background is large and overwhelms the signal .
the signal @xmath41 process , as well as the background @xmath42(g ) events are generated with the pythia @xcite program .
parton evolution and hadronization are simulated using the parton shower and the string fragmentation models .
a convolution with the luminosity distribution is performed , and kinematic cuts of @xmath43 and @xmath21 greater than 80 gev are imposed during the event generation , @xmath44 denoting the polar angle of the produced quarks .
the qcd corrections to @xmath42 have to be taken into account for a reliable background estimation .
for this purpose we use the total cross sections shown in figure [ fig_cs ] , which include the soft and hard gluon emission @xcite , virtual corrections @xcite and non - sudakov form factors @xcite to weight the event .
the event topology is given by the parton shower model .
the effective cross sections , the number of expected and generated signal and background events are presented in table [ tab_cs ] . to simulate the response of the apparatus we use simdet @xcite , a fast monte carlo program which smears the kinematics of the particles in the final state according to the detector resolutions .
the detector used in the simulation follows the proposal presented in the technical design report of the tesla @xmath45 linear collider @xcite .
however , the changes in the detector design for @xmath0 collisions will affect only the forward region @xcite which is not important for this analysis .
the analysis aims to select events with two or three multiplicity jets from the higgs boson decay .
two of these jets contain bottom quarks .
the invariant mass of the jets has to be consistent with the higgs mass .
jets are reconstructed using the durham clustering scheme @xcite with the parameter y@xmath46=0.02 .
we select events with a visible energy larger than 95 gev and the energy imbalance longitudinal to the beam direction below 10@xmath3 of the visible energy .
figures [ fig_costh]a , b shows the distribution of the cosine of the thrust angle for the signal and background events .
the s - channel signal process has an isotropic angular distribution , while the t - channel background processes are forward peaked .
we require the absolute value of this quantity to be below 0.7 .
the cross section for the continuum production of the charm quark is 16 times larger than for bottom quarks , therefore b - quark tagging is crucial in this analysis .
the b - tagging algorithm relies on a neural network with 12 inputs and 3 output nodes , described in ref .
the most important inputs comprise the impact parameter joint probability tag introduced by aleph @xcite , the @xmath47 corrected vertex invariant mass obtained with the zvtop algorithm written for the sld experiment @xcite and a one - prong charm tag using the largest and second largest track impact parameter significances in @xmath48 - @xmath49 and @xmath48 - @xmath50 , in jets where zvtop found only the primary vertex . the spectrum of the neural network output @xmath51 to discriminate b - quark jets from u- , d- , s- and c - quark jets is presented in figure [ fig_btag]a .
the performance of the neural network b - tag in z@xmath52 decays is shown in figure [ fig_btag]b .
we require at least 2 vertices to be found and ask that the @xmath51 to be greater than 0.95 .
the b - tagging efficiency is 70@xmath3 and corresponds to a purity of 98@xmath3 .
the efficiency of this selection is found to be 36@xmath3 .
a background not discussed yet is the process @xmath54hadrons , where photons interact hadronically at the interaction point . with a total hadronic cross section in @xmath0 collisions of 400 nb
, we estimate about 1.4 such events per bunch crossing .
these events obscure the interesting physics processes described in the previous sections . for this reason this class of events needs to be included in the pythia simulation for overlap .
the hades @xcite program is used for this purpose .
a large fraction of this background is distributed at small angles and we believe that it can be reduced cutting on the polar angle of the tracks @xcite .
a careful study of this hadronic background is currently being performed .
the reconstructed invariant mass for the selected signal and background events is shown in figure [ fig_mass ] . to enhance the signal a cut on the invariant mass
is tuned such that the statistical significance of the signal over background is maximised .
events in the mass region of 106 gev @xmath55 126 gev are selected .
the number of estimated signal and background events in this window are 6018 and 7111 , respectively .
the two photon decay width of the higgs boson is proportional to the event rates of the higgs signal .
the statistical error of the number of signal events , @xmath56 , corresponds to the statistical error of this measurement .
here @xmath57 is the number of observed events , while @xmath58 is the number of expected background events .
we obtain @xmath59}{[\gamma ( \rm h \to \gamma \gamma ) \times \rm br ( \rm h \to \rm b \bar{\rm b})]}=\sqrt n_{\rm obs}/({\it n } _ { \rm obs}-{\it n}_{\rm bkg})=1.9\% .\ ] ] we can improve this number by 0.1@xmath3 doing a fit with the function @xmath60 with @xmath61 being the fraction of signal events , @xmath62 and @xmath63 the background and observed number of events in each bin of the invariant mass distributins and @xmath64 a fit parameter . we conclude that for a higgs boson with a mass @xmath17=120 gev we can measure the product @xmath2 with an accuracy of 1.8@xmath3 using an integrated luminosity corresponding to one year of data taking at the tesla photon collider .
t. ohgaki , t. takahashi and i. watanabe , hep - ph/9703301 ; g. jikia and s. sldner - rembold , nucl .
inst . and meth a472 ( 2001 ) 133 ; p. niezurawski , a.f .
zarnecki and m. krawczyk , acta physica polonica b34 ( 2003 ) 177 .
p. niezurawski , a.f .
zarnecki and m. krawczyk , hep - ph/0307183 .
telnov , _ a code for the simulation of luminosities and qed backgrounds at photon colliders _ , talk presented at the 2nd ecfa / desy workshop on physics and detectors for a 90 - 800 gev linear collider , st .
malo , france , april 2002 . | search for higgs boson symmetry breaking important future colliders produced particle at @xmath4 collider singly in s - channel colliding photons if mass higgs boson light found @xmath0 collider constructed aim precise measurement of higgs properties measurement two photon decay width higgs boson important higgs boson produced in @xmath5 collisions via one - loop diagram charged particles higgs inside loop standard model dominant contributions from top and w boson loops deviation of two photon width indicates additional contribution unknown particles signature of physics beyond sm minimal extension sm predicts two photon width/ 1.2 @xcite for higgs boson with mass 120 gev supersymmetry scale 1 tev chargino mass parameters @xmath8 and @xmath9 of 300 100 gev at @xmath0 collider two photon decay width obtained by measuring higgs formation cross section in reaction @xmath10 x detected final state number detected events proportional to product @xmath11@xmath13 formation cross section productindependent measurement branching ratio@xmath14 at @xmath4 collider allows @xmath0 partial width accuracy measurement two photon decay width higgs boson mass 120 gev photon collider option tesla linear collider centre mass energy 210 gev assumed data volume corresponds @xmath0 luminosity 400 fb@xmath1 @xmath15 = 80 fb@xmath1 hard part spectrum feasibility measurement photon decay width reported by @xcite cross section higgs boson formation given by breit - wigner approximation @xmath17 higgs boson mass two photon total decay width initial photon helicities centre - of - mass energy initial photons should have equal helicities @xmath22 = 0 spin-0 resonance polarised photon beams signal cross section increased factor 2 cross section obtained folding cross section with @xmath0 collider luminosity distribution higgs boson mass 120 gev detected in @xmath23 final state event rate given by formula conversion factor @xmath25 3.8937966@xmath26 fb gev@xmath28 depends on differential luminosityluminosity distribution depends electron laser beam parameters electron laser beam energy 105 gev 1 ev maximum photon energy 75 gev higgs boson 120 gev polarisation laser electron photon spectrum maximum energy helicity combination energy @xmath22 = 0 state dominant luminosity distributions for @xmath22=0 @xmath30 = 2 provided by circe2 @xcite program shown in figure beamstrahlung secondary collisions non - linear effects @xmath0 luminosity value @xmath29 is 1.6 fb@xmath1/gev branching ratios @xmath31 @xmath32 total width 0.22@xmath3 68@xmath3 4 mev calculated with hdecay @xcite program qcd radiative corrections signal rate of 20000 events per year main background processes continuum @xmath33 @xmath34 production helicity conservation background production through opposite photon helicities equal helicity photon polarisations cross section continuum background suppressed by factor @xmath36 @xmath37 quark mass suppression to @xmath38 after gluon radiation @xmath39 system not in @xmath40 state surviving background overwhelms signalsignal @xmath41 process background @xmath42 events generated pythia @xcite program parton evolution hadronization simulated parton shower string fragmentation models convolution with luminosity distribution kinematic cuts of @xmath43 @xmath21 greater than 80 gev imposed during event generation @xmath44 polar angle produced quarks qcd corrections to @xmath42 for background estimation total cross sections figure soft hard gluon emission virtual corrections non - sudakov form factors weight event event topology parton shower model effective cross sections expected generated signal background events presented in table_cs response simdet @xcite fast monte carlo program smears kinematics particles detector resolutions detector follows tesla @xmath45 linear collider changes affect forward region not events with two three multiplicity jets from higgs boson decay two jets contain bottom quarks invariant mass consistent with higgs mass jets reconstructed durham clustering scheme parameter y@xmath46=0.02 select events with visible energy larger than 95 gev energy imbalance longitudinal to beam direction below 10@xmath3 energyfigures b distribution cosine thrust angle signal background events s - channel signal process isotropic angular distribution t - channel background processes forward peaked require absolute value quantity below 0.7 cross section continuum production charm quark 16 times larger than bottom quarks b - quark tagging crucial - tagging algorithm neural network 12 inputs 3 output nodes important inputs impact parameter joint probability tag @xmath47 corrected vertex invariant mass one - prong charm tag largest track impact parameter significances @xmath48 - @xmath49 - spectrum neural network output @xmath51 b - quark jets from u- d s c - quark jets figure_btag]a performance network b - tag in z@xmath52 decays figure]b require 2 vertices @xmath51 greater than 0.95 b - tagging efficiency 70@xmath3 purity 98@xmath3 efficiency selection 36@xmath3 background process @xmath54hadrons photons interact hadronically total hadronic cross section @xmath0 collisions 400 nb 1.4 events per bunch crossing events obscure physics processes previous sectionsclass of events in pythia simulation for overlap hades @xcite program used large fraction background distributed at small angles reduced cutting on polar angle tracks @xcite study of hadronic background performed reconstructed invariant mass for selected signal and background events shown in figure signal cut on invariant mass tuned statistical significance signal over background events in mass region of 106 gev @xmath55 126 gev selected estimated signal and background events are 6018 and 7111 two photon decay width of higgs boson proportional to event rates higgs signal statistical error of signal events @xmath56 corresponds to error measurement @xmath57 observed events @xmath58 expected background events obtain @xmath59}=1.9\% improve number by 0.1@xmath3 with function @xmath60 @xmath61 fraction of signal events @xmath62 @xmath63 background and observed number of events in each bin invariant mass distributins @xmath64 fit parameterconclude higgs boson mass @xmath17=120 gev measure product @xmath2 accuracy 1.8@xmath3 integrated luminosity one year data tesla photon collider ohgaki takahashi watanabe hep - ph/9703301 jikia sldner rembold 2001 ) 133 niezurawski zarnecki krawczyk acta physica polonica b34 2003 ) 177 hep - ph/0307183 code simulation luminosities qed backgrounds photon colliders 2nd ecfa desy workshop detectors 90 - 800 gev linear collider st malo france april 2002 | 0.436469 | 2,827 | 1,454 | , Saving $0.1 in GPT-4. |
177,083 | the ground state of a classical two - dimensional ( 2d ) system with a finite number of charges particles , trapped by two positive impurity charges localized at a distance @xmath0 from the 2d plane and separated from each other by a distance @xmath1 are obtained .
the impurities are allowed to carry more than one positive charge .
this classical system can form a 2d - like classical molecule that exhibits _ structural transitions _ and _ spontaneous symmetry breaking _ as function of the separation between the positive charges before it transforms into two 2d - like classical atoms .
we also observe structural transitions as a function of the dielectric constant of the substrate which supports the charged particles , in addition to broken symmetry states and _ unbinding of particles . _
keywords : phase transitions , low dimensions , classical molecule | classical charged particles confined in a two - dimensional ( 2d ) layer has been a subject of interest in the last few years.@xcite it is well - known that classical 2d electrons form a wigner crystal.@xcite on the other hand , confined charged particles in 2d form 2d - like classical atoms @xcite which can exhibit very different melting and structural phase transitions.@xcite recently these properties were studied for a set of charged particles confined by a parabolic and hard - wall well.@xcite they order in a ring structure and a mendeleev table for these atomic like structures was constructed .
an essentially different confinement is the one produced by a coulomb potential .
2d charges which are confined by this potential also form 2d - like classical atomic structures but they exhibit very different structural transitions as compared to the one confined by parabolic potentials.@xcite in the present paper we extend our previous work for single classical atoms @xcite and vertically coupled atoms @xcite to laterally coupled atoms which form molecular structures . the system we consider here is composed of classical charged point particles interacting through a repulsive coulomb potential and held together by two remote impurities a distance @xmath2 away from the 2d plane where the charged particles can move .
the ground state of the system is calculated by using the monte carlo simulation technique and by minimization techniques .
we show that the ground state of this 2d - like classical molecule ( 2dcm ) presents structural phase transitions and if the impurities are separated further apart it is broken into two 2d - like classical atoms ( 2dca ) .
spontaneous broken symmetry and evaporation is observed , which depend strongly on the effective impurity charges and the distance between them .
our model consists of a finite number of particles @xmath3 , each of them with negative charge @xmath4 , which move in a 2d plane .
two fixed impurities , each of them with charge @xmath5 , are localized a distance @xmath6 from the 2d plane and are separated by a distance @xmath7 along the @xmath8-axis and are placed inside a medium with dielectric constant @xmath9 .
all particles interact through a coulomb potential ( see fig .
the energy for such a classical system is given by @xmath10 where @xmath11 is the position of the @xmath12 negative particle at the @xmath13 plane and @xmath14 ( @xmath15 ) is the position of the @xmath16 positive charge , _
i.e. _ the impurity . to keep the problem as simple as possible
we have neglected in eq .
( [ eq : eq1 ] ) the effects due to the image charges .
following our previous work @xcite we will vary the dielectric constant @xmath9 through which it is possible to change the effective confinement potential in a continuous way . for convenience ,
we express the energy in units of @xmath17 where @xmath18 is the dielectric constant of vacuum and all the distances are now in units of @xmath19 .
this reduces eq .
( [ eq : eq1 ] ) to @xmath20 where @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 . in order to obtain the minimum energy configuration we perform a numerical minimization of eq .
( [ eq : eq2 ] ) with respect to the @xmath27 variables ( coordinates @xmath28 of the negative particles ) .
to perform the numerical simulation , we start from a random distribution of negative particles and obtain the minimum energy configuration by using standard minimization techniques .
next we change , _ e.g. _
@xmath1 , slightly and take as the initial configuration the last minimum obtained by our numerical minimization . to have an independent check we also use the monte carlo simulation technique starting from several different initial configurations .
different from the results obtained with the 2d classical atom @xcite we found that bounded configurations , _
i.e. _ , @xmath29 @xmath30 @xmath31 , can exist if the total effective charge @xmath32 satisfies the condition @xmath33 for @xmath34 .
in fact , we observed that the condition for bounded configurations tends to behave as a single 2-d classical atom , if the distance between the two positive charges goes to zero . to analyze the molecule - atom transition we fix the value of the dielectric constant @xmath35
, we consider a system of six electrons and put the two positive charges at the points @xmath36 and @xmath37 , each with a charge @xmath38 . in fig .
2 we show the distance @xmath39 of the six negative particles to the point @xmath40 in the minimum energy configuration as a function of the distance between the positive charges @xmath41 . we clearly observe four distinct regions . in the first region ( i ) , @xmath42 is very small and the negative particles are arranged in two equilateral triangles , inside each other ( see fig .
3(a ) ) . the two positive charges ( their positions are indicated by the crosses in fig .
3 ) are so close to each other that they behave as one central charge .
this configuration is very different from the one which was found in the case of a parabolic confinement potential @xcite where one particle would be at ( x , y)=(0,0 ) and the other 5 in a ring around it . at @xmath43
the system undergoes a small structural transition to region ( ii)(see fig .
3(b ) ) which is of first order . note that increasing @xmath1 from @xmath44 acts as a symmetry breaking field which breaks the rotational symmetry of the configuration depicted in fig .
3(a ) . with increasing @xmath42 ,
the ground state configuration changes continuously up to @xmath45 where it undergoes a structural transition and a new configuration appears in region ( iii ) .
this transition is characterized by the fact that the particles change their positions abruptly ( compare solid ( ii ) and open ( iii ) symbols in fig .
3(c ) ) . with further increase of @xmath42
the system exhibits a new structural transition to region ( iv ) at @xmath46 where the configuration clearly exhibits now inversion symmetry with respect to the center of the coordinate system . in region
( iv ) the negative particles tend to stay in two triangles , each one close to one of the positive charges . up to this region
, the system behaves as a classical molecule .
namely , all distances between the negative particles are related to the positions of both positive charges . finally , for a value @xmath47 we observe more explicitly that the negative particles split in two sets , each one correlated to one positive charge .
no structural transition is associated with this behavior but , up to this point , the system behaves as two independent 2d classical atoms .
this can be seen by the fact that , up to @xmath47 the system exhibits rotational symmetry along the @xmath48-axis for each of the two positive charges .
different from the structural transition from regions ( i ) up to ( iv ) , the one from 2d classical molecule to 2d classical atom is not characterized by a structural transition , i.e. it is continuous . in order to visualize these stable states , in fig .
3 we show typical configurations of the system in each region and in fig .
4 we plot the distance of three negative particles related to each positive charge . as shown in fig .
4 , when @xmath49 the system behaves as two independent atoms , with the negative particles in two equilater triangles and the positive charges in the center of each one , with energy @xmath50 .
in order to show the order of the structural transitions we calculated the first of the energy as a function of @xmath42 , by numerically differentiating the energy curve .
the results for the first derivative are presented in fig .
5 . as can be seen the first derivative is discontinuous at the structural transitions ( i @xmath51 ii , ii @xmath51 iii and iii @xmath51 iv ) .
it is continuous throughout the region where the classical 2d molecule transits to two 2d classical atoms .
different structural transitions are observed if we analyze the behavior of the minimum energy configuration as a function of the dielectric constant ( @xmath52 ) . to this , we consider two positive charges ( each one with @xmath53 ) fixed at the points @xmath54 and @xmath55 and six negative particles which are free to move in the xy - plane .
for @xmath56 we are in region iv of the above . in fig .
6 we show the distance of those six negative particles to the point @xmath40 for the minimum energy configuration as a function of @xmath52 .
eleven different regions are observed , of which we plot in fig .
6 only the first ten of them . in region
( i ) , @xmath57 , the minimum energy configuration corresponds to the negative particles arranged in two triangles each one close to each positive charge ( see fig .
when @xmath52 is increased a structural transition takes place at @xmath58 . up to this value of @xmath52 , region ( iii ) , the configuration changes continuously and for @xmath59 one of the negative particles becomes weakly bound and moves to infinity , i.e. it evaporates . after this value , region ( iv ) ,
the minimum energy configuration changes continuously and again presents a structural transition , at @xmath60 , to a small region ( v ) .
this region goes up to @xmath61 , where a new structural transition takes place .
now , in the new region , region ( vi ) , another negative particle starts to become unbound when the value @xmath62 is reached .
the minimum energy configuration is stable in the new region ( vii ) until the value @xmath63 and again changes its structure .
after this point , again another particle starts to move away , region ( viii ) , and goes to infinity at @xmath64 .
up to this point , region ( ix ) , only three particles remain bound until @xmath65 . in region(x ) @xmath66 only two particles remain , with each negative particle sitting almost on top of each of the positive charges in a symmetrical arrangement . finally , when @xmath67 only one particle remains bounded .
unexpectedly , this negative particle stays in an asymmetric position , in this case at a distance @xmath68 .
this fact can be confirmed by taking the minimum of the potential energy . in this case
it presents a minimum at @xmath68 and a local maximum at @xmath69 , which corresponds to a symmetrical arrangement .
the existence of region ( x ) shows that a bounded configuration can exist in a region where the condition @xmath70 for @xmath34 is not satisfied .
the molecule - atom transition and structural transitions were studied in a 2d model system of negative particles confined by a coulomb potential of two positive impurity charges .
this system presents a large variety of structural transitions as a function of its parameters , the distance between the impurity charges and the dielectric constant . as observed for a 2d like atom the 2d like molecule exhibits highly asymmetric ground state configurations which are related to broken symmetry states .
however , the condition @xmath70 for the existence of a bounded configuration is only satisfied if the positive particles are very close .
finally , our results show that a finite 2d system confined by a nonlinear potential presents a much larger wealth of structural transitions and new states as compared to the ones confined by parabolic potentials .
w. p. ferreira , g. a. farias , and h. a. carmona are supported by the brazilian agencies , fundao cearense de amparo pesquisa ( funcap ) , brazilian national research council(cnpq ) and the ministry of planning ( finep ) .
the work of f. m. peeters is supported by the flemish science foundation ( fwo - vl ) , the eu - rtn network on surface electrons on mesoscopic structures and intas .
c. c. grimes and g. adams , phys .
* 42 * , 795 ( 1979 )
. t. ando , a. b. flower and f. stern , rev .
phys . * 54 * , 437 ( 1982 ) .
r. g. clark , phys .
* t39 * , 45 ( 1991 ) .
dubin and t.m .
oneil , rev . mod . phys . * 71 * , 87 ( 1999 ) .
j.h . chu and lin
rev . lett . * 72 * , 4009 ( 1994 ) .
m. saint jean , c. even , and c. guthmann , europhys .
lett . * 55 * , 54 ( 2001 ) .
h. aref and d. l. vainchtein , nature * 392 * , 769 ( 1998 ) .
a. isihara , solid state phys .
* 42 * , 271 ( 1989 ) . v. m. bedanov and f. m. peeters , phys .
b * 49*,2667 ( 1994 ) .
i. v. schweigert , v. a. schweigert , and f. m. peeters , phys .
lett . * 82 * , 5293 ( 1999 ) .
m. kong , b. partoens , and f. m. peeters , cond - mat/0106395 ( to appear in phys .
e ( 2002 ) ) .
i. v. schweigert , v. a. schweigert , and f. m. peeters , phys .
* 84 * , 4381 ( 2000 ) .
g. a. farias and f. m. peeters , solid state commun.*100 * , 711 ( 1996 ) .
b. partoens , v. a. schweigert , and f. m. peeters , phys .
lett . * 79 * , 3980 ( 1997 ) . | classical charged particles in two - dimensional 2d layer subject interest classical 2d electrons form wigner crystal confined charged particles in 2d form 2d - like classical atoms different melting structural phase transitions properties studied for charged particles confined by parabolic hard - wall well ring structure mendeleev table for constructed different confinement by coulomb potential 2d charges confined by form 2d - like classical atomic structures different structural transitions by parabolic potentials present paper work for to laterally coupled atoms molecular structures system of classical charged point particles interacting through repulsive coulomb potential held together by two remote impurities from 2d plane ground state calculated monte carlo simulation technique minimization techniques ground state 2d - like classical molecule presents structural phase transitions if impurities separated broken into two 2d - like classical atoms spontaneous broken symmetry evaporation observed depend on effective impurity charges distance model finite number of particles with negative charge move in 2d planetwo fixed impurities with charge @xmath5 localized distance @xmath6 from 2d plane separated by distance @xmath7 along @xmath8-axis placed inside medium with dielectric constant @xmath9 particles interact through coulomb potential fig energy for classical system given by @xmath10 @xmath11 position of @xmath12 negative particle at @xmath13 plane @xmath14 position @xmath16 positive charge impurity neglected in effects due to image charges vary dielectric constant @xmath9 change effective confinement potential express energy in units of @xmath17 @xmath18 is dielectric constant of vacuum distances in units of @xmath19 reduces eq . to @xmath20 @xmath21 minimum energy configuration perform numerical minimization of eq . @xmath27 variables coordinates @xmath28 of negative particles start from random distribution of negative particles obtain minimum energy configuration standard minimization techniques change @xmath1 slightly take as initial configuration last minimum obtained by numerical minimization independent check use monte carlo simulation technique from initial configurationsresults with 2d classical atom @xcite bounded configurations @xmath29 exist if effective charge @xmath32 satisfies condition @xmath33 @xmath34 condition for bounded configurations as single 2-d classical atom if distance between two positive charges zero analyze molecule - atom transition fix dielectric constant @xmath35 consider system six electrons two positive charges at points @xmath36 and @xmath37 each with charge @xmath38 fig . 2 distance @xmath39 of six negative particles to @xmath40 minimum energy configuration function of distance between positive charges @xmath41 four distinct regions first region @xmath42 small negative particles in two equilateral triangles two positive charges close behave as one central charge configuration different from parabolic confinement potential @xcite one particle at ( x , y)=(0,0 ) 5 ring around at @xmath43 system small structural transition to region ( first order increasing @xmath1 from @xmath44 symmetry breaking field breaks rotational symmetry configuration increasing @xmath42 ground state configuration changes up to @xmath45 transition new configuration in region ( iii )transition particles change positions abruptly compare solid ii open ( iii ) symbols in fig . 3(c ) increase @xmath42 system new structural transition to region ( iv ) at @xmath46 inversion symmetry center coordinate system in region ( iv ) negative particles stay in two triangles close to positive charges system behaves as classical molecule distances between negative particles related to positions positive charges value @xmath47 negative particles split in two sets correlated to one positive charge no structural transition system behaves as two independent 2d classical atoms @xmath47 rotational symmetry along @xmath48-axis for positive charges regions ( i ) to iv ) from 2d classical molecule to 2d classical atom not transition continuous fig . 3 typical configurations each region fig . 4 distance of three negative particles related to each positive charge @xmath49 system behaves as two independent atoms negative particles in two equilater triangles positive charges in center energy @xmath50 structural transitions calculated energy function of @xmath42 numerically differentiating energy curve results first derivative in fig . 5 .first derivative discontinuous at structural transitions @xmath51 continuous throughout region where classical 2d molecule transits to two 2d atoms different structural transitions minimum energy configuration dielectric constant @xmath52 consider two positive charges @xmath53 at points @xmath54 @xmath55 six negative particles free move in xy - plane for @xmath56 in region iv fig . 6 distance of six negative particles to point @xmath40 for minimum energy configuration function @xmath52 eleven regions observed 6 first ten in region ( i ) @xmath57 minimum energy configuration to negative particles in two triangles close to positive charge @xmath52 increased structural transition at @xmath58 up to @xmath52 configuration changes continuously @xmath59 weakly bound moves to infinity evaporates after minimum energy configuration changes structural transition at @xmath60 to small region ( v ) to @xmath61 new structural transition new region vi ) another negative particle unbound when value @xmath62 reached minimum energy configuration stable in region until value @xmath63 changes structureanother particle region ( viii ) to infinity at @xmath64 region ( ix ) three particles bound until @xmath65 region(x ) @xmath66 two particles remain negative particle on top of positive @xmath67 one particle remains bounded stays asymmetric position @xmath68 minimum potential energy minimum at @xmath68 maximum at @xmath69 symmetrical arrangement region ( x ) shows bounded configuration condition @xmath70 for @xmath34 not satisfied molecule - atom transition structural transitions studied in 2d model system of negative particles confined by coulomb potential of two positive impurity charges system presents variety of structural transitions parameters distance dielectric constant molecule exhibits asymmetric ground state configurations to broken symmetry states condition @xmath70 for bounded configuration satisfied if positive particles close results finite 2d system confined by nonlinear potential presents larger wealth of structural transitions new states parabolic potentials w. p. ferreira g. a. farias h. a. carmona supported by brazilian agencies brazilian national research council ministry of planning work of f. m.peeters supported flemish science foundation eu - rtn network surface electrons mesoscopic structures grimes adams phys . 42 795 ( 1979 ) ando flower stern rev. . 54 437 ( 1982 ) clark phys . t39 45 ( 1991 ) dubin oneil rev . 71 87 ( 1999 ) chu lin rev lett 72 4009 ( 1994 ) saint jean even guthmann europhys . 55 54 ( 2001 ) h. aref l. vainchtein nature 392 769 ( 1998 ) isihara solid state phys . 42 271 ( 1989 ) bedanov peeters 49,2667 ( 1994 ). schweigert schweigert peeters phys . 82 5293 ( 1999 ). kong partoens peeters cond - mat/0106395 phys . ( 2002 ). schweigert schweigert peeters phys . 84 4381 ( 2000 ). farias f peeters solid state.*100 711 ( 1996 ) partoensschweigert peeters phys . lett 79 3980 1997 ) | 0.502025 | 3,082 | 1,559 | , Saving $0.1 in GPT-4. |
177,088 | we give lower bounds for the combinatorial complexity of the voronoi diagram of polygonal curves under the discrete frchet distance .
we show that the voronoi diagram of @xmath0 curves in @xmath1 with @xmath2 vertices each , has complexity @xmath3 for dimension @xmath4 and @xmath5 for @xmath6 . | important distance measures for polygonal curves are the , and its variant , the .
the can be computed in @xmath7 time for two polygonal curves with @xmath2 vertices @xcite and the in @xmath8 time @xcite .
consider the following scenario : a set @xmath9 of @xmath0 polygonal curves in @xmath1 , each with at most @xmath2 vertices , is given .
the task is to find for several query curves the most similar curve in @xmath9 under the ( discrete ) . in this
setting , bereg , gavrilova , and zhu @xcite propose to compute the voronoi diagram of the given set of curves under the ( discrete ) and then to locate samples of the transformed query curve in this .
the voronoi diagram of polygonal curves can be represented using the correspondence @xmath10 however , very little is known about the voronoi diagram of polygonal curves under the ( discrete ) .
recently , bereg et al .
@xcite have shown for the an upper bound of @xmath11 and a lower bound of @xmath12 for the complexity of the voronoi diagram of @xmath0 polygonal curves in @xmath1 , for @xmath13 , with at most @xmath2 vertices each .
we prove the following lower bounds : [ thm : lower - bound ] for any @xmath14 , there is a set of @xmath0 polygonal curves in @xmath1 with @xmath2 vertices each whose voronoi diagram under the discrete frchet distance has combinatorial complexity @xmath3 for @xmath4 and @xmath15 and complexity @xmath5 for @xmath6 and @xmath15 .
our lower bounds significantly improve the lower bounds of bereg et al . @xcite .
for dimension @xmath16 the bound matches ( up to @xmath17 ) their upper bound .
although bereg et al . @xcite formulate the upper bound only for dimensions @xmath16 and @xmath18 , their proof generalizes to other dimensions yielding an upper bound of @xmath19 for @xmath20 .
thus , for dimensions @xmath4 the upper and lower bounds match ( up to @xmath17 ) , while for @xmath6 a gap of @xmath21 between the lower and upper bound remains . for @xmath6 our lower bound construction is a generalization of our two - dimensional construction .
one part of the generalized construction is still inherently two - dimensional .
we assume that by finding a suitable generalization of this part or by avoiding it , the gap of @xmath21 can be closed .
therefore we conjecture that the upper bound is tight ( up to @xmath17 ) . for any @xmath22
the voronoi diagram of @xmath0 polygonal curves in @xmath1 with at most @xmath2 vertices each has combinatorial complexity @xmath3 ( as function in @xmath0 ) . in the following
, we always use the parameters @xmath23 and @xmath2 to denote @xmath0 input curves in @xmath1 , each with at most @xmath2 vertices .
we give lower bounds @xmath24 on the combinatorial complexity of the voronoi diagram by showing that it contains at least @xmath25 _
voronoi regions_. by a voronoi region we mean a set of curves with a common set of nearest neighbors under the in the given set of input curves .
we show the lower bounds in theorem [ thm : lower - bound ] first for dimension @xmath26 ( lemma [ lem : lb1d ] ) and then for dimensions @xmath27 ( lemma [ lem : lb2d ] ) . for both lower bounds
we construct a set @xmath9 of @xmath0 curves
. then we construct @xmath28 query curves which all lie in different voronoi regions of the voronoi diagram of @xmath9 .
this implies that the voronoi diagram has complexity @xmath29 .
[ lem : lb1d ] for all @xmath0 and @xmath2 , there is a set of @xmath0 polygonal curves in @xmath30 with @xmath2 vertices each whose voronoi diagram under the discrete frchet distance has at least @xmath31 voronoi regions .
we construct a set @xmath9 of @xmath0 curves with @xmath2 vertices each for @xmath32 with @xmath33 .
@xmath9 will be a union of @xmath2 sets @xmath34 of @xmath35 curves each .
we show that the voronoi diagram of @xmath9 contains @xmath36 voronoi regions . and @xmath37.,scaledwidth=95.0% ] the construction for @xmath37 is shown in figure [ fig:1d ] .
we place @xmath2 points @xmath38 with distance @xmath39 between consecutive points on the real line .
a curve in @xmath9 has the form @xmath40 for some @xmath41 and point @xmath42 close to @xmath43 .
our construction uses the following points , curves , and sets of curves
. see figure [ fig:1d ] for an illustration for @xmath37 .
+ [ cols= " > , < , < " , ] again we claim that for all @xmath44 a query curve @xmath45 exists whose set of nearest neighbors in @xmath9 is @xmath46 this will imply that there are at least @xmath47 different voronoi regions in the voronoi diagram of @xmath9 .
as second point @xmath48 of @xmath45 we choose the midpoint of the circle defined by the three points @xmath49 , and @xmath50 .
let @xmath51 be the radius of this circle .
note that @xmath52 and that this circle contains the point @xmath53 .
thus , the points @xmath54 , and @xmath55 have distance at most @xmath56 to the point @xmath48 . as first point @xmath57 of @xmath45
we choose a point that has distance @xmath56 to both @xmath58 and @xmath59 , and a larger distance to all other @xmath60 .
consider the voronoi diagram of the points @xmath61 .
consider the edge between the cells of @xmath62 and of @xmath59 . because we chose @xmath17 sufficiently small , namely @xmath63 , and because @xmath52 , there are two points in the interior of this edge with distance @xmath56 to @xmath58 and @xmath59 .
we choose @xmath57 as one of these two points .
then the distance of @xmath57 to all @xmath60 for @xmath64 is larger than @xmath56 .
now we choose the remaining points @xmath65 of @xmath45 for @xmath66 .
let @xmath67 .
there are two circles with radius @xmath56 that touch the points @xmath68 and @xmath69 .
as @xmath65 we choose the midpoint of the one circle that contains the point @xmath43 .
then a point @xmath70 or @xmath71 has distance at most @xmath56 to @xmath65 exactly if @xmath72 or @xmath73 , respectively . [ [ construction - for - mathbfd2.-1 ] ] construction for @xmath74 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the construction can be generalized to @xmath6 giving a lower bound of @xmath75 for @xmath76 curves .
the construction at @xmath58 remains the same . at @xmath53
we place @xmath77 sets of points .
then one point from each set , i.e. , @xmath77 points , define a @xmath78-ball . at @xmath43 for @xmath79 we place @xmath78 sets of points .
then one point from each set , i.e. , @xmath78 points , define a ball of the radius given by the choice of @xmath48 . in total , this gives us @xmath75 choices : @xmath35 choices at @xmath58 , @xmath80 choices at @xmath53 and @xmath81 choices each at @xmath82 .
we have shown new lower bounds for the complexity of the voronoi diagram of polygonal chains under the .
our lower bounds nearly close the gap to the known upper bounds @xcite .
however , it remains an open problem to close the gap .
we conjecture that the upper bound is tight ( up to @xmath17 ) as we have shown here for dimension @xmath83 . for closing the gap ,
consider how the construction for @xmath83 is generalized to @xmath6 in the proof of lemma [ lem : lb2d ] . while the constructions at the vertices @xmath84 are replaced by higher - dimensional analogs
, the construction at @xmath58 stays the same as in the two - dimensional case .
improving the construction at @xmath58 might close the gap . | distance measures for polygonal curves computed @xmath7 curves with @xmath2 vertices @xmath8 set @xmath0 polygonal curves in @xmath1 each @xmath2 vertices task find similar curve in @xmath9 under discrete ) bereg gavrilova zhu propose compute voronoi diagram curves under discrete locate samples transformed query curve voronoi diagram curves represented using correspondence @xmath10 little known about under discrete ) bereg et al . @xcite shown upper bound @xmath11 lower bound @xmath12 complexity of voronoi diagram @xmath0 polygonal curves in @xmath1 @xmath13 @xmath2 vertices prove lower bounds for @xmath14 set @xmath0 polygonal curves in @xmath1 with @xmath2 vertices each voronoi diagram discrete combinatorial complexity @xmath3 @xmath4 @xmath15 @xmath5 for @xmath6 lower bounds improve bereg et al . @xcite for dimension @xmath16 bound matches up to @xmath17 upper bound@xcite formulate upper bound dimensions @xmath16 proof generalizes to other dimensions upper bound @xmath19 @xmath20 dimensions @xmath4 upper lower bounds match up to @xmath17 @xmath6 gap of @xmath21 between lower upper bound remains @xmath6 lower bound construction generalization two - dimensional construction part inherently two - dimensional suitable generalization gap @xmath21 be closed conjecture upper bound tight up to @xmath17 @xmath22 voronoi diagram @xmath0 polygonal curves in @xmath1 with @xmath2 vertices each has combinatorial complexity @xmath3 use parameters @xmath23 @xmath2 denote @xmath0 input curves @xmath1 @xmath2 vertices give lower bounds @xmath24 on combinatorial complexity voronoi diagram contains least @xmath25 voronoi regions voronoi region set curves common nearest neighbors show lower bounds for dimension @xmath26 then @xmath27 construct set @xmath9 of @xmath0 curves construct @xmath28 query curves in different voronoi regions diagram @xmath9implies voronoi diagram complexity @xmath29 @xmath0 @xmath2 set @xmath0 polygonal curves @xmath30 @xmath2 vertices voronoi diagram @xmath31 voronoi regions construct set @xmath9 @xmath0 curves @xmath2 vertices @xmath32 @xmath33 @xmath9 union @xmath2 sets @xmath34 @xmath35 curves voronoi diagram @xmath9 contains @xmath36 voronoi regions @xmath37 construction @xmath37 figure fig:1d place @xmath2 points @xmath38 distance @xmath39 between points line curve @xmath9 form @xmath40 @xmath41 @xmath42 close @xmath43 construction uses points curves sets figure fig:1d illustration @xmath37 @xmath44 query curve @xmath45 exists nearest neighbors @xmath9 @xmath46 @xmath47 voronoi regions diagram @xmath9 second point @xmath48 @xmath45 midpoint circle defined points @xmath49 @xmath50 @xmath51 radius circle @xmath52 circle contains point @xmath53points @xmath54 distance @xmath56 to @xmath48 first point @xmath57 of @xmath45 choose point distance @xmath56 to @xmath58 larger to @xmath60 consider voronoi diagram points @xmath61 edge between cells @xmath62 @xmath59 chose @xmath17 small @xmath63 @xmath52 two points edge distance @xmath56 to and choose @xmath57 distance @xmath57 to @xmath60 for @xmath64 larger than @xmath56 choose remaining points @xmath65 of @xmath45 for @xmath66 @xmath67 two circles with radius @xmath56 touch points @xmath68 and @xmath69 @xmath65 choose midpoint of circle point @xmath43 point @xmath70 or @xmath71 distance @xmath56 to @xmath65 @xmath72 or @xmath73 construction mathbfd2.-1 construction for @xmath74 construction generalized to @xmath6 lower bound of @xmath75 for @xmath76 curves construction at @xmath58 same at @xmath53 place @xmath77 sets of points one point from each set @xmath77 @xmath78-ball@xmath43 @xmath79 place @xmath78 sets points one point each set @xmath78 define radius choice @xmath48 gives @xmath75 choices @xmath35 @xmath58 @xmath80 @xmath53 @xmath81 @xmath82 shown new lower bounds complexity voronoi diagram polygonal chains lower bounds close gap upper bounds @xcite open problem close gap conjecture upper bound tight @xmath17 ) shown dimension @xmath83 closing gap consider construction @xmath83 generalized @xmath6 proof lemma constructions vertices @xmath84 replaced higher - dimensional analogs construction @xmath58 stays same two - dimensional improving construction @xmath58 might close gap | 0.31201 | 2,062 | 1,125 | , Saving $0.1 in GPT-4. |
177,096 | the hadronic quark structure is investigated in the frame of high energy electron proton scattering .
a phenomenological model based on the born approximation is used to calculate the transition matrix element for the quark system forming the proton target .
a potential of electromagnetic nature is assumed for the calculation of the multiple scattering of electron with the constituent valance quarks of the proton .
it is found that the first two terms of the born approximation are sufficient to describe the experimental data of differential cross section for the electron proton system only at low momentum transfer square . on the other hand , a two body scattering amplitude with a relevant form factor may give proper agreement extended to high momentum transfer square region .
a harmonic oscillator wave function is used to reproduce the data of the experiment at low momentum transfer .
this satisfies quietly the confinement condition of the quarks in the hadron system .
however the data of the experiment at relatively high momentum transfer , show that hadronic quarks behave more freely and may be described by just a coulomb wave function . + high energy electron proton scattering in view of born approximation physics department , faculty of science , cairo university [ cols="^,^,^,^ " , ] 3 cm | the most recent data of the electron - proton ( ep ) scattering experiments at slac [ 1 - 3 ] have played a significant role in investigating the hadron structure . in the last decade , many trails have been executed to deal with the ep scattering problem .
the glauber multiple scattering approach [ 4 ] was used to present an eikonal picture for the ep scattering .
the quantum chromodynamics qcd [ 5,6 ] was developed as the most convincing present theory for the interactions between those constituent quarks , formulated in terms of the exchange of colored vector gluons . at sufficiently high values of @xmath0
, the running strong coupling constant is expected to become small enough , due to the properly of asymptotic freedom , to allow the use of perturbation theory to simplify qcd . however , there is considerable controversy as to how large a value of @xmath0 is sufficient for perturbative quantum chromodynamics pqcd [ 7 ] .
the exclusive processes such as electron proton ( e - p ) scattering are predicted to have a single dimensional scaling [ 8 ] at large @xmath0 . in this case
only the valance quark states are important and a rough idea of the @xmath0 dependence can be gained by simply counting the number of quark - gluon vertices .
elastic form factor , for example , should scale asymptotically as @xmath1 , where n is the number of valance quarks participating in the interaction . for ( e - p ) scattering n=3 thus the structure function behaves as @xmath2 .
perturbative qcd [ 7 ] predicts calculable logarithmic departures from the @xmath0 dependence of exclusive amplitudes given by the simple dimensional scaling law . the earliest efforts [ 9 ] used unrealistic symmetric distribution amplitudes and required a large multiplicative factor to normalize the results to the data at @xmath0 @xmath3 .
chernyak and zhitnitsky [ 10,11 ] proposed a model form for the nucleon distribution amplitude which satisfies the sum rules and in which the momentum balance of the valance quarks in the proton is quite asymmetric .
the result of this model is justified in the diquark model [ 12,13 ] .
the nonperturbative calculations succeeded in modeling the region @xmath4 fairly well but some difficulties have been found at large values of @xmath0 . by this article
we aim to develop the eikonal optical picture of the ( ep ) scattering on the bases of the born multiple scattering of the incident electron with the constituent valance quarks of the target hadron .
the paper is organized so that in section 2 , we present the hypothesis and the mathematical formalism of the model .
results and discussion as well the comparison with experimental data are given in section 3 .
during the electron - proton ( e - p ) scattering , we consider the proton as a bag including three valance quarks .
we proceed using the semi - classical born approximation which is a useful technique when the de broglie wavelength of the incident particle is sufficiently short compared with the distance in which the potential varies appreciably . in this approximation
we consider the multiple scattering series in which the projectile interacts repeatedly with the potential and propagates freely between two such interactions .
we expect that the born series converges if the incident particle is sufficiently fast so that it can not interact many times with the potential and/or if the potential is weak enough . starting with a plane wave as a zero order wave function of the incident electron , and by successive iteration ,
the nth order of the born scattering amplitude is given by , @xmath5 where @xmath6 , is the zero order plane wave and , @xmath7 and @xmath8 is the free green s function . let us denote by @xmath9 and @xmath10 the coordinates of the incident electron and the constituent valance quarks of the target proton . according to the first born approximation , the scattering amplitude is given by : @xmath11
@xmath12 this term represents the two - body scattering amplitude or the impulse approximation .
the second born approximation is , @xmath13 which represents the double scattering term in the scattering amplitude .
a(@xmath14 ) is the average potential acting on the incident electron due to the target valance quarks ri with initial and final states @xmath15 and @xmath16 so that , @xmath17 in the present situation we consider that the electron - quark ( eq ) potential to be a pure coulomb with the form , @xmath18 and @xmath19 is the relative coordinate @xmath20 the term @xmath21 may be expanded in terms of the spherical harmonics @xmath22 as @xmath23 where @xmath24 is the greater ( the lesser ) of @xmath14 and @xmath25 .
the radial integral of eq.(3 ) may be divided into two parts : @xmath26 \(10 ) and substitute for @xmath24 by @xmath27 in the first integral and for @xmath28 by @xmath29 in the second integral . then writing @xmath30 where @xmath31 is the average potential due to the @xmath32 quark , then @xmath33 @xmath34 @xmath35 the evaluation of @xmath36 for the three quark system differs only in their signs due to the charge situation . on the other and
the differential cross section is calculated either in terms of the multiple scattering born approximation amplitude , @xmath37 or in terms of the two body scattering amplitude with a proton form factor correction .
@xmath38 here , @xmath39 is the scattering amplitude of the electron scattering by a point - like proton . the proton form factor f(q )
is also an important physical quantity which reflects information about the particle structure .
it is defined as : @xmath40 @xmath41 as the model is a phenomenological type , it is then branches in two ways , + i ) the first one assumes that quarks are confined inside the bag by an extremely deep potential , and the quark states are reasonably represented by a harmonic oscillator wave function .
+ ii ) the second way assumes that the constituent quarks behave as almost free particles as is suggested by experiments of deep inelastic scattering [ 1,14 ] .
however , in all cases the @xmath42 scattering is treated as the collision between the high energy incident electron with a composite proton system .
proceeding with the first assumption ( i ) and assuming a harmonic oscillator wave function for the valance quarks , @xmath43 then the average scattering potential is ; @xmath44\ ] ] where @xmath45 is the incomplete gamma function of order 3/2 .
the scattering amplitude and the differential cross section are to be calculated numerically .
moreover , the form factor is calculated for the proton system as , @xmath46 considering now the second case ( ii ) of elastic scattering and that all quarks are in the 1s ground state of a coulomb wave function , @xmath47 a is the proton radius , then @xmath48 consequently , @xmath49\ ] ] @xmath50 ^ 2\ ] ] and the form factor is then , @xmath51
data used in this article are those from the experiments coded slac - e-140 [ 1 ] at low momentum transfer extended from 1 to 3 @xmath52 using electron beam of energy 8 gev , and slac - e-136 [ 2 ] at a wide range extended from 2.9 to 31.3 @xmath52 .
the later was conducted using accelerated electrons with energies from 5 to 21.5 gev were elasticaly scattered by protons in a liquid -hydrogen target at stanford linear accelerator center .
the center of mass energy of the reaction is found in terms of the energies @xmath53 of the scattered electron and proton respectively and their opening angle @xmath54 .
the differential cross section is calculated in the frame of multiple scattering using eq.(14 ) for both the experiments slac - e-136 and slac - e-140 .
the result of calculation as well as the experimental data are displayed in figs.(1 ) and ( 2 ) respectively . it is found that only the first two terms of the born series give appreciable contribution to the differential cross section .
the second term contributes for not more than 10 @xmath56 in all cases .
the calculations are carried out using harmonic oscillator wave function ( solid lines ) and coulomb wave function ( dashed lines ) for the valance quarks forming the proton system .
the experimental data are reproduced to a limited extend by the harmonic oscillator wave function only at low momentum transfer region . on the other hand
the use of the coulomb wave function shows better representation of the data at high momentum transfer .
this result reflects the fact that quarks inside the hadron are quite confined at low momentum transfer region while at high momentum transfer , quarks behave more freely and the confinement condition becomes no longer necessary .
however , the comparison with the experiment shows also that the multiple scattering approach could nt give satisfactory agreement particularly at high momentum transfer regions .
the differential cross section is recalculated in terms of the proton form factor according to eqs.(15 - 16 ) .
proceeding the same analogy , so that the proton form factor is calculated twice using the harmonic oscillator and the coulomb wave functions .
the result are displayed in fig.(3 ) compared with data of the experiment slac - nh-11 [ 3 ] as well as the prediction of the simple dipole model of the following form , @xmath57 the prediction of eq .
( 15 ) is displayed with the experimental data in figs .
( 4 ) and ( 5 ) for the experiments e-136 and e-140 respectively .
it is clear now that the two body scattering amplitude corrected with a relevant form factor may reproduce the experimental data quite well , assuming harmonic oscillator wave function for the constituent quarks . in the above calculations , we use the units where @xmath58 , hence the momentum in gev / c may be represented in units of @xmath59 with a conversion factor 0.2 in all the above cases
the potential parameters are determined by the chi- square fitting method . a root mean square radius of the proton is found to be 0.8 and 1.0 fm on using the harmonic oscillator and the coulomb wave functions respectively
. the distribution of the hadronic system produced in e - p scattering is measured by zeus collaboration [ 15 ] at center of mass energy of 296 gev .
comparison of the results with the qcd radiation has a strong influence on the characteristics of the final state .
the data are reasonably reproduced by the lund model based on a matrix element calculated in the first order of @xmath60 , followed by a appropriate parton shower , as well as by the color dipole model .
the herwing parton shower model also gives a reasonable representation of the data .
neither the first order matrix elements alone nor the lund parton shower model , without the matrix element calculation .
a.f.sill et.al . ,
phys.rev.d48 , 29 ( 1993 ) .
p. bosted et al .
68 , 3841 ( 1992 ) .
r.c.walker , phys .
224b , 353 ( 1989 ) .
hussein , 24th international cosmicray conference icrc , roma , he,135 ( 1995 ) .
n.g.stefanis , phys.rev .
d40 , 2305 ( 1989 ) .
c.e.carlson , m. gari and n.g.stefanis , phys.rev.lett .
58 , 1308 ( 1987 ) .
-s.j.brodsky and g.p.lepage , phys.rev.d22 , 2157 ( 1980 ) .
+ -g.p.lepage and s.j.brodsky , phys.rev.lett .
43,545 ( 1979 ) .
+ -a.v.radyushkin , nucl.phys .
a532 , 141c ( 1991 ) .
s.j.brodsky and g.farrar , phys.rev.d11 , 1309 ( 1975 ) .
a.duncan and a.h.muller , phys.lett 90b , 159 ( 1980 ) .
chernyak and i.r .
zhitnitsky , nucl.phys .
b246 , 52 ( 1984 ) . m.gari and n.g .
stefanis , phys.lett .
b175 , 462 ( 1986 ) . z.dziembowski and
j.frankil , phys.rev.d42 , 905 ( 1990 ) .
p.kroll , m. schurmann and w. schweiger , z.phys .
a338 , 339 ( 1991 ) .
r.g.arnold et.al . ,
phys.rev.lett.57 , 174 ( 1986 ) .
zeus collaboration , z. phys .
c59 , 231 ( 1993 ) . | recent data electron - proton scattering experiments at slac 1 - 3 hadron structure last decade trails executed ep scattering problem glauber multiple scattering approach 4 eikonal picture ep scattering quantum chromodynamics qcd [ 5,6 convincing theory interactions quarks exchange colored vector gluons high values @xmath0 coupling constant small asymptotic freedom perturbation theory simplify qcd controversy value @xmath0 for perturbative quantum chromodynamics pqcd 7 exclusive processes electron proton scattering predicted single dimensional scaling at @xmath0 valance quark states important @xmath0 dependence counting quark - gluon vertices elastic form factor scale asymptotically as @xmath1 n valance quarks e - p ) scattering n=3 structure function behaves as @xmath2 perturbative qcd [ 7 ] predicts logarithmic departures from @xmath0 dependence amplitudes dimensional scaling law earliest efforts used unrealistic symmetric distribution amplitudes required large multiplicative factor normalize results data @xmath0 @xmath3 chernyak zhitnitsky proposed model for nucleon distribution amplitude sum rules momentum balance valance quarks proton asymmetric justified in diquark model 12,13 nonperturbative calculations region @xmath4 difficulties at large values @xmath0 article aim develop eikonal optical picture of ( ep ) scattering born multiple scattering incident electron with constituent valance quarks target hadron paper section 2 hypothesis mathematical formalism model results comparison with experimental data in section 3 electron - proton scattering proton as bag three valance quarks using semi - classical born approximation when de broglie wavelength incident particle short distance potential varies multiple scattering series projectile interacts repeatedly with potential propagates freely between born series converges if incident particle fast potential weak starting with plane wave zero order wave function of incident electron nth order of born scattering amplitude given by @xmath5 @xmath6 zero order plane wave @xmath7 @xmath8 free green s function denote by @xmath9 @xmath10 coordinates of incident electron constituent valance quarks of target protonfirst born approximation scattering amplitude by @xmath11 @xmath12 two - body scattering second approximation @xmath13 double scattering@xmath14 average potential incident electron target valance quarks initial final states @xmath15 @xmath16 @xmath17 electron - quark potential pure coulomb @xmath18 @xmath19 relative coordinate @xmath21 spherical harmonics @xmath23 @xmath24 greater lesser of @xmath14 @xmath25 radial integral of eq.(3 ) divided into two parts @xmath26 ) substitute @xmath24 by @xmath27 @xmath28 @xmath29 @xmath30 @xmath31 average potential to @xmath32 quark @xmath33 @xmath34 @xmath35 evaluation @xmath36 three quark system differs charge situation differential cross section calculated multiple scattering approximation amplitude @xmath37 or two body scattering amplitude proton form factor correction @xmath38 @xmath39 scattering amplitude electron scattering point - like proton proton form factor f(q ) important physical quantity particle structuredefined @xmath40 @xmath41 model phenomenological type branches two ways first assumes quarks confined by deep potential states represented by harmonic oscillator wave function second assumes quarks behave free particles deep inelastic scattering @xmath42 scattering collision between high energy electron composite proton system first assumption harmonic oscillator wave function valance quarks @xmath43 average scattering potential @xmath44 @xmath45 incomplete gamma function order 3/2 scattering amplitude differential cross section calculated numerically form factor calculated for proton system @xmath46 second case elastic scattering quarks in 1s ground state coulomb wave function @xmath47 proton radius @xmath48 @xmath49 @xmath50 form factor @xmath51 data experiments slac - e-140 low momentum transfer 1 to 3 @xmath52 electron beam energy 8 gev slac - e-136 wide range 2.9 to 31.3 conducted accelerated electrons energies 5 to 21.5 gev elasticaly scattered by protons in liquid -hydrogen target stanford linear accelerator centercenter mass energy reaction found energies scattered electron proton opening angle @xmath54 differential cross section calculated multiple scattering using eq.(14 ) for experiments slac - e-136 slac - e-140 result experimental data displayed in figs.(1 ) ( 2 ) first two terms born series differential cross section second term contributes not than 10 @xmath56 calculations using harmonic oscillator wave coulomb wave function for valance quarks proton system experimental data reproduced harmonic oscillator low momentum transfer coulomb wave function better representation high momentum transfer quarks confined low momentum high momentum behave freely confinement comparison shows multiple scattering approach agreement high momentum transfer regions differential cross section recalculated proton form factor eqs.(15 - 16 ) proton form factor calculated twice using harmonic oscillator coulomb wave functions result displayed in fig.(3 ) compared with data experiment slac - nh-11 [ 3 ] prediction simple dipole model prediction of eq . ( 15 ) displayed with experimental data in figs . ( 4 ) ( 5 ) experiments e-136 e-140two body scattering amplitude corrected form factor reproduce experimental data assuming harmonic oscillator wave function quarks calculations use units @xmath58 momentum gev / c represented units @xmath59 conversion factor 0.2 potential parameters determined by chi- square fitting method root mean square radius proton 0.8 1.0 fm harmonic oscillator coulomb wave functions distribution hadronic system e - p scattering measured by zeus collaboration mass energy 296 gev comparison results with qcd radiation characteristics final state data reproduced by lund model matrix element calculated first order @xmath60 parton shower color dipole model herwing parton shower model representation data first order matrix elements lund parton shower model matrix element calculation.f.sill.rev.d48 , 29 1993 bosted 68 , 3841 1992.walker phys . 224b 353 1989 hussein 24th international cosmicray conference roma,135 1995.stefanis phys.rev d40 2305 1989.carlson gari.stefanis phys.rev.lett 58 , 1308 1987.brodsky.p.lepage phys.rev.d22 2157 1980.lepage s.rev.lett 43,545 1979.radyushkin a532 , 141c 1991.brodsky g.farrar.rev.d11 1309 1975 a.duncan.h.muller.lett 90b 159 1980 ) chernyak zhitnitsky.phys b246 52 1984 m.gari stefanis.lett b175 462 1986 z.dziembowski j.frankil.rev.d42 905 1990 ) p.kroll schurmann schweiger a338 339 1991 r.g.arnold phys.rev.lett.57 174 1986 zeus collaboration c59 , 231 1993 | 0.435529 | 2,829 | 1,493 | , Saving $0.1 in GPT-4. |
177,111 | we show that the liquid drop model is best suited to describe the masses of prolate deformed nuclei than of spherical nuclei . to this end
three liquid drop mass formulas are employed to describe nuclear masses of eight sets of nuclei with similar quadrupole deformations .
it is shown that they are able to fit the measured masses of prolate deformed nuclei with an rms smaller than 750 kev , while for the spherical nuclei the rms is , in the three cases , larger than 2000 kev .
the rms of the best fit of the masses of semi - magic nuclei is also larger than 2000 kev .
the parameters of the three models are studied , showing that the surface symmetry term is the one which varies the most from one group of nuclei to another . in one model , isospin dependent terms
are also found to exhibit strong changes .
the inclusion of shell effects allows for better fits , which continue to be better in the prolate deformed nuclei region . | the description of nuclear masses in terms of the liquid drop model paved the way to the basic understanding of nuclear properties , like the saturation of the nuclear force , the existence of pairing and shell effects , and the description of fission and fusion processes @xcite .
the q - values of different nuclear reactions , obtained from mass differences , must be accurately known to allow the description of the astrophysical origin of the elements @xcite .
accurate theoretical predictions of nuclear masses remain a challenge @xcite , sharing the difficulties with other quantum many - body calculations , and complicated by the absence of a full theory of the nuclear interaction .
decades of work have produced microscopic and macroscopic mass formulas @xcite . at present ,
the most successful approaches seem to be the finite range droplet model ( frdm ) @xcite , the skyrme and gogny hartee fock bogolyubov ( hfb ) @xcite , and the duflo - zuker ( dz ) mass formula @xcite .
they allow for the calculation of masses , charge radii , deformations , and in some cases also fission barriers .
they all contain a macroscopic sector which resembles the liquid drop mass ( ldm ) formula , and include deformation effects .
hfb calculations are now able to fit known nuclear masses with deviations competitive with the 95 frdm calculations , which are also being improved , while the most precise and robust nuclear mass predictions are given by the dz model @xcite .
the liquid drop mass ( ldm ) formula captures the macroscopic features of the mass dependence on the number of neutrons @xmath0 , of protons @xmath1 , and on its mass numbers @xmath2 .
it includes volume and surface terms , the coulomb interaction between protons , wigner and symmetry terms , linear and quadratic in the neutron excess @xmath3 , and a pairing term .
it is generally assumed that the liquid - drop energy of a spherical nucleus is described by a bethe - weizscker mass formula @xcite , being a common practice to describe nuclear masses and radii of spherical closed - shell nuclei in terms of a mean field and add deformation and other shell effects as corrections @xcite .
it is the purpose of this contribution to show that , when nuclei with measured masses are grouped according with their quadrupole deformations , three liquid drop mass formulas consistently allow for a fit with an rms smaller than 750 kev for the set of most prolate deformed nuclei , while they are unable to fit the masses of spherical nuclei with an rms smaller than 2000 kev . semi - magic nuclei are fit with a similarly large rms .
the parameters of the three models are studied , showing that the surface symmetry term is the one which varies the most from one group of nuclei to another . in one model , isospin dependent terms
are also found to exhibit strong changes .
shell effects refer to the differences between the experimental binding energies @xcite and the ldm predictions . in this work
they are included employing linear , quadratic @xcite , 3- and 4-body terms @xcite , functions of the number of valence nucleons , following the ideas of the duflo - zuker model @xcite.the inclusion of shell effects allows for better fits , while the smallest rms are found again in the prolate deformed nuclei region .
three liquid drop mass formulas will be employed to analyze their ability to fit nuclear masses .
one of them treats consistently volume and surface effects @xcite .
the second one incorporates explicitly isospin effects @xcite while the third one includes quadratic isospin effects , the diffuseness correction to the coulomb energy , the charge exchange correction term and the curvature energy @xcite . in ref .
@xcite an improved version of the liquid - drop mass formula with modified symmetry and coulomb terms is built , following a consistent treatment of nuclear bulk and surface effects .
the negative nuclear interaction energy is given by : @xmath4 where the pairing interaction is given by @xmath5 = 2 , 1 , and 0 for even - even , odd - mass and odd - odd nuclei , respectively .
a modification @xmath6 in the coulomb term is included @xmath7 its presence in the denominator of the coulomb term suggests that it can be viewed as a correction to the radius of the nucleus .
the symmetry term employs @xmath8 instead of @xmath9 to account for the wigner energy .
the coulomb interaction is proportional to @xmath10 to avoid the coulomb interaction of a proton with itself .
the second ldm formula we are going to analyze was introduced in ref .
@xcite by considering isospin effects . in this case the liquid - drop energy of a spherical nucleus is described by a modified bethe - weizscker mass formula @xmath11 with isospin asymmetry @xmath12 .
the pairing term is taken form @xcite @xmath13 and the symmetry energy coefficient of finite nuclei is written as , @xmath14\ ] ] based on the conventional surface - symmetry term of liquid - drop model , with a small correction term for description of isospin dependence of @xmath15 . the symmetry energy coefficient @xmath15 increases with decreasing isospin asymmetry .
this @xmath16 correction term approximately describes the wigner effect for heavy nuclei .
different mass formulae derived from the liquid drop model and the pairing and shell energies of the thomas - fermi model have been studied and compared in ref @xcite .
we selected for this study to include the diffuseness correction to the coulomb energy , the charge exchange correction term and the curvature energy . in ref .
@xcite it is reported that the coulomb diffuseness correction @xmath17 term and the charge exchange correction @xmath18 term play the main role to improve the accuracy of the mass formula .
the wigner term and the curvature energy can also be used separately but their coefficients are very unstable .
their ldm formula is @xmath19 the volume energy corresponding to the saturated exchange force and infinite nuclear matter is given by the first term .
@xmath20 is the asymmetry energy of the bethe - weizscker mass formula .
the second term is the surface energy .
its origin is the deficit of binding energy of the nucleons at the nuclear surface and corresponds to semi - infinite nuclear matter .
the following term is the curvature energy .
it results from non - uniform properties which correct the surface energy and depends on the mean local curvature .
the decrease of binding energy due to the coulomb repulsion is given by the fourth term , which has and adjustable charge radii @xmath21 .
the @xmath17 term is the diffuseness correction to the sharp radius coulomb energy , also called also the proton form - factor .
the @xmath18 term is the charge exchange correction term .
the pairing energies of the thomas - fermi model @xcite were employed .
the coefficients of the three ldm were selected to minimize the root mean square deviation ( rms ) when the predicted binding energies @xmath22 are compared with the experimental ones @xmath23 , reported in ame03 @xcite , modified so as to include more realistically the electron binding energies as explained in appendix a of lunney , pearson and thibault @xcite .
@xmath24 ^ 2}}{n_{nucl}}\right\}^{1/2}.\ ] ] @xmath25 is the number of nuclei in each group , listed in the fourth row of table [ def ] .
the minimization procedure uses the routine minuit @xcite .
the fits were performed employing the masses of nine groups of nuclei : * all nuclei whose measured masses are reported in ame03 @xcite , which have @xmath26 , * seven groups of nuclei whose quadrupole deformations , taken form the frdm @xcite , lie in the ranges listed in the second and third row of table [ def ] , * the group of all semi - magic nuclei , having z= 8 , 20 , 28 , 50 , 82 or n= 8 , 20 , 28 , 50 , 82 or 126 .
c||c|ccccccc|c group & all & 1 & 2 & 3 & 4 & 5 & 6 & 7 & semi - magic + @xmath27 min & -0.65 & -0.65 & -0.11 & 0.00 & 0.04 & 0.12 & 0.18 & 0.23 & + @xmath27 max & 0.65 & -0.11 & 0.00 & 0.04 & 0.12 & 0.18 & 0.23 & 0.65 & + @xmath25 & 2149 & 258 & 252 & 332 & 272 & 307 & 364 & 364 & 185 + [ def ] notice that group 1 contains most of the oblate nuclei , that the more spherical nuclei belong to group 3 , and that the more prolate deformed nuclei are included in groups 6 and 7 . for each ldm equation ,
nine fits were performed , one for each group of nuclei . in this way
, nine sets of parameters were obtained , which minimize the rms of each group of nuclei . employing these nine sets of parameters ,
the rms were estimated for all groups , whose rms are shown in the next subsections . here
we show the results obtained using eq .
( [ piet ] ) for the nine regions . in our calculations
we select nine sets of fixed values for the parameters @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 and @xmath33 .
the results are exhibited in tables [ tab1 ] and [ tab2 ] .
.rms ( in kev ) for the nine groups ( columns ) , for the nine sets of parameters ( rows ) , employing eq .
( [ piet ] ) [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] on the other hand , the nuclei in region 7 , the most prolate deformed , are the best fitted in the ldmm , while the spherical nuclei in region 3 and the semi - magic nuclei have the largest rms .
the inclusion of 2- , 3- and 4-body terms in ldmm seems to succeed in introducing deformation effects .
regions 1 and 4 to 7 have all rms between 562 and 623 kev .
spherical and semi - magic nuclei remain to be those with the largest rms .
along this contribution we have shown that the liquid drop model is best suited to describe the masses of prolate deformed nuclei than of spherical nuclei .
the analysis was performed employing three different liquid drop mass formulas . with them , the are nuclear masses nuclei grouped in eight sets with similar quadrupole deformations were fitted .
for the three ldm models it was found that the masses of prolate deformed nuclei can be described with remarkable precision for a ldm , with an rms smaller than 750 kev , while the masses of spherical and semi - magic nuclei are those worst described , with rms larger than 2000 kev .
the dispersion of the parameters of the three models were studied comparing the fits for the different groups of nuclei .
we found that in the three the surface symmetry term is the one which varies the most from one group of nuclei to another . in the model of ref .
@xcite , isospin dependent terms were found to exhibit strong changes , making this model the least robust of the three under this criterion .
the inclusion of shell effects allows for better fits , which continue to be better in the prolate deformed nuclei region .
the duflo - zuker model is based in a microscopic description of shell effects , and describes deformation through a change in valence occupations .
it remains a challenge to see if the dz mechanism to incorporate deformation effects can be successfully employed by other models .
this work was supported in part by conacyt , mxico , by foncicyt project 94142 , and by dgapa , unam .
9 j. duflo and a.p .
zuker , phys .
c 52 r23 ( 1995 ) r23 .
j. mendoza - temis , i. morales , j. barea , a. frank , j.g .
hirsch , j.c .
lpez - vieyra , p. van isacker and v. velzquez , nucl . phys . * a 812 * ( 2008 ) 28 . | description of nuclear masses liquid drop model paved way to basic understanding of nuclear properties like saturation nuclear force pairing shell effects fission processes q - values of nuclear reactions from mass differences must be known description astrophysical origin elements accurate theoretical predictions of nuclear masses remain challenge difficulties with quantum calculations complicated by absence of full theory of nuclear interaction decades of work produced microscopic macroscopic mass formulas successful approaches finite range droplet model duflo - zuker ( dz ) mass formula allow calculation of masses charge radii deformations fission barriers contain macroscopic sector liquid drop mass ldm ) formula include deformation effects hfb calculations fit nuclear masses with deviations competitive with 95 frdm calculations most precise nuclear mass predictions by dz model liquid drop mass ) formula captures features mass dependence on neutrons protons mass numbers includes volume surface terms coulomb interaction between protons wigner symmetry terms linear quadratic neutron excess pairing termassumed liquid - drop energy spherical nucleus described by bethe - weizscker mass formula common practice describe nuclear masses radii spherical closed - shell nuclei mean field deformation shell effects corrections purpose contribution show nuclei measured masses grouped quadrupole deformations three liquid drop mass formulas allow fit with rms smaller than 750 kev prolate deformed nuclei unable fit spherical nuclei rms smaller than 2000 kev semi - magic nuclei fit with similarly large rms parameters three models studied surface symmetry term varies most isospin dependent terms exhibit strong changes shell effects refer differences experimental binding energies ldm predictions included linear quadratic 3- 4-body terms valence nucleons duflo - zuker model inclusion shell effects allows better fits smallest rms in prolate deformed nuclei three liquid drop mass formulas analyze nuclear masses one treats volume surface effects second incorporates isospin effects third includes quadratic isospin effects diffuseness correction charge exchange correction curvature energyimproved liquid - drop mass formula modified symmetry coulomb terms nuclear bulk surface effects negative nuclear interaction energy by @xmath4 pairing interaction by @xmath5 = 2 1 0 for even - odd - mass nuclei modification @xmath6 coulomb term included @xmath7 correction radius nucleus symmetry term employs @xmath8 wigner energy coulomb interaction proportional to @xmath10 avoid proton second ldm formula introduced @xcite isospin effects liquid - drop energy spherical nucleus described modified bethe - weizscker mass formula @xmath11 isospin asymmetry @xmath12 pairing term form @xcite @xmath13 symmetry energy coefficient finite nuclei @xmath14 symmetry small correction term isospin dependence @xmath15 symmetry energy coefficient increases with decreasing isospin asymmetry @xmath16 correction describes wigner effect for heavy nuclei mass formulae liquid drop model pairing shell energies thomas - fermi model studied compared in @xcite diffuseness correction coulomb energy charge exchange correction curvature energy@xcite coulomb diffuseness correction charge exchange correction improve accuracy mass formula wigner curvature energy used separately coefficients unstable ldm formula @xmath19 volume energy saturated exchange force infinite nuclear matter first term @xmath20 asymmetry energy bethe - weizscker mass formula second term surface energy origin deficit binding energy nucleons nuclear surface semi - infinite nuclear matter following term curvature energy results non - uniform properties surface energy depends local curvature decrease binding energy coulomb repulsion fourth term adjustable charge radii @xmath21 @xmath17 term diffuseness correction sharp radius coulomb energy proton form - factor @xmath18 term charge exchange correction pairing energies thomas - fermi model employed coefficients three ldm selected minimize root mean square deviation predicted binding energies compared experimental ones @xmath23 modified include electron binding energies appendix a lunney pearson thibault @xmath24 @xmath25 number nuclei each group fourth row table minimization procedure uses routine minuitfits performed masses nine groups nuclei : all nuclei in ame03 @xcite @xmath26 seven groups nuclei quadrupole deformations @xcite in ranges second third row table group semi - magic nuclei z= 8 20 28 50 82 or n= 8 20 28 50 82 126 group all 1 2 3 4 5 6 7 semi - magic + @xmath27 min -0.65 -0.65 -0.11 0.00 0.04 0.12 0.18 0.23 @xmath27 max 0.65 -0.11 0.00 0.04 0.23 @xmath25 2149 258 252 332 272 307 364 185 group 1 contains oblate nuclei spherical nuclei group 3 prolate deformed nuclei in groups 6 and 7 each nine fits performed one each group nine sets parameters obtained minimize rms each group rms estimated for all groups shown in next subsections results using eq . ( piet ] ) for nine regionscalculations select nine sets fixed values for parameters @xmath28 results exhibited in tables tab1 tab2 .rms in kev ) for nine groups nine sets parameters employing eq piet nuclei in region 7 most prolate deformed best fitted in ldmm spherical nuclei region 3 semi - magic nuclei largest rms inclusion of 2- 3- 4-body terms in ldmm deformation effects regions 1 4 to 7 have rms between 562 and 623 kev spherical semi - magic nuclei largest rms liquid drop model best describe masses prolate deformed nuclei spherical nuclei analysis three liquid drop mass formulas nuclear masses nuclei in eight sets with similar quadrupole deformations fitted masses of prolate deformed nuclei described precision rms smaller than 750 kev masses spherical semi - magic nuclei worst described rms larger than 2000 kev dispersion of parameters three models studied comparing fits different groups nucleifound three surface symmetry term varies most nuclei model @xcite isospin dependent terms strong changes model least robust three inclusion shell effects allows better fits prolate deformed nuclei region duflo - zuker model based microscopic shell effects describes deformation through change valence occupations challenge dz mechanism deformation effects employed other models work supported by conacyt, mxico foncicyt project 94142 dgapa , unam j. duflo zuker phys . c 52 r23 ( 1995 ) mendoza - temis morales barea frank hirsch lpez - vieyra van isacker v velzquez nucl . 812 * ( 2008 ) 28 | 0.438481 | 2,734 | 1,342 | , Saving $0.1 in GPT-4. |
177,127 | on the ground of the landau criterion we study the behavior of critical velocities in a superfluid two - component bose gas . it is found that under motion of the components with different velocities the velocity of each component should not be lower than a minimum phase velocity of elementary excitations ( @xmath0 ) .
the landau criterion yields a relation between the critical velocities of the components ( @xmath1 , @xmath2 ) .
the velocity of one or even both components may exceed @xmath0 . the maximum value of the critical velocity of a given component can be reached when the other component does not move .
the approach is generalized for a two - component condensate confined in a cylindrical harmonic potential .
pacs numbers : 03.75.kk,03.75.mn | at present considerable attention is given to the study of two - component bose - einstein condensates .
the progress in cooling and trapping of atomic rarefied gases allows to obtain such condensates experimentally @xcite . from the fundamental point of view
, they are considered as objects in which some cosmological and astrophysical processes can be modelled @xcite . in two component systems
the superfluid components may flow with different velocities .
in such a situation an unusual ( nondissipative ) kind of the drag effect takes place @xcite .
it makes possible to create a controlled phase difference between two bose - condensates placed in a two - well potential , and to observe effects @xcite , similar to ones that occur in superconductive systems with josephson contacts in magnetic fields . in this article
we study a related problem , namely , the critical velocities in a two component superfluid system with components flowing with different velocities .
according to the landau criterion , the critical velocity in a single - component condensate is determined by the expression @xmath3 where @xmath4 is the excitation spectrum in an immovable condensate and @xmath5 is the wave number .
the velocity is given in the frame of reference connected with walls or obstacles .
( [ 1 ] ) can be applied to a two - component condensate only in the case when both components move with the same velocity .
the landau criterium can be reformulated as the requirement of positivity of energies of elementary excitations in the frame of reference , connected with walls or obstacles .
these energies depend on the velocities of the components @xmath6 and @xmath7 ( in the same frame of reference ) , and the landau criterium should yield some joint condition on @xmath6 and @xmath8 . to find the energy of elementary excitations
we use the gross - pitaevskii equation for the two - component system @xmath9 are the masses of the particles , @xmath10 , @xmath11 are the interaction constants ( @xmath12 , @xmath13 where @xmath14 are the scattering lengths ) .
the wave function of the component can be represented as the sum of the stationary part and the fluctuating part @xmath15 , where @xmath16 .
the stationary part of the condensate wave function can be presented in the form : @xmath17 where @xmath18 are the chemical potentials of the components .
the gradients of the phases @xmath19 are connected with the superfluid velocities by the relation @xmath20 .
the fluctuating part can be written as @xmath21 .
\end{array}\ ] ] the substitution of ( [ 14 ] ) into the linearized version of ( [ 5 ] ) leads to a system of equations for @xmath22 coefficients , whose determinant gives the dispersion equation for the spectrum of elementary excitations @xmath23 [ e_2 ^ 2 - ( e - \hbar \mathbf{v}_2\cdot \mathbf{k})^2 ] - 4 \varepsilon_1 \varepsilon_2 \gamma_{12}^2 n_1 n_2 = 0,\ ] ] where @xmath24 is the bogolyubov spectrum for the @xmath25-component ( in the absence of interaction between the components ) , and @xmath26
. for @xmath27 the equation ( [ 16 ] ) yields @xmath28 in this study we assume that the condition of stability of a two - component condensate relative to phase separation is fulfilled ( @xmath29 ) . the requirement of positivity of ( [ 18 ] ) at all @xmath30 is equivalent to the condition ( [ 1 ] ) .
this condition yields the following expression for the critical velocity @xmath31 where @xmath32 are the bare velocities of the sound modes for the components ( in the absence of interaction between the components ) . in a general case @xmath33
the landau criterium requires the existence of two positive solutions of eq .
( [ 16 ] ) at all @xmath30 .
this requirement is equivalent to the following two inequalities : @xmath34 [ e_2 ^ 2 - ( \hbar \mathbf{v}_2\cdot \mathbf{k})^2 ] - 4 \varepsilon_1 \varepsilon_2 \gamma_{12}^2
n_1 n_2 > 0,\ ] ] @xmath35 the critical values of @xmath6 and @xmath8 correspond to the case , when the inequality ( [ 30 ] ) turnes into the equality at least for one @xmath30 . in the case
considered the sufficient condition for fulfilling the inequalities ( [ 30 ] ) , ( [ 31 ] ) for all @xmath30 is their fulfillment at @xmath36 for all directions of @xmath30 , complanar to @xmath6 and @xmath8 .
therefore , the inequalities ( [ 30 ] ) , ( [ 31 ] ) can be replaced with the system of inequalities @xmath37 ( where @xmath38 is an angle between @xmath6 and @xmath8 , and @xmath39 is an angle between @xmath30 and @xmath6 ) , that should be fulfilled for all @xmath39 .
the results of the analysis of ( [ 20 ] ) is given in fig.1 if the components move in the same direction their critical velocities are related by the equation @xmath40 with the additional condition @xmath41 . according to ( [ 112 ] ) only for @xmath42
the critical velocity coincides with the velocity of the lowest hydrodynamic mode @xmath0 . in a general case one of the velocities
may exceed @xmath0 ( the second velocity should be less than @xmath0 ) . if one of the components does not move the velocity of the other component may reach the maximum value @xmath43 one can see from fig.1 that at @xmath44 both components may move with velocities that exceed @xmath0 . under the motion of the components in mutually perpendicular directions ( @xmath45 ) the velocities can simultaneously reach the maximum critical values ( [ 113 ] ) . here
we do not consider the possibility of excitation of vortices .
therefore , strictly speaking , our analysis yields only the upper bound for the critical velocities .
nevertheless , in a number of situations the estimation for the critical velocities presented in this paper is justified completely .
for example , this occurs when the superfluid flows past an obstacle with a small ( less than the healing length ) linear size @xcite .
in bose gases confined in optical or magnetic traps surface excitations have the minimum phase velocity @xcite . in such systems the process of generating of vortices is connected with the excitation of surface modes , and the critical velocity coincides with the phase velocity of the lowest surface mode @xcite . let us study the critical velocities for a two - component bose gas , confined in a harmonic cylindrical potential @xmath46 .
we will consider the case when superfluid flows are directed along @xmath47 .
for simplicity we assume @xmath48 ( @xmath49 - radial coordinate ) , @xmath50 , @xmath51 and @xmath52 .
let us consider the system whose thomas- fermi radius @xmath53^{1/2}$ ] is much large than the oscillator length of the trap . to find the spectrum of elementary excitations we pass from the gross -pitaevskii equation to the linearized system of hydrodynamic equations for the densities @xmath54 and the velocities @xmath55 of the components : @xmath56 where @xmath57 is the equilibrium density of the components ( @xmath58 is the density in the center of trap ) , @xmath59 are their superfluid velocities and @xmath60 and @xmath61 are the fluctuations of these values .
the modes we are interested in are localized near the surface and the problem considered can be reduced to the problem for the spectrum of excitations in a bose gas in a linear potential @xcite . with this simplification
we obtain the following dispersion equation @xmath62 \left [ ( \omega - k v_2)^2 - \frac{2\gamma n_0 k}{m r_{\mathrm{tf}}}\right ] - \frac{4 \gamma_{12}^2 n_0 ^ 2 k^2}{m^2 r_{\mathrm{tf}}^2}= 0.\ ] ] the region of the applicability of ( [ 24 ] ) is bounded from above by the condition @xmath63 where @xmath64 is the wave vector for which the contributions of kinetic and potential energies to the excitation energy become comparable . the value of @xmath64 can be estimated by equating the kinetic energy of the particles @xmath65 to the energy of the lowest mode ( in the hydrodynamic limit ) @xmath66^{1/2}$ ] what gives @xmath67{\displaystyle \frac{m n_0(\gamma-\gamma_{12})}{\hbar^2 r_{\mathrm{tf}}}}$ ] . at the point
@xmath68 the dependence of the excitation energy on @xmath5 has a bend and the critical velocity can be estimated by substituting of @xmath69 into the equation ( [ 24 ] ) . as it follows from ( [ 24 ] ) for @xmath70 the critical velocity is the minimum phase velocity of the lowest surface mode @xmath71 if only one component moves , the critical velocity reaches the value @xmath72 thus , the highest velocity can be reached when only one component flows .
in conclusion , let us discuss a number of possibilities to observe the predicted behavior .
one of them is to create barriers not penetrable for one of the components .
if in the absence of the barriers the components move with equal velocities the appearance of such barriers may increase of the critical velocity .
the motion of the components with different velocities ( and in different directions ) can also be realized in systems , in which the components are separated spatially , for example , in a bilayer geometry ( in that case a long - range interaction between the components is required for observing the effect predicted ) .
the effect is also may be observed in multilayer condensates of electron - hole pairs which can appear in semiconducting heterostructures with an even ( @xmath73 ) number of two - dimensional electron layers @xcite . in electron - hole condensates
the critical velocities can be easily measured since they are proportional to the value of critical currents . | attention to study two - component bose - einstein condensates progress in cooling trapping atomic rarefied gases condensates experimentally @xcite they cosmological astrophysical processes modelled in two component systems superfluid components flow different velocities unusual drag effect takes create controlled phase difference between two bose - condensates observe effects similar to superconductive systems with josephson contacts magnetic fields study problem critical velocities in two component superfluid system with components different velocities landau criterion critical velocity in single - component condensate determined by expression @xmath3 @xmath4 excitation spectrum @xmath5 wave number velocity given in frame of reference connected with walls obstacles applied to two - component condensate when both components move same velocity landau criterium requirement of positivity of energies of elementary excitations in frame energies depend on velocities components @xmath6 @xmath7 landau criterium should joint condition on @xmath6 @xmath8find energy elementary excitations use gross - pitaevskii equation two - component system @xmath9 masses particles interaction constants scattering lengths wave function component sum stationary fluctuating part @xmath15 @xmath16 stationary part condensate wave function @xmath17 @xmath18 chemical potentials components gradients phases @xmath19 connected superfluid velocities relation @xmath20 fluctuating part written @xmath21 substitution ( [ 14 ] ) into linearized version ( [ 5 ] ) leads system equations @xmath22 coefficients dispersion equation spectrum elementary excitations @xmath23 e_2 2 - - - 4_2^2 n_1 = 0 @xmath24 bogolyubov spectrum @xmath25-component @xmath26 @xmath27 equation [ 16 ] ) yields @xmath28 assume condition stability two - component condensate phase separation fulfilled @xmath29 ) requirement positivity ( [ 18 ] ) @xmath30 equivalent condition ( [ 1 ] )condition yields expression critical velocity @xmath31 @xmath32 are bare velocities sound modes components interaction @xmath33 landau criterium requires two positive solutions eq at @xmath30 equivalent to two inequalities : @xmath34 [_2 2 - - 4_2^2 > 0, @xmath35 critical values of @xmath6 @xmath8 correspond inequality ( [ 30 ] ) turnes into equality for one @xmath30 sufficient condition for fulfilling inequalities @xmath30 is fulfillment at @xmath36 all directions @xmath30 inequalities ( 30 31 ] ) replaced with system inequalities @xmath37 @xmath38 angle between @xmath6 @xmath8 @xmath39 angle between @xmath30 @xmath6 fulfilled for all @xmath39 results analysis 20 ) in fig.1 if components move same direction critical velocities related by equation @xmath40 additional condition @xmath41 for @xmath42 critical velocity coincides with velocity lowest hydrodynamic mode @xmath0 one velocities may exceed @xmath0 second velocity less than @xmath0 )one move velocity other reach maximum value @xmath43 at @xmath44 both components move velocities exceed @xmath0 under motion perpendicular directions ( @xmath45 ) velocities reach maximum critical values ( [ 113 ] ) consider excitation vortices analysis yields upper bound critical velocities estimation critical velocities justified superfluid flows past obstacle small linear size @xcite bose gases optical magnetic traps surface excitations minimum phase velocity @xcite generating vortices connected with excitation surface modes critical velocity coincides with velocity lowest surface mode @xcite study critical velocities for two - component bose gas harmonic cylindrical potential @xmath46 superfluid flows directed along @xmath47 assume @xmath48 @xmath49 @xmath50 @xmath51 @xmath52 consider system thomas- fermi radius @xmath53^{1/2}$ ] large than oscillator length trapfind spectrum elementary excitations pass from gross -pitaevskii equation to linearized hydrodynamic equations for densities @xmath54 velocities @xmath55 components @xmath56 @xmath57 equilibrium density density center trap superfluid velocities fluctuations modes interested localized near surface problem reduced to problem spectrum excitations bose gas linear potential @xcite obtain dispersion equation @xmath62 \left\omega - k v_2)^2 -{2\gamma n\mathrm{tf -{4\gamma{12^2 n_0^2{tf}}^2}= 0. region applicability ( [ 24 ] ) bounded by condition @xmath63 @xmath64 wave vector contributions kinetic potential energies excitation energy comparable value @xmath64 estimated equating kinetic energy particles @xmath65 to energy lowest mode limit @xmath66^{1/2}$ @xmath67^2 r$ point @xmath68 dependence excitation energy on @xmath5 critical velocity estimated by substituting @xmath69 into equation ( [ 24 ] )follows from ( [ 24 ] ) @xmath70 critical velocity is minimum phase velocity lowest surface mode @xmath71 if one component moves critical velocity reaches value @xmath72 highest velocity reached when one component flows discuss possibilities to observe predicted behavior create barriers not penetrable for barriers components move equal velocities increase critical velocity motion components different velocities directions realized in systems components separated spatially bilayer geometry long - range interaction required effect observed in multilayer condensates electron - hole pairs semiconducting heterostructures even @xmath73 ) number two - dimensional electron layers electron - hole condensates critical velocities measured proportional to critical currents | 0.35112 | 2,292 | 1,177 | , Saving $0.1 in GPT-4. |
177,141 | we propose the new dark matter particle candidate the `` black hole atom '' , which is an atom with the charged black hole as an atomic nucleus and electrons in the bound internal quantum states . as a simplified model we consider the the central reissner - nordstrm black hole with the electric charge neutralized by the internal electrons in bound quantum states . for the external observers these objects would look like the electrically neutral schwarzschild black holes .
we suppose the prolific production of black hole atoms under specific conditions in the early universe . | the idea of `` black hole atoms '' goes back a long way in several variations .
m. a. markov et al .
proposed and studied in detail the model of maximons ( or friedmons ) @xcite .
these objects are the particle - like gravitating systems ( semiclosed worlds ) with mass close to the planck mass @xmath0 g. they may have in principle a large gravitational mass defect .
the friedmons in connection with self - energy of elementary particles were discussed in @xcite .
maximons are interesting for cosmological applications , in particular , because they have the particle - like properties and may be the enigmatic dark matter particles
. possible role of maximon clusters in cosmology was considered in @xcite .
the idea of micro black hole caring the electric charge and having the orbiting electrons or protons at the outer ( outside the horizon ) orbits was discussed by s. hawking in @xcite .
he firstly proposed that the charged black holes may play the role similar to the atomic nuclei .
later the idea of black hole atoms was investigated in @xcite .
the possible origin of the such planck mass black hole is the final stable state of the evaporated primordial black holes ( pbh ) , see e.g. , @xcite .
the remnants of the evaporated black holes can be stable and also can serve as the dark matter candidates @xcite . in this paper
we discuss the black hole atoms , which are the atoms with the charged black hole as atomic nuclei and with electrons in the bound * internal * quantum states .
the quantum bound states of electrons may exist in principle not only outside the event horizon but also inside the cauchy horizon of the charged black hole .
so , the main new idea is that there can be configurations in which the orbiting electrons are * inside * the black hole cauchy horizon .
we propose these black hole atoms as the possible origin of dark matter particles .
the quantum levels in the gravitational field of black holes outside the event horizon were studied in @xcite .
the resulting black hole atoms can be the dark matter particles in the case of uncompensated charge ( electrons at outer levels ) as it was proposed in @xcite .
the similar idea but for the zero total charge @xmath1 and for the electrons inside a black hole was proposed in @xcite . in the latter case , the total charge of all the electrons at the inner orbits is equal to the charge of the black hole , which appears at reissner - nordstrm metric . in the case of compensated charge
these systems look for the external observer as having the schwarzschild metric .
neutral systems interact weakly with other particles , it makes them the good candidates for the dark matter particles .
the stationary quantum levels of fermions in the gravitational field of the charged black holes have been found in the work @xcite by solving the corresponding dirac equation .
the dirac equation in the riemann geometry was first derived in the papers @xcite .
the using of only the covariant generalization is not enough for derivation of the corresponding dirac equation .
it is needed the determination of the parallel spinor transport . as it was shown in @xcite , a self - consistent steady - state solution with a finite normalization integral
can exist only in the case of extreme black hole , whose charge in the appropriate units is equal to its mass @xmath2 .
let us briefly describe the method of quantum level calculations ( for the more details see @xcite ) .
the dirac equation in the general metric has the following form @xcite : @xmath3 where @xmath4 , @xmath5 are the standard dirac matrices , and @xmath6 are tetrads .
the elongated derivative has the form @xmath7 , where @xmath8 , @xmath9 is the electromagnetic 4-potential .
we consider a charged reissner - nordstrm black hole with the metric @xmath10 , where @xmath11 , @xmath12 is the black hole mass , and @xmath13 is its charge . following the method of @xcite we separate the variables by the following way @xmath14 , \label{gih}\ ] ] where @xmath15 , and @xmath16 is the angular part of the wave function .
after substitution in ( [ urd ] ) , we obtain the next system of equations : @xmath17=0},\\ \\ { \displaystyle \frac{dh}{dr}+\frac{hk}{rf^{1/2}}-\frac{g}{f^{1/2}}\left[\frac{1}{f^{1/2}}\left(e-\frac{qq}{r}\right)-m\right]=0 } , \end{array } \right . \label{itogeq2}\ ] ] where @xmath18 .
the condition for the physically acceptable solution is the finiteness of the normalization integral @xmath19 which was derived from the zero component of the fermion current @xmath20 .
this integral is finite only in the case of extremal black hole @xmath2 .
such a black hole has equal horizons radii @xmath21 .
consider the internal solution for extreme black hole near the cauchy horizon @xmath22 .
let us denote @xmath23 and @xmath24 .
as can be shown @xcite , the equation ( [ itogeq2 ] ) in this case have the regular solution @xmath25 where @xmath26 , @xmath27 , and the ( [ normint ] ) is finite in the case @xmath28 .
this solution corresponds to the energy level @xmath29 electron has the same energy ( [ energyextr ] ) for the regular external solution outside the event horizon .
equation ( [ itogeq2 ] ) can be solved numerically far away from the horizons @xcite . in the case of the non - extreme black hole ,
@xmath30 , the integral ( [ normint ] ) diverges at the horizons @xcite .
we suppose that quantum levels in the black hole interiors are stable .
these means , additionally , the supposition of the internal universes inside the charged black holes , like in the idealized eternal black holes .
the other possibility is the dynamical formation of the internal universes in the gravitational collapse @xcite . for the quantum formation of black holes at the particle accelerators
the specific extra dimensions are needed @xcite .
the classical bound stable orbits inside black holes considered , e.g. , in @xcite .
the basic condition of existence of an atom as a quantum system in the quasiclassical approximation in flat space is the geometrical condition @xmath31 , where @xmath32 is the characteristic scale of the potential hole .
strong space - tame qurvature , espesially near the horizons , changes this criterion drastically .
a physical reason for the geometrical criterion breakdown is the infinite stretching of the physical distance @xmath33 at the cauchy horizon @xmath34 .
really , the physical distance from the centre @xmath35 of the extremal reissner - nordstrm black hole ( @xmath21 ) to some @xmath36 is @xmath37 it diverges for @xmath38 .
therefore , there is enough space under the cauchy horizon for particle with any wavelength .
as seen from the exact solution of the dirac equation , the wave function of the electron just localised under the horizon , and it s occurs far from quasiclassical regime .
a question about capture of particle with large @xmath39 wavelength by a black hole can arise .
but we do nt consider here the capture of the electron inside the black hole .
instead , we propose the formation of black hole with the electron already inside due to single quantum jump .
pbhs can be formed as at the cosmological stage of radiation dominance from the adiabatic density perturbations @xcite , and in the early dust - like stages @xcite .
although the primordial black hole is formed by the classical gravitational collapse , their final stage of evaporation is mainly the quantum process .
sakharov discussed the possibility of the superheavy particles emission by the black hole at the last stage of its evaporation @xcite .
the emitted particles could have even the masses up to the planck mass , although the probability of such emission is not clear .
one can imagine the last stage of a black hole evaporation not as a gradual radiation , but as the quantum jump into a new state .
we assume that black hole atoms can result from such quantum jumps .
the discussion about the final stage of evaporation can be found in @xcite , @xcite . in the case of the jump
the charged black holes with electrons on internal quantum orbits could be born effectively just after the evaporation of the pbhs population .
note that the hawking temperature for the extremal black hole is equal to zero and , so the extremal black holes do not experience the quantum evaporation ( see @xcite , @xcite , @xcite , @xcite ) , and they are stable dark matter candidates in this sense
. really the hawking temperature of the charged black hole @xmath40 equals to zero if @xmath41 .
nevertheless , the stability of the extremal black hole atoms with respect to quantum decay remains questionable .
let us consider in more details the formation mechanism of pbh from the adiabatic density perturbations .
such perturbations can arise at inflationary stage , and the necessary condition for the effective pbh formation is the excess of perturbations power at some small scale , because the simplest near flat spectrum can not produce the sufficient amount of pbhs .
but in the case of some peak in the spectrum , the pbhs can form just in the required cosmological abundance .
now we consider the modification of the of the required peak s high by taking into account the mass loss of the bphs during their evaporation from the initial mass @xmath42 at the formation moment @xmath43 till the plank mass remnants after evaporation . with this mass loss
the contemporary cosmological parameter of the pbhs can be expressed as @xcite @xmath44 where @xmath45 is the r.m.s .
perturbation at the mass - scale @xmath42 , @xmath46 according to the analytical estimation of @xcite or @xmath47 as was obtained in the models of critical collapse , @xmath48 is the scale - factor , @xmath49 is the matter - radiation equality moment , and @xmath50 by solving iteratively the system of equations ( [ eqom ] ) and ( [ eqbet ] ) with @xmath51 we find the dependence of @xmath45 on @xmath42 , which is shown at fig .
[ gr1 ] . at the mass - scale @xmath42 ,
required for the explanations of dark matter as pbhs remnants , in dependence on the initial pbh s mass @xmath42 for value of the collapse threshold @xmath46 ( solid curve ) and @xmath47 ( dashed curve ) .
[ gr1],scaledwidth=45.0% ] for @xmath52 g there is no solutions of ( [ eqom ] ) and ( [ eqbet ] ) , and the scenario is limited only smaller @xmath42 .
therefore , to produce the cosmological abundance of pbhs plank - mass remnants @xmath53 as is required for dark matter , one needs the density perturbations spectrum with the r.m.s .
values shown at fig . [ gr1 ] .
the interaction of the neutral black hole atoms with ordinary matter via the gravitational dynamical friction effect is extremely weak , as it was first shown in @xcite .
this is due to the extremely small cross - section @xmath54 sm@xmath55 , where @xmath56 is the relative velocity , @xmath57 in the galactic halo .
one can compare it with the neutrino - nucleons interaction cross - sections @xmath58
sm@xmath55 .
the black hole can be born charged even in the classical collapse .
but the neutralization by the accretion could reduce the initial charge till the value @xmath59 @xcite .
the remaining charge can interact with the ordinary matter in the universe like heavy atomic nuclei @xcite .
so , the charged black hole atoms experience strong collisions and dissipations .
this create some difficulties for these systems as dark matter particles .
although , the solution of this problem may be in the formation of molecular - like systems with ordinary charged particles as in the composite - dark - matter scenario @xcite .
possible detection method of charged the black hole atoms was also proposed in the seminal paper @xcite .
these systems can draw straight tracks in the chambers almost without the deflection by the magnetic field due to the their large masses .
although in the case of a non - extreme black hole the normalization integral diverges , nevertheless , the irregular quasi - stationary solutions can exist @xcite .
in addition , the regular internal ( under the cauchy horizon ) and external solutions are possible @xcite .
regular internal solutions have only single energy levels inside and outside the black hole .
the internal level has the energy @xmath60 , and the energy for the outer one is @xmath61 . if electron jumps or tunnels from the external to the internal level , the energy @xmath62 can be radiated in the form of photons .
this energy could reaches the very high value , up to the energy corresponding to the ultrahigh energy cosmic rays .
note , that the radiation from the bound system of two friedmons was considered in @xcite .
the released energy is supplied by the gravitational energy of the black hole and by the energy of the electrostatic interaction of electrons with a charged black hole .
the gravitational factor is due to the fact that gravity is responsible for the localization of electrons near horizons at the quasi - stationary regular quantum orbits . in the case of the extreme black hole one
has @xmath63 . in this case
the energy release during the electron transition is absent .
for this reason the extreme black holes with @xmath1 are the very `` quiet '' , dark and non - interacting objects .
these properties are just one needs for the dark matter candidates .
in this paper we discuss the new kind of `` black hole atom '' system : the reissner - nordstrm black holes with the electrons at quantum levels under the cauchy horizon .
if the electric charge of the black hole is neutralized by the internal electrons in bound quantum states , these objects would look like the electrically neutral schwarzschild black holes for the external observers .
due to extremely small interaction cross - section these neutral systems are almost non - interacting with baryons and behave as collisionless and dissipationless gas .
this property makes them the good dark matter candidates .
the black hole atoms under consideration could form at the final stages of pbhs evaporation at early universe .
the pbhs itself may form in different scenarios : from adiabatic perturbations , during cosmological phase transitions or at the early dust - like stages @xcite .
the extremal black holes does nt evaporate in the hawking process , and they are stable in this sense .
but the existence of the internal and external quantum levels gives the possibility of the quantum transitions between the levels with radiation of the photons , and this effect makes the `` black hole atoms '' are observable in principal .
dokuchaev , yu.n .
eroshenko , `` stationary solutions of the dirac equation in the gravitational field of a charged black hole '' , _ journal of experimental and theoretical physics _
117 , p. 72
, 2013 .
v. fock , `` geometrisierung der diracschen theorie des elektrons '' , _ zeitschrift fur physik _ , vol .
57 , p. 261
, 1929 ; v.fock , d.iwanenko , ` uber eine m?gliche geometrische deutung der relativistischen quantentheorie' , _ zeitschrift fur physik _ , vol .
54 , p. 798
, 1929 .
s. grunau and v. kagramanova 2011 `` geodesics of electrically and magnetically charged test particles in the reissner - nordstrom space - time : analytical solutions '' , _ phys .
_ d , vol .
83 , p. 044009
, 2011 .
zabotin , p.d .
naselskii , a.g .
polnarev , `` high - amplitude peaks of density disturbances and the formation of primordial black - holes in the dust like universe '' , _
soviet astronomy _ , vol .
31 , p. 353 | idea of black hole atoms markov et al proposed studied model of maximons objects particle - like gravitating systems with mass close to planck mass large gravitational mass defect friedmons with self - energy particles discussed in maximons interesting for cosmological applications particle - like properties dark matter particles role of maximon clusters in cosmology considered idea micro black hole electric charge orbiting electrons protons at outer orbits discussed by s. hawking proposed charged black holes similar to atomic nuclei later idea black hole atoms investigated possible origin of planck mass black hole is final stable state of evaporated primordial black holes remnants holes stable serve as dark matter candidates paper discuss black hole atoms charged black hole as atomic nuclei electrons in bound internal quantum states bound states electrons exist outside event horizon inside cauchy horizon of charged black hole idea configurations orbiting electrons inside black hole cauchy horizon propose black hole atoms as possible origin of dark matter particles quantum levels in gravitational field of black holes outside event horizon studied in @xciteblack hole atoms dark matter particles uncompensated charge electrons outer levels proposed in @xcite similar idea for zero total charge @xmath1 electrons inside black hole proposed in @xcite total charge electrons at inner orbits equal to charge black hole at reissner - nordstrm metric compensated charge systems schwarzschild metric neutral systems interact weakly with particles good candidates for dark matter particles stationary quantum levels of fermions in gravitational field charged black holes found in @xcite by solving dirac equation geometry derived in @xcite covariant generalization not determination of parallel spinor transport self - consistent steady - state solution with finite normalization integral extreme black hole charge equal to mass @xmath2 method quantum level calculations dirac equation metric form @xcite : @xmath3 @xmath4 standard dirac matrices are tetrads elongated derivative @xmath7 @xmath8 is electromagnetic 4-potential charged reissner - nordstrm black hole with metric @xmath10 @xmath11 black hole mass @xmath13 chargemethod @xcite separate variables @xmath14{gih} @xmath15 @xmath16 angular part wave function substitution urd ] obtain system equations @xmath17=0}\frac{dh}{dr+{hk}{rf^{1/2}}-{g}{f^{1/2}}\left{qq{r\right-m\right]=0\end{array{itogeq2} @xmath18 condition solution finiteness normalization integral @xmath19 derived from zero component fermion current @xmath20 integral finite extremal black hole @xmath2 equal horizons @xmath21 internal solution extreme black hole near cauchy horizon @xmath22 denote @xmath23 @xmath24 equation itogeq2 ] regular solution @xmath25 @xmath26 @xmath27 normint ] finite @xmath28 solution corresponds energy level @xmath29 electron same energy energyextr regular external solution outside event horizon equation itogeq2 ] solved numerically far away from horizons non - extreme black hole @xmath30 integral normint ] diverges at horizons quantum levels black hole interiors stablesupposition internal universes inside charged black holes eternal possibility dynamical formation universes in gravitational collapse @xcite quantum formation black holes particle accelerators extra dimensions needed classical stable orbits inside black holes @xcite basic condition existence atom quantum system geometrical condition @xmath31 @xmath32 characteristic scale potential hole strong space - tame qurvature near horizons changes criterion reason geometrical criterion breakdown infinite stretching physical distance @xmath33 at cauchy horizon @xmath34 physical distance from centre @xmath35 - nordstrm black hole to @xmath36 @xmath37 diverges @xmath38 enough space under cauchy horizon for particle any wavelength wave function electron localised under far quasiclassical regime question capture particle large @xmath39 wavelength by black hole consider capture electron inside black hole propose formation black hole electron inside due single quantum jump pbhs formed cosmological stage radiation dominance adiabatic density perturbations early dust - like stages primordial black hole formed by classical gravitational collapse final stage evaporation quantum processsakharov discussed superheavy particles emission black hole at last stage evaporation @xcite emitted particles could have masses up to planck mass probability not clear imagine last stage black hole evaporation radiation quantum jump into new state black hole atoms result from quantum jumps discussion final stage evaporation in @xcite charged black holes with electrons on quantum orbits could be born after evaporation hawking temperature extremal black hole equal to zero experience quantum evaporation stable dark matter candidates hawking temperature charged black hole @xmath40 equals to zero if @xmath41 stability extremal black hole atoms quantum decay remains questionable consider formation mechanism of pbh from adiabatic density perturbations perturbations arise at inflationary stage for pbh formation excess of perturbations power small scale simplest flat spectrum produce pbhs peak spectrum pbhs can form in required cosmological abundance consider modification required peak s high mass loss of bphs during evaporation from initial mass @xmath42 till plank mass remnants after evaporationmass loss cosmological parameter pbhs expressed as @xcite @xmath44 @xmath45 r.m.s . perturbation at mass - scale @xmath42 @xmath46 @xmath48 scale - factor @xmath49 matter - radiation equality moment @xmath50 equations eqom eqbet @xmath51 @xmath45 on @xmath42 shown at fig . gr1 mass - scale @xmath42 explanations dark matter pbhs remnants initial pbh s mass collapse threshold @xmath46 solid curve @xmath47 dashed curve @xmath52 no solutions of eqom eqbet ] scenario limited smaller @xmath42 cosmological abundance of pbhs plank - mass remnants @xmath53 dark matter needs density perturbations spectrum r.m.s . values fig . gr1 interaction neutral black hole atoms with ordinary matter gravitational friction weak @xcite due to small cross - section @xmath54 sm@xmath55 @xmath56 relative velocity @xmath57 galactic halo neutrino - nucleons interaction cross - sections @xmath58@xmath55 black hole born charged in classical collapseneutralization by initial charge till @xmath59 remaining charge with ordinary matter like heavy atomic nuclei charged black hole atoms experience collisions dissipations difficulties for dark matter solution may in molecular - like systems with ordinary charged particles composite - dark - matter scenario detection method of charged black hole atoms proposed in seminal paper systems draw straight tracks in without deflection magnetic field to large masses non - extreme black hole normalization diverges irregular quasi - stationary solutions exist regular internal external solutions possible solutions have single energy levels inside outside black hole internal level has energy @xmath60 outer @xmath61 if electron jumps from to level energy @xmath62 radiated in photons energy high value ultrahigh energy cosmic rays radiation from bound system of two friedmons considered in released energy supplied by gravitational energy black hole electrostatic interaction of electrons with charged black hole gravitational factor due to for localization of electrons near horizons at quasi - stationary quantum orbits extreme black hole @xmath63 energy release during electron transition absentextreme black holes with @xmath1 quiet dark non - interacting for dark matter candidates new black hole system reissner - nordstrm black holes electrons under cauchy horizon if electric charge neutralized by internal electrons look like electrically neutral schwarzschild black holes for observers small interaction cross - section neutral systems almost non - interacting with baryons as collisionless dissipationless gas good dark matter candidates black hole atoms form at final stages pbhs evaporation early universe pbhs form adiabatic perturbations cosmological phase transitions early dust - like stages extremal black holes evaporate in hawking process stable internal external quantum levels quantum transitions between with radiation photons black hole atoms observable dokuchaev eroshenko stationary solutions of dirac equation in gravitational field of charged black hole journal experimental 117 p. 72 2013 v. fock geometrisierung der diracschen theorie des elektrons zeitschrift fur physik 57 p. 261 1929.iwanenko geometrische deutung relativistischen quantentheorie zeitschrift physik 54 p. 798 1929 grunaukagramanova 2011 geodesics charged particles reissner nordstrom space analytical solutions phys. vol 83 p. 044009 2011 zabotin naselskii polnarev high amplitude peaks density disturbances formation primordial black - holes universe soviet astronomy vol 31 p. 353 | 0.572447 | 3,594 | 1,807 | , Saving $0.1 in GPT-4. |
177,148 | measurement of higgs branching ratio is necessary to investigate higgs coupling to particle masses . especially , it is the most important program to measure the branching ratio of @xmath0 and @xmath1 at the international linear collider ( ilc ) .
we have studied the measurement accuracy of higgs branching ratio at ilc with @xmath2 gev by using @xmath3 events .
we obtained the higgs branching ratio with 1.1% and 13.7% accuracy for @xmath0 and @xmath1 , respectively . | in the higgs mechanism , higgs coupling is proportional to a particle mass .
for that reason , it is important to measure the higgs coupling to particle masses , i.e. higgs branching ratio , is important to confirm higgs mechanism and distinguish the standard model extensions .
especially , it is the most important program to measure the branching ratio of @xmath0 and @xmath1 at ilc @xcite with the excellent performance of the flavor tagging .
we have studied the measurement accuracy of higgs branching ratio at ilc by using @xmath3 events . in this paper
, we report the measurement accuracy of higgs branching ratio of @xmath0 and @xmath1 .
in this study , we used common generator samples in the ilc community for @xmath4 events and standard model backgrounds , which were prepared with witherd at slac @xcite . in this study ,
the higgs mass was assumed to be 120 gev .
we used the center of mass energy of @xmath2 gev and the integrated luminosity of 250 fb@xmath5 . here
, the beam energy spread was assumed as 0.3% for the electron and positron beam .
the beam polarization was set to 80% left - handed for the electron beam and 30% right - handed for the positron beam .
the signal and background events were simulated by the full simulator , mokka@xcite , where the detector model is ` ild_00 ` was implemented as the detector model @xcite .
hadronization was done by pythia6.409 , in which the higgs branching ratio is defined as shown in table [ tb : hbr ] for the higgs mass of 120 gev .
after the detector simulation , the reconstruction was performed by marlin@xcite .
.the higgs branching ratio defined in pythia6.409 . [ cols="<,>",options="header " , ]
measurement of higgs branching ratio is necessary to investigate higgs coupling to particle masses . especially , it is the most important program to measure the branching ratio of @xmath0 and @xmath1 at ilc .
we have studied the measurement accuracy of higgs branching ratio at ilc with @xmath2 gev by using @xmath3 events . for higgs mass of 120 gev and the integrated luminosity of 250 fb@xmath5
, we obtained the measurement accuracy of 1.1% and 13.7% for @xmath0 and @xmath1 , respectively , assuming that the cross section of @xmath4 is determined by other measurements .
finally , the relative branching ratio between @xmath0 and @xmath1 was obtained with 13.3% accuracy .
the authors would like to thank all the members of the ilc physics subgroup @xcite for useful discussions and ild optimization working group .
this study is supported in part by the creative scientific research grant no .
18gs0202 of the japan society for promotion of science
. 99 international linear collider reference design report , ilc global design effort and world wide study . http://ilcsoft.desy.de / portal / data_samples/. http://ilcsoft.desy.de / portal / software_packages / mokka/. http://www.ilcild.org . http://ilcsoft.desy.de / portal / software_packages / marlin/. s. catani , et al . , phys .
lett . b269 ( 1991 ) 179 ; + n. brown , w. j. stirling , z. phys .
c53 ( 1992 ) 629 .
thorsten kuhl and klaus desch , lc - phsm-2007 - 001 .
lcfivertex package reference manual , http://ilcsoft.desy.de / portal / software_packages / lcfivertex/. p. garcia - abia and w. lohmann , eur .
phys . j. direct c2 , 2 ( 2000 ) .
http://www - jlc.kek.jp / subg / physics / ilcphys/. | higgs mechanism coupling proportional to particle mass important measure higgs coupling particle masses branching ratio confirm higgs mechanism distinguish standard model extensions measure branching ratio of @xmath0 @xmath1 at ilc @xcite flavor studied higgs branching ratio at ilc @xmath3 events report higgs branching ratio @xmath0 @xmath1 used common generator samples ilc for @xmath4 events standard model backgrounds prepared with witherd slac @xcite higgs mass 120 gev used center of mass energy @xmath2 gev integrated luminosity 250 fb@xmath5 beam energy spread 0.3% for electron positron beam beam polarization 80% left - handed electron 30% right positron signal background events simulated by simulator mokka@xcite detector model ild_00 @xcite hadronization by pythia6.409 higgs branching ratio defined table for higgs mass 120 gev reconstruction by marlin@xcite higgs branching ratio defined in pythia6.409measurement higgs branching ratio higgs coupling particle masses measure branching ratio @xmath0 @xmath1 ilc studied ilc @xmath2 gev @xmath3 events higgs mass 120 gev luminosity 250 fb@xmath5 accuracy 1.1% 13.7% @xmath0 @xmath1 cross section @xmath4 relative branching ratio between @xmath0 @xmath1 obtained 13.3% accuracy ilc physics subgroup @xcite ild optimization working group study supported creative scientific research grant 18gs0202 japan society promotion science international linear collider reference design report ilc global design effort world wide study http://ilcsoft.desy.de portal data_samples software_packages mokka.ilcild.org marlin catani phys . lett . b269 ( 1991 ) 179 phys . c53 ( 1992 ) 629 thorsten kuhl klaus desch lc - phsm-2007 - 001 lcfivertex package reference manual http://ilcsoft software_packages lcfivertex. garcia - abia w. lohmannphys direct c2 2000 ) jlc.kek.jp subg physics ilcphys | 0.177525 | 883 | 465 | , Saving $0.0 in GPT-4. |
177,149 | the production and decay of heavy charged higgs bosons at a 800 gev @xmath0 linear collider have been studied . the analysis of the @xmath1 , expected to be dominant in the mssm , and @xmath2 decay modes leading to the same final state consisting of two @xmath3 bosons and four @xmath4 quarks , provides with a determination of the boson mass to 1 gev/@xmath5 and of the production cross section with 10% accuracy for 500 fb@xmath6 of data . | the study of the origin of electro - weak symmetry breaking and the exploration of the higgs sector is one of the main themes of the @xmath0 linear collider ( lc ) physics programme .
a distinctive feature of several extensions of the standard model ( sm ) , such as its minimal supersymmetric extension ( mssm ) , is the introduction of an extra higgs doublet resulting in five physical higgs boson states , two of which are charged @xmath7 . while the existence of an extended higgs sector may be indirectly revealed by a detailed study of the production and decay properties of the lightest neutral higgs boson @xmath8 @xcite , its direct manifestation will only come from the detection of pairs of heavier higgs particles , such as @xmath9 .
we report the results of a study of the sensitivity of a 800 gev @xmath0 lc , such as tesla @xcite , for the detection of pair produced heavy charged higgs bosons .
the @xmath9 production cross section depends , at tree level , only on the boson mass , @xmath10 , while the dominant @xmath7 decay mode depends on the model parameters that may modify the higgs couplings to gauge bosons and fermions .
there have been some earlier studies of @xmath11 reconstruction at the lc for the case of a rather light boson , decaying predominantly into a @xmath12 quark pair @xcite .
this study has considered the two decay processes @xmath13 , expected to be dominant in the mssm with @xmath14 for @xmath15 , and @xmath16 , leading to the same final state consisting of two @xmath3 bosons and four @xmath4 quarks , for the cases of heavier bosons , with @xmath10 = 200 and 300 gev/@xmath5 and @xmath17 = 115 gev/@xmath5 . by selecting hadronic @xmath3 decays , the complete reconstruction of the resulting eight jet final states allows to determine the @xmath7 mass , profiting of the intermediate @xmath18 ( @xmath8 ) and @xmath3 mass constraints , the production cross section and the decay branching fractions .
accurate values for the mass and the decay branching ratios may contribute to fix some of the model parameters .
recent calculations of higher order corrections to the production cross section have shown a significant model dependence @xcite that may be useful for discriminating between models , provided an accurate experimental determination of the cross section is possible at the lc .
the results reported here have been obtained from a simulation study for 500 @xmath19 of integrated luminosity at @xmath20 = 800 gev .
the signal and backgrounds events have been generated with pythia 6 @xcite , including initial state radiation and accounting for the beamstrahlung effect computed for the tesla parameters
the generated events have been passed through the parametric simdet detector simulation @xcite .
the signal cross section depends strongly on the charged higgs boson mass , varying from @xmath21 @xmath22 for 200 gev/@xmath5 to @xmath23 @xmath22 for 300 gev/@xmath5 corresponding to @xmath24@xmath25 @xmath11 pairs/500
fb@xmath6 . this abundant production at a high luminosity lc ,
allows to enforce tight requirements on the event preselection and the mass reconstruction .
most two and four fermion background processes can be effectively rejected by requiring eight jets and four @xmath4 quarks , despite their large production cross sections . the @xmath26 background which results in multi - jet final state with @xmath4 quarks
can be reduced by the both @xmath4 tagging and the mass kinematical fits .
genuine @xmath27 final states , as those originating from the higgs radiation off the top quark process @xmath28 , remain the largest source of background with an estimated cross section of 3 fb which has been reduced in this analysis by a kinematical fit . to select fully hadronic events , the visible energy of the reconstructed tracks and calorimetric clusters has been required to be above 600 gev , the missing momentum below 100 gev/@xmath29 and the energy of the most energetic lepton below 50 gev .
the hadronic system has been clustered into jets , using the camjet algorithm @xcite , varying the @xmath30 value from @xmath31 to @xmath32 .
only those events giving eight reconstructed jets , for a @xmath30 value in this range , have been further considered .
four out of the eight jets have been required to be @xmath4 tagged . in this analysis a parametrized @xmath4 tagging response , corresponding to 90% tagging efficiency with 5% mis - identification probability , has been used .
in order to efficiently distinguish the signal charged higgs production from the underlying backgrounds and to measure the boson mass , it is important to obtain a clean higgs signal in the mass distribution of the multi - jet final states .
the two decay hypotheses @xmath33 and @xmath34 have been considered .
events fulfilling the above selection criteria have been tested for the presence of two @xmath3 decaying into non-@xmath4 tagged jets .
each @xmath18 quark has then been reconstructed from a @xmath3 candidate paired with a @xmath4 tagged jet , in the case of the @xmath35 while for the case of @xmath36 , each @xmath37 decay has been reconstructed from tagged @xmath38 pairs . to further reject backgrounds as well as poorly reconstructed signal events , leading to an inaccurate measurement of @xmath10 , the compatibility of the mass of the jet combination to that of @xmath39 and @xmath8 bosons and the top quark within the measurement accuracy has been required . after completing the jet assignment , a kinematical fit , imposing energy and momentum conservation , the mass of intermediate states and equal higgs boson masses , has been applied .
this fit improves the signal mass reconstruction from by a factor of two for @xmath10 = 300 gev/@xmath5 .
the resulting fit @xmath40 has also been used to further reject background events and to distinguish between the two signal decay channels under study , based on the hypothesis giving the best @xmath40 value .
after applying these reconstruction and selection criteria , signal efficiencies in the range 2%-4% , according to the charged higgs boson mass and adopted @xmath40 cut value , has been obtained with a high signal purity . with a data set of 500 fb@xmath6 at @xmath20 = 800 gev ,
these correspond to signal samples of 150 - 600 events , with @xmath41 .
the corresponding statistical accuracies on the measurements of the boson mass and the production cross section are summarized in table 1 .
.statistical accuracy on the determination of the mass and production cross section for a charged higgs boson at @xmath20 = 800 gev with 500 fb@xmath6 [ cols="^,^,^",options="header " , ] due to the low level of background achieved , the sensitivity to a charged higgs pair production is guaranteed up to @xmath42 350 gev/@xmath5 , where the cross section falls below 5 fb , due to the @xmath43 suppression , reducing the number of signal events below the detection threshold at a @xmath20 = 800 gev collider .
the full reconstruction of the charged higgs boson @xmath1 and @xmath44 multi - jet final states has been studied for a sample of @xmath0 lc data at @xmath20 = 800 gev with @xmath10 = 200 and 300 gev/@xmath5 .
statistical accuracies of 1 gev/@xmath5 on the boson mass and of 10% on its production cross sections have been obtained .
these results can provide significant information about the parameters of an extended higgs sector in the mssm or in other sm extensions . | study origin electro - weak symmetry breaking exploration higgs sector main @xmath0 linear collider physics programme distinctive feature extensions standard model extra higgs doublet five higgs boson states two charged @xmath7 extended higgs sector revealed by study production decay properties lightest neutral higgs boson @xmath8 @xcite direct manifestation from detection heavier higgs particles @xmath9 results study sensitivity 800 gev @xmath0 lc for detection heavy charged higgs bosons @xmath9 production cross section depends on boson mass @xmath10 dominant @xmath7 decay mode depends model parameters higgs earlier studies @xmath11 reconstruction at lc light boson decaying into @xmath12 quark pair study considered two decay processes @xmath13 dominant mssm @xmath14 @xmath15 @xmath16 leading same final state two @xmath3 bosons four @xmath4 quarks heavier bosons @xmath10 = 200 300 gev/@xmath5 @xmath17 = 115 gev/@xmath5 selecting hadronic @xmath3 decays reconstruction eight jet final states @xmath7 mass intermediate @xmath18 @xmath3 mass constraints production cross section decay branching fractions accurate mass decay branching ratios fix model parameters calculations production cross section model dependence @xcite discriminating models results from simulation study for 500 @xmath19 integrated luminosity at @xmath20 = 800 gev signal backgrounds events generated with pythia 6 @xcite initial state radiation beamstrahlung effect computed tesla parameters events passed through parametric simdet detector simulation @xcite signal cross section depends on charged higgs boson mass varying from @xmath21 200 to 300 gev/ abundant production at high luminosity requirements event preselection mass reconstruction fermion background processes rejected eight jets four @xmath4 quarks large production cross sections @xmath26 background multi - jet final state with @xmath4 quarks reduced by @xmath4 tagging mass kinematical fits@xmath27 final states from higgs radiation largest source background cross section 3 fb reduced by kinematical fit select hadronic events visible energy reconstructed tracks calorimetric clusters above 600 gev missing momentum below 100 gev/@xmath29 energetic lepton below 50 gev hadronic system clustered into jets camjet algorithm varying @xmath30 value from to @xmath32 events eight reconstructed jets for @xmath30 value considered four eight jets @xmath4 tagged parametrized @xmath4 tagging response 90% tagging efficiency 5% mis - identification probability used distinguish signal charged higgs production measure boson mass clean higgs signal in mass distribution multi - jet final states decay hypotheses @xmath33 and @xmath34 considered events tested for two @xmath3 decaying into non-@xmath4 tagged jets each @xmath18 quark reconstructed from @xmath3 candidate @xmath4 tagged jet @xmath37 decay reconstructed from @xmath38 pairsreject backgrounds poorly reconstructed signal events inaccurate measurement @xmath10 compatibility mass @xmath39 @xmath8 bosons top quark measurement accuracy required assignment kinematical fit energy momentum conservation mass intermediate states equal higgs boson masses applied improves signal mass reconstruction @xmath10 = 300 gev/@xmath5 fit @xmath40 reject background events distinguish signal decay channels @xmath40 value signal efficiencies 2%-4% charged higgs boson mass @xmath40 cut value obtained high signal purity data set 500 fb@xmath6 @xmath20 = 800 gev signal samples 150 - 600 events @xmath41 statistical accuracies boson mass production cross section summarized table 1 accuracy mass production cross section charged higgs boson @xmath20 = 800 gev 500 fb@xmath6 low level background sensitivity charged higgs pair production guaranteed to @xmath42 350 gev/@xmath5 cross section below 5 fb @xmath43 suppression reducing signal events below detection threshold @xmath20 = 800 gev colliderreconstruction charged higgs boson @xmath1 - jet states studied @xmath0 data 800 gev 200 300 gev/@xmath5 statistical accuracies 1 gev/@xmath5 boson mass 10% production cross sections obtained results provide information extended higgs sector mssm sm extensions | 0.309016 | 1,753 | 907 | , Saving $0.1 in GPT-4. |
177,153 | + here diffusion and ware pinch are analyzed as opposed effects for plasma confinement , when instabilities are not considered . in this work
it is studied the equilibrium inductive electric field where both effects annul each other in the sense that the average normal velocity is zero , that is , marginal velocity confinement is reached . the critical electric field defined in that way
is studied for different values of elliptic elongation , shafranov shift and triangularity .
a short complementary analysis is also performed of the variation of the poloidal magnetic field along a magnetic line .
magnetohydrodynamic transport theory in the collisional regime is used as in recent publications .
axisymmetric and up - down symmetry are assumed .
+ -0.5 cm 0.2 cm epsf | the h - mode is characterized by the suppression of anomalous transport in tokamaks , because of low plasma turbulence induces by internal barriers[1 - 3 ] . as result neoclassical transport calculations becomes very important in this mode .
diffusion in the collisional regime depends of the pressure gradient at the 95% surface adjacent to the scrape of layer or sol , as well as the inductive electric field @xmath0 .
the diffusion due to the gradient pressure is opposed to ware the pinch effect due to @xmath0 .
+ in previous papers was shown that neoclassical diffusion can be treated with great simplicity in the case of arbitrary plasma configuration , using a new kind of tokamaks coordinates described there[4 - 6 ] .
however , suitable direct numerical calculations for different values of elongation , shafranov shift and triangularity have not been presented until now . some previous results on this theme using these coordinates are incomplete , and they are not using the right parameters for the tokamaks in operation at present .
+ here calculations are presented in a different way , since we look for the values of @xmath0 in the marginal velocity confinement , that is , when the average velocity on 95 % surface is zero the results corresponding to marginal confinement flux , that is , the transition from the outgoing to ingoing flux , could be more interested , but much more difficult to calculate and a suitable and simple treatment describing this process for any plasma configuration seem that they have not been publishing until now .
+ in the calculations now presented there are first a suitable normalization procedure absent in previous calculations as well as an adequate selection of tokamak parameters .
the normalization used here allows us to get results , which can be useful for a diversity of different tokamak plasma configurations , with different values of ellipticity , shafranov shift and triangularity .
the collisonal transport treatment presented in previous paper for toroidal axisymmetric plasmas can be written in a more convenient way using dimensionless integrals and variables , as + @xmath1 - \frac{\tilde{e}_{\varphi 1}}{\gamma_{1 } \hat{i}_{0 } } ( \hat{i}_{7 } - \frac{\hat{i}_{1 } \hat{i}_{6}}{\hat{i}_{4 } } + { \gamma}^{2}_{1 } \hat{i}_{5 } ) \ \ , \end{aligned}\ ] ] where @xmath2 is the dimensionless normal velocity derived from the velocity @xmath3 along the magnetic surface normal , and @xmath4 is a dimensionless electric field defined as @xmath5 her @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 are respectively the major radius , toroidal and poloidal magnetic field , inductive electric field and pressure gradient at the mid - plane external point @xmath11 of the magnetic cross section .
the plasma resistivity are @xmath12 and @xmath13 in the directions perpendicular and parallel to the magnetic field lines and @xmath14 is the minor axis radius .
the dimensionless quantity @xmath15 is the ratio @xmath16 between poloidal and toroidal magnetic field at point @xmath11 . on the other hand the new dimensionless integrals
@xmath17 , @xmath18 to @xmath19 , are defined as + @xmath20 @xmath21 @xmath22 @xmath23 @xmath24 @xmath25 @xmath26 @xmath27 where all the integrals are around a magnetic surface and @xmath28 is a function , depending of the curvature @xmath29 of the orthogonal line family , giving by @xmath30 this results are obtained using mhd equations and assuming toroidal axisymmetry . + in order to get numerical as well as analytic results it is useful to express the family of magnetic cross sections by the equations @xmath31 \ \ \ , \end{aligned}\ ] ] @xmath32 \ \ \ , \end{aligned}\ ] ] where @xmath33 , @xmath34 and @xmath35 are respectively ellipticity , triangularity and shafranov shift distortions . the parameter @xmath36 in this equations labels each magnetic surface .
the previous equations are a generalization of the equations presented by roach , et al[7 ] .
the quantities @xmath33 , @xmath34 and @xmath35 are dimensionless , however for the analysis and calculations are more useful the new dimensionless quantities @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 , defined respectively as @xmath42 the well know shafranov shift @xmath43 is connected to the previous parameters by @xmath44 where @xmath45 is the size of the 95 % magnetic surface measurement at the mid - plane , or in different words , @xmath46 is the horizontal half - width of the plasma .
the equations of the cross section magnetic lines will be now @xmath47 \ \ \ , \end{aligned}\ ] ] @xmath48 \ \ \ .\end{aligned}\ ] ] denoting by @xmath49 the value of the parameter @xmath36 generating the 95 % magnetic surface with general coordinates @xmath50 , @xmath51 , then @xmath52 \ \ \ , \end{aligned}\ ] ] @xmath53 \ \ \ .\end{aligned}\ ] ] the largest and smallest values of @xmath50 are respectively @xmath54 and @xmath55 , and its values are @xmath56 \ \ \ , \end{aligned}\ ] ] @xmath57 \ \ \ .\end{aligned}\ ] ] the radius @xmath58 of the center of the plasma and plasma size @xmath46 will be respectively @xmath59 \ \ \ , \end{aligned}\ ] ] @xmath60 it seems also convenient to connect that parameters with the aspect ratio @xmath61 , such that @xmath62 as in our previous papers , @xmath63 is the maximum value of @xmath51 .
the elliptic elongation @xmath64 and triangularity are connected to our previous parameters as @xmath65 where @xmath66 and @xmath67 are @xmath68 and @xmath69 \ \ \ , \end{aligned}\ ] ] where @xmath70 is the radius of the point with maximum z. +
the dimensionless variables defined previously simplifies the computation , because quantities we need for the calculation are : the ratio between poloidal and toroidal magnetic fields @xmath71 , the horizontal half - width of the plasma @xmath46 , the aspect ratio @xmath61 , the ratio @xmath72 , the shafranov shift , the ellipticity @xmath64 and triangularity @xmath73
. from @xmath64 and @xmath73 , the values of @xmath33 and @xmath38 are determined .
the shafranov shift , the value of @xmath46 and the calculated value @xmath38 together allow the calculation of @xmath74 and @xmath75 .
+ in this way the family of magnetic surfaces can be determined . following the procedures described in previous works the family of orthogonal lines
can be also determined , which allows to obtain the curvature function @xmath76 and the function @xmath77 .
all the integrals needed to obtain the normal dimensionless velocity @xmath2 can also be performed once the value of @xmath71 is given .
the velocity @xmath2 can also be obtained if the ratio @xmath78 is given , and it is a linear function of the dimensionless toroidal electric field @xmath79 .
the intersection of that value with the axis of abscissa allows to determine the critical dimensionless electric field @xmath80 , for marginal velocity confinement .
this field will be later determine for different values of ellipticity @xmath64 , dimensionless shafranov shift @xmath81 , and triangularity @xmath73 .
+ following the above described procedure , the ellipticity @xmath64 and triangularity @xmath73 , in figure 1 , are given as @xmath82 and @xmath83 .
these correspond to values of @xmath33 and @xmath38 : @xmath84 , @xmath85 .
the real procedure we use was a little different .
we first select values of @xmath33 and @xmath38 , in such way @xmath64 and @xmath73 become about their values in jet tokamak[8 ] .
this procedure is simpler for us , and it is the same idea .
+ in figure 1 , several cross sections magnetic lines have been drawn for different values of @xmath36 in the interval from zero to @xmath49 , giving in eq.(24 ) .
the characteristic values for that figure are : elipticity @xmath86 ; relative shafranov shift @xmath87 ; triangularity @xmath88 ; horizontal half width @xmath89 ( 95 % surface ) ; aspect ratio @xmath90 and minor magnetic axis ratio @xmath91 . from the previous data ,
the following parameters are determined : elliptic dispersion @xmath92 , see eq.(25)to ( 27 ) , relative triangularity dispersion @xmath93 ; relative shafranov dispersion @xmath94 ; relative @xmath36-parameter at the 95 % surface @xmath95 ; the outward radius @xmath96 , @xmath97 = 3.38 \ m $ ] ; the inward radius @xmath98 = 1.45 \ m$ ] ; center plasma radius @xmath99 and radius at the maximum z , @xmath100 , @xmath101 .
+ in figure 2 , the function @xmath77 is show as a function of @xmath102 .
this function allows to determine poloidal field around a magnetic surface , once the value @xmath8 at the outward point @xmath11 is measured or determined . in order to find @xmath103 ,
it is necessary to determine the curvature of the family of orthogonal lines .
the procedure has been explained elsewhere[4 ] .
the function @xmath103 appears in most of the integrals needed to determine the average normal velocity @xmath104 first calculated by pfirsch - schlter for cross - sections of circular magnetic surfaces[9 ] .
the function to be used in this work is the central function , plain line .
however , two other @xmath105-functions are also shown to illustrate that function .
the upper curve correspond to the case where the triangularity and shafranov shift are zero .
the lower curve is just the case of zero triangularity , but the same shafranov shift than in the main curve ( central one ) , where the triangularity is @xmath106 , as in figure 1 .
+ since the function @xmath77 shown essentially the behavior of @xmath107 , the upper curve illustrate that this product is constant at the inward and outward points , when there is not triangularity and shafranov shift
. however , the value of @xmath108 , decreases at the uppest and lowest points because of the ellipticity elongation @xmath86 .
the shafranov shift modifies this pattern in the way that the values at the inward area are almost constant , and very low compared with those at the outward point @xmath11 .
introducing triangularity as in the case of the central line does not modify the pattern , but the values at the flat part of the curve are not so low . + in figure 3 ,
the dimensionless average normal velocity is shown as a function of the dimensionless toroidal electric field @xmath4 at point @xmath11 , the line here found is the straight line , because of the normalization here used using the value of @xmath109 at the point @xmath11 .
the intersection of that line with the abscissa define the critical dimensionless toroidal field @xmath4 for marginal velocity confinement .
this critical electric field is show as a function of the elliptic elongation @xmath64 , where all the other variables are kept fixed at the values shown in figure 1 , that is , @xmath110 and @xmath111 . to determine the integrals in order to draw figure 3 and 4 ,
the value of @xmath71 is chosen as 0.3 , which correspond for instance to toroidal magnetic field @xmath112 and poloidal magnetic field @xmath113 .
the ratio of perpendicular to parallel resistivity , @xmath72 has been considered as 1.97 as in page 669 of ref.(8 ) . as a way of completion @xmath114
is taken as @xmath115 .
+ the figure 3 shown that the outward velocity due to diffusion is opposite by the inward effect due to the inductive electric fields @xmath0 . after a critical value @xmath116 , the characteristic ware pinch effect[10 ] becomes more important , and an average inward velocity @xmath117 is produced , in such a way that the plasma appears confined as long as instabilities are not considered .
the critical dimensionless electric field increase with the elliptic elongation @xmath64 if all the parameters are kept fixed , however , the variation is not so large , from 0.8 a factor to about 1.1 as it is shown in figure 4 . in this figure
has been normalized with a second procedure .
first a normalized critical toroidal electric field is selected as reference and denoted by @xmath118 , which in this case corresponds to that given in fig .
3 , where the characteristic values of the parameters above given . the value of this critical dimensionless toroidal electric field is @xmath119 , and correspond also to the horizontal line through one , which it is also show by a dot line .
this kind of normalization is also performed in figure 5 and 6 .
the toroidal electric fields with the second normalization explain above are denoted by @xmath120 .
+ the changes due to shafranov shift with all the other parameters fixed , are more significative as illustrates the figure 5 , where the critical electric field could be one forth or 3 times the value of that shown in figure 3 .
furthermore the curve in figure 4 is almost linear , but not that in figure 5 , which seems somewhat as a parabola . finally , in figure 6
, the change of @xmath121 with triangularity are shown .
there also @xmath120 changes strongly with @xmath73 .
more important than this it is that the changes are very significative for low values of @xmath73 , and @xmath120 could be lower by a factor 5 , and with values of triangularity no so large , as @xmath122 . here
also the curve is strongly not linear , however the concavity of the curve is opposite to that in figure 6 . +
in most tokamak operation the inductive magnetic field effect exceeds the plasma diffusion effect and the ware pinch effect contract the toroidal plasma column . here the critical point where both effect becomes almost equals is studied .
this equilibrium situation is consider as that where the average normal velocity becomes null or void , which will be defined as the marginal velocity confinement .
a suitable normalization procedure allows to extend our analysis to a large amount of different situations in tokamak plasma configurations .
the critical toroidal electric field changes very little with the ellipticity of the plasma . however , the changes are very strong with the shafranov shift and triangularity .
the curves in those last cases seems parabolas , but with opposite curvature .
very large changes for small values of triangularity have been found , producing changes with a factor 5 for modest triangularity values as @xmath106 . | h - mode anomalous transport in tokamaks low plasma turbulence internal barriers[1 - 3 neoclassical transport calculations important diffusion in collisional regime depends pressure gradient at 95% surface scrape layer inductive electric field @xmath0 diffusion due to gradient pressure opposed to pinch effect due to @xmath0 previous papers neoclassical diffusion treated arbitrary plasma configuration using new tokamaks coordinates[4 - 6 suitable direct numerical calculations for values elongation shafranov shift triangularity not presented now previous results incomplete not using right parameters for tokamaks here calculations presented different look for values of @xmath0 in marginal velocity confinement average velocity on 95 % surface zero results marginal confinement flux transition to ingoing flux difficult to calculate simple treatment plasma configuration not until now calculations suitable normalization procedure absent previous adequate selection of tokamak parameters normalization allows results useful for different tokamak plasma configurations different values of ellipticity shafranov shift triangularitycollisonal transport treatment toroidal axisymmetric plasmas using dimensionless integrals variables @xmath1 -_{1 }{i}_{0_{7 -{1{6}}_{4 }\gamma{2}_{1 }\hat{i}_{5 } @xmath2 dimensionless normal velocity velocity @xmath3 magnetic surface @xmath4 dimensionless electric field @xmath5 @xmath6 major radius toroidal poloidal magnetic field inductive electric field pressure gradient at mid - plane external point @xmath11 magnetic cross section plasma resistivity @xmath12 @xmath13 directions perpendicular parallel magnetic field lines @xmath14 minor axis radius dimensionless quantity @xmath15 ratio between poloidal toroidal magnetic field at point @xmath11 new dimensionless integrals @xmath17 @xmath19 defined as + @xmath20 @xmath21 integrals around magnetic surface @xmath28 function curvature @xmath29 orthogonal line family @xmath30 results obtained using mhd equations assuming toroidal axisymmetrynumerical analytic results express magnetic cross sections by equations @xmath31 @xmath32 @xmath33 are ellipticity triangularity shafranov shift distortions parameter @xmath36 labels each magnetic surface previous equations generalization of roach et al[7 quantities @xmath33 @xmath35 dimensionless for analysis calculations useful new dimensionless quantities @xmath37 @xmath41 defined as @xmath42 shafranov shift @xmath43 connected to previous parameters by @xmath44 @xmath45 is size of 95 % magnetic surface measurement at mid - plane @xmath46 is horizontal half - width of plasma equations of cross section magnetic lines @xmath47 {aligned} @xmath48 denoting @xmath49 value of parameter @xmath36 generating 95 % magnetic surface with general coordinates @xmath50 @xmath51 @xmath52 @xmath53 largest smallest values of @xmath50 are @xmath54 @xmath55 values @xmath56 @xmath57 {aligned} radius @xmath58 center plasma size @xmath46 @xmath59 @xmath60 connect parameters with aspect ratio @xmath61 @xmath62 @xmath63 maximum value @xmath51 elliptic elongation @xmath64 triangularity connected previous parameters @xmath65 @xmath66 @xmath67 @xmath68 @xmath69 @xmath70 radius point maximum z dimensionless variables computation quantities ratio between poloidal toroidal magnetic fields @xmath71 horizontal half - width plasma @xmath46 aspect ratio @xmath61 ratio @xmath72 shafranov shift ellipticity @xmath64 triangularity @xmath73 from @xmath64 @xmath73 values @xmath33 @xmath38 determined shafranov shift value @xmath46 calculated value @xmath38 allow calculation @xmath74 @xmath75 family magnetic surfaces determined family orthogonal lines determined curvature function @xmath76 function @xmath77 integrals normal dimensionless velocity @xmath2 performed value @xmath71 given velocity @xmath2 obtained if ratio @xmath78 given linear function of dimensionless toroidal electric field @xmath79intersection value with axis abscissa critical dimensionless electric field @xmath80 for marginal velocity confinement field values ellipticity @xmath64 dimensionless shafranov shift triangularity @xmath73 ellipticity triangularity in figure 1 as @xmath82 and @xmath83 correspond to values @xmath33 @xmath38 procedure different select values of @xmath33 @xmath64 @xmath73 values in jet tokamak procedure simpler in figure 1 cross sections magnetic lines drawn for values of @xmath36 zero to @xmath49 characteristic values are elipticity @xmath86 relative shafranov shift triangularity horizontal half width aspect ratio minor magnetic axis ratio @xmath91 previous data parameters determined : elliptic dispersion @xmath92 relative triangularity dispersion @xmath93 shafranov dispersion @xmath94 @xmath36-parameter at 95 % surface @xmath95 outward radius @xmath96 = 3.38 inward radius @xmath98 = 1.45 center plasma radius @xmath99 radius at maximum zfigure 2 function @xmath77 function of @xmath102 allows poloidal field around magnetic surface value @xmath8 at outward point @xmath11 measured find @xmath103 determine curvature of orthogonal lines procedure explained elsewhere function @xmath103 appears in integrals average normal velocity @xmath104 calculated by pfirsch - schlter for cross - sections circular magnetic surfaces function used central function plain line two other @xmath105-functions illustrate function upper curve triangularity shafranov shift zero lower curve zero triangularity same shafranov shift main curve central triangularity @xmath106 figure 1 function @xmath77 behavior of @xmath107 upper curve product constant at inward outward points triangularity shafranov shift value of @xmath108 decreases at uppest lowest points ellipticity elongation @xmath86 shafranov shift modifies pattern values inward area almost constant low outward point @xmath11 introducing triangularity central line modify pattern values at flat part curve not lowfigure 3 dimensionless average normal velocity function toroidal electric field @xmath4 at point @xmath11 line straight line normalization @xmath109 intersection with abscissa critical dimensionless toroidal field @xmath4 for marginal velocity confinement electric field function of elliptic elongation @xmath64 variables fixed at values figure 1 @xmath110 and @xmath111 integrals figure 3 4 value @xmath71 0.3 to toroidal magnetic field @xmath112 and poloidal magnetic field @xmath113 ratio perpendicular to parallel resistivity @xmath72 1.97 page 669 @xmath114 as @xmath115 figure 3 outward velocity due to diffusion opposite by inward effect inductive electric fields @xmath0 after critical value @xmath116 characteristic ware pinch effect important average inward velocity @xmath117 produced plasma appears confined instabilities critical dimensionless electric field increase with elliptic elongation @xmath64 if parameters fixed variation not large from 0.8 to 1.1 figure 4 figure normalized second procedure normalized critical toroidal electric field denoted by @xmath118 corresponds to fig . 3 characteristic values parametersvalue critical dimensionless toroidal electric field is @xmath119 horizontal line dot line normalization in figure 5 and 6 toroidal electric fields second normalization denoted by @xmath120 changes due to shafranov shift parameters fixed significative figure 5 critical electric field could be 3 times value figure 3 curve in figure 4 almost linear not figure 5 parabola figure 6 change of @xmath121 with triangularity @xmath120 changes with @xmath73 changes significative for low values of @xmath73 @xmath120 lower by factor 5 values triangularity large @xmath122 curve not linear concavity opposite to figure 6 inductive magnetic field effect exceeds plasma diffusion effect ware pinch effect contract toroidal plasma column critical point where almost equals studied equilibrium situation average normal velocity becomes null or void marginal velocity confinement normalization procedure analysis to different situations plasma configurations critical toroidal electric field changes little with ellipticity plasma changes strong with shafranov shift triangularity curves parabolas opposite curvature large changes for small values triangularity found changes factor 5 for modest triangularity values as @xmath106 | 0.533001 | 3,385 | 1,786 | , Saving $0.1 in GPT-4. |
177,161 | we propose a method to improve quantum state transfer in transmission lines . the idea is to localize the information on the last qubit of a transmission line , by dynamically varying the coupling constants between the first and the last pair of qubits .
the fidelity of state transfer is higher then in a chain with fixed coupling constants .
the effect is stable against small fluctuations in the system parameters . | efficient short - distance quantum state transfer is an important problem in the field of quantum computing .
one of the most promising solutions is to use chains constructed from qubits that are statically coupled to each other .
the idea to use quantum spin chains was initially put forward by bose @xcite and then developed in a number of papers .
these proposals exploit the unitary time evolution governed by the system hamiltonian .
the state is initialized / encoded at the sender part of the chain and then , after a certain time , measured / decoded at the receiving part of the chain . the major advantage of this method is its simplicity : it does not require controllable coupling constants between the qubits or complicated gating schemes .
it was shown @xcite that for short - length chains the fidelity of state transfer is high , i.e. , close to one .
but the fact that it is substantially reduced with the length of the chain triggered the search of methods that allow to increase the fidelity or even to obtain perfect state transfer , in the absence of decoherence and relaxation processes .
the main reason for imperfect transfer is the dispersion of the initial information over the whole chain .
therefore it was proposed to use spatially varying coupling constants to `` refocus '' the information at the receiving part of the chain @xcite .
another possibility is to encode the information in gaussian wave packets ( with low dispersion ) spread over several spins @xcite .
chains where the first and the last qubits are only weakly coupled to the rest of the chain provide a very high fidelity @xcite , because the intermediate spins are only slightly excited , which means that dispersion is small .
this method has the major disadvantage that the time required for the transfer is long compared to the qubit decoherence / relaxation times in present experimental setups .
the idea of so - called conclusive transfer , providing perfect state transfer using parallel quantum channels @xcite , is very promising .
it can be realized using almost any spin chain and it is stable against fluctuations of the chain parameters @xcite .
almost all the proposals mentioned above have one common disadvantage : the time interval for which the fidelity is high is very small for physical qubits and realistic qubit coupling parameters .
for example , for a chain of flux qubits @xcite with realistic experimental parameters @xcite , the half - width of the first fidelity maximum is about @xmath0ns . at these time scales state readout and manipulation is impossible using current experimental technology . here
we show that by dynamically varying the coupling constants only between the first and the last pair of qubits we can solve this problem and also increase the fidelity of state transfer . in real chains the state to be transmitted
is initialized in the first qubit , and this process must not influence the fidelity and dynamics of the chain .
the most natural idea for a full transferring protocol is as follows : initialize the state in the first qubit , that is decoupled from the rest of the chain , then adiabatically couple it , wait a certain time and then adiabatically decouple the last qubit from the chain .
this method requires two controllable gates like one of the proposals for achieving perfect state transfer @xcite . in this paper ,
the main purpose of the gates is to localize the state on the last qubit where it can be manipulated during times that are comparable to the decoherence / relaxation times . in the following
we use the terms spin and qubit as equivalent .
state @xmath1 in qubit language ( which we will also call `` excitation '' ) corresponds to spin - up in spin language , and state @xmath2 corresponds to spin - down .
we consider the xxz - hamiltonian with time dependent coupling constants between the first and the last pair of qubits : @xmath3 this type of hamiltonian or some of its special cases is used in most of the papers mentioned above .
the xx - part of the hamiltonian describes the tunneling of the excitation from one site to another and is a necessary requirement for quantum state transfer . and
@xmath4 as functions of time and coupling parameters.,scaledwidth=50.0% ] ) for a chain with constant coupling parameters ( dashed line ) and time dependent coupling parameters ( solid line ) , @xmath5 , @xmath6 , @xmath7 , @xmath8.,scaledwidth=50.0% ] the physical systems described by this type of hamiltonian include josephson arrays of charge@xcite and persistent - current@xcite ( flux ) qubits , connected by josephson - junctions / capacitors .
the time - dependent coupling constants can be realized by varying the gate voltages on the @xmath9st/@xmath10nd and @xmath11th/@xmath12th qubits for the flux qubit chain , or by replacing the josephson junction between the charge qubits with a squid and varying the flux through it . as a model
we use `` fermi - function like '' coupling constants : @xmath13 with @xmath14 these are smooth functions that vary from @xmath15 ( no coupling ) to @xmath16 ( full coupling ) and vice versa , see fig .
[ fig0a ] .
the time scale of the coupling / decoupling procedure is determined by @xmath17 .
instant coupling / decoupling corresponds to @xmath18 .
our goal is to calculate the fidelity of the state transfer , the quantity that characterizes the quality of the transmission line .
let us assume that the chain is initialized in the state @xmath19 .
then , the first qubit is prepared in the state @xmath20 , i.e. the total state of the array is @xmath21 .
this is not an eigenstate of the hamiltonian ( [ xxz ] ) , therefore the system will evolve in time . after a time @xmath22 the state of the last qubit
is read out .
following bose@xcite , we average the fidelity over all pure input states on the bloch sphere @xmath23 to obtain a quantity @xmath24 that measures the quality of transmission independent of @xmath20 . here
@xmath25 is the reduced density matrix of the last qubit .
fidelity one corresponds to the perfect state transfer . by numerically solving the schrdinger equation for the time - dependent hamiltonian ( [ xxz ] ) we get the fidelity of the state transfer as a function of time and the coupling parameters @xmath17 , @xmath26 and @xmath27 .
the fidelity has a complex oscillating behavior .
our goal is to find the coupling parameters that allow us to localize the state at the last qubit by decoupling it from the rest of the chain such that the fidelity is maximal . in comparing this fidelity with the static case ,
we concentrate on the first maximum : higher maxima appear only after times much longer than the time at which the first one occurs@xcite .
the typical behavior of @xmath28 for the static chain in the vicinity of the first maximum is shown in fig .
[ fig0 ] ( dashed line ) . and
@xmath27 , @xmath5 , @xmath6.,scaledwidth=50.0% ] figure 2 also shows the fidelity in the presence of time - dependent coupling constants ( solid line ) .
one can see that at large time the state is localized at the last qubit with a fidelity @xmath29 that is higher than for static coupling constants .
the time at which the maximum is achieved is slightly larger .
this is natural since in the presence of the coupling / decoupling procedure the transmission of the information from the first qubit to the chain and then to the last qubit is slower . after decoupling ,
the localized state can be manipulated during a time interval comparable with the decoherence and relaxation times for the qubit , which are several orders of magnitude longer then the half - width of the first fidelity maximum in the static case in present experimental setups .
we would like to mention that the first fidelity maximum in the case of dynamical coupling constants is even higher than the stationary value of the fidelity after decoupling .
numerical calculations show that it can exceed the value 0.99 ( but , in this case , after the full decoupling the fidelity will go down to about 0.9 ) .
figure [ fig1 ] shows the fidelity of the state transfer after completely decoupling the last qubit from the rest of the chain for @xmath30 as a function of the parameters @xmath17 and @xmath27 ( for @xmath6 ) . there is a region where the fidelity for the localized state is higher than in the time - independent case ( up to @xmath31 ) . , @xmath5 , @xmath6 , @xmath32 , @xmath7 .
@xmath33 is the first fidelity maximum for the ideal chain with static coupling constants .
inset : distribution of the fidelity difference between the dynamical and statical cases in the presence of equal disorder.,scaledwidth=45.0% ] the origin of this phenomenon is similar to the effect described in ref . . by dynamically varying the coupling constant between the first and the second qubit , the information about the state enters the chain as a wave packet that has small dispersion .
this corresponds to some sort of filtering , an interpretation in agreement with the fact that the fidelity is higher in the case of equal `` profiles '' for the coupling and decoupling functions .
if we use dynamical decoupling only at the end of the chain and employ instant coupling to initialize the chain , the maximal possible fidelity for a chain of @xmath5 qubits drops from about @xmath34 to @xmath35 ( but it is still higher than the fidelity for the time - independent case , which is around @xmath36 ) .
an intuitive explanation is as follows : during the dynamical decoupling , the information , that is still dispersed in the chain , will arrive at the last qubit .
therefore , slow decoupling allows more information to be gathered before the full decoupling occurs .
experimental qubit arrays are always inhomogeneous , so in the rest of the paper we will discuss the effect of static disorder in @xmath16 and dynamical fluctuations in the coupling / decoupling functions . for charge qubit arrays ,
the most important source of inhomogeneity is the variance of the josephson energies of the junctions ( about @xmath37 ) . in the case of the flux - qubit chain with capacitive coupling
, @xmath16 is a complicated function of the josephson and charging energies as well as the capacitance of the coupling capacitor , see ref . .
a rough estimate using realistic parameters leads to a variance of @xmath38 .
we have performed numerical simulations to evaluate the time evolution of the system . as a result
we find that the phenomena described above , are stable to static disorder and dynamical fluctuations in the coupling functions , see figs .
[ fig2],[fig4 ] .
figure [ fig2 ] shows the distribution of the fidelity after complete decoupling in the presence of disorder in the coupling constants .
its half - width is quite small : even in the worst case the fidelity is higher than the fidelity of the ideal chain without disorder .
the graph was constructed using a numerical simulation for an ensemble of @xmath39 chains where the coupling constants were of the form @xmath40 , @xmath41 .
the quantity @xmath42 was a random number with uniform distribution in the interval @xmath43 $ ] .
the inset of fig .
[ fig2 ] shows the difference between the fidelities for different realizations of the chains with constant and time - dependent couplings .
this difference is around @xmath44 , so the effect of increased fidelity persists . in each realization
both chains have the same randomized coupling constants and the only difference is that @xmath45 and @xmath4 are not multiplied by coupling functions for the time - independent chain . , @xmath32 , @xmath7 .
@xmath46 is the fidelity after decoupling in the absence of fluctuations.,scaledwidth=45.0% ] figure [ fig4 ] shows the influence of fluctuations in the coupling / decoupling functions . here
the coupling constants @xmath16 are the same for all realizations and the coupling / decoupling functions are of the form @xmath47 the quantities @xmath48 are stepwise stochastic processes of step width @xmath49 , the step heights are uniformly distributed in the interval @xmath50 $ ] .
one can see that the influence of these fluctuations is small .
we would like to mention , that the fidelity in the presence of dynamical fluctuations in the coupling functions is always decreased .
this is in agreement with the filtering idea described above . ,
@xmath51.,scaledwidth=45.0% ] finally , to check that all the effects described above are not the consequence of our special choice of coupling functions ( [ eq1 ] ) , we also did the calculation for another type of dynamical coupling / decoupling : @xmath52 \\
j_{xy } & t>\tau \end{array } \right.\ ] ] @xmath53 \\ 0 & t >
t_f+\tau \end{array } \right.\ ] ] these functions vary from @xmath15 to @xmath16 ( and vice versa ) , and we have chosen @xmath6 . the parameters @xmath54 and @xmath17 describe the shape and timescale of the coupling / decoupling function .
the first maxima of the fidelity for different @xmath55 $ ] are shown in fig . [ fig5 ] . here , as in fig .
[ fig4 ] , @xmath17 and @xmath27 are chosen to maximize it .
one can see that this type of dynamical coupling also allows us to have better state transfer than for the chain with constant couplings ( where the height of the first maximum is @xmath33 ) .
in general , wave packets with bigger width have lower dispersion .
therefore we expect that every smooth monotonic coupling / decoupling function with equal profiles will allow us to improve the fidelity of state transfer .
in the past , a number of quantum transmission line systems was proposed to achieve a perfect or almost perfect state transfer .
a common disadvantage of most of these proposals is the very short time interval , for which the fidelity of the state transfer is high .
manipulating the state in such short time intervals is impossible using current experimental technology . in this paper
we have proposed a method that allows to localize the transferred state on the last qubit of the transmission line , by varying the coupling constants between the first and the last pair of qubits .
we have also shown that this method increases the fidelity of the state transfer and that this effect is stable to static disorder in the coupling constants and dynamical fluctuations in the coupling / decoupling functions .
we acknowledge fruitful discussions with s. bose , d. burgarth , r. fazio , and f.g .
this work was supported by the european union under contract ist-3 - 015708-ip eurosqip , by the swiss nsf , and the nccr nanoscience . | short - distance quantum state transfer important problem quantum computing promising chains from qubits statically coupled idea quantum spin chains by bose @xcite developed in papers proposals exploit unitary time evolution system hamiltonian state initialized / encoded at sender part chain measured / decoded at receiving part advantage method simplicity controllable coupling constants or complicated gating schemes for short - length chains fidelity of state transfer high close to reduced with length chain triggered search of methods to increase fidelity or obtain perfect state transfer decoherence reason for imperfect transfer dispersion of initial information over chain proposed to use spatially varying coupling constants to refocus information at receiving part possibility encode information in gaussian wave packets low dispersion spread over several spins chains where first and last qubits weakly coupled provide high fidelity intermediate spins slightly excited dispersion small method disadvantage time for transfer long compared to qubit decoherence / relaxation times idea of conclusive transfer perfect state transfer using parallel quantum channels promising can be realized using any spin chain stable against fluctuations chain parametersproposals have disadvantage time interval fidelity high small for physical qubits realistic qubit coupling parameters for chain of flux qubits @xcite with experimental parameters half - width first fidelity maximum is about @xmath0ns time scales state readout manipulation impossible using current experimental technology dynamically varying coupling constants between first and last qubits problem increase fidelity state transfer in real chains state initialized in first qubit process influence fidelity dynamics chain natural idea for full transferring protocol initialize state in first qubit decoupled adiabatically couple wait time decouple last qubit method requires two controllable gates for perfect state transfer purpose gates to localize state on last qubit during times comparable to decoherence / relaxation times terms spin and qubit as equivalent state @xmath1 corresponds to spin - up @xmath2 to spin - down consider xxz - hamiltonian with time dependent coupling constants between first last qubits : @xmath3 hamiltonian used in papers xx - part describes tunneling excitation necessary for quantum state transfer @xmath4 functions of time and coupling parameters,scaledwidth=50.0% chain with constant coupling line time dependent solid line @xmath5=50.0% physical systems hamiltonian include josephson arrays charge@xcite persistent - current@xcite flux qubits connected by josephson - junctions capacitors time - dependent coupling constants varying gate voltages @xmath9st qubits flux replacing josephson junction between charge qubits with squid varying flux model use fermi - function coupling constants @xmath13 @xmath14 vary from @xmath15 no to @xmath16 full coupling fig time scale coupling / decoupling determined by @xmath17 instant coupling decoupling to @xmath18 goal calculate fidelity of state transfer quality transmission line chain initialized in state @xmath19 first qubit prepared in state @xmath20 total state array is @xmath21 not eigenstate hamiltonian system in time after time @xmath22 state last qubit read out average fidelity over pure input states bloch sphere @xmath23 obtain quantity @xmath24 quality transmission independent of @xmath20@xmath25 reduced density matrix last qubit fidelity one corresponds perfect state transfer solving schrdinger equation for time - dependent hamiltonian fidelity state transfer function of time coupling parameters @xmath17 @xmath27 fidelity complex oscillating behavior goal to find coupling parameters localize state at last qubit by decoupling from chain fidelity maximal comparing fidelity with static case concentrate on first maximum higher maxima after times longer than time first typical behavior of @xmath28 static chain first maximum shown in fig . @xmath27 @xmath5 @xmath6.=50.0% figure 2 shows fidelity time - dependent coupling constants at large time state localized at last qubit fidelity @xmath29 higher than static coupling constants time maximum achieved slightly larger natural coupling / decoupling procedure transmission information first qubit last slower after decoupling localized state manipulated during time interval comparable with decoherence relaxation times qubit longer then first fidelity maximum in static case first fidelity maximum dynamical coupling constants higher than stationary value fidelity after decoupling can exceed value 0.after full decoupling fidelity to 0.9 figure shows fidelity state transfer after decoupling last qubit from chain for @xmath30 function of parameters @xmath17 and @xmath27 @xmath6 region where fidelity for localized state higher than time - independent case up to @xmath31 ) @xmath5 first fidelity maximum for ideal chain with coupling constants distribution of fidelity difference between dynamical and statical cases in equal disorder=45.0% origin phenomenon similar to effect dynamically varying coupling constant between first second qubit information state enters chain as wave packet small dispersion corresponds to filtering fidelity higher in equal profiles for coupling and decoupling functions dynamical decoupling at end chain instant coupling to maximal fidelity for chain @xmath5 qubits drops from @xmath34 to @xmath35 still higher than fidelity time - independent case around @xmath36 during dynamical decoupling information dispersed at last qubit slow decoupling allows more information before full decouplingqubit arrays inhomogeneous discuss effect static disorder in @xmath16 dynamical fluctuations in coupling / decoupling functions charge qubit arrays important source inhomogeneity variance josephson energies junctions @xmath37 ) flux - qubit chain capacitive coupling @xmath16 complicated function of josephson charging energies capacitance coupling capacitor rough estimate leads to variance of @xmath38 performed numerical simulations time evolution system phenomena stable to static disorder dynamical fluctuations in coupling functions fig2 shows distribution fidelity after decoupling disorder coupling constants half - width small worst case fidelity higher than ideal chain without disorder graph constructed numerical simulation for @xmath39 chains coupling constants @xmath40 , @xmath41 @xmath42 random number uniform distribution interval @xmath43 difference between fidelities for different realizations constant time - dependent couplings difference around @xmath44 effect increased fidelity persists chains have same randomized coupling constants @xmath45 @xmath4 not multiplied by coupling functions for time - independent chain@xmath46 fidelity after decoupling absence fluctuations,scaledwidth=45.0% figure fig4 shows influence fluctuations in coupling / decoupling functions coupling constants @xmath16 same for all realizations coupling decoupling functions form @xmath47 quantities @xmath48 stepwise stochastic processes step width @xmath49 step heights uniformly distributed in interval @xmath50 $ influence fluctuations small fidelity dynamical fluctuations coupling functions always decreased agreement with filtering idea @xmath51.,scaledwidth=45.0% effects not consequence special choice coupling functions calculation for another type dynamical coupling / decoupling @xmath52 \\ @xmath53 functions vary from @xmath15 to @xmath16 chosen @xmath6 parameters @xmath54 @xmath17 describe shape timescale of coupling / decoupling function first maxima fidelity for different @xmath55 $ shown in fig . fig5 ] @xmath17 and @xmath27 chosen to maximize dynamical coupling allows better state transfer than chain constant couplings height first maximum @xmath33wave packets bigger width have lower dispersion expect smooth monotonic coupling / decoupling function equal profiles improve fidelity state transfer past quantum transmission line systems proposed achieve perfect state transfer disadvantage short time interval fidelity state transfer high manipulating state short impossible using current technology paper proposed method localize transferred state on last qubit varying coupling constants between first last qubits shown method increases fidelity state transfer effect stable to static disorder coupling constants dynamical fluctuations acknowledge discussions with s. bose d. burgarth r. fazio f.g work supported by european union - 015708 eurosqip swiss nsf nccr nanoscience | 0.552273 | 3,258 | 1,601 | , Saving $0.1 in GPT-4. |
177,168 | we study transport properties of two entangled photons which are initially injected into two nearest - neighbor coupling cavities in a one - dimensional coupled - cavity array ( cca ) .
it is found that photonic transport dynamics in the two - photon cca exhibits entanglement - enhanced two - photon delocalization ( tpd ) phenomenon .
it is shown that the cca can realize the localization - to - delocalization transition for two entangled photons . | in recent years , much attention has been paid to coupled - cavity arrays ( ccas ) @xcite due to their extremely rich physics as well as their wide potential applications to work as an effective platform to realize quantum information processing and to carry out photonics tasks . for instance , ccas stand out as an attractive controllable test bed for strongly correlated many - body models and quantum simulators of many - body physics @xcite .
transport properties of photons and various excitations in ccas embedded with natural or artificial atoms are also receiving considerable interest @xcite .
in particular , it has been demonstrated that ccas with a cyclic three - level system or a single atom open the possibility to quantum routing of single photons @xcite .
ccas typically consist of an arrangement of low - loss cavities with nearest - neighbor coupling allowing photon hopping between neighboring cavities .
an appealing feature of ccas is the available high control of each single cavity in terms of performable measurements , quantum - state engineering , and dynamical parameter tuning .
this feature can be used to realize local addressing in quantum information processing .
progress in the fabrication techniques make ccas experimentally accessible @xcite .
although quantum entanglement is an essential resource in quantum information processing , propagation of spatially entangled photons in ccas without natural or artificial atoms has not been studied extensively heretofore .
a natural question is how quantum entanglement of initial input photons affects photonic transport properties in ccas . in this paper , we study transport properties of two initially entangled photons which are injected into two nearest - neighbor coupling cavities in a one - dimensional cca . we show that photonic transport dynamics in the two - photon cca exhibits the entanglement - enhanced two - photon delocalization ( tpd ) phenomenon .
we investigate tpd dynamics by introducing the concept of the tpd degree when two nearest - neighbor coupling cavities are initially in two - photon entangled state @xmath0 .
we demonstrate that the cca can realize the localization - to - delocalization transition for two entangled photons .
the rest of this paper is organized as follows . in sec
ii , we present our physical model and its solution . in sec iii ,
we analyze photonic transport properties and tpd in the two - photon cca , and show the entanglement - enhanced tpd and the localization - to - delocalization transition for two entangled photons .
finally , we shall conclude our paper with discussions and remarks in the last section .
let us consider a finite - length cca in which nearest - neighbor cavities are coupled together so as to allow for photon hopping .
we assume that all the cavities are identical .
each cavity sustains a single field mode of frequency @xmath1 .
the hamiltonian of the full system ( here and throughout we take @xmath2 ) is written as @xmath3 where @xmath4 is a photonic creation ( annihilation ) operator at @xmath5-th cavity .
the second term in above equation represents the hopping interaction with the intercavity coupling strength @xmath6 .
the hamiltonian of the cca given by eq .
( 1 ) can be diagonalized in terms of the normal modes defined by @xmath7 with the following inverse transformation @xmath8 where the transformation matrix is given by @xmath9 making use of eqs .
( 2)-(4 ) , one can obtain the diagonalized form of hamiltonian ( 1 ) where the transformation matrix is given by @xmath10 from heisenberg equations on eq .
( 5 ) and making use of eq . ( 3 ) , one can obtain exact solution of the cca in the heisenberg representation with the following form @xmath11 where the explicit form of the green s function is given by @xmath12 in the following section , we shall use above exact solution to investigate quantum dynamics of photonic transport and tpd in the two - photon cca .
the quantum mechanical properties of light are observed when correlations between the propagating photons are considered . in this section , we study photonic transport properties and quantum dynamics of two - photon delocalization in the two - photon cca when two entangled photons are injected into the cca by investigating the evolution of the photon - number correlation function @xmath13 which represents the probability of detecting a two - photon coincidence across cavity @xmath14 and cavity @xmath15 .
the probability to detect both photons at the same cavity @xmath15 is given by @xmath16 .
we consider such a situation in which the two photons are initially in @xmath17-th cavity and @xmath18-th cavity with the following two - photon noon - type states @xmath19 where for the simplicity , we take @xmath20 throughout the paper .
it is interesting to note that the two photons in the two - photon entangled states ( 9 ) are in the same cavity , i.e. , cavity @xmath17 or cavity @xmath18 .
the other cavities in the cca are in vacuum state . hence , the two - photon entangled states given by eq .
( 9 ) are two - photon localization states .
the amount of entanglement of the two - photon states in eq.(9 ) can be measured by concurrence @xcite with the following expression @xmath21 obviously , the concurrence vanishes when @xmath22 , and @xmath23 while the concurrence reaches the maximum value @xmath24 when @xmath25 .
substituting eqs .
( 6 ) and ( 9 ) into eq .
( 8) , we can obtain the probability of detecting a two - photon coincidence across cavity @xmath14 and cavity @xmath15 with the following expression @xmath26 as an example , we consider the photonic propagation in the cca for the two - photon entangled state given by eq .
( 9 ) which is injected initially into two neighboring cavities , say @xmath27 and @xmath28 .
assume that the cca under our consideration consists of 29 cavities ( @xmath29 ) .
quantum evolution of the calculated correlation function for this case is presented in fig . 1 for different amount of initial entanglement .
fig.1(a ) , fig.1(c ) , and fig.1(e ) are the initial - state correlation function when the amount of initial entanglement @xmath30 and @xmath24 , respectively .
fig.1(b ) , fig.1(d ) , and fig.1(f ) are the correlation function at time @xmath31 when the amount of initial entanglement @xmath30 and @xmath24 , respectively . from fig .
1 we can see that the correlation represented by the diagonal elements of the two - photon correlation function @xmath32 completely vanishes while the correlation represented by the off - diagonal elements of the two - photon correlation function @xmath32 is enhanced with the increase of the initial amount of entanglement from vanishing entanglement ( @xmath33 ) to the maximal entanglement ( @xmath34 ) . accordingly , the two photons will always tend to separate and to emerge in different cavities . of simultaneously detecting photons in cavity @xmath14 and @xmath15 after a temporal evolution @xmath31 .
the evolution time is scaled by @xmath35 , and other parameters are @xmath36 , @xmath29.,width=316,height=377 ] in order to well understand transport dynamics of the two photons in the cca , we introduce the concept of the two - photon delocalization ( tpd ) . when the two photons are in two different cavities in the two - photon cca we call it as the tpd while when the two photons are in the same cavity we call it as the two - photon localization ( tpl ) .
the degree of the tpd can be defined in terms of the sum of probability @xmath37 of detecting photons in every cavity @xmath38 which represents the probability of the two photons being not at the same cavity in the cca simultaneously .
especially , when @xmath39 , the two photons are in the complete tpd state in which the two photons are completely in different cavities . in this case , it is impossible that the two photons are in the same cavity . hence , the two photons in the tpd state tend to stay at different cavities . in this sense
, the tpd degree can reflect repulsive - statistics or anti - bunching property of two photons in the cca . obviously , the tpd degree of the two - photon noon - type state defined by eq .
( 9 ) is zero .
hence , the state given by eq .
( 9 ) is not a tpd state , it is a tpl state .
the larger the value of the parameter @xmath40 is , the stronger the tpd phenomenon becomes .
we consider the two - photon cca formed by 29 cavities .
when the two photons are injected into @xmath41-th and @xmath42-th cavities with the input state ( 9 ) , the photon - number second - order correlation function changes considerably in the time evolution as shown in fig .
1 . the most obvious feature is the appearance of the tpd in which the diagonal elements of the two - photon probability , i.e. , @xmath37 , are suppressed considerably in the time evolution . in particular , for the case of the maximally entangled state with the concurrence @xmath34 we can observe almost completetpd with @xmath43 in the process of the time evolution as shown in fig .
1(e ) and fig .
1(f ) . in the weak hopping regime for three different values of quantum entanglement of the two - photon state given by eq .
the evolution time is scaled by @xmath35 .
the photon hopping parameter takes @xmath44 .
the number of the cavities in the cca is @xmath29 .
the dashed curve represents the result of the maximally entangled state with @xmath34 , the solid line corresponds to the initial input state with the concurrence @xmath34 , and the dot line shows the result of the initial unentangled state with @xmath33 .
, width=321,height=226 ] we now investigate quantum dynamics of the tpd and the influence of quantum entanglement of the initial state in the weak hopping regime of @xmath45 . in fig .
2 , we have plotted the time evolution of the tpd degree for three different values of quantum entanglement of the initial state ( 9 ) , @xmath46 , and @xmath24 , respectively .
we here take the photon hopping strength @xmath44 and the number of the cavities in the cca @xmath29 .
the dashed curve represents the case of the maximally entangled state with @xmath34 , the solid line corresponds to the initial input state with the concurrence @xmath47 , and the dot line represents the case of the initial unentangled state with @xmath33 . from fig
2 we can see that the two photons in the cca rapidly tend to the tpd with the time evolution in a short - time regime .
this clearly indicates the transition from the tpl state to the tpd state in the two - photon cca .
the tpd degree can be preserved in the process of the long time evolution afterwards except some small oscillations . on the other hand
, fig.2 indicates that the tpd is enhanced with the increase of quantum entanglement of the initial state ( 9 ) .
hence , we can conclude that quantum entanglement can essentially affect the tpd . in the strong hopping regime for three different values of quantum entanglement of the two - photon state given by eq .
the evolution time is scaled by @xmath35 .
the photon hopping parameter takes @xmath48 .
the number of the cavities in the cca is @xmath29 .
the dashed curve represents the result of the maximally entangled state with @xmath34 , the solid line corresponds to the initial input state with the concurrence @xmath34 , and the dot line shows the result of the initial unentangled state with @xmath33 .
, width=321,height=226 ] we then turn to quantum dynamics of the tpd in the strong hopping regime of @xmath49 . in .
2 , we have plotted the time evolution of the tpd degree @xmath40 for three different values of quantum entanglement of the initial state @xmath50 .
the photon hopping parameter takes @xmath48 .
the number of the cavities in the cca is @xmath29 .
the dashed curve represents the case of the maximally entangled state with @xmath34 , the solid line corresponds to the initial input state with the concurrence @xmath34 , and the dot line represents the case of the initial unentangled state with @xmath33 . from fig .
3 we can clearly observe the entanglement - enhanced tpd phenomenon . in comparison with the quantum dynamics of the tpd in the weak hopping regime of @xmath51 , from fig.2 and fig .
3 we can see that the tpd in the strong hopping regime becomes faster than that in the weak hopping regime in the short time regime .
the oscillations of the tpd degree in the process of the time evolution become stronger with the increasing the photon hopping strength .
however , the tpd degree is over 90 percent in the most time .
the short time behavior of the tpd degree is indicated in the inset in fig .
3 , from which we can see that the transition from the tpl to the tpd happens in shorter time in the strong hopping regime than the weak hopping regime .
in conclusion , we have studied photonic transport properties and tpd dynamics in the two - photon cca when two entangled photons are initially injected into two nearest - neighbor coupling cavities in the form of two - photon noon - type states which are superposition states of two - photon states @xmath52 and @xmath53 .
we have investigated the tpd dynamics by introducing the concept of the tpd degree for the initial two - photon entangled states @xmath54 .
it has been found that photonic transport dynamics in the two - photon cca exhibits the entanglement - enhanced tpd phenomenon .
it has demonstrated that the cca can realize the localization - to - delocalization transition for two entangled photons .
it is worthwhile to point out that the entanglement - enhanced tpd phenomenon is a two - photon cooperative effect induced by quantum interference of two entangled photons due to the initial quantum entanglement .
the entanglement - enhanced tpd vanishes in the absence of entanglement between two photons . | attention paid to coupled - cavity arrays ( ccas @xcite to rich physics potential applications quantum information processing photonics tasks ccas test bed for many - body models quantum simulators transport properties of photons excitations in ccas with natural artificial atoms interest ccas with cyclic three - level system single atom open quantum routing of photons ccas low - loss cavities nearest - neighbor coupling photon hopping high control of each cavity measurements quantum - state engineering dynamical parameter tuning local addressing in quantum information processing fabrication techniques make ccas experimentally accessible quantum entanglement essential propagation of spatially entangled photons in ccas without atoms not studied quantum entanglement affects photonic transport properties in ccas transport properties of two initially entangled photons injected into nearest - neighbor coupling cavities in one - dimensional cca photonic transport dynamics in two - photon cca entanglement - enhanced two - photon delocalization ( tpd ) phenomenon dynamics concept tpd degree cavities two - photon entangled statedemonstrate cca localization - to - delocalization transition for two entangled photons paper organized sec ii present physical model solution analyze photonic transport properties tpd two - photon cca show entanglement - enhanced tpd localization - delocalization transition photons conclude paper discussions remarks last section consider finite - length cca nearest - neighbor cavities coupled photon hopping assume cavities identical each cavity sustains single field mode frequency @xmath1 hamiltonian full system @xmath2 written @xmath3 @xmath4 photonic creation annihilation operator at @xmath5-th cavity second term represents hopping interaction intercavity coupling strength @xmath6 hamiltonian cca diagonalized normal modes @xmath7 inverse transformation @xmath8 transformation matrix @xmath9 obtain diagonalized form hamiltonian 1 ) transformation @xmath10 heisenberg equations obtain exact solution cca heisenberg representation form @xmath11 explicit form green s function given by @xmath12 use solution investigate quantum dynamics of photonic transport tpd in two - photon cca quantum mechanical properties light observed correlations between propagating photons consideredstudy photonic transport properties quantum dynamics of two - photon delocalization in cca injected evolution photon - number correlation function @xmath13 probability detecting two - photon coincidence across cavity @xmath14 @xmath15 probability both photons same cavity @xmath15 given by @xmath16 two photons in @xmath17-th @xmath18-th cavity two - photon noon - type states @xmath19 two photons in entangled states in same cavity @xmath17 @xmath18 other cavities in vacuum state two - photon entangled states 9 localization states entanglement measured by concurrence @xcite @xmath21 concurrence vanishes when @xmath22 @xmath23 reaches maximum @xmath24 @xmath25 substituting eqs ( 6 ) 9 ) into eq . ( 8) probability detecting two - photon coincidence across cavity @xmath14 @xmath15 with expression @xmath26 photonic propagation in for two - photon entangled state eq 9 ) injected into two neighboring cavities @xmath27 @xmath28 cca consists of 29 cavities @xmath29quantum evolution calculated correlation function presented in fig . 1 different initial entanglement fig.1(a(c(e ) initial - state correlation function initial entanglement @xmath30 @xmath24 fig(b(d(f correlation function time @xmath31 initial entanglement @xmath30 @xmath24 correlation diagonal elements - photon @xmath32 vanishes off - diagonal elements @xmath32 enhanced with increase initial entanglement from vanishing to maximal entanglement @xmath34 ) two photons separate emerge in different cavities simultaneously detecting photons in cavity @xmath14 @xmath15 after temporal evolution @xmath31 evolution time scaled by @xmath35 other parameters @xmath36 , @xmath29 understand transport dynamics two photons concept two - photon delocalization ( tpd ) two photons in different cavities photon cca tpd same cavity two - photon localization tpl ) degree tpd defined sum of probability @xmath37 detecting photons in every cavity @xmath38 probability two photons not at same cavity simultaneously@xmath39 two photons in complete tpd state different cavities impossible same cavity stay at different cavities tpd degree repulsive - statistics anti - bunching property of two photons cca tpd degree two - photon noon - type state eq. 9 ) is zero not tpd tpl state larger value parameter @xmath40 stronger tpd phenomenon two - photon cca by 29 cavities two photons injected into @xmath41-th and @xmath42-th cavities input state photon - number second - order correlation function changes in time evolution fig . 1 appearance tpd diagonal elements of two - photon probability @xmath37 suppressed in time evolution maximally entangled state @xmath34 completetpd with @xmath43 time evolution fig . 1(e ) 1(f ) weak hopping regime for three values quantum entanglement two - photon state evolution time scaled by @xmath35 photon hopping parameter takes @xmath44 number cavities in cca is @xmath29 dashed curve maximally entangled state with @xmath34 solid line initial input state @xmath34 dot line initial unentangled state with @xmath33=321 investigate quantum dynamics tpd influence quantum entanglement initial state in weak hopping regime @xmath45 fig . 2 plotted time evolution tpd degree for three values quantum entanglement initial state @xmath24 photon hopping strength @xmath44 number cavities in cca @xmath29 dashed curve maximally entangled state @xmath34 solid line initial input state @xmath47 dot line initial unentangled state @xmath33 two photons cca tend to tpd time evolution short - time regime transition from tpl to tpd state two - photon cca tpd degree preserved long time evolution except small oscillations fig.2 tpd enhanced with increase quantum entanglement initial state quantum entanglement tpd strong hopping regime three values quantum entanglement two - photon state evolution time scaled by @xmath35 photon hopping parameter takes @xmath48 number cavities in cca @xmath29dashed curve maximally entangled state with @xmath34 solid line initial input state concurrence @xmath34 dot line initial unentangled state with @xmath33 width=321,height=226 quantum dynamics tpd in strong hopping regime @xmath49 plotted time evolution of tpd degree @xmath40 for three values quantum entanglement initial state @xmath50 photon hopping parameter takes @xmath48 cavities in cca is @xmath29 dashed curve maximally entangled state with @xmath34 solid line initial input state concurrence @xmath34 dot line initial unentangled state with @xmath33 fig . 3 entanglement - enhanced tpd phenomenon comparison with quantum dynamics tpd weak hopping regime @xmath51 tpd in strong hopping regime faster weak short time regime oscillations tpd stronger photon hopping strength tpd over 90 percent most time short time behavior indicated in fig . 3 transition from tpl to tpd shorter in strong hopping regimestudied photonic transport tpd dynamics in two - photon cca two photons injected into coupling cavities two - photon noon - type states superposition states of @xmath52 @xmath53 investigated tpd dynamics tpd degree for initial two - photon states @xmath54 photonic transport dynamics exhibits entanglement - enhanced tpd phenomenon cca localization - to - delocalization transition for two entangled photons - enhanced tpd phenomenon two - photon cooperative effect induced by quantum interference initial entanglement vanishes absence entanglement | 0.54184 | 3,160 | 1,623 | , Saving $0.1 in GPT-4. |
177,169 | we introduce a fluid dynamics algorithm that performs with nearly spectral accuracy , but uses finite - differences instead of ffts to compute gradients and thus executes 10 times faster .
the finite differencing is not based on a high - order polynomial fit .
the polynomial scheme has supurb accuracy for low - wavenumber gradients but fails at high wavenumbers .
we instead use a scheme tuned to enhance high - wavenumber accuracy at the expense of low wavenumbers , although the loss of low - wavenumber accuracy is negligibly slight .
a tuned gradient is capable of capturing all wavenumbers up to 80 percent of the nyquist limit with an error of no worse than 1 percent .
the fact that gradients are based on finite differences enables diverse geometries to be considered and eliminates the parallel communications bottleneck . | the spectral fluid algorithm ( canuto et .
al . 1987 ) uses fast fourier transforms ( ffts ) to compute gradients , the most precise means possible .
finite - difference gradients based on a polynomial fit execute faster than ffts but with less accuracy , necessitating more grid zones to achieve the same resolution as the spectral method .
the loss of accuracy outweighs the gain in speed and the spectral method has more resolving power than the finite - difference method .
we introduce an alternative finite - difference formula not based on a polynomial fit that executes as quickly but with improved accuracy , yielding greater resolving power than the spectral method . in section [ finite ]
we derive high - wavenumber finite difference formulas and exhibit their effect on the resolving power of a turbulence simulation in section [ flop ] . in section [ mhd ]
we apply finite differences for the purpose of mimicking the spectral algorithm , and then proceed to other applications in section [ applications ] .
define a function @xmath0 on a set of grid points @xmath1 with j an integer . then construct a gradient @xmath2 at @xmath3 from sampling a stencil of grid points with radius ( or order )
@xmath4 on each side .
the familiar result for the gradient on a radius-1 stencil is @xmath5 which is obtained from fitting a polynomial of degree @xmath6 to @xmath7 for a degree 4 polynomial on a radius-2 stencil , @xmath8 for a stencil of order @xmath4 , @xmath9 where @xmath10 consider the finite - difference error at x=0 for a fourier mode @xmath11 cosine modes can be ignored because they do nt contribute to the derivative at @xmath12 note that the wavenumber @xmath13 is scaled to grid units so that @xmath14 corresponds to the maximum wavenumber expressible on the grid , also known as the nyquist limit .
the finite difference formula ( [ stencil ] ) gives @xmath15 whereas the correct value should be @xmath16 define an error function @xmath17 figure @xmath18 shows @xmath19 for stencils of radius 1 through 24 .
the first order finite difference formula is quite lame , delivering 1 percent accuracy only for @xmath13 less than @xmath20 brandenburg ( 2001 ) recognized that higher - order finite differences can significantly extend the wavenumber resolution of the polynomial - based finite difference .
the 8th order finite difference accuracy is better than 1 percent up to @xmath21 and the 16th order finite difference up to @xmath22 nevertheless these are still far from the nyquist limit of @xmath14 and even higher - order finite - differences yield little further progress .
a fourier transform gives the correct gradient for all @xmath13 up to unity .
this is why the spectral algorithm delivers the most resolution per grid element .
the resolution limit is set by the maximum @xmath13 for which gradients can be captured accurately .
although the 8th order finite - difference formula involves considerably fewer floating point operations than an fft , the loss of resolution still renders it less efficient than the spectral method .
polynomial - based finite differences have high accuracy at low @xmath13 but fail at large @xmath23 we can instead construct a more practical scheme to improve high-@xmath13 accuracy at the expense of low-@xmath13 accuracy , yet the loss of low-@xmath13 accuracy is negligibly small . from equation [ eqerror ] , we see that the problem of computing accurate gradients reduces to optimizing , or tuning " the coefficients @xmath24 to minimize the error over a suitable range of k , or equivalently to construct a sine series that best mimics a linear function .
a set of tuned coefficients appear in table 2 with the associated error functions in figure 1 .
the error in the radius-8 tuned finite difference is less than 1 percent up to @xmath25 a dramatic improvement over the radius-8 polynomial .
an algorithm based on tuned gradients still has a lower maximum @xmath13 than the spectral algorithm but due to its increased speed it has greater resolving power ( section [ flop ] ) .
henceforth we denote these tuned gradients as hypergradients . "
minimizing the error function involves a multiparameter optimization of the coefficients @xmath26 - a problem - dependent task with multiple legitimate options .
in fact , a high degree of customization is possible for the form of the error function . for this application
we proceed as follows .
define a target @xmath27 and an indicator for the quality of the tuned coefficients : @xmath28 ^ 4.\ ] ] then , perform a multi - dimensional optimization on @xmath29 the use of a fourth power ensures an approximately uniform error within @xmath30 although a weight function could be added to further customize the form of the error function .
@xmath27 is then adjusted until the error is 1 percent .
the procedure is repeated for each order @xmath4 to yield the coefficients in table 2 .
it is worth noting that the radius-8 tuned coefficients are similar to the radius-24 polynomial coefficients .
this is not surprising because the polynomial coefficients are too small to matter outside of radius 8 .
[ cols="<,<,<,<,<,<,<,<,<,<",options="header " , ] _ table 7 : index of simulations .
all simulations have @xmath31 , @xmath32 and @xmath33 except for z677 , which has @xmath34 @xmath35 indicates that the fft divergence correction is never applied .
all finite - difference simulations utilize a radius-8 stencil and a phase - shift dealiasing correction . _
the most accurate and most expensive interpolation procedure is drect evaluation of the fourier series .
tuned finite differences provide a less expensive interpolation high - wavenumber interpolation .
for example , in 2d , the centered interpolation ( equation [ eqinterp ] ) provides function values halfway between grid points , and another interpolation along the diagonal yields the values at the grid centers .
we have thus doubled the resolution of the grid , which we can do again if we wish .
note that we can do this for the entire grid or just for a subsection .
after a few doublings , a simple linear interpolation serves to provide the function value at any point in space , yielding a 2d interpolation with the same wavenumber resolution as the component 1d interpolation .
this procedure generalizes to arbitrary dimension . as if it was nt enough trouble to run large simulations on thousands of cpus , one is next confronted with analyzing large data cubes that are too big to fit in the memory of one machine .
tuned operators allow for the construction of local local alternatives to global functions like the fft .
these include derivatives , dealiasing , filtering , and grid interpolation .
large output files can be stored in small easy - to - manage segments and operated on successively . for the purpose of data analysis , we have provided
radius-16 tuned operators in table 3 that are accurate to 0.3 percent .
we thank eric blackman , benjamin chandran , peggy varniere , yoram lithwick , and tim dennis for useful discussions , and yoram lithwick for the fft benchmark results .
the simulations were run at the pittsburgh supercomputing center s alpha es45 lemieux " with the national resource allocation committee grant mca03s010p , and at the national center for supercomputing applications intel itanium titan " with the grant ast020012n .
the author was supported by dr .
eric blackman s doe grant de - fg02 - 00er54600 at the university of rochester , and by dr .
benjamin chandran s nsf grant ast-0098086 and doe grants de - fg02 - 01er54658 and de - fc02 - 01er54651 , at the university of iowa . | spectral fluid algorithm canuto 1987 uses fast fourier transforms gradients finite - difference gradients polynomial fit execute faster less accuracy more grid zones loss of accuracy outweighs gain speed spectral method more resolving power alternative finite - difference formula executes quickly improved accuracy greater resolving power derive high - wavenumber finite difference formulas exhibit effect resolving power turbulence simulation apply finite differences spectral algorithm proceed other applications define function @xmath0 on grid points @xmath1 j integer construct gradient @xmath2 at @xmath3 grid points radius @xmath4 each side familiar result gradient radius-1 stencil is @xmath5 obtained from fitting polynomial of degree @xmath6 to @xmath7 degree 4 polynomial radius-2 stencil @xmath8 stencil order @xmath4 @xmath9 consider finite - difference error at x=0 fourier mode @xmath11 cosine modes ignored contribute derivative @xmath12 wavenumber @xmath13 scaled to grid units corresponds to maximum wavenumber expressible on grid nyquist limitfinite difference formula gives @xmath15 correct @xmath16 error function @xmath17 @xmath18 shows @xmath19 for stencils radius 1 through 24 first order formula lame 1 percent accuracy for @xmath13 less than @xmath20 higher - order differences extend wavenumber resolution 8th order accuracy better than 1 percent up to @xmath21 16th order up to @xmath22 far from nyquist limit of @xmath14 little progress fourier transform gives correct gradient for @xmath13 up to unity spectral algorithm delivers most resolution per grid element resolution limit set by maximum @xmath13 8th order formula fewer floating operations loss of resolution less efficient than spectral polynomial - based finite differences high accuracy at low @xmath13 fail at large @xmath23 improve high-@xmath13 accuracy low-@xmath13 accuracy loss small problem computing accurate gradients to optimizing coefficients @xmath24 to minimize error over range k sine series linear function tuned coefficients in table 2 error functions figure 1error radius-8 tuned finite difference less than 1 percent up to @xmath25 improvement over radius-8 polynomial algorithm tuned gradients lower maximum @xmath13 than spectral algorithm increased speed greater resolving power tuned gradients hypergradients minimizing error involves multiparameter optimization coefficients @xmath26 problem - dependent options high customization possible for error function define target @xmath27 indicator quality coefficients @xmath28 perform multi - optimization on @xmath29 fourth power ensures uniform error @xmath30 weight function added customize error @xmath27 adjusted until error 1 percent procedure repeated each order @xmath4 coefficients table 2 radius-8 tuned coefficients similar to radius-24 polynomial coefficients polynomial coefficients too small matter outside radius 8 table 7 index of simulations all simulations have @xmath31 @xmath32 @xmath33 except z677 @xmath34 @xmath35 divergence correction never applied simulations utilize radius-8 stencil phase - shift dealiasing correction most accurate expensive interpolation procedure drect evaluation fourier seriestuned finite differences provide less expensive high - wavenumber interpolation in 2d centered interpolation provides function values halfway between grid points another interpolation yields values at grid centers doubled resolution grid can again entire grid or subsection after doublings simple linear interpolation function value at any point 2d interpolation same wavenumber resolution as 1d procedure generalizes to arbitrary dimension large simulations analyzing large data cubes too big memory tuned operators allow local alternatives to global functions include derivatives dealiasing filtering grid interpolation large output files stored in small segments operated successively provided radius-16 tuned operators in table 3 accurate to 0.3 percent thank eric blackman benjamin chandran peggy varniere yoram lithwick tim dennis for discussions for fft benchmark results simulations run at pittsburgh supercomputing center alpha national resource allocation committee grant mca03s010p national center for supercomputing applications itanium grant author supported by dr eric blackman s doe grant - 00er54600 university of rochester dr benjamin chandran grant ast-0098086 doe grants 01er54658 01er54651 university iowa | 0.325501 | 1,690 | 866 | , Saving $0.0 in GPT-4. |
177,179 | the determination of smuon and neutralino masses in smuon pair production is an important part of the program of spectroscopic studies of supersymmetry at a high energy linear collider . in this note
we report the first results of a study of @xmath0 in a high - mass , cosmology - motivated supersymmetric scenario at 3 tev at clic .
this process is a good example to study requirements on the beam energy spectrum and polarisation and the track momentum resolution in a simple final state .
we discuss the expected accuracy on the mass measurements as a function of the momentum resolution , luminosity spectrum , beam polarisation and time stamping capability .
results obtained at generator level are validated by comparison to full simulation and reconstruction .
preliminary requirements for the detector performances and beam polarisation are presented . | one of the main objectives of linear collider experiments is the precision spectroscopy of new particles predicted in theories of physics beyond the standard model ( sm ) , such as supersymmetry ( susy ) . since some , or most , of these particles may have masses of @xmath1
, these studies may be central to the physics program of a multi - tev @xmath2 linear collider , such as clic . in this note , we study the production of the supersymmetric partners of the muon in a specific susy scenario , where we assume r parity conservation within the so - called constrained minimal supersymmetric extension of the sm ( cmssm ) . in this model the neutralino ( ) is the lightest supersymmetric particle and the specific parameters of the benchmark point @xcite are chosen to make it compatible with current collider and cosmology data .
in particular , the properties of the lightest neutralino are such that it generates the correct amount of relic dark matter density in the universe , as obtained from the analysis of the wmap data @xcite . scalar muons ( @xmath3 and @xmath4 ) are the supersymmetric partners of the right- and left - handed charged muons .
smuons are produced in pair through @xmath5-channel @xmath6 exchange in the process @xmath0 and each decay into an ordinary muon and a neutralino , .
the neutralino , being weakly - interacting , escapes detection .
therefore , the experimental signature of the process is two oppositely charged muons plus missing energy .
this study concentrates on the lightest smuon , @xmath3 , which , for the chosen model parameters , has a mass of 1108.8 gev , while the mass of the lightest neutralino is 554.3 gev .
the accurate determination of their masses is an essential part of the spectroscopy study of a high mass susy scenario at clic .
a study of the variation of the predicted relic dark mass density in the universe @xmath7 with the lightest neutralino mass in the cmssm shows that a @xmath81.0 gev uncertainty on its mass corresponds to a @xmath80.05 relative uncertainty on @xmath7 , i.e. the current accuracy from cosmic microwave background observations @xcite .
the main aim of this study is to assess the requirements for a detector at clic operating at a centre - of - mass energy , @xmath9 , of 3 tev as a function of the track momentum resolution , luminosity spectrum and beam polarisation .
the reconstruction of the particle masses through the endpoints of the muon momentum spectrum is a good example for these requirements since the analysis is particularly simple and can be carried out using a simple momentum smearing on generator - level observables .
results are validated using full simulation and reconstruction with the clic - ild detector model .
the simulation is performed for the cmssm parameters of point k of ref . @xcite . in the cmssm
the mass parameters are defined at the gut scale .
the subsequent evolution to the electro - weak scale is performed using the renormalisation group equations of isasugra 7.69 @xcite .
signal events are generated using pythia 6.125 @xcite . at 3 tev , the production cross section for the process @xmath10 , for unpolarised beams , is 0.71 fb .
beamstrahlung effects on the luminosity spectrum are included using results of the clic beam simulation for the 2008 accelerator parameters @xcite .
initial state radiation ( isr ) is included in the event generation in pythia .
the following background processes have been included in the background calculation : [ h ] [ cols= " < , < " , ] [ tab : summary ] this study allows us to draw some conclusions on the potential of a 3 tev clic collider in susy spectroscopic measurements and some of the requirements on the detector and the beams . because of the tiny production cross section in the chosen high - mass scenario , background subtraction is the dominant source of statistical uncertainty .
electron beam polarisation at @xmath11 80 % gives an equivalent luminosity gain of a factor of six and is essential to recover precision .
positron polarisation is desirable , since it gives an additional gain of a factor of two in equivalent luminosity and it also allows us to disentangle the contributions of @xmath12 and @xmath13 .
smuon and neutralino masses of 1108.8 gev and 554.3 gev , respectively can be extracted from the muon kinematics , in events with two oppositely charged muons and missing energy , with a relative statistical accuracy @xmath14 0.5 % with 2 ab@xmath15 of integrated luminosity and both beams polarised .
in addition , the signal production cross section of 0.7 fb can be determined with a relative statistical uncertainty of 2.0 % .
since a major source of smearing of the kinematic edges of the muon momentum spectrum is beamstrahlung and isr , the track momentum resolution does not appear to be critical for the measurement of the smuon mass , as long as a resolution @xmath16 gev@xmath15 can be achieved , though it remains important for the neutralino mass .
it is important to have a good control of the luminosity spectrum and desirable to limit the beamstrahlung not significantly beyond that corresponding to the 2008 clic parameters .
finally , the effect of the overlay of @xmath17 events from machine - induced background does not lead to any significant degradation of the signal selection efficiency for a detector with time stamping capability of 10 ns .
we are grateful to daniel schulte for making the luminosity spectrum and generated @xmath18 events available and to dieter schlatter for his careful reading of this note .
100 m. battaglia , a. de roeck , j. r. ellis , f. gianotti , k. a. olive and l. pape , eur .
j. c * 33 * ( 2004 ) 273 [ arxiv : hep - ph/0306219 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl .
* 180 * ( 2009 ) 330 [ arxiv:0803.0547 [ astro - ph ] ] .
e. komatsu _ et al .
_ , arxiv:1001.4538 [ astro-ph.co ] . f. e. paige , s. d. protopopescu , h. baer and x. tata , arxiv : hep - ph/0312045 .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * ( 2006 ) 026 [ arxiv : hep - ph/0603175 ] .
h. braun _ et al . _
[ clic study team ] , clic - note-764 ( 2008 ) .
e. boos _ et al .
_ [ comphep collaboration ] , nucl .
instr . and meth . a * 534 * ( 2004 ) , 250 . g. a. moortgat - pick _
_ , phys .
* 460 * ( 2008 ) 131 [ arxiv : hep - ph/0507011 ] .
j. l. feng and d. e. finnell , phys . rev .
d * 49 * ( 1994 ) 2369 [ arxiv : hep - ph/9310211 ] .
h. u. martyn and g. a. blair , arxiv : hep - ph/9910416 .
f. james and m. roos , comput .
commun .
* 10 * ( 1975 ) 343 .
d. schulte , tesla note 97 - 08 .
+ the results of the beam simulation have been provided by daniel schulte .
m. skrzypek and s. jadach , z. phys .
c * 49 * ( 1991 ) 577 .
t. ohl , comput .
commun .
* 101 * ( 1997 ) 269 [ arxiv : hep - ph/9607454 ] .
m. battaglia , s. jadach and d. bardin , in proc . of the snowmass 2001 workshop , n. graf ( ed . ) , slac - r-599 ( 2002 ) , p. e3015 . | main linear collider experiments precision spectroscopy of new particles beyond standard model supersymmetry susy particles masses @xmath1 studies central to physics program multi - tev @xmath2 linear collider study production supersymmetric partners muon in specific susy scenario r parity conservation within constrained minimal supersymmetric extension sm neutralino lightest supersymmetric particle parameters benchmark point @xcite compatible with current collider cosmology data properties lightest neutralino generates correct relic dark matter density universe obtained from wmap data scalar muons @xmath3 and @xmath4 supersymmetric partners of right- left - handed charged muons produced in pair through @xmath5-channel @xmath6 exchange decay into ordinary muon neutralino neutralino weakly - interacting escapes detection experimental signature two oppositely charged muons plus missing energy study concentrates on lightest smuon @xmath3 mass 1108.8 gev mass lightest neutralino 554.3 gev accurate determination masses essential spectroscopy study high mass susy scenario at clicstudy predicted relic dark mass density universe @xmath7 lightest neutralino mass cmssm shows @xmath81.0 gev uncertainty corresponds to @xmath80.05 relative uncertainty @xmath7 current accuracy from cosmic microwave background observations aim requirements detector clic centre - of - mass energy 3 tev track momentum resolution luminosity spectrum beam polarisation reconstruction particle masses muon momentum spectrum analysis simple momentum smearing generator - level observables results validated full simulation reconstruction clic - ild detector model simulation performed cmssm parameters point k ref. mass parameters defined at gut scale evolution to electro - weak scale renormalisation group equations isasugra 7.69 signal events generated pythia 6.125 at 3 tev production cross section process @xmath10 unpolarised beams 0.71 fb beamstrahlung effects luminosity spectrum included clic beam simulation 2008 accelerator parameters initial state radiation isr ) included in event generation pythia background processes included calculation study conclusions on potential 3 tev clic collider susy spectroscopic measurements requirements detector beamstiny production cross section high - mass scenario background subtraction dominant statistical uncertainty electron beam polarisation at @xmath11 80 % luminosity gain factor six essential precision positron polarisation desirable additional gain factor two luminosity contributions @xmath12 @xmath13 smuon neutralino masses 1108.8 gev 554.3 gev extracted from muon kinematics oppositely charged muons missing energy accuracy @xmath14 0.5 % 2 ab@xmath15 integrated luminosity both beams polarised signal production cross section 0.7 fb determined statistical uncertainty 2.0 % major smearing kinematic muon momentum spectrum beamstrahlung isr track momentum resolution critical for measurement smuon mass resolution @xmath16 gev@xmath15 important for neutralino mass control luminosity spectrum limit beamstrahlung beyond 2008 clic parameters overlay of @xmath17 events from machine - induced background degradation signal selection efficiency for detector time stamping capability 10 ns grateful to daniel schulte luminosity spectrum @xmath18 events dieter schlatter for reading. battaglia de roeck ellis gianotti olive pape33 2004 ) 273 hep - ph/0306219 komatsu wmap collaboration astrophys 180 2009 ) 330 arxiv:0803.0547 astro - ph komatsu arxiv:1001.4538 astro-ph.co paige protopopescu baer x. tata hep - ph/0312045 sjostrand mrenna skands 0605 2006 ) 026 hep - ph/0603175 braun clic study team clic - note-764 ( 2008 ) boos comphep collaboration nucl . instr . meth 534 2004 ) 250 moortgat - pick . 460 2008 ) 131 hep - ph/0507011. feng e. finnell . rev 49 1994 ) 2369 hep - ph/9310211 martyn a. blair hep - ph/9910416 james m. roos comput . commun 10 ( 1975 ) 343 schulte tesla note 97 - 08 results beam simulation daniel schulte skrzypek s. jadach phys . 49 * 1991 ) 577 ohl comput . commun 101 1997 ) 269 hep - ph/9607454. battaglia s. jadach d. bardinsnowmass 2001 workshop slac - r-599 2002 p. e3015 | 0.315575 | 1,921 | 968 | , Saving $0.1 in GPT-4. |
177,189 | this brief report describes the search for gravitational - wave inspiral signals from short gamma - ray bursts . since these events are probably created by the merger of two compact objects , a targeted search with a lower threshold can be made .
the data around 22 short gamma - ray bursts have been analyzed .
gamma ray bursts ( grb ) are intense flashes of @xmath0-rays that occur in two classes ; long duration bursts ( duration @xmath1 ) which are generally associated with hypernova explosions in star - forming galaxies @xcite and short duration bursts ( duration @xmath2 ) , thought to originate primarily from the coalescence of a neutron star ( ns ) with another compact object@xcite , like a neutron star or a black hole ( bh ) . such coalescences produce gravitational waves ( gw ) which are searched for , and because the grbs give the time and sky location the search can be performed with reduced threshold , increasing the sensitivity compared to untriggered searches .
this is a brief report on a search for gravitational wave signals associated with short grbs ( sgrb ) during ligo s fifth science run s5 and virgo s first science run , vsr1 .
s5 took place from 4 november 2005 to 30 september 2007 and vsr1 from 18 may 2007 to 30 september 2007 .
s5 is comprised of three detectors ( h1 and h2 at hanford , wa , and l1 at livingston , la ) and vsr1 is comprised of the virgo detector near pisa , italy
. there are 212 grbs in total observed during this time , of which 33 are sgrbs . because we require coincident data from at least two detectors ( which are the ones with the best directional sensitivity to the known sky location of the grb ) ,
only 22 sgrbs have been selected for the analysis .
there are 9 grbs in h1h2 coincident time , 11 in h1l1 , 1 in h2l1 , and 1 grb in h1l1v1 coincident time .
the latter grb ( 070923 ) is a special case because it uses data from three spatial separated detectors , which all have about the same directional sensitivity to grb 070923 .
worth special mention is also grb 070201 , which has been already analyzed in a high - priority search because the progenitor was possibly located in m31 , a galaxy only @xmath3780 kpc away .
no gravitational - wave signal was found and a coalescence scenario could be ruled out with @xmath499% confidence at that distance @xcite . this grb is being reanalyzed in the current work because of the use of a lower threshold and improved analysis methods .
the coalescence model suggests that the time delay between the arrival of the gravitational wave and the subsequent electromagnetic burst , referred to as trigger time , is of order one second or less .
if the initial spike of grbs is created by so - called internal shocks ( produced by the impact of a faster outmoving shells on a slower moving shells in a grb ) , the time - delay is on the millisecond scale @xcite or at least below 1 second @xcite . with a more semi - analytical description of the final stage of a ns-bh merger
it has been calculated that the majority of matter plunges on the bh within 1 second @xcite and numerical simulations on the mass transfer suggest a timescales of milliseconds @xcite or some seconds at maximum @xcite .
these considerations prompted the use of a long time window ( _ on - source _ ) around the trigger time ( from -5 to + 1 ) , in which we expect the gw signal to be .
the data around the on - source segment is chosen to estimate the background and to perform software injections . to prevent a possible loud on - source signal biasing the background estimation
, we discard on each side of the on - source segment , and split the remaining data in up to 324 segments , each long .
the number of used off - source segments varies because some might be vetoed due to bad data quality @xcite .
the data are analyzed using the standard compact binary search pipeline described in detail in [ ] , using a matched - filter technique to filter the data against post - newtonian approximants of the expected gw signal .
the mass - range of binaries searched for spans the range @xmath5 in total mass .
after having applied signal - based vetoes and performing the coincidence criterion , we end up with a list of _ candidate events _ ; see [ ] for details . to obtain the search efficiency and to calculate the likelihood of a candidate event , simulated waveform signals ( _ software injections _ )
are added to the off - source data and processed with the pipeline described above .
the efficiency is determined by the fractional number of recovered injections , made over a wide range of parameters ( i.e. masses , spins etc . ) .
the likelihood ratio of a candidate is then the efficiency divided by the false - alarm probability , which is obtained by the relative number of off - source trials yielding a statistical significance equal or greater than the on - source measurement .
the analysis yielded no evidence for a gravitational - wave signal in coincidence with any grb in our sample . with the null observations and the large number of injection trials we are able to constrain the distance to each grb assuming it was caused by a compact binary coalescence consisting of a neutron star and a neutron star or a black hole .
detailed results on the exclusion distances are given in [ ] .
in addition to the individual detection searches , we assess the presence of a gravitational wave signal too weak to stand out above background separately , but which is significant when the entire sample of analyzed grbs are taken together .
this is being done with the non - parametric wilcoxon - mann - whitney u test @xcite , to answer the question of whether both samples ( significances of the on - source and off - source segments ) are drawn from the same population .
this test calculates a u statistic by the sums of the ranks of both samples , and compares it to the expected value if both samples were drawn from the same distribution .
since this statistic is approximately gaussian distributed , a one - sided probability @xmath6 can be calculated that either shows the two samples are indeed drawn from the same distribution ( @xmath7 ) or contain a population of weak signals in the sample ( @xmath8 ) .
a large value for @xmath6 would hint of a problem with the analysis , since it implies more signicant candidates in the off - source than in the on - source .
the full details of this analysis and discussion of the results can be found in [ ] . | the authors gratefully acknowledge the support of the united states national science foundation for the construction and operation of the ligo laboratory , the science and technology facilities council of the united kingdom , the max - planck - society , and the state of niedersachsen / germany for support of the construction and operation of the geo600 detector , and the italian istituto nazionale di fisica nucleare and the french centre national de la recherche scientifique for the construction and operation of the virgo detector .
the authors also gratefully acknowledge the support of the research by these agencies and by the australian research council , the council of scientific and industrial research of india , the istituto nazionale di fisica nucleare of italy , the spanish ministerio de educacin y ciencia , the conselleria deconomia hisenda i innovaci of the govern de les illes balears , the foundation for fundamental research on matter supported by the netherlands organisation for scientific research , the royal society , the scottish funding council , the scottish universities physics alliance , the national aeronautics and space administration , the carnegie trust , the leverhulme trust , the david and lucile packard foundation , the research corporation , and the alfred p. sloan foundation .
this document has been assigned ligo document number p0900307 . | authors acknowledge support united states science foundation ligo laboratory science technology facilities council united kingdom max - planck society state niedersachsen germany geo600 detector italian istituto fisica nucleare french centre national recherche scientifique virgo detector support australian research council council scientific industrial research india istituto nazionale fisica nucleare spanish ministerio educacin ciencia conselleria deconomia illes balears foundation fundamental research netherlands organisation scientific research royal society scottish funding council scottish universities physics alliance national aeronautics space administration carnegie trust leverhulme trust david lucile packard foundation research corporation alfred p. sloan foundation document assigned ligo document number p0900307 | 0.072333 | 279 | 145 | , Saving $0.0 in GPT-4. |
177,190 | zero - range effective interactions are commonly used in nuclear physics and in other domains to describe many - body systems within the mean - field model .
if they are used within a beyond - mean - field framework , contributions to the total energy that display an ultraviolet divergence are found . we propose a general strategy to regularize this divergence and we illustrate it in the case of the second - order corrections to the equation of state ( eos ) of uniform symmetric matter . by setting a momentum cutoff @xmath0
, we show that for every ( physically meaningful ) value of @xmath0 it is possible to determine a new interaction such that the eos with the second - order corrections reproduces the empirical eos , with a fit of the same quality as that obtained at the mean - field level .
in many - body systems contact interactions are reasonable approximations to the realistic finite - range forces that can be employed in cases where the interaction range is smaller than the typical length scale represented by the interparticle distance .
the main advantage of using zero - range interactions is that the equations to handle are greatly simplified .
two examples of commonly used contact interactions are the skyrme forces which are quite popular in nuclear physics @xcite , and the contact interactions with coupling strengths depending on the @xmath1-wave scattering length which are employed for dilute atomic gases ( see , e.g. , ref .
@xcite ) . in general ,
effective interactions contain parameters that must be adjusted to reproduce a given set of observables ( for instance , in the nuclear case , binding energies and radii of a few selected nuclei , and bulk properties of nuclear matter ) .
usually this is done at the self - consistent mean - field level . when going beyond this level ,
it is likely that the effective interaction has to be redetermined .
however , trying to discuss this issue is impossible in the case of contact interactions , since going beyond mean field , e.g. , by including second - order corrections , implies dealing with contributions to the total energy which contain an ultraviolet divergence .
many authors have studied the perturbation series for the energy of the electron gas @xcite and of nuclear systems @xcite . in all cases ,
the interaction is finite - range and no ultraviolet divergence appears . in this paper
, we apply to a simple case a strategy to handle the contact interactions up to second order .
we include a momentum cutoff @xmath0 among the parameters of the interaction , and we show that for every value of @xmath0 the other parameters can be determined in such a way that the total energy of the system with second - order contributions remains the same .
this strategy , and the formulas we will show , are quite general and can be applied to different fermi systems . for the numerical application
we restrict ourselves to symmetric nuclear matter , treated with a simplified zero - range interaction .
the second - order terms that contribute to the total energy in uniform matter and diverge in the case of contact forces are shown in the lower part of fig .
[ figdiag1 ] .
the divergence is caused by the integration on @xmath2 and is somehow unphysical since the high - momentum states are certainly outside the scale at which effective forces are to be used .
in the case of effective interactions between point - nucleons , the cutoff @xmath0 must certainly be smaller than the momentum associated with the nucleon size , i.e. , smaller than @xmath3 2 @xmath4 .
in fact , these interactions are used to describe giant resonances or rotational bands of nuclei and consequently the scale should be even smaller , perhaps around 0.5 @xmath4 .
however , our procedure is tailored on the basic idea of renormalizable quantum - field theories : so , in principle , for any value of @xmath0 a new set of parameters is found which leads to the same equation of state ( eos ) -including the second - order contributions- .
we show below that this is mathematically possible owing to the fact that the second - order correction is well - behaved as a function of the density .
thus , we imitate the qed idea that the bare electron mass and charge can be chosen for every different energy cutoff in such a way that the physical values of mass and charge are always obtained .
an ultraviolet divergence appears already at the mean - field level in the hartree - fock - bogoliubov ( hfb ) @xcite or bogoliubov - de gennes ( bdg ) @xcite models when zero - range forces are employed in the pairing channel . in this case ,
sophisticated regularization schemes exist in which the irregular term of the pairing field is suppressed and the dependence on the energy cutoff is eliminated .
these techniques are commonly adopted for example in atomic physics in bdg models @xcite , and have been also employed in nuclear physics @xcite , but they can not be directly applied to the case of interest studied in the present work .
let us write the zero - range force as @xmath5 to make contact with the skyrme interactions @xcite the strength @xmath6 is written as @xmath7 and this corresponds to the so - called @xmath8 model that is a simplification of the usual skyrme model where the spin - dependent , velocity - dependent and spin - orbit terms are dropped . if the skyrme force is viewed as a @xmath9 matrix ( thus including ladder diagrams ) , the introduction of second - order contributions would in principle imply a double counting . however , our attitude is to consider as a matter of fact our interaction as phenomenological , and our framework as an effective theory where parameters are readjusted according to which diagrams are explicitly introduced . normalizing the single - particle wave functions within a box of volume @xmath10 , the hartree - fock ( hf ) potential - energy contribution ( upper part of fig .
[ figdiag1 ] ) is equal to @xmath11 where @xmath12 , @xmath13 being the level degeneracy ( 4 in the case of symmetric nuclear matter ) , and @xmath14 .
the energy per particle or eos is obtained by adding the kinetic contribution and reads in our case @xmath15 eq .
( [ eq5 ] ) coincides with the eos obtained with the skp parameter set @xcite , in which no contribution coming from the velocity - dependent terms appears and the effective mass coincides with the bare mass . since skp , as all the skyrme sets , has been fitted to reproduce within hf the basic features of the nuclear eos , one can consider its associated energy per particle as a benchmark which must be reproduced with reasonable accuracy for every value of @xmath0 when the second - order correction is included .
this second - order correction is given by @xmath16 where @xmath17 for symmetric nuclear matter , and @xmath18 reads @xmath19 some details about the evaluation of @xmath18 and @xmath20 with a zero - range force are recalled in the appendix ; finally , we write @xmath21 as @xmath22 , with @xmath23 where @xmath24 .
when the cutoff @xmath0 is introduced , the last integral has @xmath25 as upper limit and the corresponding quantity is denoted by @xmath26 .
the expressions for @xmath27 and @xmath28 are given in the appendix .
the analytical expression of @xmath26 is @xmath29 as a function of the density and for different values of the cutoff @xmath0 .
the skp mean - field eos ( solid black line ) is shown for comparison .
( b ) correction @xmath30 for different values of @xmath0.,width=245 ] in fig .
2(a ) , @xmath31 is plotted for different values of @xmath0 and compared with the skp mean - field values of @xmath32 ( solid black line ) .
the correction @xmath30 is also shown in fig .
2(b ) .
one observes that the second - order correction causes a shift of the saturation point to lower densities .
the curves are calculated using the skp parameters ( cf .
table i ) . for @xmath33 1.5 @xmath4 ,
the maximum correction is already comparable with the energy per particle at the saturation point , i.e. , @xmath3 15 mev . to better understand the behavior of @xmath30 , we plot @xmath34 and @xmath26 separately in fig .
[ fig3 ] .
the ultraviolet divergence is visible at all the values of the density , by comparing the trends in panels ( b ) , ( c ) and ( d ) . at the same time
, one observes also a divergent behavior of @xmath35 for @xmath36 that is dictated by the upper limit @xmath37 of the second integral in eq .
( [ eq8 ] ) .
it can be seen in the analytical expression of eq .
( [ eq11 ] ) . @xmath38 multiplies the coefficient @xmath39 which goes to zero like @xmath40 when the density goes to zero .
one may worry that for large values of @xmath0 , the divergent behavior of the integral dominates when @xmath41 0 , so that no regularization is possible for very low density .
however , this happens for values of @xmath0 which are physically meaningless , namely for @xmath42 250 - 300 @xmath4 . coefficient .
other panels : the values of @xmath43 for @xmath33 0.5 ( b ) , 1 ( c ) and 1.5 ( d ) @xmath4.,width=245 ] for each value of @xmath0 we can perform a least square fit to determine a new parameter set skp@xmath44 , such that the eos including the second - order correction matches rather well the one obtained with the original force skp at the mean - field level .
this is our main result and it is illustrated in fig .
4(a ) .
the refitted parameters are listed in table i ( together with the saturation point ) . in fig .
4(b ) we display , for purely mathematical illustration , the refit done with the extreme value of @xmath33 350 @xmath4 .
350 @xmath4.,width=245 ] in summary we have shown that , for any physically meaningful value of the cutoff @xmath0 it is possible to find a contact interaction that can be used in a mean - field plus second - order corrections framework and can describe satisfactorily the empirical eos . the quality of the fits , that can be judged from fig .
4 and the @xmath45-values in table i , demonstrates that the strategy we have outlined in a general fashion works , in practice , for the case of symmetric nuclear matter treated with a simplified contact force .
.from the second line , columns 2 , 3 and 4 : parameter sets obtained in the fits associated with different values of the cutoff @xmath0 compared with the original set skp ( first line ) . in the fifth column the @xmath46-value ( @xmath45 divided by the number of fitted points ) associated to each fit is shown . in columns 6 and 7
the saturation point is shown . [
cols="^,^,^,^,^,^,^ " , ] one can foresee various applications to the studies of strongly correlated fermion systems in all the domains of many - body physics where zero - range forces are used and where beyond - mean - field theories are necessary for a more accurate treatment of complex correlations . in nuclear physics , models beyond mean - field theories where ultraviolet divergences appear owing to the use of a zero - range interaction are : ( i ) models where multiparticle - multihole configurations are introduced @xcite , ( ii ) second random - phase - approximation models @xcite , and ( iii ) models that take into account the coupling between single - particle degrees of freedom and collective vibrations @xcite . in these cases , a further development of our technique may be envisaged . _
aknowledgments . _
the authors thank daniel pea for useful discussions .
one of the authors ( g.c . ) acknowledges partial support from the italian research project ( prin ) named `` many - body theory of nuclear systems and implications on the physics of neutron stars '' . | @xmath18 in eq . ( [ eq7 ] ) can be rewritten by expressing all wave vectors in units of @xmath47 : @xmath48 where the domains @xmath49 are @xmath50 . by introducing @xmath51 , and the unit vector @xmath52 , @xmath53 \2 ) second case : @xmath60 .
one can apply the technique of ref .
@xcite , with this change of variables : @xmath61 where @xmath62 and @xmath13 is a unit vector @xmath63 . in this case , eq .
( [ eq16 ] ) becomes @xmath64 ^ 2 .
\label{eq17n}\end{aligned}\ ] ] the energy correction can be written as @xmath65 with the change @xmath66 we obtain @xmath21 as @xmath22 [ cf . eqs .
( [ eq7n ] ) and ( [ eq8 ] ) ] , with @xmath67 @xmath68 10 [ 2]#2 [ 1]#1 [ 1]#1 t.h.r .
skyrme , phil . mag . * 1 * , 1043 ( 1956 ) ; nucl . phys .
* 9 * , 615 ( 1959 ) ; d. vautherin and d.m .
brink , phys .
c * 5 * , 626 ( 1972 ) .
j. d * 7 * , 433 ( 1999 ) .
m. gell - mann and k.a .
brckner , phys . rev . *
106 * , 364 ( 1957 ) .
k. sawada _
et al . _ ,
rev . * 108 * , 507 ( 1957 ) .
p. nozires and d. pines , phys . rev . *
111 * , 442 ( 1958 ) .
h. euler , z. physik * 105 * , 553 ( 1937 ) .
r. huby , proc .
( london ) a * 62 * , 62 ( 1949 ) .
thouless , phys . rev . * 107 * , 559 ( 1957 ) .
et al . _ ,
phys . rev . *
119 * , 230 ( 1960 ) .
p. ring and p. schuck , _ the nuclear many - body problem _ , springer - verlag berlin heidelberg new york , 1980 . p.g
. de gennes , _ superconductivity of metals and alloys _
, perseus books publishing , l.l.c .
1966 . m. grasso and m. urban , phys .
a * 68 * , 033610 ( 2003 ) .
a. bulgac and y. yu , phys .
lett . * 88 * ,
042504 ( 2002 ) . j. dobaczewski _
et al . _ ,
a * 422 * , 103 ( 1984 ) .
l. bonneau _
et al . _ ,
c * 76 * , 014304 ( 2007 ) .
d. gambacurta _ et al .
c * 81 * , 054312 ( 2010 ) . v. bernard and n. van giai , nucl .
a * 348 * , 75 ( 1980 ) .
et al . _ , nucl phys .
a * 752 * , 345 ( 2005 ) .
et al . _ , rev .
55 , 287 ( 1983 ) .
e. litvinova _
et al . _ ,
c * 75 * , 064308 ( 2007 ) .
dubois , ann .
( n.y . ) * 8 * , 24 ( 1959 ) . | @xmath18 rewritten expressing wave vectors units @xmath47 @xmath48 domains @xmath49 @xmath50 introducing @xmath51 unit vector @xmath52 @xmath53 second case @xmath60 apply technique @xcite change variables @xmath61 @xmath62 @xmath13 unit vector @xmath63 eq eq16 becomes @xmath64 energy correction written @xmath65 change @xmath66 obtain @xmath21 @xmath22 eqs eq7n eq8 @xmath67 @xmath68 10#2 skyrme phil . mag . 1 1043 ( 1956 ) nucl . 9 615 ( 1959 ) d vautherin brink phys . 5 626 ( 1972 ) 7 433 ( 1999 ) gell - mann brckner phys . rev . 106 364 ( 1957 ) sawada al rev 108 507 ( 1957 ) nozires d pines phys . rev . 111 442 ( 1958 ) euler 105 553 ( 1937 ) huby proc. london 62 62 ( 1949 )thouless phys . rev 107 559 ( 1957 ) et 119 230 ( 1960 ) p. ring schuck nuclear many - body problem springer - verlag berlin heidelberg new york 1980 de gennes superconductivity metals alloys perseus books publishing 1966 m. grasso urban phys . 68 033610 ( 2003 ) bulgac yu . 88 042504 ( 2002 ) dobaczewski 422 103 ( 1984 ) bonneau 76 * 014304 ( 2007 ) gambacurta 81 * 054312 ( 2010 ). bernard n. van giai nucl 348 75 ( 1980 ) phys 752 345 ( 2005 ) rev 55 287 ( 1983 ) litvinova c * 75 * 064308 ( 2007 ) dubois ann 8 * 24 ( 1959 ) | 0.134594 | 854 | 456 | , Saving $0.0 in GPT-4. |
177,198 | we suggest that a high proportion of brown dwarfs are formed by gravitational fragmentation of massive extended discs around sun - like stars .
we argue that such discs should arise frequently , but should be observed infrequently , precisely because they fragment rapidly . by performing an ensemble of radiation - hydrodynamic simulations ,
we show that such discs typically fragment within a few thousand years to produce mainly brown dwarfs ( including planetary - mass brown dwarfs ) and low - mass hydrogen - burning stars .
subsequently most of the brown dwarfs are ejected by mutual interactions .
we analyse the properties of these objects that form by disc fragmentation , and compare them with observations .
address = school of physics & astronomy , cardiff university , 5 the parade , cardiff cf24 3aa , uk address = school of physics & astronomy , cardiff university , 5 the parade , cardiff cf24 3aa , uk | gravitational fragmentation of massive , extended discs provides a way to form a large number of brown dwarfs per sun - like star ( stamatellos et al .
we examine the consequences of this mechanism on the observed properties of brown dwarfs and compare them with observations .
more specifically we perform an ensemble of 13 simulations with different numerical realisations of the same star and disc initial conditions , and for the objects produced we compute ( i ) their mass spectrum , ( ii ) their radial distribution from the primary star , ( iii ) their distribution of the orbital planes , ( iv ) the distribution of the masses and sizes of the brown dwarf discs , and ( v ) their binary properties .
we consider a 0.7 m@xmath0 star with a 0.7 m@xmath0 disc that extends out to 400 au .
the disc surface density drops as @xmath1 and the temperature drops as @xmath2 , where @xmath3 is the distance from the star on the disc midplane ( stamatellos & whitworth , 2008b ) .
the simulations are performed in two phases .
( i ) until @xmath4% of the disc has been accreted onto the stars ( this happens within 20,000 yr ) the evolution of the system is followed using sph , with a new method to capture the thermal and radiative effects when protostellar gas fragments ( stamatellos et al .
the method uses a detailed equation of state for a gas mixture of hydrogen and helium.the effects of the rotational and vibrational degrees of freedom of h@xmath5 , of h@xmath5 dissociation , of h ionisation , and the first and second ionisation of he are included .
moreover the opacity changes due to e.g. ice mantle melting , the sublimation of dust , molecular and h@xmath6 contributions , are also taken into account .
the method also includes irradiation by the central star ( see stamatellos & whitworth 2008a , b ) .
( ii ) afterwards the evolution of the system is followed up to 200,000 yr using an n - body code .
the discs quickly ( within a few thousand years ) fragment ( see fig . [
fig : sim ] ) forming gravitationally bound objects outside 50 au , where the conditions for fragmentation are right ( i.e. the disc is toomre unstable _ and _ can cool fast enough ) in accordance with our previous simulations and analytic studies ( whitworth & stamatellos 2006 ; stamatellos et al . 2007 ; stamatellos et al .
2008a , b ) .
a total of 96 objects are produced in all of our simulations .
more than half of them are ejected into the field .
a typical outcome is a sun - like star with a close low - mass star companion , a wide low - mass star companion , and a wide brown dwarf or brown dwarf binary companion .
disc around a 0.7-m@xmath0 star .
snapshots of the logarithm of the column density are presented for 3 instances ( as marked on each graph ) .
this particular simulation produces 4 brown dwarfs , and 3 low - mass hydrogen burning stars .
after 200,000 years , 4 of these objects have been ejected into the field.,title="fig : " ] disc around a 0.7-m@xmath0 star .
snapshots of the logarithm of the column density are presented for 3 instances ( as marked on each graph ) .
this particular simulation produces 4 brown dwarfs , and 3 low - mass hydrogen burning stars .
after 200,000 years , 4 of these objects have been ejected into the field.,title="fig : " ] disc around a 0.7-m@xmath0 star .
snapshots of the logarithm of the column density are presented for 3 instances ( as marked on each graph ) .
this particular simulation produces 4 brown dwarfs , and 3 low - mass hydrogen burning stars .
after 200,000 years , 4 of these objects have been ejected into the field.,title="fig : " ]
[ [ mass - distribution ] ] mass distribution + + + + + + + + + + + + + + + + + most of the objects produced by disc fragmentation ( 75% ) are brown dwarf stars ( i.e. with masses below the h - burning limit ) , and 4% of them are planetary - mass brown dwarfs ( i.e. have mass below the d - burning limit ) .
the rest 25% of the objects produced are low - mass hydrogen - burning stars .
more than half ( 55% ) of the objects produced by disc fragmentation are ejected into the field .
[ [ radial - distribution - and - the - brown - dwarf - desert ] ] radial distribution and the brown dwarf desert + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the inner disc can not cool fast enough so fragmentation happens only outside 50 au from the primary star .
objects that form closer to the primary star are generally the most massive ones formed in the disc as ( i ) they form first , and ( ii ) they form in a region where there is more material to accrete .
hence , they become h - burning stars .
brown dwarfs form farther out in the disc . due to dynamical interactions
the low - mass h - burning stars move further in populating the region close to the primary star whereas the brown dwarfs move outwards ( see fig .
[ fig : radius ] ) .
hence the mechanism of disc fragmenation provides a natural explanation for the brown dwarf desert ( marcy & butler 2000 ) .
our model predicts that the frequency of brown dwarf companions to sun - like stars is 8% within 100 au from the star and 16% within 400 au from the star .
[ [ orbital - planes - of - companions - to - sun - like - stars ] ] orbital planes of companions to sun - like stars + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + initially the orbits of the objects formed in the disc are almost coplanar with the disc of the primary star .
their orbital planes lie within @xmath7 from the disc mid - plane . however due to 3 body interactions , in 3-dimensions ,
after 20,000 yr only 70% of the orbital planes lie within @xmath7 of the disc midplane , and by 200,000 yr this has dropped to 22% .
hence coplanarity , or the absence of it , can not confirm or rule out formation by disc fragmentation .
[ [ properties - of - brown - dwarf - discs ] ] properties of brown dwarf discs + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + most brown dwarfs form with discs .
these discs have masses of a few m@xmath8 and radii of a few tens of au .
these disc masses and radii are consistent with observations , which show discs to be very common around bds ( klein et al .
2003 ; luhman 2004 ; luhman et al .
2005 ; scholz et al . 2006 ; guieu et al . 2007 ; riaz & gizis 2007 , 2008 ) .
most of the ejected brown dwarfs do not retain their discs .
we predict that brown dwarfs that are companions to sun - like stars are more likely to have discs than brown dwarfs in the field .
[ [ binary - properties ] ] binary properties + + + + + + + + + + + + + + + + + 13 low - mass binaries are formed in our simulations .
27% of the objects that form are in binaries corresponding to a low - mass binary fraction of 16% .
this is comparable with the low - mass binary fraction in taurus - auriga ( @xmath9 , kraus et al .
2006 ) , chamaeleon i ( @xmath10 , ahmic et al .
2007 ) , and the field ( e.g. @xmath11 , gizis et al . 2003 ) .
these binaries include star - star ( 4/13 ) , star - brown dwarf ( 4/13 ) , brown dwarf -brown dwarf ( 4/13 ) and brown dwarf - planetary mass object binaries ( 1/13 ) . 4 of these binaries are ejected into the field , including both close and wide binaries .
we predict that for the binaries remaining bound to the central star , the total mass of the low - mass binary tends to decrease as the distance from the central star increases .
most of the low - mass binaries ( 55% ) have components with similar masses ( q>0.7 ) .
our model suggests that brown dwarfs that are companions to sun - like stars are more likely to be in brown dwarf- brown dwarf binaries ( binary frequency 25% ) than brown dwarfs in the field ( binary frequency 8% ) .
burgasser et al .
( 2005 ) report a binary fraction of @xmath12 for bd companions to sun - like stars , and a binary fraction of only @xmath13 for bds in the field .
[ [ free - floating - planetary - mass - objects ] ] free - floating planetary - mass objects + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 3 planetary - mass brown dwarfs form in our simulations , and all three of them are ejected into the field .
their masses remain below the d - burning limit because they are ejected from the disc very soon after formation . according to our simulations planetary - mass brown dwarfs in the field
should be outnumbered by d - burning brown dwarfs by a factor of @xmath14 .
[ [ caveats ] ] caveats + + + + + + + the simulations presented here correspond only to one set of disc and star parameters ( i.e. disc density and temperature profiles , star mass , disc mass and disc radius ) .
hence , the results need to be verified / updated for a wider set of initial conditions
. they also do not account for the disc formation around the sun - like star and disc growth due to accretion of material from the infalling envelope ( see attwood et al .
disc fragmentation is a robust mechanism for the formation of brown dwarfs , as well as planetary - mass objects and low - mass hydrogen - burning stars .
it explains successfully properties that are not satisfactorily explained by other formation mechanisms , for example the brown dwarf desert and the statistics of low - mass binary systems .
we estimate that even if only a fraction of sun - like stars host the required massive extended discs needed for the fragmentation scenario to work , this mechanism can produce all the free floating planetary - mass objects , most of the brown dwarfs , and a significant proportion of low - mass hydrogen burning stars . | gravitational fragmentation of massive extended discs large brown dwarfs per sun - like star stamatellos et al examine consequences on properties brown dwarfs compare observations perform 13 simulations with star disc conditions compute mass spectrum radial distribution from star orbital planes masses sizes of brown dwarf discs binary properties consider 0.7 m@xmath0 star with 0.7 m@xmath0 disc extends to 400 au disc surface density drops as @xmath1 temperature drops as @xmath2 @xmath3 distance from star on disc midplane simulations in two phases until @xmath4% disc accreted onto stars within 20,000 yr ) evolution followed using sph new method capture thermal radiative effects when protostellar gas fragments method uses detailed equation of state for gas mixture of hydrogen helium effects of rotational vibrational degrees of freedom of h@xmath5 dissociation ionisation first second ionisation included opacity changes due to ice mantle melting sublimation of dust molecular h@xmath6 contributions method includes irradiation by central starevolution system followed to 200,000 yr using n - body code discs quickly fragment see fig . forming gravitationally bound objects outside 50 au conditions fragmentation right disc unstable cool fast previous simulations analytic studies whitworth & stamatellos 2006 stamatellos 2007 2008a 96 objects produced simulations half ejected into field typical outcome sun - like star with close low - mass star companion wide low - mass star companion wide brown dwarf binary companion disc around 0.7-m@xmath0 star snapshots logarithm column density for 3 instances simulation produces 4 brown dwarfs 3 low - mass hydrogen burning stars after 200,000 years 4 objects ejected field disc around 0.7-m star snapshots logarithm column density 3 instances simulation produces 4 brown dwarfs 3 low - mass hydrogen stars after 200,000 years 4 objects ejected into field disc around 0.7-m star snapshots logarithm column density 3 instances simulation produces 4 brown dwarfs 3 low - mass hydrogen starsafter 200,000 years 4 objects ejected into field.,title="fig : " mass - distribution ] objects produced by disc fragmentation ( 75% ) are brown dwarf stars masses below h - burning limit 4% planetary - mass brown dwarfs below d - burning limit rest 25% are low - mass hydrogen - burning stars than half ( 55% ) objects ejected field radial - distribution - brown - dwarf - desert ] radial distribution brown dwarf desert inner disc not cool fragmentation outside 50 au from primary star objects form closer to primary star most massive first more material become h - burning stars brown dwarfs form farther out disc dynamical interactions low - mass h - burning stars move further region close to primary star brown dwarfs move outwards ( see fig . radius mechanism of disc fragmenation provides explanation for brown dwarf desert ( marcy & butler 2000 ) model predicts frequency of brown dwarf companions to sun - like stars 8% within 100 au from star 16% within 400 au from starorbital - planes - of - companions - to - sun - like - stars ] orbital planes of companions to sun - stars + + + initially orbits of objects in disc almost coplanar with disc primary star orbital planes within @xmath7 from disc mid - plane due to 3 body interactions after 20,000 yr only 70% orbital planes within @xmath7 of disc midplane by 200,000 yr dropped to 22% coplanarity can not confirm or rule out formation by disc fragmentation properties - of - brown - dwarf - discs ] properties of brown dwarf discs most brown dwarfs form with discs discs have masses of few m@xmath8 radii of few tens of au disc masses radii consistent with observations show discs common around bds ( ejected brown dwarfs do not retain discs predict brown dwarfs companions to sun - like stars more likely to have discs than in field [ binary - properties ] ] binary properties + 13 low - mass binaries formed in simulations 27% of objects form are in binaries low - mass binary fraction of 16% comparable with low - mass binary fraction in taurus - auriga chamaeleon i field ( binaries include star - star star - brown dwarf brown dwarf -brown dwarf brown dwarf - planetary mass object binaries ( 1/13 4 binaries ejected into field close wide binaries bound to central star total mass as distance from central star increases low - mass binaries ( 55% ) have components with similar masses ( q>0.7 ) model suggests brown dwarfs companions to sun - like stars more likely in brown dwarf- brown dwarf binaries frequency 25% ) than in field 8% ) burgasser et al 2005 report binary fraction of @xmath12 for bd companions to sun - like stars @xmath13 for bds in field - floating planetary - mass objects 3 planetary - mass brown dwarfs form simulations three ejected into field masses remain below d - burning limit ejected soon after formation planetary - mass brown dwarfs in field should be outnumbered by d - burning brown dwarfs by factor of @xmath14 caveats simulations correspond to one disc star parameters disc density temperature profiles star mass disc mass radius results need for wider initial conditions account for disc formation around sun - like star disc growth due to accretion material from infalling envelope attwood et al disc fragmentation for formation brown dwarfs planetary - mass objects low - mass hydrogen - burning stars explains properties other brown dwarf desert statistics low - mass binary systems fraction of sun - like stars host massive extended discs for fragmentation can produce free floating planetary - mass objects brown dwarfs significant low - mass hydrogen burning stars | 0.357662 | 2,430 | 1,176 | , Saving $0.1 in GPT-4. |
177,218 | the possibility of improving the description of the semi - inclusive deep inelastic electron scattering off polarized @xmath0he , that provides information on the neutron single spin asymmetries , is illustrated .
in particular , the analysis at finite momentum transfers in a poincar covariant framework is outlined and a generalized eikonal approach to include final state interaction is presented .
= 1 example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore | as it is well known , the quark helicity takes into account at most one third of the proton spin .
this motivates the great effort on both experimental and theoretical sides to accurately determine the contributions from the quark orbital angular momentum ( @xmath1 ) and from the gluons .
analogous efforts are carried out to study the neutron . in view of this , the investigations of reactions like the electron semi - inclusive deep inelastic scattering ( sidis ) by a polarized @xmath0he target play a very relevant role @xcite . such processes yield information on the so - called transverse - momentum distributions ( tmd ) of a polarized quark inside a polarized neutron , such that an unprecedented amount of details on the quark dynamics inside the neutron can be achieved . our aim @xcite is to extend a previous analysis of sidis by polarized @xmath0he target @xcite , that addressed the extraction of the neutron single - spin asymmetries ( ssas ) , within a plane wave impulse approximation ( pwia ) framework and in the bjorken limit .
one can improve such an approach by : i ) dealing with the relativistic effects through a poincar covariant description of the nuclear dynamics ( see , e.g. , @xcite ) and ii ) taking into account the final state interaction ( fsi ) by adopting @xcite the so - called generalized eikonal approach ( gea ) ( see , e.g. @xcite ) .
a polarized @xmath0he nucleus is an ideal target to study the neutron , since at a 90% level it is equivalent to a polarized neutron . for disentangling the nucleon structure from the dynamical nuclear effects
, one can adopt an approach based on the spin - dependent spectral function of @xmath0he , @xmath2 , ( see , e.g. @xcite ) that yields the probability distribution to find a nucleon with given missing energy , three - momentum and polarization inside the nucleus . by using this formalism
, one can safely extract @xcite the neutron longitudinal asymmetry , @xmath3 , from the corresponding @xmath0he observable , @xmath4 , obtained from the reaction @xmath5 in dis regime , i.e. ( a^exp_3 - 2 p_p f_p a^exp_p ) /(p_n
f_n ) [ dis]with @xmath6 the neutron ( proton ) effective polarization inside the polarized @xmath0he , and @xmath7 , the dilution factor
. realistic values of @xmath8 and @xmath9 are @xmath10 , @xmath11 ( see , e.g. , @xcite ) . in @xcite ,
an analogous extraction was applied to the ssa of a transversely polarized @xmath12 target , obtained from the process @xmath13 , in order to obtain the ssa of a transversely polarized neutron . in pwia and adopting the bjorken limit , the ssas of @xmath14 are a convolution of @xmath15 , and the nucleon ssas , that in turn are convolutions of suitable tmds and fragmentation functions , phenomenologically describing the hadronization of the hit quark . to improve the previous description , taking into account relativistic effects in the actual experimental kinematics and developing a poincar covariant framework for analyzing the ssa of @xmath0he , one can adopt the light - front ( lf ) relativistic hamiltonian dynamics ( rhd ) , combined with the bakamjian - thomas construction of the poincar generators .
then , one gets the following expression for the @xmath0he hadronic tensor @xcite ( for details see @xcite ) _ , _ .. _ ^min_s^^max_s d _ s _ m^2_n^(m_x - m_s)^2 dm^2_f _ _ l^_u d^2 ( 1-)(2)^3 _ p^m_^p^m_d p_sin ( p^++q^+- h^+ ) w^_(,q , h , p ) p^_(,_s , s_he )
where @xmath16 , @xmath17 is the hadronic nucleon tensor and @xmath18 the lf spin - dependent spectral function , related to the instant - form spectral function , @xmath19 through the unitary melosh rotations , @xmath20}}$ ] , as follows _ _ 1 _1 d^1 2 [ r_m^(k)]__1 s^__1_1(*k*,_s , s_he ) d^1 2 [ r_m ( k)]__1 [ sfrhd ] in our approach , @xmath21 can be approximated by @xmath22 , obtained within a non relativistic framework , since the constraints imposed by the poincar covariance can be fully satisfied with our assumptions , namely the lf rhd framework completed by the bakamjian - thomas construction @xcite .
the second ingredient to be added is gea ( see , e.g. , @xcite and references therein quoted ) , devised for taking into account the fsi effects .
in particular , fsi effects to be considered are due to the propagation of the debris , formed after the @xmath23 absorption by a target quark , and the subsequent hadronization , both of them influenced by the presence of a fully - interacting @xmath24 spectator system .
clearly , such fsi effects represent a very complicated many - body problem and their evaluation from first principles is a hard challenge .
therefore it is necessary to develop model approaches for evaluating their impact on the extraction of neutron ssas .
the approximation based on gea has been recently applied for describing the spectator sidis by a polarized @xmath0he target @xcite .
the key quantities are the intrinsic overlaps defined as follows ^_a(fsi)_(*p*_n , e ) = ( ^*_a-1 ) _ gl \ { _ ^*_a-1,,*p*_n}| s_a , _ a s_a , _ a| _ gl\ { _ ^*_a-1,,*p*_n } ( e+ m_a - m_n - m^*_a-1-t_a-1 ) [ overlaps ] where i ) @xmath25 is the usual missing energy @xmath26 , with @xmath27 ( @xmath28 ) the energy ( state density ) of the spectator system and @xmath29 the binding energy of the target nucleus , ii ) @xmath30 the three - momentum of the struck nucleon and iii ) @xmath31 represents the debris - nucleon eikonal scattering s - matrix , that depends upon the relative coordinates only , and it is given by @xmath32 $ ] , where @xmath33 is the perpendicular ( parallel ) component of @xmath34 ( remind that @xmath35 ) , with respect to the direction of the propagation of the debris @xmath36 . in the dis limit @xmath37 ( @xmath38 is the three - momentum transfer ) and the eikonal s - matrix is defined with respect to @xmath39 . the profile function , @xmath40 , is @xcite ( * b*_1i , z_1i ) = , [ eikonal ] where @xmath41 , @xmath42 , and @xmath43 , with @xmath44 the effective number of pions which are produced .
indeed , the effective cross sections , @xmath45 depends also on the total energy of the debris , @xmath46 , but such a dependence is weak , if the energy is not too large and the hadronization process develops inside the nuclear environment . hence , the effective cross section can be approximated as @xmath47 @xcite .
@xmath48 successful applications of gea to unpolarized sidis can be found in refs .
@xcite . in conclusion ,
the embedding of the gea overlaps , eq .
( [ overlaps ] ) , in a poincar covariant framework can be achieved through a generalization of the relation shown in eq .
( [ sfrhd ] ) @xcite .
the impact of the fsi treatment can be appreciated by considering the following quantity , needed in the evaluation of sidis processes and strictly related to the nuclear dynamics , namely @xmath49 , viz _ ||(fsi)^12 ( * p*_n , e)=12 indeed , the evaluation was carried out for the spectator sidis , where a deuteron is detected in the final state .
figure [ figsparperp ] illustrates the effects of fsi , evaluated within the gea , by showing the absolute value of @xmath50 , for the above mentioned spectator sidis , as a function of the missing momentum , @xmath51 , in both the parallel ( @xmath52 ) , and perpendicular ( @xmath53 and @xmath54 ) , kinematics .
it is interesting to note that a wise choice of the kinematical variables can minimize the fsi effects , as already shown in the unpolarized case @xcite . as shown in fig .
[ figsparperp ] , fsi is negligible at low values of @xmath55 ( in this case one has a fast final debris , given @xmath56 ) , while it starts to be sizable for @xmath57 , where the equal sign holds for the perpendicular kinematics ( cf the right panel ) . to illustrate the influence of the fsi effects on the extraction procedures of the neutron information ,
let us consider the preliminary comparison , for a jlab12 kinematics @xcite , between the pwia values of both dilution factors and nucleon polarizations that appear in eq .
( [ dis ] ) and the corresponding calculations with fsi effects taken into account . as shown in table 1
, fsi can sizably modify the overlaps given in eq .
( [ overlaps ] ) , that are necessary for evaluating the quantities in eq .
( [ dis ] ) .
in particular , about a @xmath58 depolarization effect of the nucleons in @xmath59 is produced .
it should be pointed that , even if both polarizations and dilution factors in eq .
( [ dis ] ) are affected by the presence of fsi , fortunately their product does not .
therefore the extraction procedure , can be safely applied .
this seems a quite encouraging message , from the phenomenological point of view , and motivates the application of such an approach , combined with the the poincar covariant treatment of the nucleus tensor , to the extraction procedure of the neutron ssas from the @xmath59 ssas , to be considered elsewhere @xcite .
3 cates g. et al , e12 - 09 - 018 , jlab approved experiment , hallaweb.jlab.org/collab/pac/pac38//e12-09-018-sidis.pdf ; x. jiang et al , www.jlab.org/expprog/generated/apphalla.html ; j. arrington et al , www.jlab.org/@xmath64jinhuang/12gev/12gevlongitudinalhe3.pdf .
kaptari , a. del dotto , e. pace , g. salm and s. scopetta : distorted spin - dependent spectral function of an a=3 nucleus and sidis processes .
c. ciofi degli atti and l. p. kaptari : sidis off few - nucleon systems : tagging the emc effect and hadronization mechanisms with detection of slow recoiling nuclei .
* c 83 * , 044602 ( 2011 ) . c. ciofi degli atti ,
e. pace and g. salm : spin dependent spectral function of @xmath0he and the asymmetry in the process @xmath65 .
phys . rev .
* c 46 * , 1591 ( 1992 ) . c. ciofi degli atti , e. pace , g. salm and s. scopetta : nuclear effects in deep inelastic scattering of polarized electrons off polarized @xmath0he and the neutron spin structure functions .
* c 48 * , r968 ( 1993 ) . c. ciofi degli atti , l. p. kaptari and b. z. kopeliovich .
final state interaction effects in semiinclusive dis off the deuteron .
j. a * 19 * , 145 ( 2004 ) . w. cosyn and m. sargsian .
final - state interactions in semi - inclusive deep inelastic scattering off the deuteron .
rev.c * 84 * , 014601 ( 2011 ) | quark helicity one third of proton spin motivates effort to determine contributions from quark orbital angular momentum gluons analogous efforts study neutron investigations of reactions electron semi - inclusive deep inelastic scattering by polarized @xmath0he target relevant yield information on transverse - momentum distributions of polarized quark inside neutron unprecedented details on quark dynamics inside neutron aim to extend previous analysis of sidis by polarized @xmath0he target addressed neutron single - spin asymmetries plane wave impulse approximation framework bjorken limit improve approach by dealing with relativistic effects through poincar covariant description of nuclear dynamics final state interaction adopting generalized eikonal approach polarized @xmath0he nucleus ideal target to study neutron at 90% level equivalent to polarized neutron for disentangling nucleon structure from dynamical nuclear effects approach based on spin - dependent spectral function of @xmath0he @xmath2 probability distribution to find nucleon with missing energy three - momentum polarization inside nucleususing formalism extract neutron longitudinal asymmetry @xmath3 from @xmath0he observable @xmath4 reaction @xmath5 dis regime a^exp_3 - 2 p_p f_p^exp_p ) /(p_n f_n ) @xmath6 neutron polarization inside polarized @xmath0he @xmath7 dilution factor values of @xmath8 @xmath9 are @xmath10 @xmath11 analogous extraction applied ssa transversely polarized @xmath12 target from process @xmath13 ssa polarized neutron ssas @xmath14 convolution of @xmath15 nucleon ssas convolutions of tmds fragmentation functions describing hadronization hit quark improve description relativistic effects kinematics poincar covariant framework analyzing ssa @xmath0he adopt light - front relativistic hamiltonian dynamics bakamjian - thomas construction poincar generatorsexpression @xmath0he hadronic tensor @xcite ^min_s^^max_s m^2_n^(m_x - m_s)^2 dm^2_f l^_u d^2 1-)(2)^3 p^m_^p^m_d p_sin p^++q^+- h^+ ) w^_(,q h p ) p^_(_s s_he @xmath16 @xmath17 hadronic nucleon tensor @xmath18 lf spin - dependent spectral function instant - form spectral function @xmath19 unitary melosh rotations @xmath20}} d^1 2 [ r_m^(k)]__1 s^__1_1(*k*_s s_he ) d^1 2 [ r_m k)]__1 sfrhd @xmath21 approximated by @xmath22 non relativistic framework constraints poincar covariance satisfied assumptions lf rhd framework bakamjian - thomas construction @xcite second ingredient gea fsi effects propagation debris @xmath23 absorption target quark hadronization influenced - interacting @xmath24 spectator system fsi effects complicated many - body problem evaluation hard challengenecessary develop model approaches evaluating impact extraction neutron ssas approximation gea applied describing spectator sidis polarized @xmath0he target @xcite key quantities intrinsic overlaps defined ^_a*_n = ^*_a-1 ) m_a - m_n - m^*_a-1-t_a-1 ) overlaps @xmath25 missing energy @xmath26 @xmath27 energy spectator system @xmath29 binding energy target nucleus @xmath30 three - momentum struck nucleon @xmath31 represents debris - nucleon eikonal scattering s - matrix depends relative coordinates given by @xmath32 @xmath33 perpendicular component of @xmath34 direction propagation debris @xmath37 @xmath38 three - momentum transfer eikonal s - matrix defined @xmath39 profile function @xmath40 is @xcite ( * b*_1i , z_1i ) = [ eikonal ] @xmath41 @xmath42 @xmath43 @xmath44 effective number of pions producedeffective cross sections @xmath45 depends on total energy debris @xmath46 weak if energy not large hadronization develops inside nuclear environment effective cross section approximated as @xmath47 @xcite @xmath48 successful applications of gea to unpolarized sidis in refs @xcite embedding of gea overlaps in poincar covariant framework achieved through generalization of relation in eq. sfrhd impact of fsi treatment considering quantity sidis nuclear dynamics @xmath49 (fsi)^12 ( * p*_n e)=12 evaluation for sidis deuteron detected in final state figure illustrates effects of fsi gea absolute value of @xmath50 for spectator sidis function of missing momentum @xmath51 in parallel @xmath52 and perpendicular @xmath53 kinematics wise choice of kinematical variables minimize fsi effects shown in unpolarized case @xcite in fig .fsi negligible at low values @xmath55 fast final debris @xmath56 sizable for @xmath57 equal sign holds for perpendicular kinematics right panel influence fsi effects on extraction procedures neutron information preliminary comparison jlab12 kinematics @xcite between pwia values dilution factors nucleon polarizations in eq . calculations with fsi effects table 1 fsi modify overlaps in eq . necessary for evaluating quantities @xmath58 depolarization effect nucleons in @xmath59 produced polarizations dilution factors affected by fsi product does not extraction procedure safely applied encouraging message motivates application approach poincar covariant treatment nucleus tensor to extraction procedure neutron ssas from @xmath59 ssas cates g. et al , e12 - 09 - 018 jlab approved experiment.jlab.org/collab/pac/pac38//e12-09-018-sidis.pdf ; x. jiang et al www.jlab.org/expprog/generated/apphalla.html j. arrington et al www.jlab.org/@xmath64jinhuang/12gev/12gevlongitudinalhe3.kaptari del dotto pace salm scopetta distorted spin - dependent spectral function a=3 nucleus sidis processes ciofi atti kaptari sidis few - nucleon systems emc effect hadronization mechanisms slow recoiling nuclei 83 044602 2011 ) pace salm spin dependent spectral function @xmath0he asymmetry process @xmath65 46 1591 1992 ) pace salm scopetta nuclear effects deep inelastic scattering polarized electrons @xmath0he neutron spin structure functions 48 r968 1993 ) kaptari kopeliovich final state interaction effects semiinclusive dis deuteron 19 145 ( 2004 ) cosyn sargsian final - state interactions semi - inclusive deep inelastic scattering deuteron 84 014601 ( 2011 | 0.434019 | 2,727 | 1,449 | , Saving $0.1 in GPT-4. |
177,225 | we show that the conductance of spin field effect transistors ( spinfet ) [ datta and das , appl .
phys .
lett .
, , 665 ( 1990 ) ] is affected by a single ( non - magnetic ) impurity in the transistor s channel .
the extreme sensitivity of the amplitude and phase of the transistor s conductance oscillations to the location of a single impurity in the channel is reminiscent of the phenomenon of universal conductance fluctuations in mesoscopic samples and is extremely problematic as far as device implementation is concerned . | in a seminal paper published in 1990 , datta and das @xcite proposed a gate controlled electron spin interferometer which is an analog of the standard electro - optic light modulator .
their device ( later dubbed `` spin field effect transistor '' or spinfet ) consists of a one - dimensional semiconductor channel with ferromagnetic source and drain contacts ( fig .
[ structure](a ) ) .
electrons are injected into the channel from the ferromagnetic source with a definite spin orientation , which is then controllably precessed in the channel with a gate - controlled rashba interaction @xcite , and finally sensed at the drain . at the drain end ,
the electron s transmission probability depends on the relative alignment of its spin with the drain s ( fixed ) magnetization . by controlling the angle of spin precession in the channel with a gate voltage
, one can modulate the relative spin alignment at the drain end , and hence control the source - to - drain current ( or conductance ) .
this is the principle of the spinfet .
is the exchange splitting energy in the contacts .
@xmath0 is the height of the potential barrier between the energy band bottoms of the semiconductor and the ferromagnetic electrodes .
@xmath0 takes into account the effects of the quantum confinement in the y- and z - directions .
also shown as dashed lines are the resonant energy states above @xmath0 .
peaks in the conductance of the electron spin interferometer are expected when the fermi level in the contacts lines up with the resonant states .
the barriers at the ferromagnet / semiconductor interface are modeled as simple one - dimensional delta - potentials.,width=240 ] there have been some studies of spin transport in such a device .
mireles and kirczenow @xcite carried out a study of _ ballistic _ spin transport , but they overlooked two crucial features that are always present in a real device structure .
first , there is an axial magnetic field along the channel caused by the ferromagnetic source and drain contacts which are magnetized in the same direction
. this field can be quite strong ( @xmath1 1 tesla ) @xcite and dramatically alters the dispersion relations of the subbands in the channel , causes spin mixing , and has a serious effect on spin transport .
second , there will always be a few impurities in the channel ( even if they are remote impurities ) associated with channel doping or unintentional defects .
we show that even a single ( non - magnetic ) impurity can cause spin relaxation in the presence of the axial magnetic field .
we first consider the quasi one - dimensional semiconductor channel of a spinfet in the absence of any impurities .
the channel is along the x - axis ( fig .
[ structure](a ) ) and the gate electric field is applied along the y - direction to induce a rashba spin - orbit coupling in the channel .
this system is described by the single particle effective - mass hamiltonian @xcite @xmath2 \label{hamiltonian}\end{aligned}\ ] ] where @xmath3 is the unit vector along the y - direction in fig .
[ structure](a ) and @xmath4 is the vector potential due to the axial magnetic field @xmath5 along the channel ( x - direction ) caused by the ferromagnetic contacts . in ( 1 ) , @xmath6 is the bohr magneton ( @xmath7 ) and @xmath8 is the effective land g - factor of the electron in the channel .
the quantity @xmath9 is the rashba spin - orbit coupling strength which can be varied with the gate potential .
the confining potentials along the y- and z - directions are denoted by @xmath10 and @xmath11 , with the latter being parabolic in space . in ( 1 )
, @xmath12 represents an interfacial potential barrier between the ferromagnetic contacts and the semiconducting channel .
if the contact neighborhood consists of heavily doped semiconductor material in close proximity to a metallic ferromagnet , the schottky barriers at the interface will be very narrow and electrons from the contacts can tunnel fairly easily into semiconducting channel resulting in a nearly - ohmic contact .
we model these ultra - narrow schottky barriers as delta - barriers given by : @xmath13 where @xmath14 and @xmath15 are assumed equal ( @xmath16 = @xmath17 = @xmath18 ) . in practice , the strength of the barrier depends on the ferromagnetic materials and also on the doping level in the channel .
these barriers have a beneficial effect ; they can facilitate coherent spin injection across a metallic ferromagnet and a semiconducting paramagnet interface @xcite which is crucial for the spinfet . in ( 1 )
, we have neglected a few effects for the sake of simplicity .
we have neglected the normal elliott - yafet interaction @xcite because it is weak in quasi one - dimensional structures ( where elastic scattering is strongly suppressed ) .
we have also neglected the dresselhaus interaction @xcite since it does not relax spin when the initial spin polarization is along the axis of the wire @xcite .
the dresselhaus interaction can however be easily included in the hamiltonian and is left for future work .
the choice of the landau gauge @xmath4 = ( 0 , -bz , 0 ) allows us to decouple the y - component of the hamiltonian in ( 1 ) from the x - z component .
furthermore , in the absence of any impurity scattering potential ( @xmath19 ) , the hamiltonian in the semiconducting channel is translationally invariant in the x - direction and the wavevector @xmath20 is a good quantum number .
the eigenstates of the system can then be determined using plane waves traveling in the x - direction @xcite .
the two - dimensional hamiltonian in the plane of the channel ( x - z plane ) is therefore given by @xmath21 where @xmath22 is the curvature of the confining potential in the z - direction , @xmath23 , and @xmath0 is the potential barrier between the ferromagnet and semiconductor .
we assume that @xmath0 includes the effects of the quantum confinement in the y - direction ( fig.1(a ) ) .
to model localized non - magnetic impurities ( i.e. , which do not by themselves flip the spin ) we use a standard model of delta - scatterers @xmath24 for an impurity located at a distance @xmath25 from the left ferromagnet / semiconductor interface with scattering strength @xmath26 ( assumed to be spin independent ) . in our numerical examples
, we consider the case of both attractive ( @xmath26 negative ) and repulsive ( @xmath26 positive ) impurities . while ( 1 ) and ( 3 ) represent a ballistic channel with no scattering , addition of the scattering potential in ( 4 ) to ( 1 ) or ( 3 ) will result in a hamiltonian describing a weakly disordered channel in which impurity scattering takes place .
the eigenstates of this ( spin - dependent ) hamiltonian can then be found using a transfer matrix technique to extract the electron wavefunction . from this wavefunction
, we can calculate the ( spin - dependent ) transmission probability through the channel and ultimately the ( spin - dependent ) channel conductance .
the details of the calculations have been presented elsewhere @xcite .
the linear response conductance of the spin interferometer ( for injection from either the + x or -x polarized bands in the left contact ) is found from the landauer formula .
we model the ferromagnetic contacts by the stoner - wohlfarth model ( fig.1.(b ) ) .
the + x - polarized spin ( majority carrier ) and -x - polarized spin ( minority carrier ) band bottoms are offset by an exchange splitting energy @xmath27 .
the solutions of the schrdinger equation for injection from minority and majority spins from the left magnetic contact then be written throughout the entire device leading to the corresponding transmission probabilities and the conductance of the interferometer based on the landauer formula ( see @xcite ) .
the strength of the barrier at the ferromagnet / semiconductor interface is characterized by the following parameter @xmath28 typical values of @xmath29 vary in the range of 0 to 2 @xcite . using @xmath30 and @xmath31 = 1.05x@xmath32 @xmath33 , we get a barrier strength @xmath18 = 16 ev@xmath34 for z = 2 .
we consider a spin interferometer consisting of a quasi one - dimensional inas channel between two ferromagnetic contacts .
the electrostatic potential in the z - direction is assumed to be harmonic ( with @xmath35 = 10 mev in ( 3 ) .
a zeeman splitting energy of 0.34 mev is used in the semiconductor channel assuming a magnetic field b = 1 tesla along the channel .
this corresponds to a @xmath36 factor of 3 and an electron effective mass @xmath37 which is typical of inas - based channels @xcite .
the fermi level @xmath38 and the exchange splitting energy @xmath39 in the ferromagnetic contacts are set equal to 4.2 and 3.46 ev , respectively @xcite .
the rashba spin - orbit coupling strength @xmath40 is typically derived from low - temperature magnetoresistance measurements ( shubnikov - de haas oscillations ) in 2deg created at the interface of semiconductor heterostructures @xcite . to date , the largest reported experimental values of the rashba spin - orbit coupling strength @xmath40 has been found in inas - based semiconductor heterojunctions . for a normal hemt @xmath41 heterojunction , sato et al .
have reported variation of @xmath40 from 30- to 15 @xmath42 ev - m when the external gate voltage is swept from 0 to -6 v ( the total electron concentration in the 2deg is found to be reduced from 5- to 4.5@xmath43 over the same range of bias ) .
tuning the gate voltage varies both the potential energy barrier @xmath0 and the rashba spin - orbit coupling strength @xmath9 . both of these variations lead to distinct types of conductance oscillations .
the variation of @xmath0 causes the fermi - level in the channel to sweep through the resonant energies in the channel , causing the conductance to oscillate .
these are known as ramsauer oscillations and have been examined by us in detail in @xcite .
the variation of @xmath9 , on the other hand , causes spin precession in the channel leading to the type of conductance oscillation which is the basis of the spin interferometer , as originally visualized by datta and das @xcite . in @xcite
we found that the ramsauer oscillations are much stronger and can mask the oscillations due to spin precession , unless the structure is designed with particular care to eliminate ( or reduce ) the ramsauer oscillations . in the numerical examples below
, we assumed that the rashba spin - orbit coupling strength @xmath40 is constant ( i.e. , independent of the gate potential ) and equal to the maximum reported value of 30 @xmath42 ev - m ( best case scenario ) .
we also used a value of z = 0.25 corresponding to a value of @xmath14 and @xmath17 in ( 2 ) equal to 2 ev-@xmath34 .
the scattering strength of the single impurity in the channel was set equal to @xmath44 = @xmath45 0.25 ev-@xmath34 ( the plus and minus sign corresponding to repulsive and attractive impurity , respectively ) .
first , we consider the case of a single repulsive scatterer ( impurity ) whose location is varied from the left to the rigth side of the channel in steps of 10 @xmath34 .
figure 2 shows that variation of the conductance with the impurity location for three different values of @xmath46 in the channel , i.e , gate potential on the spinfet .
it is seen that the condutance of the interferometer is a strong function of the impurity location .
figure 2 shows that the condutance modulation ( difference between maximum and minimum conductance values ) of the interferometer is about 0.9 , 0.21 , and 0.2 @xmath47 , for @xmath0 equal to 4.191 , 4.188 , and 4.185 ev , respectively .
the same calculation were repeated for the case of an attractive impurity with the same magnitude of the scattering strength and the conductance modulation as a function of the impurity location are shown in fig.3 . in this case , the condutance modulation ( difference between maximum and minimum conductance values ) of the interferometer is about 0.76 , 0.16 , and 0.17@xmath47 , for @xmath0 equal to 4.191 , 4.188 , and 4.185 ev , respectively .
in this paper , we have shown how the conductance of gate controlled spin interferometers proposed in @xcite is strongly dependent on the location of a single impurity in the semiconducting channel ( figs . 2 and 3 ) .
the extreme sensitivity of the amplitude and phase of conductance oscillations of a spinfet to impurity location is reminiscent of the phenomenon of universal conductance fluctuations of mesoscopic samples @xcite .
this will hinder practical applications of electron spin interferometers since it will lead to such problems as large threshold variability , random device characteristics , and general irreproducibility .
j. nitta , t. akazaki , h. takayanagi , and t. enoki , phys .
lett . , * 78 * , 1335 ( 1997 ) ; th .
schpers , j. nitta , h.b .
heersche , and h. takayanagi , phys .
b , * 64 * , 125314 ( 2001 ) ; y. sato , t. kita , s. gozu and s. yamada , j. appl .
phys . , * 89 * , 8017 ( 2001 ) ; y. sato , s. gozu , t. kita and s. yamada , physica e , * 12 * , 399 ( 2002 ) .
long . the strength of the delta - scatterer @xmath26 is set equal to 0.25 ev@xmath34 .
the three curves correspond to different values of the potential barrier height @xmath0 which is controlled by the gate voltage .
case 1 , 2 , and 3 corresponds to a value of @xmath0 equal to 4.188 , 4.185 , and 4.191 ev , respectively .
the calculations are for absolute zero temperature.,width=240 ] | paper 1990 datta das @xcite proposed gate controlled electron spin interferometer analog standard electro - optic light modulator device dubbed spin field effect transistor spinfet consists one - dimensional semiconductor channel ferromagnetic source drain contacts fig structure electrons injected channel from source definite spin orientation precessed channel with gate - controlled interaction sensed at drain electron s transmission probability depends on alignment spin with drain s magnetization controlling angle spin precession with gate voltage spin alignment drain end source - to - drain current conductance principle spinfet exchange splitting energy in contacts @xmath0 height potential barrier between energy band bottoms semiconductor ferromagnetic electrodes quantum confinement y- z - directions dashed lines resonant energy states above @xmath0 peaks in conductance expected when fermi level contacts lines up with resonant states barriers at ferromagnet / semiconductor interface modeled as one - dimensional delta - potentials studies of spin transport mireles kirczenow overlooked two features axial magnetic field along channel caused by ferromagnetic source drain contacts magnetized same directionfield strong @xmath1 1 tesla ) @xcite alters dispersion relations subbands causes spin mixing spin transport impurities in channel associated with channel doping unintentional defects single non - magnetic impurity cause spin relaxation axial magnetic field consider quasi one - dimensional semiconductor channel spinfet impurities channel along x - axis gate electric field applied y - direction rashba spin - orbit coupling system described by single particle effective - mass hamiltonian @xcite @xmath2 @xmath3 unit vector y - direction @xmath4 vector potential due to axial magnetic field @xmath5 channel ferromagnetic contacts @xmath6 bohr magneton @xmath7 ) @xmath8 effective land g - factor electron @xmath9 rashba spin - orbit coupling strength varied with gate potential confining potentials y- z - directions by @xmath10 @xmath11 latter parabolic in space @xmath12 interfacial potential barrier between ferromagnetic contacts semiconducting channelcontact neighborhood heavily doped semiconductor material metallic ferromagnet schottky barriers interface narrow electrons tunnel into semiconducting channel nearly - ohmic contact model - narrow barriers as delta - barriers @xmath13 @xmath14 @xmath15 assumed equal @xmath16 @xmath18 strength barrier depends on ferromagnetic materials doping level channel barriers facilitate coherent spin injection across metallic ferromagnet semiconducting paramagnet interface crucial for spinfet neglected effects normal elliott - yafet interaction weak in quasi one - dimensional structures dresselhaus interaction not relax spin initial spin polarization along axis wire dresselhaus interaction included in hamiltonian left for future work landau gauge @xmath4 = ( 0 , -bz , 0 ) decouple y - component hamiltonian from x - z component impurity scattering potential @xmath19 hamiltonian in semiconducting channel translationally invariant in x - direction wavevector @xmath20 good quantum number eigenstates system determined using plane waves x - direction @xcitetwo - dimensional hamiltonian in channel x - z plane given by @xmath21 @xmath22 curvature confining potential z - direction @xmath23 @xmath0 potential barrier between ferromagnet semiconductor @xmath0 includes effects quantum confinement y - direction fig.1(a ) model localized non - magnetic impurities not flip spin use standard model delta - scatterers @xmath24 impurity distance @xmath25 from left ferromagnet / semiconductor interface scattering strength @xmath26 spin independent attractive negative and repulsive positive impurities ( 1 ) ( 3 ) represent ballistic channel no scattering scattering potential in ( 4 ) to 1 ) or 3 ) weakly disordered channel impurity scattering eigenstates of spin - dependent hamiltonian found using transfer matrix technique electron wavefunction calculate spin - dependent transmission probability channel conductance details calculations presented linear response conductance of spin interferometer -x polarized bands found from landauer formula model ferromagnetic contacts by stoner - wohlfarth model ( fig.1.(b ) + band bottoms offset by exchange splitting energy @xmath27solutions schrdinger equation injection minority majority spins left magnetic contact written device transmission probabilities conductance interferometer landauer formula strength barrier ferromagnet semiconductor interface parameter @xmath28 values @xmath29 vary 0 to 2 @xmath30 @xmath31 = 1.05x barrier strength @xmath18 = 16 ev@xmath34 z = 2 spin interferometer quasi one - inas channel ferromagnetic contacts electrostatic potential z - direction harmonic @xmath35 = 10 mev zeeman splitting energy 0.34 mev semiconductor channel magnetic field b = 1 tesla @xmath36 factor 3 electron effective mass @xmath37 typical inas - based channels fermi level @xmath38 exchange splitting energy ferromagnetic contacts equal 4.2 3.46 ev rashba spin - orbit coupling strength @xmath40 derived low - temperature magnetoresistance measurements oscillations semiconductor heterostructures largest experimental values spin orbit coupling strength inas - based semiconductor heterojunctions normal @xmath41 heterojunctionreported variation @xmath40 30- to 15 @xmath42 ev - m external gate voltage swept 0 to -6 v total electron concentration 2deg reduced 5- to 4.5@xmath43 same bias tuning gate voltage varies potential energy barrier @xmath0 rashba spin - orbit coupling strength @xmath9 variations lead to conductance oscillations variation @xmath0 causes fermi - level sweep through resonant energies conductance known ramsauer oscillations examined @xcite variation @xmath9 causes spin precession conductance oscillation basis spin interferometer @xcite ramsauer oscillations stronger mask oscillations spin precession unless structure designed eliminate reduce ramsauer oscillations assumed rashba spin - orbit coupling strength @xmath40 constant independent gate potential equal maximum reported value 30 @xmath42 ev - m used value z = 0.25 value @xmath14 and @xmath17 2 equal to 2 ev-@xmath34 scattering strength single impurity channel equal to @xmath44 = @xmath45 0.25 ev-@xmath34 plus minus consider case single repulsive scatterer location varied left to rigth side channel steps 10 @xmath34 figure 2 shows variation conductance with impurity location for three values @xmath46 channel gate potential spinfet condutance interferometer function impurity location condutance modulation about 0.9 0.21 0.2 @xmath47 @xmath0 equal to 4.191 4.188 4.185 ev same calculation repeated for attractive impurity same magnitude scattering strength conductance modulation impurity location shown in.3 0.76 0.16 0.17@xmath47 @xmath0 equal to 4.191 4.188 4.185 ev conductance of gate controlled spin interferometers in @xcite dependent on location single impurity in semiconducting channel ( figs . 2 3 ) extreme sensitivity of amplitude phase conductance oscillations spinfet to impurity location universal conductance fluctuations mesoscopic samples @xcite applications electron spin interferometers problems large threshold variability random device characteristics irreproducibility j. nitta t. akazaki h takayanagi t enoki phys . 78 , 1335 ( 1997 ) ; schpers nitta heersche takayanagi 64 125314 ( 2001 )sato t kita s gozu yamada appl.. 89, 8017 ( 2001 ) sato gozu yamada physica 12 * , 399 ( 2002 ) strength delta - scatterer @xmath26 equal 0.25 ev@xmath34 three curves potential barrier height @xmath0 controlled gate voltage case 1 2 3 @xmath0 equal 4.188 4.185 4.191 ev calculations absolute zero temperature | 0.504179 | 3,246 | 1,696 | , Saving $0.1 in GPT-4. |
177,227 | for the stationary advection - diffusion problem the standard continuous galerkin method is unstable without some additional control on the mesh or method .
the interior penalty discontinuous galerkin method is stable but at the expense of an increased number of degrees of freedom .
the hybrid method proposed in @xcite combines the computational complexity of the continuous method with the stability of the discontinuous method without a significant increase in degrees of freedom .
we discuss the implementation of this method using the finite element library ` deal.ii ` and present some numerical experiments . | we consider the advection - diffusion equation @xmath0 with @xmath1 , @xmath2 , @xmath3 and @xmath4 . for simplicity
we assume the region @xmath5 is polygonal .
we also assume @xmath6 and then we have a weak solution @xmath7 .
it is well known that this problem can exhibit boundary or internal layers in the convection dominated regime and that for the standard continuous galerkin ( cg ) formulation these layers cause non - physical oscillations in the numerical solution .
several adaptations to the cg method are effective but space does not allow their discussion here . we refer readers to @xcite for a full description of these approaches .
discontinuous galerkin ( dg ) methods also offer a stable approach for approximating this problem .
however the number of degrees of freedom required for dg methods is in general considerably larger than for cg methods .
we describe an alternative approach also studied in @xcite .
a dg method is applied on the layers and a cg method away from the layers .
we call this approach the continuous - discontinuous galerkin ( cdg ) method .
the hypothesis is that provided the layers are entirely contained in the dg region the instability they cause will not propagate to the cg region .
note that in our formulation there are no transmission conditions at the join of the two regions .
here we present the cdg method and discuss its implementation using the ` deal.ii ` finite element library .
we additionally provide some numerical experiments to highlight the performance of the method .
assume that we can identify a decomposition of @xmath8 where it is appropriate to apply the cg and dg methods respectively .
we do not consider specific procedures to achieve this here , but generally it will be that we wish all boundary and internal layers to be within @xmath9 . identifying these regions can be done a priori in some cases or a posteriori based on the solution of a dg finite element method . consider a triangulation @xmath10 of @xmath5 which is split into two regions @xmath11 and @xmath12 where we will apply the cg and dg methods respectively . for simplicity
we assume that the regions @xmath11 and @xmath12 are aligned with the regions @xmath13 and @xmath9 and the set @xmath14 contains edges which lie in the intersection of the two regions .
call the mesh skeleton @xmath15 and the internal skeleton @xmath16 .
define @xmath17 as the union of boundary edges and the inflow and outflow boundaries by @xmath18 where @xmath19 is the outward pointing normal .
define @xmath20 ( resp .
@xmath21 ) to be the intersection of @xmath17 with @xmath11 ( resp .
@xmath12 ) . by convention
we say that the edges of @xmath14 are part of the discontinuous skeleton @xmath22 and @xmath23 . with this convention
there is potentially a discontinuity of the numerical solution at @xmath14 .
elements of the mesh are denoted @xmath24 , edges ( resp .
faces in 3d ) by @xmath25 and denote by @xmath26 and @xmath27 the diameter of an element and an edge , defined in the usual way . the jump @xmath28 and average @xmath29 of a scalar or vector function on the edges in @xmath15 are defined as in , e.g. , @xcite .
define the cdg space to be @xmath30 where @xmath31 is the space of polynomials of degree at most @xmath32 supported on @xmath24 .
this is equivalent to applying the usual cg space on @xmath13 and a dg space on @xmath9 .
we may now define the interior penalty cdg method : find @xmath33 such that for all @xmath34 @xmath35 where @xmath36 \nonumber \\ & \qquad + \sum_{{\ensuremath{e \in { \ensuremath{{\mathcal{e}_h}}}}}}\left [ \int_e \sigma \frac{{\varepsilon}}{h_e } { \ensuremath{{\ensuremath{\llbracket u_h \rrbracket } } } } { \ensuremath{\cdotp}}{\ensuremath{{\ensuremath{\llbracket v_h \rrbracket } } } } - \int_e \left ( { \ensuremath{{\ensuremath{\ { \ ! \ ! \ { { \varepsilon}\nabla u_h \ } \ ! \ ! \ } } } } } { \ensuremath{\cdotp}}{\ensuremath{{\ensuremath{\llbracket v_h \rrbracket } } } } + { \vartheta}{\ensuremath{{\ensuremath{\ { \ ! \ ! \ { { \varepsilon}\nabla v_h \ } \ ! \ ! \ } } } } } { \ensuremath{\cdotp}}{\ensuremath{{\ensuremath{\llbracket u_h \rrbracket}}}}\right ) \right ] \nonumber \\ b_a(u_h , v_h ) & = \sum_{{\ensuremath{e \in { \ensuremath{{\mathcal{e}_h^o}}}}}}\int_e b { \ensuremath{\cdotp}}{\ensuremath{{\ensuremath{\llbracket v_h \rrbracket } } } } u_h^- + \sum_{e \in { \ensuremath{\gamma^{\text{out } } } } } \int_e ( b { \ensuremath{\cdotp}}n ) u_h v_h \nonumber \end{split}\ ] ] and @xmath37 - \sum_{e \in { \ensuremath{\gamma^{\text{in } } } } } \int_e ( b { \ensuremath{\cdotp}}n)v_hg . \nonumber\ ] ] here @xmath38 is the penalization parameter and @xmath39 .
note that through the definition of @xmath40 the edge terms are zero on @xmath41 and the method reduces to the standard cg fem . if we take @xmath42 , i.e. , the entire triangulation as discontinuous , we get the interior penalty ( ip ) family of dg fems ( see , e.g. , @xcite ) .
the work of @xcite shows that the cg method is the limit of the dg method as @xmath43 .
a reasonable hypothesis is that the solution to the cdg method is the limit of the solutions to the dg method as the penalty parameter @xmath43 on @xmath44 , i.e. , super penalising the edges in @xmath41 .
call @xmath45 and @xmath46 the penalty parameters for edges in @xmath41 and @xmath22 respectively .
call the numerical solution for the cdg problem @xmath47 .
the solution to the pure dg problem on the same mesh is denoted @xmath48 where @xmath49 is the usual piecewise discontinuous polynomial space on @xmath10
. then we have [ thm : convergence ] the dg solution converges to the cdg solution of - as @xmath50 , i.e. , @xmath51 we do not prove this result here but direct readers to @xcite for a full discussion .
although this result does not imply stability of the cdg method ( indeed , for the case where the @xmath13 region is taken to be the whole of @xmath5 it shows that the cdg method has the same problems as the cg method ) , but does indicate that investigation of the cdg method as an intermediate stage between cg and dg is justified .
hence , it aids into building an understanding of the convergence and stability properties of the cdg method , based on what is known for cg and dg .
this , in turn , is of interest , as the cdg method offers substantial reduction in the degrees of freedom of the method compared to dg .
the cdg method poses several difficulties in implementation .
one approach is to use the super penalty result of theorem [ thm : convergence ] to get a good approximation to the cdg solution .
however this will give a method with the same number of degrees of freedom as dg .
we therefore present an approach to implement the cdg method with the appropriate finite element structure .
we discuss this approach with particular reference to the ` deal.ii ` finite element library @xcite .
this is an open source library designed to streamline the creation of finite element codes and give straightforward access to algorithms and data structures .
we also present some numerical experiments .
the main difficulty in implementing a cdg method in ` deal.ii ` is the understandable lack of a native cdg element type . in order to assign degrees of freedom to a mesh in ` deal.ii ` the code must be initialised with a ` triangulation ` and then instructed to use a particular finite element basis to place the degrees of freedom .
although it is possible to initialise a ` triangulation ` with the dg and cg regions set via the ` material_id ` flag , no appropriate element exists . in the existing ` deal.ii ` framework it would be difficult to code an element with the appropriate properties . a far more robust approach is to use the existing capabilities of the library and therefore allow access to other features of ` deal.ii ` .
for instance without the correct distribution of degrees of freedom the resulting sparsity pattern of the finite element matrix would be suboptimal , i.e. , containing more entries than required by the theory and therefore reducing the benefit of shrinking the number of degrees of freedom relative to a dg method .
the ` deal.ii ` library has the capability to handle problems with multiple equations applied to a single mesh such as the case of a elastic solid fluid interaction problem . in our case
we wish to apply different methods to the same equation on different regions of the mesh , which is conceptually the same problem in the ` deal.ii ` framework .
in addition we will use the @xmath52 capability of the library .
the ` deal.ii ` library has the capability to create collections of finite elements , ` hp::fecollection ` . here
multiple finite elements are grouped into one data structure .
as the syntax suggests the usual use is for @xmath52 refinement to create a set of finite elements of the same type ( e.g. , scalar lagrange elements ` fe_q ` or discontinuous elements ` fe_dgq ` ) of varying degree .
unfortunately it is not sufficient to create a ` hp::fecollection ` of cg and dg elements as the interface between the two regions will still be undefined . in order to create an admissible collection of finite elements we use ` fe_nothing ` .
this is a finite element type in ` deal.ii ` with zero degrees of freedom . using the @xmath53 class we create two vector - valued finite element types @xmath54 and @xmath55 and
combine them in a ` hp::fecollection ` .
we apply the first @xmath53 on the cg region , and the second on the dg region .
now when we create a ` triangulation ` initialised with the location of cg and dg elements the degrees of freedom can be correctly distributed according to the finite element defined by ` hp::fecollection ` . when assembling the matrix for the finite element method we need only be careful that we are using the correct element of ` hp::fecollection ` and the correct part of @xmath53 .
the most difficult case is on the boundary @xmath14 where from a dg element we must evaluate the contribution from the neighbouring cg element ( note that in the cdg method a jump is permissible on @xmath14 ) .
if we implement the cdg method in ` deal.ii ` in this way we create two solutions : one for the ` fe_q`-`fe_nothing ` component and another for the ` fe_nothing ` -`fe_dgq ` component .
consider a domain @xmath56 in @xmath57 , @xmath58 and @xmath59 .
the dirichlet boundary conditions and the forcing function @xmath60 are chosen so that the analytical solution is @xmath61 this solution exhibits an exponential boundary layer along @xmath62 and @xmath63 of width @xmath64 , @xmath65 .
we solve the finite element problem on a 1024 element grid and fix @xmath66 .
this is larger than is required for stability ( see example 1 below ) but shows the behaviour of ` fe_nothing ` more clearly .
we show each of the components of ` fe_system ` and the combined solution . for comparison we also show the dg finite element solution for the same problem . one advantage of following the
` deal.ii ` framework is that the data structures will allow the implementation of @xmath52 methods .
in fact we can envisage the implementation of a @xmath67 method where at each refinement there is the possibility to change the mesh size , polynomial degree or the element type .
we propose no specific scheme here but simply remark that implementing a @xmath67 method is relatively straightforward with the ` fe_nothing ` approach .
we present two numerical experiments highlighting the performance of the cdg method .
both examples present layers when @xmath68 is small enough . in each case
we fix the region where the continuous method is to be applied then vary @xmath68 .
this causes the layer to steepen . in the advection dominated regime ,
i.e. , @xmath68 large and no steep layer present , we see the cdg solution approximates the true solution well . as we make @xmath68 smaller the layer forms and
extends into the continuous region .
as @xmath68 becomes smaller still the layer leaves the continuous region and the performance of the dg and cdg method is indistinguishable . in each experiment
we pick the @xmath13 and @xmath9 regions so that with the given refinement the region @xmath12 consists of exactly one layer of elements and coincides with @xmath9 .
consider again the problem with true solution presented above .
we solve the finite element problem on a 1024 element grid and fix @xmath69 so exactly one row of elements is in @xmath9 .
as we vary @xmath68 the layer sharpens and moves entirely into the dg region .
as we can see from figure [ fig : example1a ] before the layer has formed the two methods perform well .
as the layer begins to form with decreasing @xmath68 it is not entirely contained in the discontinuous region and the error peaks . as the layer sharpens further it
is entirely contained in the discontinuous region and the difference between the two solutions becomes negligible .
now we look at a problem with an internal layer .
let the advection coefficient be given by @xmath70 and pick the boundary conditions and right hand side @xmath60 so that the true solution is @xmath71 where @xmath72 is the error function defined by @xmath73 in figure [ fig : example2a ] we notice same the same behaviour as in example 1 .
if the layer exists it must be contained within the discontinuous region for the two methods to perform equivalently . in figure
[ fig : example2b ] we can see the cdg solution for various @xmath68 with the oscillations clearly visible when @xmath78 . when the layer is sharpened , the oscillations disappear . | consider advection - diffusion equation @xmath0 with @xmath1 assume region @xmath5 polygonal assume @xmath6 weak solution @xmath7 problem boundary internal layers in convection dominated regime standard continuous galerkin cg ) formulation layers cause non - physical oscillations in numerical solution adaptations to cg method effective discussion refer to @xcite for full description discontinuous galerkin ( dg ) methods offer stable approach degrees of freedom required for dg methods larger than cg methods alternative approach in @xcite dg method applied on layers cg method away from layers continuous - discontinuous galerkin ( cdg ) method layers contained in dg region instability propagate to cg region no transmission conditions at join of two regions present cdg method discuss implementation using ` deal.ii ` finite element library provide numerical experiments performance identify decomposition of @xmath8 where apply cg and dg methods procedures wish all boundary internal layers within @xmath9 identifying regions a priori or a posteriori based on solution dg finite element methodtriangulation @xmath10 of @xmath5 split into regions @xmath11 apply cg dg methods regions aligned with @xmath13 @xmath9 @xmath14 contains edges intersection regions mesh skeleton @xmath15 internal skeleton @xmath16 define @xmath17 union of boundary edges inflow outflow boundaries @xmath18 @xmath19 outward pointing normal @xmath20 intersection of @xmath17 with @xmath11 edges of @xmath14 part of discontinuous skeleton @xmath22 @xmath23 discontinuity numerical solution at @xmath14 elements mesh denoted @xmath24 edges @xmath25 @xmath26 @xmath27 diameter element edge jump @xmath28 average @xmath29 of scalar vector function on edges in @xmath15 defined @xcite cdg space @xmath30 @xmath31 space of polynomials of degree @xmath32 @xmath24 equivalent to applying cg space on @xmath13 dg space on @xmath9define interior penalty cdg method find @xmath33 @xmath34 @xmath35 @xmath36 \nonumber \qquad + \sum_{{\ensuremath{e\mathcal{e}_h}}}}}}\left\sigma \frac{{\varepsilon}}{h_e \ensuremath{{{\llbracket u_h \rrbracket\ensuremath{\cdotp}}\llbracket v_h \rrbracket\int_e \left \ensuremath{{\varepsilon}\nabla u_h\cdotp}}\llbracket v_h \rrbracket\vartheta}\ensuremath{{\varepsilon}\nabla v_h\ensuremath{\cdotp}}\llbracket u_h \rrbracket}}}}\right \nonumber \\ b_a(u_h , v_h ) \sum_{{\ensuremath{e \in\ensuremath{{\mathcal{e_h^o}}}}}}\int_e \ensuremath{\cdotp}}\ensuremath{{{\llbracket v_h \rrbracket u_h^- + \sum_{e \in \ensuremath{\gamma^{\text{out }\int_e b \ensuremath{\cdotp}}n ) u_h v_h\nonumber\end{split} @xmath37 - \sum_{e \ensuremath{\gamma^{\text{in }_e b\ensuremath{\cdotp}}n)v_hg\nonumber @xmath38 penalization parameter @xmath39 definition @xmath40 edge terms zero on @xmath41 method reduces standard cg fem @xmath42 triangulation discontinuous interior penalty ip family dg fems @xcite cg method limit dg method @xmath43 hypothesis solution cdg method limit solutions dg method penalty parameter @xmath43 @xmath44 penalising edges @xmath41 @xmath45 @xmath46 penalty parameters for edges @xmath41 @xmath22 numerical solution cdg problem @xmath47 solution pure dg problem @xmath48 @xmath49 piecewise discontinuous polynomial space @xmath10 convergence dg solution converges to cdg solution @xmath50 @xmath51 prove result direct @xcite full discussionresult imply stability cdg method @xmath13 region @xmath5 shows cdg method same problems as cg method investigation cdg method intermediate stage between cg and dg justified aids understanding convergence stability properties cdg method based on cg dg cdg method offers reduction in degrees freedom compared to dg cdg method poses difficulties in implementation approach use super penalty result theorem convergence approximation to cdg solution method same degrees of freedom as dg present approach to implement cdg method with appropriate finite element structure discuss approach ` deal.ii ` finite element library open source library creation finite element codes access to algorithms data structures present numerical experiments difficulty implementing cdg method in ` deal.ii ` lack of native cdg element type to assign degrees freedom mesh code must initialised with triangulation ` use particular finite element basis degrees freedom possible initialise triangulation with dg cg regions via material_id ` flag no appropriate element exists in existing deal.ii ` framework difficult to code element with appropriate properties robust approach use existing capabilities library allow access to other features ` deal.ii `without correct distribution of degrees freedom sparsity pattern finite element matrix suboptimal more entries than reducing benefit shrinking degrees freedom relative to dg method ` deal.ii ` library multiple equations single mesh elastic solid fluid interaction problem apply different methods to same equation on different regions mesh same problem framework use @xmath52 capability library collections of finite elements hp::fecollection ` multiple elements grouped into one data structure use for @xmath52 refinement create finite elements same type scalar lagrange discontinuous elements varying degree not sufficient to create hp::fecollection ` of cg and dg elements interface between regions undefined admissible collection use ` fe_nothing ` finite element type with zero degrees freedom using @xmath53 class two vector - valued finite element types @xmath54 and @xmath55 combine in ` hp::fecollection ` apply first @xmath53 on cg region second on dg region triangulation ` location cg and dg elements degrees of freedom correctly distributed according to finite element defined by hp::fecollection `assembling matrix for finite element method careful using correct element hp::fecollection correct part @xmath53 difficult case boundary @xmath14 dg element evaluate contribution neighbouring cg element cdg method jump permissible on @xmath14 implement cdg method in deal.ii ` create two solutions fe_q`-`fe_nothing fe_nothing -`fe_dgq consider domain @xmath56 @xmath57 @xmath59 dirichlet boundary conditions forcing function @xmath60 chosen analytical solution is @xmath61 exponential boundary layer along @xmath62 @xmath63 width @xmath64 @xmath65 solve finite element problem on 1024 element grid fix @xmath66 larger than required for stability shows behaviour ` fe_nothing ` show components of fe_system ` combined solution dg finite element solution same problem deal.ii ` framework data structures allow implementation @xmath52 methods envisage implementation @xmath67 method change mesh size polynomial degree element type no implementing @xmath67 method straightforward with ` fe_nothing ` approach present two numerical experiments performance cdg method present layers when @xmath68 small fix region where continuous method applied vary @xmath68 causes layer steepen advection dominated regime @xmath68 large no steep layer cdg solution approximates true solution make @xmath68 smaller layer forms extends continuous region @xmath68 smaller layer leaves continuous region performance dg cdg method indistinguishable experiment pick @xmath13 @xmath9 regions region @xmath12 one layer coincides with @xmath9 consider problem true solution solve finite element problem 1024 element grid fix @xmath69 one row elements in @xmath9 vary @xmath68 layer sharpens moves dg region before layer formed two methods perform well layer begins decreasing @xmath68 not entirely contained discontinuous region error peaks layer sharpens entirely contained discontinuous region difference solutions negligible problem internal layer advection coefficient given by @xmath70 pick boundary conditions right hand side @xmath60 true solution @xmath71 @xmath72 error function defined by @xmath73 figure example2a same behaviour example 1 layer exists must contained within discontinuous region two methods perform cdg solution for @xmath68 oscillations visible when @xmath78 layer sharpened oscillations disappear | 0.503901 | 3,319 | 1,757 | , Saving $0.1 in GPT-4. |
177,228 | solving analytic systems using inversion can be implemented in a variety of ways .
one method is to use lagrange inversion and variations . here
we present a different approach , based on dual vector fields . for a function analytic in a neighborhood of the origin in the complex plane , we associate a vector field and its dual , an operator version of fourier transform .
the construction extends naturally to functions of several variables .
we illustrate with various examples and present an efficient algorithm readily implemented as a symbolic procedure in maple while suitable as well for numerical computations using languages such as c or java . | we introduce the operator calculus necessary to present our approach to ( local ) inversion of analytic functions . it is important to note that this is different from lagrange inversion and
is based on the flow of a vector field associated to a given function .
it appears to be theoretically appealing as well as computationally effective .
acting on polynomials in @xmath0 , define the operators @xmath1 and @xmath2 .
they satisfy commutation relations @xmath3=i$ ] , where @xmath4 , the identity operator , commutes with both @xmath5 and @xmath6 .
abstractly , the heisenberg - weyl algebra is the associative algebra generated by operators @xmath7 satisfying @xmath8=c$ ] , @xmath9=[b , c]=0 $ ] .
the _ standard _ hw algebra is the one generated by the realization @xmath10 , @xmath11 , @xmath12 .
an _ appell system _ is a system of polynomials @xmath13 that is a basis for a representation of the standard hw algebra with the following properties : 1 .
@xmath14 is of degree @xmath15 in @xmath0 ; 2 .
@xmath16 . in several variables , @xmath17 , with multi - indices @xmath18 , the corresponding monomials are @xmath19 .
denote the partial derivative operators by @xmath20 and the corresponding multiplication operators by @xmath21 .
then @xmath22=\delta_{ij}\,i$ ] .
an appell system is a system of polynomials @xmath23 in the variables @xmath24 such that 1 .
the top degree term of @xmath25 is a constant multiple of @xmath26 ; 2 .
@xmath27 , where @xmath28 has all components zero except for 1 in the @xmath29 position .
rota @xcite is well - known for his _ umbral calculus
_ development of special polynomial sequences , called _ basic sequences . _ from our perspective , these are canonical polynomial systems " in the sense that they provide polynomial representations of the heisenberg - weyl algebra , in realizations different from the standard one .
our idea @xcite is to illustrate explicitly the rle of vector fields and their duals , using operator calculus methods for working with the latter ( in our volumes this viewpoint is prefigured in @xcite ) .
the main feature of our approach is that the action of the vector field may be readily calculated while the action of the dual vector field on exponentials is identical to that of the vector field .
then we note that acting iteratively with a vector field on polynomials involves the complexity of the coefficients , while acting iteratively with the dual vector field always produces polynomials from polynomials .
so we can switch to the dual vector field for calculations .
specifically , fix a neighborhood of 0 in @xmath30 .
take an analytic function @xmath31 defined there , normalized to @xmath32 , @xmath33 .
denote @xmath34 and @xmath35 the inverse function , i.e. , @xmath36 , @xmath37 .
then @xmath38 is defined by power series as an operator on polynomials in @xmath0 and @xmath39=v'(d)$ ] so that @xmath40=i$ ] . in other words ,
@xmath41 and @xmath42 generate a representation of the hw algebra on polynomials in @xmath0 .
the basis for the representation is @xmath43 , i.e. , @xmath44 is a _ raising operator_. and @xmath45 so that @xmath46 is the corresponding _ lowering operator . _
the @xmath47 form a system of _ canonical polynomials _ or generalized appell system .
the operator of multiplication by @xmath0 is given by @xmath48 , which is a _ recursion operator _ for the system .
we identify vector fields with first - order partial differential operators .
consider a variable @xmath49 with corresponding partial differential operator @xmath50 .
given @xmath46 as above , let @xmath51 be the vector field @xmath52 .
then we observe the following identities @xmath53 as any operator function of @xmath5 acts as a multiplication operator on @xmath54 .
the important property of these equalities is that @xmath44 and @xmath51 commute , as they involve independent variables .
so we may iterate to get @xmath55 on the other hand , we can solve for the left - hand side of this equation using the method of characteristics .
namely , if we solve @xmath56 with initial condition @xmath57 , then for any smooth function @xmath58 , @xmath59 thus @xmath60 to solve equation ( [ eq : vf ] ) , multiply both sides by @xmath61 and observe that we get @xmath62 integrating yields @xmath63 or , writing @xmath64 for @xmath65 , we have @xmath66 we can set @xmath67 to get @xmath68 on the one hand while @xmath69 in summary , we have the expansion of the exponential of the inverse function @xmath70 or @xmath71 this yields an alternative approach to inversion of the function @xmath31 rather than using lagrange s formula .
we see that the coefficient of @xmath72 yields the expansion of @xmath73 .
in particular , @xmath35 itself is given by the coefficient of @xmath0 on the right - hand side . specifically , we have : the coefficient of @xmath72 in @xmath74 is equal to @xmath75 , each giving the coefficient of @xmath76 in the expansion of @xmath77 .
expand both sides of equation ( [ eq : iterate ] ) , using @xmath64 for @xmath65 , in powers of @xmath0 and @xmath64 , and let @xmath67 : @xmath78 and compare with equation ( [ eq : expansion ] ) .
the same idea works in several variables .
we have @xmath79 analytic in a neighborhood of @xmath80 in @xmath81 .
denote the jacobian matrix @xmath82 by @xmath83 and its inverse by @xmath84 .
the variables @xmath85 commute and act as raising operators for generating the basis @xmath86 .
namely , @xmath87 . and @xmath88 , @xmath89 are lowering operators : @xmath90 .
denote @xmath91 by @xmath92 . with variables @xmath93 and corresponding partials @xmath94 , define the vector fields @xmath95 for a vector field @xmath96 , we have the identities @xmath97 the method of characteristics applies as in one variable and as in equation ( [ eq : evf ] ) @xmath98 thus , we have the expansion @xmath99 in particular , the @xmath100 component , @xmath101 , of the inverse function is given by the coefficient of @xmath102 in the above expansion .
an important feature of our approach is that to get an expansion to a given order requires knowledge of the expansion of @xmath84 just to that order .
the reason is that when iterating @xmath103 , at step @xmath15 it is acting on a polynomial of degree @xmath104 , so all terms of the expansion of @xmath105 of order @xmath15 or higher would yield zero acting on @xmath106 .
this allows for streamlined computations . for polynomial systems
@xmath107 , @xmath83 will have polynomial entries , and @xmath84 will be rational in @xmath108 . hence raising operators will be rational functions of @xmath109 , linear in @xmath24 .
thus the coefficients of the expansion of the entries @xmath110 of @xmath84 would be computed by finite - step recurrences .
note that to solve @xmath111 for @xmath112 near @xmath113 , with @xmath114 , apply the method to @xmath115 , so that @xmath116 .
the inverse is @xmath117
. then @xmath118 .
in this section we focus on the one - variable case .
we illustrate the method with examples , and then present an algorithm suitable for symbolic computation . in one variable , solving
a cubic is interesting as the expansion of @xmath84 can be expressed in terms of chebyshev polynomials .
. then @xmath120 .
thus @xmath121 where @xmath122 are chebyshev polynomials of the second kind .
specializing @xmath123 provides interesting cases .
for example , let @xmath124 , or @xmath125 .
then the coefficients in the expansion of @xmath84 are periodic with period 8 and , in fact , @xmath126 the coefficient of @xmath0 in the polynomials @xmath14 yield the coefficients in the expansion of the inverse @xmath127 . here
are some polynomials starting with @xmath128 , @xmath129 : @xmath130 this gives to order 6 : @xmath131 this expansion will give approximate solutions to @xmath132 for @xmath64 near @xmath80 .
inversion of the chebyshev polynomial @xmath133 can be used as the basis for solving general cubic equations ( @xcite ) .
to get started we have , with @xmath134 , @xmath135 so @xmath136 , @xmath137 , @xmath138 , etc .
we find @xmath139 in this case , we can find the expansion analytically . to solve @xmath140 ,
write @xmath141 invert to get , for integer @xmath142 , @xmath143 , with @xmath144 denoting the principal branch
. then @xmath145 we want a branch with @xmath146 corresponding to @xmath147 . with @xmath148 ,
we want the argument of the cosine to be @xmath149 , for some integer @xmath150 .
this yields the condition @xmath151 . taking @xmath152 , we get @xmath153 , with the minus sign .
namely , @xmath154 using hypergeometric functions ( see next example ) and rewriting , we find the form @xmath155 if we generate the polynomials @xmath14 , we can find the expansion of @xmath77 to any order .
a similar approach is interesting for the chebyshev polynomial @xmath156 .
@xmath157 satisfies the hypergeometric differential equation @xmath158 which can be written in the form @xmath159f=\lambda^2\,f\ ] ] with here @xmath160 denoting @xmath161 . for integer @xmath162 ,
this is the differential equation for the corresponding chebyshev polynomial . in general , these are _ chebyshev functions_. as noted above , for @xmath163 , we take @xmath164 , and , as above , we require @xmath165 with @xmath166 , we have the solution @xmath167 for symbolic computation using maple , one can use the ore_algebra package . 1 .
first fix the degree of approximation .
expand @xmath84 as a polynomial to that degree .
2 . declare the ore algebra with one variable , @xmath0 , and one derivative , @xmath5 .
3 . define the operator @xmath103 in the algebra .
iterate starting with @xmath128 using the applyopr command .
extract the coefficient of @xmath72 to get the expansion of @xmath77 .
here is a matrix approach that can be implemented numerically . fix the order of approximation @xmath15 . cut off the expansion @xmath168 at @xmath169 . let the matrix @xmath170 define the auxiliary diagonal matrices @xmath171 q&=&\begin{pmatrix}1/\gamma(1)&0&\ldots&0\cr 0&1/\gamma(2)&\ldots&0\cr \vdots&\vdots&\ddots&\vdots\cr 0&0&\ldots&1/\gamma(n)\cr\end{pmatrix}.\end{aligned}\ ] ]
note that @xmath172 . denoting @xmath173
, we have the recursion @xmath174 = [ c_1^{(k)},c_2^{(k)},\ldots , c_n^{(k ) } ] pwq.\ ] ] the condition @xmath175 gives @xmath128 . then @xmath176 yields @xmath177
. we see that @xmath178 for @xmath179 .
we iterate as follows : \1 . start with @xmath180 times the unit vector @xmath181 $ ] of length @xmath15 .
multiply by @xmath84 .
iterate , multiplying on the right by @xmath182 at each step .
finally , multiply on the right by @xmath183
. the top row will give the coefficients of the expansion of @xmath35 to order @xmath15 .
here is a simple @xmath184 system for illustration .
@xmath185 so @xmath186 the raising operators are @xmath187 expanding @xmath188 yields , with @xmath189 , @xmath190 thus @xmath191 so @xmath192
for any given order @xmath15 , the polynomials of degree @xmath15 are an invariant subspace for the operator @xmath44 up until the last step .
we can formulate an alternative matrix computation as follows .
let @xmath193 and @xmath194 denote the matrices of the operators of differentiation and multiplication by @xmath0 respectively on polynomials of degree less than or equal to @xmath15 .
the space is invariant under differentiation , and we cut off multiplication by @xmath0 to be zero on @xmath195 . we get @xmath196 with the first row of @xmath194 all zeros .
we then compute the matrix @xmath194 times @xmath197 , where @xmath197 is computed as a matrix polynomial by substituting in @xmath198 up to order @xmath15 .
then @xmath44 has a matrix representation , @xmath199 , on the space and we iterate multiplying by @xmath200 acting on the unit vector @xmath201 .
these give the coefficients of the polynomials @xmath14 .
in several variables , one constructs matrices for @xmath202 and @xmath21 using kronecker products of @xmath193 and @xmath194 with the identity . for example , @xmath203 with @xmath193 in the @xmath204 spot .
similarly for @xmath205 .
then one has explicit matrix representations for the dual vector fields and the polynomials can be found accordingly .
this approach is explicit , but seems to much slower than using the built - in ore_algebra package . | introduce operator calculus local inversion functions different from lagrange inversion based on flow vector field function theoretically appealing computationally effective acting polynomials @xmath0 define operators @xmath1 @xmath2 satisfy commutation relations @xmath3=i$ @xmath4 operator commutes with @xmath5 @xmath6 heisenberg - weyl algebra associative algebra operators @xmath7 @xmath8=c$ @xmath9= c $ standard hw algebra generated @xmath10 @xmath12 appell system system polynomials @xmath13 basis representation standard hw algebra properties @xmath14 degree @xmath15 @xmath0 @xmath16 variables @xmath17 multi - indices @xmath18 monomials @xmath19 denote partial derivative operators @xmath20 multiplication operators @xmath21 @xmath22=\delta_{ij}\,i$ ] appell system polynomials @xmath23 variables @xmath24 top degree term @xmath25 constant multiple @xmath26 @xmath27 @xmath28 components zero except 1 @xmath29 @xcite known umbral calculus development special polynomial sequences basic sequencesthese are canonical polynomial systems provide representations of heisenberg - weyl algebra realizations different from standard idea illustrate rle of vector fields duals using operator calculus methods prefigured in action of vector field calculated dual vector field on exponentials identical to acting iteratively with vector field on polynomials involves complexity coefficients dual vector field produces polynomials from switch to dual vector field for calculations fix neighborhood of 0 in @xmath30 analytic function @xmath31 normalized to @xmath32 @xmath33 denote @xmath34 @xmath35 inverse function @xmath36 @xmath38 defined by power series as operator on polynomials in @xmath0 @xmath39=v'(d)$ @xmath40=i$ @xmath41 and @xmath42 generate representation of hw algebra on polynomials in @xmath0 basis for representation is @xmath43 @xmath44 raising operator lowering operator @xmath47 form system of canonical polynomials operator of multiplication by @xmath0 by @xmath48 recursion operatoridentify vector fields with partial differential operators consider variable @xmath49 partial differential operator @xmath50 @xmath46 @xmath51 be vector field @xmath52 observe identities @xmath53 function @xmath5 multiplication operator on @xmath54 @xmath44 @xmath51 commute involve independent variables iterate to get @xmath55 solve left - hand side equation method characteristics solve @xmath56 initial condition @xmath57 smooth function @xmath58 @xmath59 @xmath60 solve equation multiply both sides by @xmath61 @xmath62 yields @xmath63 @xmath64 @xmath65 @xmath66 set @xmath67 to get @xmath68 @xmath69 expansion exponential inverse function @xmath70 or @xmath71 alternative approach inversion function @xmath31 lagrange formula coefficient @xmath72 yields expansion @xmath73 @xmath35 given by coefficient @xmath0 side coefficient @xmath72 @xmath74 equal to @xmath75 coefficient @xmath76 expansion @xmath77expand sides equation iterate using @xmath64 for @xmath65 powers @xmath0 let @xmath67 @xmath78 compare equation idea works several variables @xmath79 analytic @xmath80 @xmath81 denote jacobian matrix @xmath82 by @xmath83 inverse @xmath84 variables @xmath85 commute raising operators basis @xmath86 @xmath87 @xmath88 @xmath89 lowering operators @xmath90 denote @xmath91 by @xmath92 variables @xmath93 partials @xmath94 define vector fields @xmath95 vector field @xmath96 identities @xmath97 method characteristics applies one variable equation @xmath98 expansion @xmath99 @xmath100 component @xmath101 inverse function given coefficient @xmath102 expansion expansion order requires knowledge expansion @xmath84 iterating @xmath103 step @xmath15 acting on polynomial degree @xmath104 expansion @xmath105 @xmath15 higher yield zero @xmath106 streamlined computations polynomial systems @xmath107 @xmath83 polynomial entries @xmath84 rational in @xmath108raising operators rational functions @xmath109 linear @xmath24 coefficients expansion entries @xmath110 @xmath84 computed by finite - step recurrences solve @xmath111 @xmath112 @xmath113 @xmath114 apply method @xmath115 @xmath116 inverse @xmath117 @xmath118 focus one - variable case illustrate method examples present algorithm symbolic computation variable solving cubic expansion @xmath84 expressed chebyshev polynomials @xmath120 @xmath121 @xmath122 chebyshev polynomials second kind specializing @xmath123 cases @xmath124 @xmath125 coefficients expansion @xmath84 periodic period 8 @xmath126 coefficient @xmath0 polynomials @xmath14 yield coefficients expansion inverse @xmath127 polynomials starting @xmath128 @xmath129 @xmath130 order 6 @xmath131 expansion approximate solutions @xmath132 @xmath64 near @xmath80 inversion chebyshev polynomial @xmath133 basis solving general cubic equations @xmath134 @xmath135 @xmath136 find @xmath139 expansion analyticallysolve @xmath140 write @xmath141 invert get integer @xmath142 @xmath143 @xmath144 principal branch @xmath145 branch @xmath146 @xmath147 @xmath148 argument cosine @xmath149 integer @xmath150 yields condition @xmath151 @xmath152 get @xmath153 minus sign @xmath154 hypergeometric functions rewriting form @xmath155 generate polynomials @xmath14 expansion @xmath77 order similar approach chebyshev polynomial @xmath156 @xmath157 satisfies hypergeometric differential equation @xmath158 form @xmath159f=\lambda^2\,f\ ] @xmath160 @xmath161 integer @xmath162 differential equation corresponding chebyshev polynomial chebyshev functions @xmath163 take @xmath164 require @xmath165 @xmath166 solution @xmath167 symbolic computation maple use ore_algebra package fix degree approximation expand @xmath84 polynomial declare ore algebra one variable @xmath0 derivative @xmath5 define operator @xmath103 algebra iterate @xmath128 applyopr commandextract coefficient @xmath72 expansion @xmath77 matrix approach implemented numerically fix order approximation @xmath15 cut expansion @xmath168 @xmath169 matrix @xmath170 define auxiliary diagonal matrices @xmath171{pmatrix\gamma(1){pmatrix}{aligned} @xmath172 @xmath173 recursion @xmath174 = [ c_1^{(k)},c_2^{(k)},\ldots c_n^{(k condition @xmath175 gives @xmath128 @xmath176 yields @xmath177 @xmath178 @xmath179 iterate start @xmath180 times unit vector @xmath181 length @xmath15 multiply @xmath84 iterate multiplying right @xmath182 each step multiply right @xmath183 top row coefficients expansion @xmath35 order @xmath15 simple @xmath184 system illustration@xmath185 @xmath186 raising operators @xmath187 expanding @xmath188 yields @xmath189 @xmath190 @xmath191 @xmath192 order @xmath15 polynomials degree @xmath15 invariant subspace operator @xmath44 last step alternative matrix computation @xmath193 @xmath194 matrices operators differentiation multiplication @xmath0 polynomials degree less equal @xmath15 space invariant differentiation cut multiplication @xmath0 zero @xmath195 @xmath196 first row @xmath194 zeros compute matrix @xmath194 @xmath197 @xmath197 matrix polynomial substituting @xmath198 @xmath15 @xmath44 matrix representation @xmath199 space iterate multiplying @xmath200 unit vector @xmath201 coefficients polynomials @xmath14 constructs matrices @xmath202 @xmath21 products @xmath193 @xmath194 @xmath203 @xmath193 @xmath204 spot @xmath205 explicit matrix representations dual vector fields polynomials found approach slower than ore_algebra package | 0.497617 | 3,252 | 1,780 | , Saving $0.1 in GPT-4. |
177,240 | the observational study of star - formation laws is paramount to disentangling the physical processes at work on local and global scales in galaxies . to this aim
we have expanded the sample of extreme starbursts , represented by local lirgs and ulirgs , with high - quality data obtained in the 1 - 0 line of hcn .
the analysis of the new data shows that the star - formation efficiency of the dense molecular gas , derived from the fir / hcn luminosity ratio , is a factor 3 - 4 higher in extreme starbursts compared to normal galaxies .
we find a duality in the kennicutt - schmidt laws that is enhanced if we account for the different conversion factor for hcn ( @xmath0 ) in extreme starbursts and correct for the unobscured star - formation rate in normal galaxies .
we find that it is possible to fit the observed differences in the fir / hcn ratios between normal galaxies and lirgs / ulirgs with a common constant star - formation rate per free - fall time ( sfr@xmath1 ) if we assume that hcn densities are @xmath212 orders of magnitude higher in lirgs / ulirgs , and provided that sfr@xmath1@xmath20.005 - 0.01 and/or if @xmath0 is a factor of a few lower than our favored values . | based on simple theoretical arguments , ( * ? ? ?
* schmidt ( 1959 ) ) proposed that the star - formation rate volume density ( @xmath3 ) in galaxies should scale as a power law of the corresponding gas volume density ( @xmath4 ) : @xmath5 . if the relevant time scale for star formation is the _ local _ free - fall time ( t@xmath1 ) , then @xmath6/@xmath7 . a similar relation between the surface densities ( @xmath8 , @xmath9 ) follows provided that the gas scale - height remains constant in different galaxy populations , a widely adopted yet far from trivial assumption .
observers have used different proxies for @xmath8 and @xmath9 to derive star - formation relations in normal galaxies and extreme starbursts .
* kennicutt et al .
( 1998 ) ) found a power law index @xmath10=1.40@xmath110.15 using a compilation of galaxy - averaged co+hi data obtained in a sample of @xmath2100 galaxies .
more recently the co surveys of ( * ? ? ?
* daddi et al . ( 2010 ) ) and ( * ? ? ?
* genzel et al . ( 2010 ) ) presented evidence of bimodality in galaxy - averaged star - formation laws : normal galaxies show 410 longer depletion time - scales compared to mergers
. observations of hcn lines have also been used to study the star - formation laws for the dense molecular gas .
the first studies by ( * ? ? ?
* gao & solomon ( 2004 ) ) found that the star - formation efficiency measured with respect to the dense molecular gas content ( sfe@xmath12 ) was about constant in galaxies .
this scenario was later questioned by ( * ? ? ? * graci - carpio et al .
( 2008 ) ) , who observed a sample of lirgs / ulirgs and found that sfe@xmath13 is a factor @xmath223 higher in ir luminous targets compared to normal galaxies .
these observations suggested that galaxies classified as lirgs may represent the transition point in the star - formation laws . to improve the statisitics in this luminosity range ,
we have recently expanded up to 28 the number of lirgs with high - quality hcn data obtained with the iram 30 m telescope . out of these , fourteen were extracted from the sample of nearby lirgs studied by ( * ? ? ?
* alonso - herrero et al . ( 2006 ) ) .
the results of this work , published by ( * ? ? ?
* garca - burillo et al . ( 2012 ) ) , have been used to study if the _ observed _
bimodality of star - formation laws in galaxies derived from co can be extended to the higher density regime probed by hcn .
we have also compared our observations with the predictions of theoretical models in which the efficiency of star formation is determined by the ratio of a _ constant _ star - formation rate per free - fall time ( sfr@xmath1 ) to the _ local _ free - fall time t@xmath1 ( ( * ? ? ?
* krumholz & mckee 2005 ) ) .
as illustrated in fig .
[ lfir - lhcn]a , with the addition of the new lirg sample we confirm that the @xmath14-@xmath15 correlation is superlinear ( power law index @xmath21.23@xmath110.05 ) .
furthermore we corroborate on a more solid statistical basis , compared to the work of @xcite , that the average star - formation efficiency of the dense gas is a factor @xmath223 higher in lirgs / ulirgs ( @xmath21400@xmath11100@xmath16/(kkms@xmath17pc@xmath18 ) ) compared to normal galaxies ( @xmath2600@xmath1170@xmath16/(kkms@xmath17pc@xmath18 ) ) ( see fig .
[ lfir - lhcn]b ) .
if we use universal conversion factors for hcn and fir luminosities , a fit to the full sample yields a power index @xmath19 for the implied kennicutt - schmidt laws relating @xmath8 and @xmath20 .
however , lirgs / ulirgs and normal galaxies are not fully overlapping in this scatter plot ( see fig .
6 of ( * ? ? ?
* garca - burillo et al . 2012 ) ) . in order to quantify
if a dual fit applies to the power law , we split the sample into normal and ir luminous galaxies .
a two - function power law fit with indices close to unity qualifies as a much better description of the star - formation relation for the dense gas .
this result is seen to be solid against the statistical biases inherent to this analysis ( see discussion in sect . 5 of ( * ? ? ?
* garca - burillo et al . 2012 ) ) . the validity of a universal @xmath21
@xmath15 conversion factor ( @xmath0 ) for normal galaxies and lirgs / ulirgs has been questioned by a number of works , which favor lower @xmath0 in extreme starbursts compared to normal galaxies ( e.g. , ( * ? ? ? * graci - carpio et al .
2008 ) ; see , however , ( * ? ? ?
* papadopoulos et al . 2012 ) ) .
if we account for the different conversion factors for hcn in extreme starbursts and for the unobscured star - formation rate in normal galaxies the duality in star - formation laws is enhanced , as illustrated in fig .
[ ks - revised]a . at the high end of @xmath20 values
( @xmath2a few 10@xmath22m@xmath23pc@xmath24 ) the normal galaxy law underpredicts @xmath8 in ir luminous galaxies by about an order of magnitude .
furthermore , within the range of gas surface densities shared by normal galaxies and lirgs / ulirgs , @xmath20@xmath22 10@xmath252 10@xmath26m@xmath23pc@xmath24 , the factor 3 to 5 disagreement between the two laws is a factor @xmath234 times larger than the typical uncertainty on @xmath8 .
this result extends the more extreme bimodal behavior of star - formation laws , derived from co molecular lines by two recent surveys ( ( * ? ? ?
* daddi et al .
2010 ) , ( * ? ? ?
* genzel et al . 2010 ) ) , to the higher molecular densities probed by hcn lines . on average ,
the revised depletion time scales for the dense molecular gas show a significant difference between lirgs / ulirgs ( t@xmath2714@xmath111.4 myr ) and normal galaxies ( t@xmath2750@xmath115 myr ) ( fig .
[ ks - revised]b ) .
* garca - burillo et al . ( 2012 ) ) have used the new hcn observations presented above to test a key prediction of the star - formation model of @xcite : the constancy of sfr@xmath1 , which is expected to be @xmath20.02 up to the density regime explored by hcn observations .
figure [ sfrff]a represents sfr@xmath1 as a function of l@xmath28 obtained from the sfe@xmath29 values derived with universal conversion factors for l@xmath28 and l@xmath30 . to render a constant sfr@xmath1 compatible with the order of magnitude increase in sfe@xmath13 , the density of hcn clouds has to change notably from normal galaxies to lirgs / ulirgs
nevertheless , the values derived for sfr@xmath1 and/or n@xmath30(h@xmath31 ) are well below the expected range both for normal galaxies and lirgs / ulirgs .
the use of revised conversion factors alleviates the problem only in lirgs / ulirgs ( fig .
[ sfrff]b ) .
an overall satisfactory solution , which accounts for the observed trend in the l@xmath28/l@xmath30 luminosity ratios , is only found provided that : 1 ) sfr@xmath1@xmath20.0035 ( i.e. , a factor @xmath26 lower than the canonical value of @xmath22@xmath32 ) , 2 ) n@xmath30(h@xmath31 ) increases with l@xmath28 from @xmath210@xmath22@xmath33 in normal galaxies to @xmath210@xmath34@xmath33 in extreme starbursts , and 3 ) @xmath0 is @xmath224 times lower in lirgs / ulirgs . in summary : there is new observational evidence that the overall efficiency of the dense molecular gas in galaxies is significantly lower than initially postulated by ( * ? ? ? * krumholz & mckee ( 2005 ) ) .
furthermore , an agreement between observations and models requires a dramatic increase in the average gas densities from normal galaxies to extreme starbursts .
multi - line observations are a key to constrain the conversion factors of the different molecular tracers that are commonly used as proxies for @xmath9 .
these observations can also yield an estimate of volume gas densities . as recently argued by (
* krumholz et al . ( 2012 ) ) a simple , local , volumetric star - formation law with sfr@xmath350.01 can dissolve the apparent bimodality found by ( * ? ? ?
* genzel et al .
( 2010 ) ) if different scale - heights are assumed for the gas disks in normal galaxies and extreme starbursts .
while this is a plausible explanation , the change in scale - heights required to make bimodality disappear in star formation laws remains to be observationally validated . | schmidt 1959 proposed star - formation rate volume density @xmath3 in galaxies power law gas volume density @xmath4 @xmath5 relevant time scale for star formation local free - fall time@xmath1 @xmath6/@xmath7 similar relation between surface densities @xmath8 @xmath9 follows gas scale - height constant in galaxy populations observers used proxies for @xmath8 @xmath9 star - formation relations in normal galaxies extreme starbursts kennicutt et al 1998 found power law index @xmath10=1.40@xmath110.15 galaxy - averaged co+hi data @xmath2100 galaxies daddi et genzel presented bimodality in galaxy - star - formation laws normal galaxies show 410 longer depletion time - scales mergers observations of hcn lines star - formation laws for dense molecular gas studies gao & solomon 2004 star - formation efficiency dense molecular gas content sfe@xmath12 ) constant in galaxies questioned by graci - carpio et al 2008 sfe@xmath13 factor @xmath223 higher in ir luminous targets normal galaxiesobservations suggested galaxies lirgs may represent transition point in star - formation laws improve statisitics expanded to 28 lirgs with high - quality hcn data with iram 30 m telescope fourteen extracted from sample nearby lirgs studied by alonso - herrero et al . 2006 results work published by garca - burillo et al . 2012 used to study if bimodality of star - formation laws in galaxies from co be extended to higher density regime probed by hcn compared observations with predictions theoretical models efficiency star formation determined by ratio of constant star - formation rate per free - fall time to local free - fall time & 2005 in with addition of new lirg sample confirm @xmath14-@xmath15 correlation is superlinear ( power law index corroborate average star - formation efficiency of dense gas is factor @xmath223 higher in lirgs / ulirgs compared to normal galaxies see fig . - lhcnuniversal conversion factors for hcn fir luminosities fit full sample yields power index @xmath19 for kennicutt - schmidt laws @xmath8 @xmath20 lirgs / ulirgs normal galaxies not in scatter plot fig . 6 garca - burillo 2012 quantify dual fit power law split sample into normal ir luminous galaxies two - function power law fit with indices close to unity better description star - formation relation for dense gas result solid against statistical biases discussion 5 2012 validity of universal @xmath21 @xmath15 conversion factor @xmath0 for normal galaxies lirgs / ulirgs questioned works lower @xmath0 in extreme starbursts papadopoulos 2012 account different conversion factors for hcn in extreme starbursts unobscured star - formation rate in normal galaxies duality in star - formation laws enhanced illustrated in fig . high end of @xmath20 values normal galaxy law underpredicts @xmath8 in ir luminous galaxies order magnitudegas surface densities normal galaxies lirgs / ulirgs factor 3 to 5 disagreement between laws is factor @xmath234 times larger than uncertainty on @xmath8 extends extreme bimodal behavior of star - formation laws to higher molecular densities by hcn lines revised depletion time scales for dense molecular gas show significant difference between lirgs / ulirgs@xmath2714@xmath111.4 myr and normal galaxies@xmath115 myr ) garca - burillo et al . 2012 used new hcn observations test prediction star - formation model @xcite constancy of sfr@xmath1 expected to be @xmath20.02 up to density regime figure represents sfr@xmath1 function of l@xmath28 from sfe@xmath29 values constant sfr@xmath1 compatible with increase sfe@xmath13 density of hcn clouds from normal galaxies to lirgs / ulirgs values for sfr@xmath1 n@xmath30 below expected range for normal galaxies and lirgs / ulirgsrevised conversion factors alleviates problem in lirgs / ulirgs fig . sfrff]b satisfactory solution trend l@xmath28/l@xmath30 luminosity ratios found sfr@xmath1@xmath20.0035 factor lower than @xmath22@xmath32 n@xmath30(h@xmath31 ) increases with l@xmath28 from normal to extreme starbursts @xmath0 @xmath224 times lower in lirgs / ulirgs new evidence efficiency of dense molecular gas in galaxies lower than postulated by krumholz & mckee 2005 agreement between observations models requires increase average gas densities from normal galaxies to extreme starbursts multi - line observations key to constrain conversion factors of molecular tracers @xmath9 yield estimate volume gas densities simple volumetric star - formation law with sfr@xmath350.01 can dissolve bimodality by if different scale - heights assumed for gas disks in normal galaxies and extreme starbursts change in scale - heights bimodality star formation | 0.318049 | 2,217 | 1,098 | , Saving $0.1 in GPT-4. |
177,243 | bone age assessment is a task performed daily in hospitals worldwide .
this involves a clinician estimating the age of a patient from a radiograph of the non - dominant hand .
our approach to automated bone age assessment is to modularise the algorithm into the following three stages : segment and verify hand outline ; segment and verify bones ; use the bone outlines to construct models of age . in this paper
we address the final question : given outlines of bones , can we learn how to predict the bone age of the patient ?
we examine two alternative approaches .
firstly , we attempt to train classifiers on individual bones to predict the bone stage categories commonly used in bone ageing .
secondly , we construct regression models to directly predict patient age .
we demonstrate that models built on summary features of the bone outline perform better than those built using the one dimensional representation of the outline , and also do at least as well as other automated systems .
we show that models constructed on just three bones are as accurate at predicting age as expert human assessors using the standard technique .
we also demonstrate the utility of the model by quantifying the importance of ethnicity and sex on age development .
our conclusion is that the feature based system of separating the image processing from the age modelling is the best approach for automated bone ageing , since it offers flexibility and transparency and produces accurate estimates . bone age assessment , automated tanner - whitehouse , shapelet , elastic ensemble | bone age assessment typically involves estimating the age of a patient from a radiograph by quantifying the development of the bones of the non - dominant hand .
it is used to evaluate whether a child s bones are developing at an acceptable rate , and to monitor whether certain treatments are affecting a patient s skeletal development . currently , this task is performed manually using an atlas based system such as greulich and pyle ( gp ) @xcite or a bone scoring method like tanner and whitehouse ( tw ) @xcite .
atlas methods such as gp involve comparing the query image to a set of representative hand radiographs taken from subjects at a range of ages .
scoring systems assign each bone to one of several predefined stages , then combine these stage classifications to form an age estimate .
manual procedures are time consuming and often inaccurate .
automated systems for bone age assessment have previously been proposed .
these either attempt to recreate the tw or gp methods @xcite , or construct regression models for chronological age @xcite .
our approach is modular and feature based , and can be used to either recreate tw scores or predict age directly . to predict tw bone stages we train a range of classifiers on three transformations of the outline .
the first uses an ensemble technique described in @xcite that uses elastic distance measures directly on a one dimensional representation of the bone outline .
the second technique finds discriminatory subsequences of the one dimensional series ( called shapelets ) through a transformation described in @xcite and constructs classifiers in the shapelet feature space .
finally , we derive a set of summary shape features based on the tw descriptors .
we conclude that the classifiers built on shape features are significantly better on at least one bone and provide greater explanatory power . to predict age directly , we perform linear and non linear regressions from the shape feature space to age . we evaluate this process on a data set of images taken from @xcite in the age range 218 .
we show that , given the correct outline , we can accurately recreate tw stages and , using just three bones , can predict chronological age as accurately as clinical experts .
this stepwise , feature driven approach to automated bone ageing is transparent and explicable to clinicians . by separating out the feature extraction from the segmentation and regression we retain the potential for quickly and simply constructing new models for regional populations .
this offers the possibility of producing age estimates tailored to local demographics based on data stored locally in film free hospitals .
the rest of this paper is structured as follows . in section [ sec : baa ] , we review the current manual methods , and describe previous attempts at automated bone age assessment . in section [ sec : extraction ] we describe how we format the segmented bones into outlines and shape features and in section [ classifiers ] we provide an overview of the classification and regression techniques we use to predict bone age . in sections [ sec :
twstages ] and [ regression ] we present our results .
finally , we discuss our conclusions and describe the future direction of this work in section [ sec : conclusions ] .
bone age assessment is a task performed in hospitals worldwide on a daily basis .
the skeletal development of the hand is most commonly assessed using one of two methods : greulich and pyle ( section [ sec : gp ] ) @xcite or tanner and whitehouse ( section [ sec : tw ] ) @xcite .
the bone age estimate obtained by one of these methods is compared with the chronological age to determine if the skeletal development is abnormal .
if there is a significant difference between the patient s bone age and chronological age then the paediatrician may , for example , diagnose the patient with a disorder of growth or maturation @xcite .
the greulich and pyle ( gp ) method uses an atlas of representative hand radiographs taken from subjects at a range of ages .
the latest ( second ) edition of the gp atlas was released in 1959 @xcite which included new images for four new age points .
the final atlas consists of 31 standard radiographs of males from newborn to the age of 19 years and 27 standard radiographs of females from newborn to the age of 18 years . along with each standard
, there is a piece of text describing the development . to use the atlas ,
the clinician checks the patient s radiograph against each of the example radiographs of the appropriate sex .
the key features to check are the development of the epiphysis ( the region at the end of particular bones ) and the presence of certain carpal bones .
the age estimate is the age of the subject who provided the representative image selected as the closest by the clinician .
this process is clearly somewhat subjective and a large variation between clinicians has been observed @xcite .
the representative images are from a very restricted sample taken over 60 years ago , and variation , changes in diet , healthcare and culture may mean this sample is no longer representative .
another criticism of gp is that the method implies an assumption that the ossification process happens in an linear fashion , which may not be true . in 1975 , tanner _
@xcite published a scoring system commonly referred to as tw2 .
two separate methods of calculating bone age are described .
the first method uses the radius , ulna and short bones ( rus ) ( the short bones cover the metacarpals and phalanges of fingers one , three and five ) .
the second method uses just the carpal bones .
the rus method has been found to outperform the carpal bone technique and is easier to use .
each bone has various stages associated with it and each stage has certain descriptors to use in the classification .
table [ tab : twstage ] shows an image of each stage for the distal phalange of the middle finger with the associated criteria .
once all bones have been awarded a score , these scores are summed to find the skeletal maturity score ( sms ) .
the distribution of ages for the sms is described by a centile chart , from which a point estimate of bone age can be derived .
the tw3 method was published in 2001 @xcite .
the basic maturity stages and scores remained the same but the centile charts have been updated to adapt to the modern population .
[ tab : twstage ] .the various tw stages of distal phalange three @xcite [ cols="^,^,<",options="header " , ] one of the benefits of adopting a linear regression model is the ease with which we can perform an exploratory analysis of the feature relevance .
the epiphysis models include 15 - 17 terms , including a large number of interactions .
the model for the proximal phalange is @xmath0 the index of each variable corresponds to the feature number given in table [ tab : features ] .
our first observation is that the model consists almost exclusively of epiphyseal features ( the exception is @xmath1 , metaphysis to width ratio ) .
this is true for the other two bones also .
this implies that future image processing efforts should focus more on accurately extracting and summarising the epiphysis .
secondly , the features epiphysis width ( @xmath2 ) and epiphysis distance to phalanx ( @xmath3 ) are common to all models and are the first to enter the stepwise forward selection .
clearly they are the most important factors , and alone account for approximately 80% of the variability in age ( @xmath4 on two variable linear model ) .
the model constructed on just the proximal phalangeal features , epiphyseal width and epiphyseal distance to phalanx has a mae of 0.98 and rmse of 1.67 .
this implies that it would be very easy to construct a simple , practical model that would give a fairly accurate estimate based on two measurements that can be quickly performed by a non - specialist directly from the image .
this offers the potential for screening for abnormality at very low cost .
a further benefit of the linear model is that , if the regression assumptions hold , we can construct confidence and prediction intervals for new data .
this improves utility of the model in the decision making process because the decision of whether development is abnormal can be phrased as a hypothesis test where the null hypothesis is that the difference between bone age and chronological age is zero .
a linear model also offers a simple way of determining whether there are differences in age model between populations .
we address the question of whether the models for male and female are significantly different by adding a factor to indicate sex . with regressors @xmath5 and @xmath6 , where @xmath7 if the subject is male and @xmath8 if female , we find that we can not reject the null hypothesis that the @xmath6 coefficient is zero and the resulting model is @xmath9 if we fit a stepwise model , sex is the fourth variable to enter the model , and the interaction with @xmath2 is also significant . clearly , sex is a predictive variable and future models should include the term . the other demographic variable we have available is ethnicity . to test the significance of ethnicity ,
we include for factors to model whether the subject was asian ( a ) , african - american ( f ) or hispanic ( h ) .
the only significant factor we find is whether subject is asian .
this is significant in the simplified model and in the stepwise model it is the second most important variable .
clearly there is a different development process at work for the patients with asian ethnicity used in this study .
if we include both sex and ethnicity , the _ dmp _ epiphysis model with three bones has a rmse of 0.855 ( compared to the human raters whose estimates had rmse of 0.89 ) .
we describe three alternative approaches for using bone outlines to classify bone age stage and conclude that the shape features based on tw descriptors is the most appropriate .
we then use these shape features to construct regression models of chronological age . the results for models predicting both tanner - whitehouse stage and chronological age are at least as good as those reported for other automated bone ageing systems .
furthermore , with just three bones , we produce age estimates that are as accurate as expert human assessors using the whole image .
data and code for each of these classification problems is available from @xcite .
in addition to the predictive accuracy , there are several benefits of using feature based regression models .
firstly , we can explain the importance of individual variables .
just two variables account for about 80% of the variation in age , and this relationship offers the potential for fast screening for abnormalities . secondly ,
when the regression assumptions are valid , we can construct confidence and prediction intervals .
this can aid diagnosis and implies that increasing the training set size will incrementally improve the models ( by reducing the variance ) .
finally , we can test the importance of alternative demographic variables and construct models tailored to specific populations .
we demonstrate this through an assessment of sex and ethnicity to the model .
this offers several interesting possibilities : film free hospitals could enhance the quality of the general model through including their own data ; geographic and demographic effects on bone development can be studied ; historic data could be mined to quantify the effects of development drugs .
there are many potential applications for an accurate age model constructed on a diverse and expanding database of images .
there are several obvious ways of improving our models .
we shall include more bones and examine the effect of intensity features .
we can investigate alternative segmentation and outline classification algorithms .
we can attempt to screen for full maturity to improve the no epiphysis models .
we can estimate age via the full tw methodology rather than directly .
our conclusion from the research we have conducted to date is that the feature based system of separating the image processing from the age modelling is the best approach .
it offers flexibility , transparency and produces accurate estimates .
s. mahmoodi , b. sharif , e. chester , j. owen , r. lee , skeletal growth estimation using radiographic image processing and analysis , ieee transactions on information technology in biomedicine 4 ( 4 ) ( 2000 ) 292297 .
r. bull , p. edwards , p. kemp , s. fry , i. hughes , bone age assessment : a large scale comparison of the greulich and pyle , and tanner and whitehouse ( tw2 ) methods , archives of disease in childhood 81 ( 2 ) .
e. pietka , a. gertych , s. pospiech , f. cao , h. huang , v. gilsanz , computer - assisted bone age assessment : image preprocessing and epiphyseal / metaphyseal roi extraction , ieee transactions on medical imaging 20 ( 8) ( 2001 ) 715729 . | bone age assessment estimating age patient from radiograph quantifying development bones non - dominant hand used to child s bones developing acceptable rate monitor treatments affecting skeletal development performed manually using atlas system or bone scoring method tanner whitehouse atlas methods query image to representative hand radiographs ages scoring systems assign each bone to stages combine classifications form age estimate manual procedures time consuming inaccurate automated systems for proposed recreate tw or gp methods or construct regression models for chronological age our approach modular feature based recreate tw scores or predict age directly to predict tw bone stages train classifiers on three transformations of outline first uses technique measures second technique finds discriminatory subsequences constructs classifiers in shapelet finally derive summary shape features based on tw descriptors conclude classifiers built on shape features better on one bone provide greater explanatory power to predict age directly perform linear and non linear regressions from shape feature space to age process on data set images from @xcite in age range 218 .show correct outline recreate tw stages using three bones predict chronological age accurately as clinical experts stepwise feature driven approach to automated bone ageing transparent explicable to clinicians separating feature extraction from segmentation regression potential for constructing new models for regional populations offers possibility producing age estimates tailored to local demographics based on data in film free hospitals paper structured section review current manual methods describe previous attempts automated bone age assessment section extraction format segmented bones into outlines shape features overview classification regression techniques predict bone age sections present results discuss conclusions describe future direction section bone age assessment task performed in hospitals worldwide daily skeletal development of hand assessed using two methods greulich and pyle or tanner and whitehouse bone age estimate compared with chronological age determine if skeletal development abnormal significant difference bone age chronological age paediatrician may diagnose patient with disorder of growth or maturation greulich and pyle method uses atlas of representative hand radiographs range of ages latest edition gp atlas released in 1959 included new images for four new age pointsfinal atlas consists of 31 standard radiographs males to 19 years 27 females newborn to 18 years each standard text describing development clinician checks patient s radiograph against example radiographs appropriate sex key features are development of epiphysis bones presence carpal bones age estimate is age subject representative image selected process subjective large variation between clinicians observed representative images from restricted sample taken over 60 years ago variation changes in diet healthcare culture may mean sample no longer representative criticism method implies assumption ossification process linear may not be true in 1975 tanner _ @xcite published scoring system tw2 two methods calculating bone age described first method uses radius ulna short bones second method uses carpal bones rus method carpal bone technique easier to use each bone has stages stage descriptors classification table [ twstage ] shows image of each stage for distal phalange middle finger with associated criteria once all bones awarded score scores summed to find skeletal maturity score ( sms ) distribution of ages for sms described by centile chart point estimate of bone age derived tw3 method published in 2001maturity stages scores same centile charts updated modern population tab : twstage tw stages of distal phalange three cols=" linear regression model ease exploratory analysis feature relevance epiphysis models include 15 - 17 terms large interactions model for proximal phalange is @xmath0 index each variable corresponds to feature number table tab features model exclusively of epiphyseal features exception @xmath1 metaphysis to width ratio true for other two bones implies future image processing focus on extracting summarising epiphysis features epiphysis width @xmath2 epiphysis distance to phalanx @xmath3 ) common to models first enter stepwise forward selection most important factors account for 80% variability in age model on proximal phalangeal features width distance phalanx has mae of 0.98 rmse of 1.67 easy to construct simple model accurate estimate based on two measurements non - specialist potential for screening for abnormality at low cost benefit linear model if regression assumptions hold construct confidence prediction intervals for new dataimproves utility model decision making decision development abnormal hypothesis test null hypothesis difference between bone age chronological age zero linear model determining differences age between populations address models male female by adding factor sex regressors @xmath5 @xmath6 @xmath7 male @xmath8 female can reject null hypothesis @xmath6 coefficient zero resulting model @xmath9 stepwise model sex fourth variable interaction with @xmath2 significant sex predictive variable future models should include term other demographic variable ethnicity test significance ethnicity include subject asian african - american hispanic only significant factor asian significant in simplified model stepwise model second most important variable different development process for patients with asian ethnicity include sex ethnicity _ dmp _ epiphysis model with three bones has rmse of 0.855 compared human raters 0.89 ) describe three alternative approaches bone outlines classify bone age stage conclude shape features based on tw descriptors most appropriate use shape features construct regression models of chronological age results models predicting tanner - whitehouse stage chronological age as good as other automated bone ageing systems three bones produce age estimates as accurate as human assessors whole imagedata code for classification problems from @xcite benefits of feature based regression models explain importance of individual variables two variables account for 80% variation in age potential for fast screening for abnormalities regression assumptions valid construct confidence prediction intervals diagnosis increasing training set size models reducing variance test importance of alternative demographic variables construct models to specific populations through assessment of sex ethnicity model possibilities film free hospitals enhance quality model data geographic demographic effects on bone development studied historic data quantify effects of development drugs potential applications for accurate age model diverse database images improving models include more bones examine effect of intensity features investigate alternative segmentation classification algorithms screen for full maturity no epiphysis models estimate age via full tw methodology conclusion feature based system separating image processing from age modelling best approach offers flexibility transparency produces accurate estimates s. mahmoodi b. sharif e. chester j. owen r. lee skeletal growth estimation using radiographic image processing analysis information technology in biomedicine 4 ( 4 ) 2000 ) 292297hughes bone age assessment comparison greulich pyle tanner whitehouse methods archives disease childhood 81 pietka gertych pospiech cao huang gilsanz computer assisted bone age assessment image preprocessing epiphyseal metaphyseal extraction medical imaging 20 8) 2001 715729 | 0.441735 | 2,707 | 1,290 | , Saving $0.1 in GPT-4. |
177,248 | ultracam is an ultra - fast , triple - beam ccd camera which has been designed to study one of the few remaining unexplored regions of observational parameter space
high temporal resolution .
the camera will see first light in spring 2002 , at a total cost of @xmath0300 k , and will be used on 2-m , 4-m and 8-m class telescopes to study astrophysics on the fastest timescales .
# 1 1.25 in .125 in .25 in | the fastest timescale variations likely to be observed in an astrophysical environment are milliseconds , corresponding to the innermost orbits around neutron stars and black holes . variations of faint sources on timescales longer than a few seconds have already been explored by conventional ccd instruments .
ultracam will explore the region of observational parameter space which lies between these two extremes , namely photometry of faint objects on timescales of seconds to milliseconds .
the resulting scientific applications will include the study of pulsars , occultations , xrbs ( e.g. searching for the optical analogue of the kilohertz qpos ) , speckle imaging , precise timing studies ( e.g. measuring eclipses ) , echo mapping , eclipse mapping of cvs , asteroseismology , and flickering and oscillations in cv accretion discs .
the five essential requirements for such work are : 1 . the capability of taking short exposures ( from milliseconds to seconds ) with essentially no dead - time between exposures ; 2 .
multi - channels ( 3 or more ) covering a wide wavelength range ( @xmath1 to @xmath2 ) in order to distinguish a blackbody spectrum from a stellar spectrum ; 3 .
simultaneous measurement of the different wavelength bands to avoid problems due to stochastic variations ; 4 .
imaging capability in order to simultaneously measure target , sky and comparison stars and to avoid the need for fixed apertures ; 5 . high efficiency and the capability of being mounted on various large aperture telescopes .
these five requirements result in the instrument design presented here .
* optics . * light from the telescope is first collimated and then split into 3 bandpasses ( @xmath1 , @xmath3 and one of @xmath4 ) using dichroic beamsplitters and sdss filters .
the light from each filter is then re - imaged onto one of 3 ccds at a plate scale of 0.3 arcseconds / pixel , giving a field of view of 5 arcminutes .
this design ensures that : 1 .
there is an 80% probability of finding an r=12 magnitude comparison star in the field ; 2 .
there is no ghosting due to the dichroics ; 3 .
only the collimator needs to be changed to mount ultracam on a different telescope .
* mechanics . *
the opto - mechanical chassis is a double - octopod constructed of aluminium plates and carbon fibre struts , giving a stiff , compact ( 75 cm long ) , lightweight ( 75 kg ) structure which is relatively insensitive to temperature variations .
these characteristics make ultracam highly portable and mountable on both small and large aperture telescopes .
* detectors .
* ultracam will use 3 marconi ( formerly eev ) 47 - 20 aimo , back - thinned , ar - coated ccds .
these will be grade 0 devices ( i.e. of the highest cosmetic quality ) and contain 1024x1024 image pixels , each of 13 microns , with 1024x1024 masked pixels ( i.e. they are frame - transfer devices ) .
the chips exhibit exceptionally high qe ( up to 97% ) and low noise ( 3 e@xmath5/pixel readout noise and @xmath6 e@xmath5/pixel / s dark current , achieved by cooling to 233 k using a 3-stage peltier stack and water cooling ) . * data acquisition and reduction . * data from the 3 ccds is first read by an sdsu controller and then passed to a dual - processor pc running rtlinux .
each exposure is time - stamped by a gps receiver to an accuracy of 0.01 milliseconds and then written to a raid array and archived to tape .
pipeline data reduction software then fully reduces each data frame .
such a data acquisition and reduction system will enable us to observe for a whole night at the highest data rates ( 3.6 mbytes / s ) without stopping , building up light curves in real time .
frame transfer chips allow a completed exposure to be rapidly shifted from the imaging area to the masked area and then read out whilst the next exposure is in progress .
this , in conjunction with the high - speed data acquisition system described above , results in a full - frame readout time of 5 milliseconds ( as long as the exposure time is @xmath7 s ) . by using small windows and stacking the windows in the masked area ( a procedure known as ` drift mode ' ) ,
it is possible to obtain dead - times of 0.05 milliseconds ( as long as the exposure time is @xmath7 milliseconds ) , thereby meeting our requirement of millisecond exposure times with negligible dead time .
the optics and coatings in ultracam have been designed with throughput as the highest priority .
this , in conjunction with the exceptional qe performance of the ccds , means that we expect to detect objects as faint as @xmath8th magnitude at a signal - to - noise ratio of 3 in a 1 millisecond exposure on an 8-m class telescope . | fastest timescale variations in are milliseconds to orbits around neutron stars black holes variations of faint sources on timescales longer than seconds explored by ccd instruments ultracam between these two extremes photometry of faint objects on timescales seconds to milliseconds scientific applications include study of pulsars occultations xrbs speckle imaging precise timing studies echo mapping eclipse mapping of asteroseismology flickering oscillations in cv accretion discs five essential requirements are of short exposures from milliseconds to seconds no dead - time multi - channels ( 3 wide wavelength range to to distinguish blackbody spectrum from stellar spectrum simultaneous measurement of wavelength bands avoid stochastic variations 4 imaging capability to measure target sky comparison stars avoid fixed apertures 5 high efficiency mounted on large aperture telescopes requirements result in instrument design light from telescope collimated split into 3 bandpasses using dichroic beamsplitters sdss filterslight from each filter re - imaged onto 3 ccds at 0.3 arcseconds / pixel field of view 5 arcminutes design ensures 80% probability finding r=12 magnitude comparison star no ghosting collimator mount ultracam on different telescope mechanics opto - mechanical chassis double - octopod aluminium plates carbon fibre struts compact 75 lightweight ( 75 kg structure insensitive to temperature variations ultracam portable mountable on small large aperture telescopes detectors ultracam 3 marconi 47 - 20 aimo back - thinned ar - coated ccds grade 0 devices 1024x1024 image pixels 13 microns 1024x1024 masked pixels chips high qe 97% low noise cooling to 233 k 3-stage peltier stack water cooling data acquisition reduction data from 3 ccds read by sdsu controller passed to dual - processor pc running rtlinux exposure time - stamped by gps receiver 0.01 milliseconds written to raid array archived to tape pipeline data reduction software reduces each data framedata acquisition reduction system night highest data rates 3.6 mbytes / s ) without stopping building light curves real time frame transfer chips allow completed exposure imaging to masked area read out next exposure high - speed data acquisition system results full - frame readout time 5 milliseconds exposure time @xmath7 s using small windows stacking in masked area drift mode dead - times 0.05 milliseconds exposure time @xmath7 milliseconds meeting requirement exposure negligible dead time optics coatings ultracam designed throughput highest priority exceptional performance ccds detect objects @xmath8th magnitude signal - to - noise ratio 3 in 1 millisecond exposure on 8-m class telescope | 0.185369 | 1,066 | 526 | , Saving $0.0 in GPT-4. |
177,250 | this paper applies the major computational tool from topological data analysis ( tda ) , persistent homology , to discover patterns in the data related to professional sports teams .
i will use official game data from the north - american national hockey league ( nhl ) 2013 - 2014 season to discover the correlation between the composition of nhl teams with the currently preferred offensive performance markers .
specifically , i develop and use the program teamplex ( based on the javaplex software library ) to generate the persistence bar - codes .
teamplex is applied to players as data points in a multidimensional ( up to 12-d ) data space where each coordinate corresponds to a selected performance marker .
the conclusion is that team s offensive performance ( measured by the popular characteristic used in nhl called the corsi number ) correlates with two bar - code characteristics : greater _ sparsity _ reflected in the longer bars in dimension 0 and lower _ tunneling _ reflected in the low number / length of the 1-dimensional classes .
the methodology can be used by team managers in identifying deficiencies in the present composition of the team and analyzing player trades and acquisitions .
we give an example of a proposed trade which should improve the corsi number of the team .
the hockey world used to be old fashioned .
managers would recruit players strictly from what they could see with the naked eye .
now , hockey analytics is becoming an influential cog in managing many nhl teams @xcite .
the toronto maple leafs have already hired an assistant manager who is a well - known proponent of data analytics @xcite . in the next five years
, every team is predicted to have at least one `` stat analysis guru '' working with them . unlike baseball , where it is easy to have solid position - specific stats , hockey is a faster and more fluid game .
hockey positions are much more dynamic and fluid , they are better described as roles that the players play .
these are shifting roles between players , especially for forwards .
the nhl keeps track of puck possession , turnovers for and against , hits for and against , shots blocked for and against , face - offs won , and scoring opportunities .
all of that in addition to some standard stats that are easy to record such as shots on goal and , of course , goals scored , goals saved , scores of games , etc .
the dallas stars general manager , jim nill , uses a computer program from the work of 100 college students to measure corsi numbers , turnovers , and scoring opportunities to generate statistical data for his team @xcite .
the sheer size of the data available is impressive , as can be viewed at the hockey analytics website ` extraskater.com ` .
even more data is publicly available on the official nhl website ` nhl.com ` .
it is very unclear how to assemble this information into good use . only recording player stats
does not give the manager the tools to assemble an effective team .
the next step should be the development of tools to analyze this data .
topology is a proper tool for taking information about individual players and generating conclusions about the team .
this is known as a local - to - global transition .
this is where the topological data analysis ( tda ) could be useful .
this paper seems to be the first attempt to apply the major computational tool from tda , the persistent homology , to the data collected by the nhl . | the classical homology theory in topology is a way to describe , in algebraic ways , the presence and number of holes ( of some dimension ) in the given geometric shape or even topological space .
the precise definition and the foundations can be found in munkres @xcite . the simplest algebraic invariant in a given dimension is the betti number which expresses the `` number of holes '' .
the computations are the simplest with the coefficients from a finite field . for our purposes we will always use the field with two elements .
when the shape is a discrete collection of disjoint points , there is only 0-dimensional homology .
but the points might approximate some interesting shape . for example , if the points are in the plane then they might be tracing out some circle which can be extrapolated by a person . in three dimensions
, the points might be lying densely on some sphere . in these cases , the circle has non - zero 1st dimensional homology , and the sphere has non - zero 2nd dimensional homology .
persistent homology is an idea that allows us to recognize these homology classes from the given set of disjoint points .
the articles @xcite give good introductions to this subject .
i will only illustrate how this works in the plane with illustrations called bar - codes created using the program teamplex i developed for this purpose , which extends the javaplex software library from the stanford applied topology research group . in the following pictures
one can see the fattening of the data points happening during the construction of the complex .
the details of this construction are given in section .
the emerging multiple intersections correspond to simplices in the complex . just for intuitive understanding
, one can imagine solid disks merging together into a connected figure .
the hole that you see in figures 3 and 4 represents the 1-dimensional persistent class that we will soon see in the javaplex diagram .
this whole process of fleshing out the circle from the eight points is easy to see in the 2-dimensional plane .
there are two levels of complications that appear in the application in this paper .
even though the number of points which represent players will not increase that much , there will be at most 20 players in the data sets , the number of properties of the players will increase to 12 .
this number is the number of coordinates that describe the points .
so the dimension of the space will increase from 2 to 12 .
the properties of such data sets are impossible to visualize and predict . , _ from 5 to 13 : redefining the positions in basketball _ , 2012 eos / alpha award winnning presentation at at the 2012 mit sloan sports analytics conference . ` | classical homology theory in topology presence number of holes in geometric shape or topological space precise definition foundations in munkres @xcite simplest algebraic invariant in is betti number expresses number of holes computations simplest with coefficients from finite field use field with two elements shape discrete collection of disjoint points 0-dimensional homology points might approximate interesting shape if points in plane tracing out circle extrapolated by person in three dimensions points densely on sphere circle has non - zero 1st dimensional homology sphere has non - zero 2nd dimensional homology persistent homology recognize homology classes from disjoint points articles @xcite introductions illustrate in plane with illustrations bar - codes using program teamplex javaplex software pictures fattening of data points during construction complex details in section . emerging intersections correspond to simplices in complex imagine solid disks merging into connected figure hole in figures 3 and 4 represents 1-dimensional persistent class in javaplex diagram process of fleshing out circle from eight points easy in 2-dimensional plane two levels of complications in application in paperpoints players not increase much most 20 players in data sets properties increase to 12 number is coordinates points dimension space from 2 to 12 properties data sets impossible to visualize predict from 5 to 13 redefining positions in basketball 2012 eos / alpha award presentation 2012 mit sloan sports analytics conference | 0.130143 | 565 | 282 | , Saving $0.0 in GPT-4. |
177,252 | we present results of femtoscopic analysis with @xmath0 , k@xmath1k@xmath1 , pp , @xmath2 , @xmath3 and @xmath4 in pb
pb collisions at @xmath5 tev and @xmath0 in pp collisions at @xmath6 tev registered by alice at the lhc .
the femtoscopic radius is extracted from the one - dimensional correlation functions .
the analysis reveals that the emission source sizes of pions , kaons and protons measured in heavy - ion collisions exhibit approximate transverse mass scaling which is consistent with hydrodynamic collectivity prediction .
furthermore , baryon - antibaryon correlations are investigated to study the influence of final state interactions ( annihilation ) .
it may contribute to lower multiplicities of protons observed at lhc energies , with respect to predictions from thermal models . | the method of two - particle correlations at low relative momenta ( commonly referred to as _ femtoscopy _ ) allows to extract space - time characteristics of the particle emitting source , created in heavy - ion collisions @xcite , @xcite .
correlations of identical pions are usually used to perform this study . in particular , the alice collaboration recently carried out the analysis of two - pion correlations in pp and central pb
pb collisions @xcite , @xcite .
the main goal of kaon and proton femtoscopic study is to complement information about the source size deduced from pion correlations .
the analysis provides an extension of the transverse momentum range .
therefore , it is possible to verify whether the `` homogeneity lengths '' mechanism @xcite is valid for mesons as well as baryons .
furthermore , one can test the transverse mass scaling of the source size which is interpreted as a signature of collective behaviour @xcite .
baryon - antibaryon correlations might be also used to study final state interactions ( fsi ) which can influence particle yields . it is observed that thermal models are not able to reproduce the p/@xmath7 ratio simultaneously with other hadron production ratios @xcite , @xcite . it is argued that rescattering in the hadronic phase should be taken into account while determining particle yields @xcite , @xcite , @xcite .
if the assumption is valid , it ought to be reflected as anticorrelation in baryon - antibaryon correlation functions , due to the annihilation channel .
the dataset of pb pb collisions at @xmath8 tev and pp collisions at @xmath6 tev consist of roughly 30 and 300 million events , respectively .
tracks were reconstructed with the time projection chamber ( tpc ) @xcite .
particle identification of k@xmath1 , p and @xmath9 was performed with the tpc and the time - of - flight detector @xcite in @xmath10 .
primary tracks were selected based on the distance of closest approach to the primary vertex ( dca ) .
@xmath11 ( @xmath12 ) candidates were extracted by identifying daughter tracks using tpc and tof detectors .
the decay channel @xmath13 was used for identification of neutral kaons .
more details about @xmath0 analysis can be found in @xcite .
particles from the same event were paired to form the signal distribution of relative momenta @xmath14 , from two different events to form a mixed background distribution .
pair cuts based on the ratio of clusters ( groups of detector signals ) shared by two tracks to all clusters of both tracks and the angular distance inside tpc between two tracks were used to account for fake low - q pairs ( splitting ) and two - track inefficiency ( merging ) .
after pair selection , the resulting signal distribution was divided by the background distribution to form the correlation function .
@xmath0 correlations were fitted with a parametrization which includes bose - einstein statistics as well as strong fsi @xcite .
femtoscopic radii from charged kaon correlations were obtained using the bowler - sinyukov formula @xcite .
the transverse mass dependence of the radii extracted from @xmath0 and @xmath15 correlations in pp collisions ( presented in the left panel of fig . [ kzspp ] ) appears to be weak . also , the radii increase with increasing event multiplicity . the comparison of the radii obtained from k@xmath1k@xmath1 and @xmath0 femtoscopy in pb
pb collisions is shown in the right panel of fig .
[ kzspp ] .
the results from the two analyses are consistent with each other .
the radii clearly increase with the event multiplicity and decrease with pair transverse momentum ( @xmath16 ) which is compatible with the hydrodynamic description .
the value of the correlation strength @xmath17 is of the order of @xmath18 and it does not show any @xmath19 dependence .
tev @xcite .
right panel : the comparison of @xmath20 and @xmath17 vs. @xmath19 for neutral and charged kaons correlations in pb
pb collisions at @xmath21 tev.,title="fig:",scaledwidth=47.0% ] tev @xcite .
right panel : the comparison of @xmath20 and @xmath17 vs. @xmath19 for neutral and charged kaons correlations in pb
pb collisions at @xmath21 tev.,title="fig:",scaledwidth=52.0% ]
the p@xmath9 correlations ( left panel of fig . [ cfbab ] ) show a maximum for the low relative momentum due to coulomb attraction , then a wide minimum caused by the annihilation part of the strong fsi .
@xmath4 correlation functions ( right panel of fig .
[ cfbab ] ) show a clear suppression at low relative momentum for three centralities ranges .
it is consistent with annihilation in the strong fsi between baryons and antibaryons .
the strength of the correlation effect increases with more peripheral events , an indication that the emitting source size is shrinking for those events .
( left panel ) and @xmath4 ( right panel ) correlation functions from the pb pb collisions at @xmath5 tev.,title="fig:",scaledwidth=49.5% ] ( left panel ) and @xmath4 ( right panel ) correlation functions from the pb
pb collisions at @xmath5 tev.,title="fig:",scaledwidth=49.5% ] the correlations of pp and @xmath2 pairs ( left panel of fig .
[ mtscaling ] ) are due to a combination of fermi - dirac statistics , coulomb and strong fsi which results in a distinct maximum for @xmath22 mev/@xmath23 @xcite .
the height of this peak is related to the femtoscopic radius . due to the fact that feed - down from weak decays can not be neglected in high - energy heavy - ion collisions , the residual correlations from p@xmath11 system in pp correlations have to be taken into account .
since @xmath11 decays into p and @xmath24 with small decay momentum with respect to the mass of p , femtoscopic correlations between a primary p and a @xmath11 might still be detected for a pair consisting of the primary p and the p from @xmath11 decay .
the method of simultaneous fitting of @xmath25 ( @xmath2 , @xmath3 ) and @xmath26 ( @xmath27 , @xmath28 ) correlations was used ( left panel of fig .
[ mtscaling ] ) .
transverse mass scaling of the femtoscopic radii is considered as a signature of collective behaviour of the system created in heavy - ion collisions .
the right panel of fig .
[ mtscaling ] shows transverse mass dependence of the invariant radii scaled by kinematic factor obtained from @xmath29 , k and p femtoscopy .
as shown , the radii seem to follow the same curve which is in agreement with hydrodynamic prediction .
correlation functions , taking into account contribution from residual correlations ( see text for details ) .
right panel : @xmath30 dependence of the radius parameter extracted from correlations of pions , charged kaons , neutral kaons and protons.,title="fig:",scaledwidth=49.0% ] correlation functions , taking into account contribution from residual correlations ( see text for details ) .
right panel : @xmath30 dependence of the radius parameter extracted from correlations of pions , charged kaons , neutral kaons and protons.,title="fig:",scaledwidth=50.0% ]
the femtoscopic parameters for the radius of the emission source sizes of kaons and protons are extracted from one - dimensional correlation functions .
results show an increase of the radius with increasing multiplicity and slight decrease of the radius with increasing pair transverse momentum . 00 g. i. kopylov , m. i. podgoretsky , sov . j. nucl
15 , 219 - 223 ( 1972 ) .
g. i. kopylov , v. l. lyuboshits , m. i. podgoretsky , jinr - p2 - 8069 k. aamodt et al .
( alice collaboration ) , phys.rev .
d84 ( 2011 ) 112004 , arxiv:1101.3665 [ hep - ex ] k. aamodt et al .
( alice collaboration ) , phys.lett .
b696 ( 2011 ) 328 - 337 , arxiv:1012.4035 [ nucl - ex ] s. v. akkelin , y. .m .
sinyukov , phys .
b356 , 525 - 530 ( 1995 ) .
m. a. lisa , s. pratt , r. soltz and u. wiedemann , ann .
55 , 357 ( 2005 ) .
r. preghenella for the alice collaboration , arxiv:1203.5904 [ hep - ex ] l. milano for the alice collaboration , these proceedings k. werner et al .
, phys.rev .
c85 ( 2012 ) 064907 , arxiv:1203.5704 [ nucl - th ] i. karpenko et al . , arxiv:1204.5351 [ nucl - th ] j. steinheimer et al . , arxiv:1203.5302 [ nucl - th ] k. aamodt , et al . , the alice experiment at the cern lhc , jinst 3 ( 2008 ) s08002 b. abelev et al .
( alice collaboration ) phys.lett .
b717 ( 2012 ) pp .
151 - 161 , arxiv:1206.2056 [ hep - ex ] r. lednicky , v. l. lyuboshitz , sov.j.nucl.phys .
35 ( 1982 ) 770 | method two - particle correlations at low momenta femtoscopy space - time characteristics particle emitting source in heavy - ion collisions correlations of identical pions used alice collaboration two - pion correlations in central pb pb collisions goal kaon proton femtoscopic study information source size from pion correlations analysis extension transverse momentum range verify homogeneity lengths mechanism for mesons baryons test transverse mass scaling of source size signature collective behaviour baryon - antibaryon correlations study final state interactions particle yields thermal models reproduce p/@xmath7 ratio with other hadron production ratios argued rescattering in hadronic phase particle yields if assumption valid anticorrelation in baryon - antibaryon correlation functions annihilation channel dataset of pb pb collisions at @xmath8 pp collisions at @xmath6 tev 30 and 300 million events tracks reconstructed with time projection chamber @xcite particle identification of k@xmath1 p @xmath9 performed with tpc time - of - flight detector @xcite @xmath10tracks selected distance closest approach to primary vertex @xmath11 ( @xmath12 ) candidates extracted daughter tracks using tpc tof detectors decay channel @xmath13 used for identification neutral kaons details @xmath0 analysis in @xcite particles from same event paired signal distribution relative @xmath14 events mixed background distribution pair cuts on ratio of tracks to angular distance between tracks for fake low - q pairs two - track inefficiency after signal distribution divided by background distribution correlation function @xmath0 correlations fitted with parametrization bose - einstein statistics strong fsi @xcite femtoscopic radii from charged kaon correlations obtained using bowler - sinyukov formula transverse mass dependence of radii from @xmath0 @xmath15 correlations in pp collisions weak radii increase with event multiplicity comparison radii @xmath0 femtoscopy in collisions shown in right panel fig results consistent radii increase with event multiplicity decrease with pair transverse momentum compatible with hydrodynamic description correlation strength @xmath17 order of @xmath18 @xmath19 dependenceright comparison @xmath20 @xmath17 vs @xmath19 for neutral charged kaons correlations pb collisions at @xmath21 tev:",scaledwidth=47.0% right panel comparison @xmath20 @xmath17 vs @xmath19 neutral charged kaons correlations collisions @xmath21 tev% p@xmath9 correlations left show maximum low relative momentum due to coulomb attraction minimum annihilation strong fsi @xmath4 correlation functions show suppression at low relative momentum three centralities ranges consistent with annihilation in strong fsi between baryons antibaryons strength correlation effect increases with more peripheral events emitting source size shrinking left panel @xmath4 right correlation functions from collisions @xmath5 tev=49.5% correlations of pp @xmath2 pairs left due to fermi - dirac statistics coulomb strong fsi distinct maximum for @xmath22/@xmath23 @xcite height peak related to femtoscopic radiusfeed - down from weak decays in high - energy heavy - ion collisions residual correlations from p@xmath11 system in pp @xmath11 decays into p @xmath24 with small decay momentum mass p femtoscopic correlations between primary p and @xmath11 detected for simultaneous fitting of @xmath25 and @xmath26 correlations used left transverse mass scaling of femtoscopic radii signature of collective behaviour system in heavy - ion collisions right panel shows transverse mass dependence of invariant radii scaled by kinematic factor from @xmath29 , k and p femtoscopy radii follow same curve agreement with hydrodynamic prediction correlation functions contribution residual correlations @xmath30 dependence of radius parameter from correlations of pions charged kaons neutral protons.0% correlation @xmath30 dependence of radius parameter from correlations pions protons.0% femtoscopic parameters for radius emission source sizes of kaons protons extracted from one - dimensional correlation functionsincrease radius multiplicity decrease transverse momentum kopylov podgoretsky nucl 15 219 - 223 ( 1972 ) kopylov lyuboshits podgoretsky p2 - 8069 aamodt alice collaboration.rev d84 2011 ) 112004 arxiv:1101.3665 hep - ex.lett b696 2011 ) 328 - 337 arxiv:1012.4035 nucl - ex akkelin sinyukov phys b356 525 - 530 ( 1995 ) lisa pratt soltz wiedemann ann 55 357 ( 2005 ) preghenella alice collaboration arxiv:1203.5904 hep - ex milano werner phys.rev c85 2012 ) 064907 arxiv:1203.5704 nucl - th karpenko:1204.5351 nucl - steinheimer arxiv:1203.5302 nucl aamodt alice experiment cern lhc jinst 3 2008 ) s08002 abelev alice collaboration phys.lett. b717 2012 ) 151 - 161 arxiv:1206.2056 hep - ex lednicky lyuboshitz.nucl.phys . 35 ( 1982 ) 770 | 0.318209 | 2,204 | 1,108 | , Saving $0.1 in GPT-4. |
177,266 | we present the calculation of the forward jet vertex associated to a rapidity gap ( coupling of a hard pomeron to the jet ) in the bfkl formalism at next - to - leading order ( nlo ) .
real emission contributions are computed via lipatov s effective action .
the nlo jet vertex turns out to be finite within collinear factorization and allows , together with the nlo non - forward gluon green s function , to perform nlo studies of jet production in diffractive events ( e.g. mueller - tang dijets ) .
address = department of physics , brookhaven national laboratory , upton , ny 11973 , usa .
address = institut de physique thorique , cea saclay , f-91191 gif - sur - yvette , france .
address = infn , grupo collegiato di cosenza , i-87036 arcavacata di rende , cosenza , italy .
address = instituto de fsica terica uam / csic , nicols cabrera 15 + & facultad de ciencias , universidad autnoma de madrid , c.u .
cantoblanco , e-28049 madrid , spain .
, altaddress = cern , geneva , switzerland . | more than twenty years ago , mueller and tang @xcite proposed the study of the cross - section for dijet production with an associated large rapidity gap as an ideal environment to test the resummation effects associated to the bfkl pomeron @xcite . in comparison to other observables that test the multi - regge limit of qcd like angular decorrelation of mueller - navelet jets
@xcite , the growth of the proton structure functions with decreasing bjorken-@xmath0 @xcite or the total inclusive @xmath1 cross - section @xcite , hard exclusive diffraction processes with large momentum transfer are sensitive not only to the pomeron intercept but also to its slope , probing the bfkl green s function at @xmath2 , currently known at nlo @xcite . + next -
to - leading logarithmic corrections are known to be sizeable in bfkl physics .
this is expected to be the case at the level of the universal green s function , i.e. the amplitude for pomeron exchange , since nlo corrections incorporate the dependence on the renormalization , factorization and reggeization scales which formally correspond to subleading terms in the lo approximation . however , it turns out that nlo corrections coming from the process - dependent impact factors ( the couplings of the pomeron to the projectile and target ) are also quite important ( see , e.g. @xcite ) .
therefore , in order to provide a reliable description of rapidity gap phenomenology within the bfkl approach , it is mandatory to go beyond the approximations used in previous works @xcite , where nlo corrections were only included ( partially ) at the level of the green s function . +
a quantitative understanding of the mueller - tang cross - section faces a number of challenges . on the one hand ,
the configurations of interest with color singlet exchange , which do not generate any emission into the gap , compete with sudakov - suppressed color exchange contributions where emissions are allowed up to a scale set by the experimental resolution @xmath3 of the rapidity gap definition .
moreover jet - gap - jet events are affected by soft rescatterings of the proton remnants which destroy the gap .
we will focus on the singlet exchange configuration while assuming that the latter contribution can be taken into account through a rapidity gap survival factor . on the other hand , even at the level of the calculation of the singlet exchange contribution ,
we are faced with interesting and challenging questions , for instance : does the exclusive nature of our observable preclude the applicability of collinear factorization ? moreover , does it make sense ( at nlo ) the mueller - tang prescription @xcite to couple the pomeron to a single quark , thus enabling the computation within collinear factorization ? at the impact factor level , we find rather non - trivially that all collinear singularities in the jet are cancelled by dglap renormalization of the pdfs . in any case , a more conclusive answer to these questions can only be given after considering the convolution with the green s function . a promising way to do
this is the monte - carlo implementation of the nlo bfkl equation directly in @xmath4 space already used in @xcite .
in multi - regge kinematics , multi - particle production amplitudes are naturally factorized into a simple expression given by the product of _ reggeon _ propagators and lipatov effective vertices @xcite . in a seminal work @xcite
( see also @xcite ) , lipatov showed that this high - energy factorization can be preserved in a gauge - invariant fashion also when considering logarithmically- ( non - power- ) suppressed contributions in the high - energy limit ( quasi - multi - regge kinematics ) . to do this , it is necessary to introduce an infinite number of induced vertices that follow from the action @xmath5-a_-(x))\partial_\perp^2
a_+(x)+\{+\leftrightarrow -\}],\quad \partial_\pm a_\mp(x)=0.\ ] ] here @xmath6 is the gluon field , @xmath7 the _ reggeon _ field , and @xmath8=-\frac{1}{g}\partial_\pm{\cal p}\exp\left\{\-\frac{g}{2}\int_{-\infty}^{x^\mp}{\rm d}z^\pm v_\pm ( z)\right\}$ ] , the wilson line coupling whose perturbative expansion generates all induced vertices ( fig .
[ induced ] ) .
+ in the original formulation it is tacitly implied that the usual qcd interactions are restricted to the interactions of particles with similar rapidities , while non - local in rapidity interactions are mediated through reggeon exchange , so that no overcounting of degrees of freedom occurs . in recent years
, we devised a manifestly gauge - invariant procedure to avoid this double - counting and make sense of the new appearing rapidity divergences when one considers amplitudes beyond tree level @xcite , and used it to provide an independent computation of the 1-loop corrections to the forward jet vertex and the 2-loop gluon regge trajectory @xcite .
+ lipatov s action thus provides a highly efficient tool to compute radiative corrections for high - energy processes . in this framework ,
the ( bare ) pomeron appears as a state of two reggeons projected into a color singlet state ( fig .
[ fig : schematic ] ) .
high - energy factorization suggests absorbing in the impact factor definition the integral over longitudinal components of the reggeon loop momenta ; therefore , after bfkl resummation of @xmath9 enhanced terms , we have @xmath10 here @xmath11 is the partonic c.o.m .
energy squared , @xmath12 the momentum transfer and @xmath13 the green s function .
we assume that , like in the lo case @xcite , the expression after full resummation encounters no divergences in the integrations over the transverse loop momenta in the direct ( @xmath14 ) and complex conjugate ( @xmath15 ) amplitudes . at lo , the computation of the impact factor ( in @xmath16 ) is very simple ( fig .
[ fig : schematic ] ) , and in the gluon - initiated case we have @xmath17 with @xmath18 and @xmath19 . in a similar way , in the quark - initiated case we have @xmath20 . + channel . ] at nlo , the computation is much more involved .
virtual corrections were already computed in @xcite .
the missing ingredient is the evaluation of the quasielastic emission corrections , like those appearing in fig .
[ quasi ] for the @xmath21 channel .
similarly , we must also compute the diagrams corresponding to the @xmath22 and @xmath23 channels . at the level of the differential partonic impact factor ,
very compact expressions are found @xcite : @xmath24{formula}}\ ] ] where @xmath25 ( resp .
@xmath26 ) is the longitudinal momentum fraction carried by the outgoing parton with momentum @xmath27 ( @xmath28 ) .
the phase space integration in reveals singularities that are manifested as poles in the dimensional regularization parameter @xmath29 . as usual , in order to define an ir and collinear safe cross - section , we need to convolute the partonic cross - section with a jet function distribution @xmath30 @xcite : @xmath31 + for a general jet algorithm , one can then prove that , by using an explicit phase slicing parameter @xmath32 which isolates the singular regions in the integrations , one obtains a finite expression for the nlo corrections to the effective jet vertex @xmath33 in @xmath34 after including the virtual corrections and taking into account the renormalization of the coupling and the dglap renormalization of the parton densities @xcite .
then , it is possible to write within collinear factorization @xmath35 we refer the reader to @xcite for the final expressions for @xmath33 .
we plan to carry out in the near future the numerical implementation of this vertex which , as already remarked , is a necessary step towards reliable phenomenological studies of diffractive jet production at high energies .
we would like to thank the organizers of diffraction 2014 for the oportunity to present our research in such a pleasant environment . m.h .
acknowledges support from u.s .
department of energy ( de - ac02 - 98ch10886 ) and bnl grant ldrd 12 - 034 .
j.d.m . is supported by advanced investigator grant erc - ad-267258 .
a.s.v . acknowledges support from european commission under contract lhcphenonet ( pitn - ga-2010 - 264564 ) , madrid regional government ( hephacosesp-1473 ) , spanish government ( micinn ( fpa2010 - 17747 ) ) and spanish mineco centro de excelencia severo ochoa programme ( sev-2012 - 0249 ) .
v. s. fadin , e. a. kuraev , and l. n. lipatov , _
phys.lett _ * b60 * , 50 ( 1975 ) ; _ sov.phys
. jetp _ * 45 * , 199 ( 1977 ) ; l. n. lipatov , _ sov.j.nucl.phys . _
* 23 * , 338 ( 1976 ) ; ia .
balitsky , and l. n. lipatov , _ sov.j.nucl.phys . _
* 28 * , 822 ( 1978 ) ; l. n. lipatov , _
* 63 * , 904 ( 1986 ) .
a. h. mueller , and h. navelet , _ nucl.phys .
_ * b282 * , 727 ( 1987 ) ; v. del duca , and c.r .
schmidt , _ phys.rev . _ * d49 * , 4510 ( 1994 ) ; a. sabio vera , and f. schwennsen , _
_ * b776 * , 170 ( 2007 ) ; d. colferai , f. schwennsen , l. szymanowski and s. wallon , _ jhep _ * 1012 * , 026 ( 2010 ) ; m. angioni , g. chachamis , j. d. madrigal and a. sabio vera , _ phys.rev.lett . _ * 107 * , 191601 ( 2011 ) ; f. caporale , b. murdaca , a. sabio vera , and c. salas , _ nucl.phys . _ * b875 * , 134 ( 2013 ) ; b. duclou , l. szymanowski , and s. wallon , _ phys.rev.lett . _ * 112 * , 082003 ( 2014 ) ; f. caporale , d. yu .
ivanov , b. murdaca , and a. papa , ` arxiv:1407.8431 ` .
s. j. brodsky , v. s. fadin , v. t. kim , l. n. lipatov , and g. b. pivovarov , _ jetp lett . _ * 70 * , 155 ( 1999 ) ; g. a. chirilli , and yu . v. kovchegov , _ jhep _ * 1405 * , 099 ( 2014 ) ; d. yu .
ivanov , b. murdaca , and a. papa , ` arxiv:1407.8447 ` .
g. chachamis , m. dak , a. sabio vera , and p. stephens , _ nucl.phys .
_ * b849 * , 28 ( 2011 ) ; g. chachamis , and a. sabio vera , _ phys.lett . _ * b709 * , 301 ( 2012 ) ; _ ibid_. * b717 * , 458 ( 2012 ) ; g. chachamis , a. sabio vera , and c. salas , _ phys.rev . _ * d87 * , 016007 ( 2013 ) ; f. caporale , g. chachamis , j. d. madrigal , b. murdaca , and a. sabio vera ,
_ phys.lett . _ * b724 * , 127 ( 2013 ) . m. hentschinski , and a. sabio vera , _ phys.rev . _ * d85 * , 056006 ( 2012 ) ; m. hentschinski , _
_ * b859 * , 129 ( 2012 ) ; g. chachamis , m. hentschinski , j. d. madrigal , and a. sabio vera , _ phys.part.nucl . _ * 45 * , 788 ( 2014 ) . | twenty years ago mueller tang proposed study cross - section dijet production large rapidity gap test resummation effects bfkl pomeron comparison other observables multi - limit decorrelation mueller - navelet jets growth proton structure bjorken- cross - section hard diffraction processes large momentum transfer sensitive to pomeron intercept slope probing bfkl green s function at @xmath2 nlo @xcite next - to - leading logarithmic corrections sizeable in bfkl physics expected at level universal green s function amplitude pomeron exchange nlo corrections incorporate dependence renormalization factorization reggeization scales nlo corrections from process - dependent impact factors couplings pomeron projectile target important reliable description rapidity gap phenomenology bfkl mandatory to go beyond approximations previous works @xcite nlo corrections included at level green s function quantitative understanding mueller - tang cross - section faces challenges configurations with color singlet exchange generate emission into gap compete with sudakov - suppressed color exchange contributions emissions allowed up to scale set experimental resolution @xmath3 rapidity gap definitionjet - gap events affected by soft rescatterings proton remnants destroy gap focus on singlet exchange configuration contribution through rapidity gap survival factor calculation singlet exchange contribution with questions exclusive nature of observable preclude collinear factorization ? mueller - tang prescription couple pomeron to single quark computation within collinear factorization ? all collinear singularities in jet cancelled by dglap renormalization pdfs conclusive answer after considering convolution with green s function promising monte - carlo implementation of nlo bfkl equation in @xmath4 space used in @xcite in multi - regge kinematics multi - particle production amplitudes factorized into expression by product of reggeon _ propagators and lipatov effective vertices @xcite lipatov showed high - energy factorization preserved gauge - invariant considering logarithmically- suppressed contributions in high - energy limit quasi - multi - regge kinematics necessary to introduce infinite number of induced vertices from action @xmath5-a_-(x)\partial_\perp^2(x)=0@xmath6 gluon field @xmath7 reggeon field @xmath8\frac{1}{g}\partial_\pm{ p}\exp\left\frac{g}{2}\int_-\infty}^{x^\mp}{}z^\pm wilson line coupling perturbative expansion generates induced vertices fig . original formulation qcd interactions restricted particles similar rapidities non - local rapidity interactions mediated through reggeon exchange no overcounting freedom devised gauge - invariant procedure avoid double - counting rapidity divergences amplitudes beyond tree level independent computation 1-loop corrections forward jet vertex 2-loop gluon regge trajectory lipatov s action efficient tool compute radiative corrections high - energy processes pomeron appears two reggeons color singlet state fig high - energy factorization impact factor definition integral longitudinal components reggeon loop momenta @xmath9 @xmath10 @xmath11 partonic c.o.m . energy squared @xmath12 momentum transfer @xmath13 green s functionassume lo case @xcite expression after resummation no divergences integrations transverse loop momenta direct @xmath14 complex conjugate @xmath15 ) amplitudes lo computation impact factor @xmath16 ) simple fig . schematic gluon - initiated case @xmath17 @xmath18 @xmath19 quark - initiated case @xmath20 channel nlo computation involved virtual corrections computed @xcite missing evaluation quasielastic emission corrections fig . @xmath21 channel compute diagrams @xmath22 @xmath23 channels differential partonic impact factor compact expressions found @xcite @xmath24 @xmath25 @xmath26 longitudinal momentum fraction outgoing parton with momentum @xmath27 @xmath28 ) phase space integration reveals singularities poles dimensional regularization parameter @xmath29define ir collinear safe cross - section convolute partonic cross - section jet function distribution @xmath30 @xcite @xmath31 general jet algorithm explicit phase slicing parameter @xmath32 isolates singular regions obtains finite expression nlo corrections effective jet vertex @xmath33 @xmath34 virtual corrections renormalization coupling dglap renormalization parton densities @xcite write collinear factorization @xmath35 refer reader @xcite final expressions @xmath33 plan numerical implementation vertex necessary step reliable phenomenological studies diffractive jet production high energies thank organizers diffraction 2014 research support u.s . department of energy bnl grant ldrd 12 - 034 j.d.m. supported advanced investigator grant erc - ad-267258 support european commission madrid regional government spanish government spanish mineco centro de excelencia severo ochoa programme - 0249 v s. fadin e a. kuraev l. n. lipatov phys.lett b60 * , 50 ( 1975 ) sov.phys . jetp 45 * , 199 ( 1977 ) lipatov sov.j.nucl.phys 23 338 ( 1976 ) balitsky lipatov sov.nucl.phys 28 822 ( 1978 ) lipatov 63 904 ( 1986 ) mueller h. navelet nucl.phys b282 727 ( 1987 ) v. del duca c schmidt.rev . d49 4510 ( 1994 ) sabio vera f. schwennsen b776 170 ( 2007 ) colferai schwennsen. szymanowski s. wallon 1012 * 026 ( 2010 ). angioni. chachamis madrigal sabio vera.rev.lett 107 * 191601 ( 2011 ) caporale murdaca sabio vera c. salas nucl.phys . b875 134 ( 2013 ) duclou l. szymanowski s. wallon phys.rev.lett 112 * 082003 ( 2014 ). caporale ivanov murdaca a. papa arxiv:1407.8431 j. brodsky fadin. kim lipatov pivovarov 70 * 155 ( 1999 ). chirilli kovchegov 1405 * 099 ( 2014 )ivanov murdaca papa arxiv:1407.8447 chachamis dak sabio vera p. stephens nucl.phys. b849 28 ( 2011 ) chachamis sabio vera.lett b709 301 ( 2012 ) b717 458 ( 2012 ) chachamis. sabio vera c. salas.rev d87 016007 ( 2013 ) f. caporale. chachamis. madrigal. murdaca. sabio vera.lett b724 * 127 ( 2013 ) hentschinski a. sabio vera.rev d85 056006 ( 2012 ) hentschinski b859 129 ( 2012 ) chachamis hentschinski. madrigal. sabio vera phys.part.nucl 45 788 ( 2014 ) | 0.435287 | 2,849 | 1,507 | , Saving $0.1 in GPT-4. |
177,270 | we present the first numerical results on the binary coalescence of a quark star with a black hole , obtained with a 3-d newtonian smooth particle hydro ( sph ) code .
the star is initially represented by 17,000 particles modeling a self gravitating fluid with the equation of state @xmath0 , and the black hole by a point mass with an absorbing boundary at the schwarzschild radius . as in similar calculations carried out for a stiff polytrope , the stellar core survives the initial episode of mass transfer , but here an accretion disk is clearly formed as well .
[ 1999/12/02 1.01 ( pwd ) ] | binary coalescence of compact objects must give rise to gravitational radiation emission which is expected to be observed with such instruments as ligo or virgo ( e.g. thorne 1995 ) .
neutron stars coalescing with black holes may also be the source of r process nuclei and of gamma
ray bursts ( lattimer and schramm 1976 ) .
coalescence of any compact objects ( black holes , neutron stars , quark stars ) may release enough energy to power a fireball giving rise to some cosmological gamma
ray bursts ( paczyski 1986 ) , and the coalescence of strange ( quark ) stars may be especially promising ( haensel et al .
1991 ) , provided that no quark matter is spilled into the environment with dire consequences for the birth rate of neutron stars ( caldwell and friedman 1991 , kluniak 1994 ) .
more detailed motivation and further references can be found in lee and kluniak ( 1999a ) , where the sph coalescence of a tidally locked neutron star and a black hole is discussed .
lee and kluniak ( 1999b ) discuss the coalescence of a tidally locked stiff polytrope with a black hole , while sph simulations of non rotating neutron stars in binary coalescence with black holes have been presented by lee ( 2000 ) .
we have used the smooth particle hydrodynamics ( sph ) method ( monaghan 1992 ) for our calculations .
our newtonian code is three dimensional and we include a term in the equations of motion that simulates the loss of angular momentum to gravitational radiation in the binary ( this is calculated in the quadrupole approximation for a point
mass binary ) .
we also compute the gravitational radiation waveforms emitted during the coalesence , in the quadrupole approximation .
the strange star is modeled with @xmath1 sph particles , and has mass @xmath2 and radius @xmath3 km .
the black hole is modeled as a point mass producing a newtonian potential , and we model accretion onto it by placing an absorbing boundary at the schwarzschild radius @xmath4 . for the strange star
we take the equation of state to be @xmath5 , with @xmath6 kg m@xmath7 . to perform a dynamical coalescence calculation
, we place the binary components in a close keplerian orbit at a small separation ( a few stellar radii ) and give them the corresponding orbital velocities for point masses .
the strange star is initially spherical and has no spin as seen from an external , inertial reference frame .
the orbit decays through the emission of gravitational waves and because of tidal instabilites of purely newtonian origin ( lai , rasio & shapiro 1993 ) .
here we present the result of one calculation , with an initial mass ratio @xmath8 and initial separation @xmath9 .
the dynamical evolution of the system is followed for approximately 7 ms .
at the start of the simulation a tidal bulge appears on the strange star as a result of the presence of the black hole companion .
the orbital decay causes the star to overflow its roche lobe , beginning mass transfer within the first orbit .
the star is tidally stretched into a tube of practically uniform thickness , and some matter that moves towards the black hole winds around it and produces a torus ( see figure 1 ) .
the outer parts of the star move away from the hole as the stream connecting the two becomes thinner .
small condensations form at regular intervals along the narrow stream ( see panel ( d ) in figure 1 ) , reminiscent of the results of a black hole
neutron star coalescence for a stiff equation of state ( lee 2000 ) .
the mass accretion rate onto the black hole shows a large initial peak at the initial mass transfer episode ( see figure 2 ) , and a subsequent decay , due to the presence of the accretion torus . the small spikes in the curve ( such as the ones visible at @xmath10 ms ) indicate the accretion of small knots along the matter stream coming from the star . in the accretion disk around the black hole
, the density has dropped below @xmath11 and so this matter behaves as pressure free dust in orbit about the hole ( for reference , in figure 1 @xmath12 ) . however , in the surviving core , flung to a large distance during the tidal encounter , and in the small condensations seen in the matter stream connecting the two binary components , the density remains above this value , and thus the matter continues to feel hydrodynamic effects . by @xmath13
ms , there are 0.15 solar masses in the system excluding the black hole , and this is roughly evenly divided between matter with densities above and below @xmath11 .
at the end of the calculation we have attempted to obtain an estimate for the amount matter that could be dynamically ejected from the system , by computing the total mechanical energy ( kinetic + gravitational potential ) of the fluid .
we find that @xmath14 could leave the system , and would originate from the outermost section of the largest blob formed at the end of the tidal tail .
the rest of the matter ( @xmath15 ) is gravitationally bound to the black hole , located practically at the center of mass of the system , and which has acquired a kick velocity from the asymmetric interaction ( @xmath16 km s@xmath17 ) .
some of the matter in the accretion torus has enough angular momentum so as not to be accreted immediatly by the black hole ( i.e. it can remain in orbit in a stable fashion ) , amounting to @xmath18 .
this fluid would presumably be accreted onto the black hole on the viscous time scale as angular momentum is redistributed in the disk .
this work supported in part by kbn grant 2p03d 004 18 , conacyt ( 28987e)and dgapa unam ( in119998 ) | coalescence compact objects gravitational radiation emission observed with ligo virgo thorne 1995 neutron stars coalescing with black holes source r process nuclei gamma ray bursts lattimer schramm 1976 coalescence compact objects black holes neutron stars quark stars release energy fireball cosmological gamma ray bursts paczyski 1986 coalescence of strange quark stars promising haensel 1991 ) no quark matter spilled caldwell friedman 1991 kluniak 1994 detailed motivation references in lee kluniak ( 1999a coalescence tidally locked neutron star black hole coalescence tidally locked stiff polytrope with black hole simulations of non rotating neutron stars coalescence with black holes by lee ( 2000 ) used smooth particle hydrodynamics sph method monaghan 1992 for calculations newtonian code three dimensional term equations loss angular momentum to gravitational radiation calculated quadrupole gravitational radiation waveforms during coalesence strange star modeled with @xmath1 sph particles mass @xmath2 radius @xmath3 kmblack hole modeled as point mass producing newtonian potential model accretion absorbing boundary at schwarzschild radius @xmath4 strange star equation of state @xmath5 with @xmath6 kg m@xmath7 dynamical coalescence calculation place binary components in close keplerian orbit small separation few stellar radii corresponding orbital velocities for point masses strange star initially spherical no spin orbit decays through gravitational waves tidal instabilites newtonian origin result of calculation initial mass ratio @xmath8 separation @xmath9 dynamical evolution system followed for 7 ms tidal bulge appears on strange star black hole companion orbital decay causes star roche lobe beginning mass transfer first orbit star tidally stretched into tube uniform thickness matter towards black hole winds produces torus figure 1 outer parts star move away from hole stream thinner small condensations form along narrow stream panel d figure black hole neutron star coalescence mass accretion rate onto black hole shows large initial peak initial mass transfer episode 2 subsequent decay due to accretion torussmall spikes in curve at @xmath10 ms indicate accretion of knots matter stream from star in accretion disk around black hole density dropped below @xmath11 matter behaves as pressure free dust in orbit figure 1 @xmath12 in surviving core during tidal encounter in small condensations in stream density remains above value matter hydrodynamic effects by @xmath13 ms 0.15 solar masses in system excluding black hole divided between matter with densities above and below @xmath11 attempted estimate for matter ejected from total mechanical energy @xmath14 could leave system from outermost section of largest blob tidal tail rest of matter ( @xmath15 gravitationally bound to black hole at center of mass acquired velocity from asymmetric interaction ( @xmath16@xmath17 some matter in accretion torus has angular momentum not accreted by black hole remain in orbit to @xmath18 fluid accreted onto black hole on viscous time scale as momentum redistributed work supported by kbn grant 2p03d 004 18 conacyt ( dgapa unam ( in119998 ) | 0.189424 | 1,314 | 665 | , Saving $0.0 in GPT-4. |
177,276 | we propose hmdap , a hybrid framework for large - scale data analytical processing on spark , to support multi - paradigm process ( incl .
olap , machine learning , and graph analysis etc . ) in distributed environments .
the framework features a three - layer data process module and a business process module which controls the former
. we will demonstrate the strength of hmdap by using traffic scenarios in a real world . | data analysis has become a useful technique to organize , process , and analyze large amounts of data in order to obtain useful knowledge effectively such as hidden patterns , implicit correlations , future trends , customer preferences , valuable business information etc @xcite .
olap ( _ online analytical processing _ ) @xcite , as a key technology to provide rapid access to data ( mostly relational data ) for analysis via multidimensional structures , enables users ( e.g. , analysts , managers , executives etc . ) to gain useful knowledge from data in a fast , consistent , interactive accessing way .
there are many popular enterprise database management systems for supporting olap .
for example , oracle olap @xcite is oracle s current computing engine for online analytical processing .
ibm company based on the db2 database proposes the ibm db2 olap server @xcite which can analyze the relational database quickly and directly .
microsoft also provides sql server analytic services ( ssas ) @xcite supporting for olap to analyze information , tables , and files scattered across multiple databases .
the characteristics of big data is not confined to only volume and velocity ; it is also referred by the variety , variability and complexity of the data @xcite . due to
the volume , variety and velocity at which the data grows , it is extremely difficult for organisations to process this data for timely and accurate decisions @xcite .
for this challenge , big data analysis @xcite has become a tool to slove the problem .
the primary goal of big data analysis is to help companies make more informed business decisions by enabling data scientists , predictive modelers and other analytics professionals to analyze large volumes of transaction data , as well as other forms of data that may be untapped by conventional business intelligence programs @xcite .
recently , many techniques have been successfully developped for providing big data analysis in various applications .
for example , oracle bigdata @xcite builds on hadoop @xcite through oracle direct connector connecting hadoop and oracle databases .
sql server 2012 @xcite provides the extension service of olap and business intelligence on hadoop to support big data analysis .
ibm smartcloud provides a hadoop - based analytical software infosphere biginsights @xcite which can connect with ibm db2 .
however , those existing techniques of big data analysis are mostly based on olap which is not effective to process data in various models ( e.g. , semi - structure @xcite ) , they do not always bring highly accurate analysis due to the variety and variability of big data in a complicated application for example , the real - time data on the performance of traffic applications or of mobile applications .
besides , how to process big data analysis efficiently is always an important problem when the scale of big data grows exponentially @xcite . in this demonstration
, we propose a hybrid framework for big data analysis on apache spark @xcite ( a high - performance computing architecture ) which builds on hdfs of hadoop .
the framework features a three - layer data process module and a business process module which controls the former . within this framework ,
we can support multi - paradigm data process ( i.e. , a technical connectivity between various disparate process @xcite ) in order to improve the accuracy of analysis , where various big data analysis techniques ( incl .
olap , machine learning , and graph analysis etc . )
are interoperated to process the analysis of various applications of big data ( incl .
data cube @xcite , intelligent prediction , and complex network etc . )
respectively . moreover , our proposed framework built on spark can process large - scale data efficiently .
finally , we implement hmdap and demonstrate the strength of hmdap by using traffic scenarios in a real world .
in figure [ fig : architecture ] , we depict the architecture of our framework consisted of four parts : _ the storage management _ , _ the resource scheduling _ , _ the query analysis _ and _ the business process_. in the following sections , we will introduce each part in detail . in figure
[ fig : storage ] , there are two parts , the physical storage and the logical storage .
the rapid growth of data makes the physical storage of data from single source storage to distributed storage . in order to solve the storage of multi - source data , we adopt the existing distributed file system . in our framework
, it is hdfs ( hadoop distributed file system @xcite ) .
besides , it products many types of data due to the different needs of applications , such as tables , texts , rcfile(the file type of hive ) and sequence data . in order to use these different types of data , we compose the abstract relational views by designing the metadata with semantics to convert data types to the relational data we can handle .
+ in our framework , the development is based on spark and the module of the resource scheduling is assigned to spark .
the figure [ fig : scheduling ] depicts the resource scheduling in our framework .
we use mysql @xcite to query over relational database .
the part of mllib is spark machine learning library .
we call the functions in the library to compute .
graphx is the graph query module of spark .
we user it to query graphs and it provides a possibility to transform the different data formats to graph to query .
+ the module of the query and analysis is located on the top of the framework .
it is not only the entrance to provide services , but also provides the standard syntax and semantic specification of multi - paradigm data analytical processing . at present , hiveql is similar to the standard sql , which is oriented to the classic olap task , and does not deal with the query language based on ml analysis and graph data analysis .
on the basis of not changing the existing query language syntax standard , we develop a multi paradigm for large data fusion analysis query language expanded of machine learning(ml ) and graph analysis .
our big data analysis and processing of the query language is based on the improvement of the fusion of sql and hiveql in multi - paradigm .
first of all , we analyze the support of hiveql and sql respectively and count the amount of operations which can be supported by the traditional relational algebra model . on the basis of the relational algebra model ,
we add other necessary operators to construct an extension of the algebraic language model , which can fully support the operation of hiveql and standard sql .
for the operator with higher complexity , it is split into smaller sub operator or used other methods to optimize it . for the ml analysis ,
we count the commonly used analytical processing methods , such as classification and clustering , and define the abstract interfaces for common ml analysis processing methods . for the graph analysis processing
, we also count the commonly used analytical processing methods , such as the shortest path algorithm , and define the abstract interfaces for them . in this module
, the framework also relates to the implementation of the olap on the relational database and ml and graph data processing tasks on the distributed framework .
the traditional relational database query optimization method is no longer applicable to this situation . according to the different characteristics of relational storage management query engine and distributed file system of computing engine , we summarize the query information and optimize the performance .
firstly , we investigate the statistical index system used in traditional database and analyze the interaction between each index and the index in the system .
then , for each index in the index system of statistical information , we design efficient and accurate sampling methods to calculate the cost model in query optimization . according to the above statistics , we can also design a storage and maintenance programs which is easy to update and manage . and we may use the cost model in the traditional relational database to design a new cost model which can reflect the query cost of the mixed data . figure
[ fig : query ] displays the query analysis .
the main architectural components of the query analysis are _ query _ and _ data analysis process tools(dap tools)_. in the first part , we can query by sql or the function user defined as specified format .
the dap tools contain classical olap , dap on machine learning and dap on graph .
+ our framework provides an analysis method for the large scale data analysis process .
but in the face of complex business processes in different fields , we need the domain knowledge and according to the domain knowledge , we can design the multi - paradigm fusion of analysis task .
we can draw lessons from the method of service composition in service oriented architecture design . in this module , we need to do two things : developing a multi paradigm fusion analysis process orchestration language syntax and the complex business process scheduling method . in the first part
, we need to analyze the patterns and characteristics of service orchestration language in service oriented architecture design and design an abstract model of the executable process . on the basis of the abstract model , we summarize the basic activities of complex business process analysis .
finally , we define the grammar of the business process . in the semantic
, we need to research and analysis the meanings of basic business activities and define the start point , end point and the basic command . in the second part
, we need to study and analyze complex business processes in practical applications .
then , we build complex business process models and refine the way to exchange messages in public business processes . after that
, we need to control the interaction of each part of the resources through the interaction sequence of messages , achieving a reasonable call for each resource service .
we still need to investigate the applicability of existing object - oriented design patterns .
for the analysis of complex business process integration model , we design data business processes .
we refine the design patterns in complex business processes based on the advantages and principles of existing design patterns . in the real world , the business process model is complex and it takes a lot of time to analyze . the figure [ fig : business ] illustrates the details of the business process in our framework .
the user needs to write the configuration files before he or she submits the query .
the format of configuration files are shown in section [ sec : demonstration ] .
when the user submits a query to the framework , the query and analysis module in the framework starts to parse the user s query .
this module parses queries according to predefined semantics , such as xml(extensible markup language ) .
the module transforms the user s query to two parts , the query over relational databases and the query in machine learning .
we default that the user s query including the query over relational databases and the module determines whether or not to carry out the query in machine learning .
we think that when the result of the query over relational databases is null , the framework begins to query in machine learning .
after the analysis module , the framework uses the query over relational databases and the information about the databases which is read from the configuration files to query the relational databases .
then , the framework runs the query in machine learning .
the input of machine learning is the result of querying by relational databases which the query statement is stored in the configuration files . and
the parameters of the machine learning algorithm is also stored in the configuration files .
when the framework gets the information of the machine learning algorithm , it starts to train and calculate and the parameters of the training of the machine learning also comes from the configuration files .
finally , the framework makes a join of the results of two parts .
in this section , we present the interface of hmdap based in javascript , which communicate with the service in java .
we show the screenshot of hmdap in figure [ fig : screenshot ] and the configuration file we mentioned above in figure [ fig : ml ] .
the interface is composed as follows : configuration of machine learning : it is a text to input the path of the configuration file of the machine learning algorithm , such as parameters .
configuration of relation database : it is a text to input the path of the configuration file of the relational databases , such as the user name .
results : it is a text to display the results of the background . run : it is a button to start the program and when the program runs over , the results are shown in the _
results_. save : it is a button to save the context from _ results _ in text file and at the same time , empty all the text box contents .
cancel : cancel the running of this program and empty all the text box contents . and
the details of the configuration file is as follows : configuration : it is the beginning of the configuration file . input : it is the training dataset of the machine learning algorithm .
database : it indicates that the input dataset comes from the relational database as following information url , user , password : they are the parameters to connect to the relational database , the location of the database , the user name and the password of the user .
sql : it is the statement to query the relational database .
parameter : the contents under this label are the parameters of the machine learning algorithms except the input parameter . value : a series of these labels are the values of the parameters .
algorithm : it is the name of algorithm .
for example , the value of _ algorithm _ is _ kmeans _ and our framework runs the algorithm named
_ kmeans _ which is defined in our library .
user can customize the algorithm and give the location of the algorithm in this label . before running the interface , the user should write two configuration files , the configuration of machine learning algorithms as figure [ fig : ml ] and the configuration of relational databases that the contexts are the parts of _ < database > _ in figure [ fig : ml ] . when the user writes two files , he or she should write the paths of the files in the texts on the interface
. then , click the button _
run_. if the user wants to save the results , he or she clicks the button of _
save_. if the user do nt need the results , he or she clicks the button of _ cancel_.
in this demonstration , we proposed hmdap , a hybrid framework for large - scale data analytical processing to support multi - paradigm process on spark .
the multi - paradigm processing mechanism of hmdap can provide the interoperability of data analytical process techniques to process data which might be not effectively handled if we only apply single data analytical process technique . on the other hand , hmdap takes advantage of the high - performance of spark in processing large - scale data .
we believe that hmdap provides a new approach to big data analysis in a multi - paradigm way .
this work is supported by the programs of the key technology research and development program of tianjin ( 16yfzcgx00210 ) , the national key research and development program of china ( 2016yfb1000603 ) , the national natural science foundation of china ( nsfc ) ( 61672377 ) , and the open project of key laboratory of computer network and information integration , ministry of education ( k93 - 9 - 2016 - 05 ) .
xiaowang zhang is supported by tianjin thousand young talents program .
gray j. , chaudhuri s. , bosworth a. , layman a. , reichart d. , venkatrao m. , pellow f. , and pirahesh h. ( 1997 ) .
data cube : a relational aggregation operator generalizing group - by , cross - tab , and sub - totals . , 1(1 ) : 29 - 53 .
meng x. , k.bradley j. , yavuz b. , r.sparks e. , venkataraman s. , liu d. , freeman j. , b.tsai d. , amde m. , owen s. , xin d. , xin r. , m. franklin m. , zadeh z. , zaharia m. , and talwalkar a. ( 2016 ) .
mllib : machine learning in apache spark . , 17:17 .
sheth a. , kochut k. , a. miller j. , worah d. , das s. , lin c. , palaniswami d. , lynch j. , and shevchenko i. ( 1996 ) .
supporting state - wide immunisation tracking using multi - paradigm workflow technology . in : _ proc .
of vldb96 _ , pp . 263273 . | data analysis useful technique to organize process analyze large amounts data obtain useful knowledge hidden patterns implicit correlations future trends customer preferences valuable business information olap online analytical processing key technology rapid access to data mostly relational data for analysis via multidimensional structures enables users analysts managers executives gain useful knowledge fast consistent interactive popular enterprise database management systems olap oracle olap @xcite is s computing engine for online analytical processing ibm proposes ibm db2 olap server @xcite relational database quickly microsoft provides sql server analytic services @xcite for olap analyze information files across multiple databases characteristics big data not volume velocity variety variability complexity data difficult for organisations to process data for timely accurate decisions big data analysis tool problem primary goal analysis help companies make informed business decisions data scientists predictive modelers analytics professionals analyze large volumes transaction data other forms data untapped by conventional business intelligence programs techniques developped for providing big data analysis various applications oracle bigdata @xcite builds on hadoop @xcitesql server 2012 @xcite provides extension service olap business intelligence on hadoop support big data analysis ibm smartcloud provides hadoop - based analytical software infosphere biginsights @xcite with ibm db2 existing techniques big data analysis based on olap not effective process data in various models always bring accurate analysis due to variety variability big data in complicated application process big data analysis efficiently important problem when scale big data grows @xcite propose hybrid framework for big data analysis on apache spark @xcite high - performance computing architecture builds on hdfs hadoop framework features three - layer data process module business process module support multi - paradigm data process technical connectivity between improve accuracy analysis big data analysis techniques olap machine learning graph analysis interoperated process analysis various applications big data proposed framework built on spark process large - scale data efficiently implement hmdap using traffic scenarios in real world figure [ fig : architecture architecture framework four parts storage management resource scheduling query analysis business process introduce each part in detail in figure [ fig : storage ] two parts physical storage logical storagerapid growth of data physical storage from single to distributed adopt existing distributed file system it is hdfs ( hadoop distributed file system @xcite ) products many types of data tables texts rcfile sequence data compose abstract relational views metadata with semantics to convert data types to relational data development based on spark resource scheduling assigned to spark figure scheduling resource scheduling use mysql @xcite to query over relational database mllib is spark machine learning library call functions to compute graphx is graph query module of spark query graphs transform data formats to graph query module of query and analysis on top of framework entrance to services provides standard syntax semantic specification of multi - paradigm data analytical processing hiveql similar to standard sql oriented to classic olap task deal with query language based on ml analysis graph data analysis language develop multi paradigm for large data fusion analysis query language expanded of machine learning graph analysis big data analysis processing query language based on improvement of fusion of sql and hiveqlanalyze support of hiveql and sql count operations supported by traditional relational algebra model add necessary operators construct extension algebraic language model support operation hiveql standard sql operator higher complexity split into smaller sub operator or other methods optimize for ml analysis count commonly used analytical processing methods classification clustering define abstract interfaces graph analysis count commonly used methods shortest path algorithm define abstract interfaces framework relates to implementation olap on relational database ml graph data processing tasks on distributed framework traditional relational database query optimization method no longer applicable characteristics relational storage management query engine distributed file system computing engine summarize query information optimize performance investigate statistical index system in traditional database analyze interaction between each index design efficient accurate sampling methods to calculate cost model in query optimization design storage maintenance programs easy to update manage use cost model in traditional relational database design new cost model reflect query cost of mixed data figure query ] displays query analysis main architectural components are _ query _ data analysis process tools(dap tools) query by sql or function user defined specified formatdap tools contain classical olap machine learning graph framework provides analysis method for large scale data analysis complex business processes in fields need domain knowledge design multi - paradigm fusion analysis task draw lessons from service composition in service oriented architecture design module multi paradigm fusion analysis process orchestration language syntax complex business process scheduling method analyze patterns characteristics service orchestration language in design design abstract model of executable process summarize activities of complex business process analysis define grammar of business process research meanings of basic business activities define start point end point basic command second part study analyze complex business processes in practical applications build complex business process models refine exchange messages in business processes control interaction of resources through sequence messages reasonable call for each resource service investigate applicability of existing object - oriented design patterns for design data business processes refine design patterns based on advantages principles existing design patterns business process model complex takes time to analyze figure business ] illustrates details business process in framework user needs write configuration files before query format of configuration files shown in section demonstration ]user submits query to framework query analysis module starts query module parses queries according predefined semantics xml(extensible markup language module transforms user s query to two parts query over relational databases and query in machine learning default user s query including query over relational databases module determines query in machine learning when result query over relational databases null framework begins query in machine learning after analysis module framework uses query over relational databases information databases from configuration files to query relational databases framework runs query in machine learning input machine learning is result of querying relational databases stored in configuration files parameters of machine learning algorithm stored in configuration files framework gets information machine learning algorithm starts calculate parameters training machine from configuration files framework of results of two parts interface of hmdap based in javascript with service in java screenshot of hmdap in figure [ fig : configuration file interface composed configuration of machine learning text to input path of configuration file algorithm configuration of relation database text input path configuration file relational databases user name results text display results background. run button to start program results shown in results_.save : button to save context from _ results _ in text file empty text box contents cancel cancel running program empty text box contents details of configuration file configuration beginning configuration file input training dataset of machine learning algorithm database input dataset from relational database information url user password parameters to connect database location user name password sql statement to query relational database parameter : contents under label are parameters machine learning algorithms except input parameter value : series of labels are values of parameters algorithm name of algorithm value of algorithm _ is _ kmeans _ framework runs algorithm _ kmeans _ defined in library user can customize algorithm give location algorithm in label before running interface user write two configuration files configuration of machine learning algorithms figure fig : ml ] configuration relational databases database user writes files write paths of files in texts on interface click button _ run_ if user save results clicks button _ save_ if results clicks button _ cancel_ proposed hmdap hybrid framework for large - scale data analytical processing support multi - paradigm process on spark multi paradigm processing mechanism interoperability of data analytical process techniques process datahmdap high - performance spark processing large - scale data provides new approach to big data analysis supported by key technology research development program tianjin national key research development program china national natural science foundation china open project key laboratory computer network information integration ministry of education - 9 - 2016 - 05 ) xiaowang zhang supported by tianjin thousand young talents program gray j chaudhuri bosworth layman reichart venkatrao pellow pirahesh h 1997 ) data cube relational aggregation operator generalizing group - by cross - tab sub - totals 1(1 ) : 29 - 53 meng x yavuz liu freeman amde owen xin franklin zadeh zaharia talwalkar ( 2016 ) mllib machine learning in apache spark 17:17 sheth kochut miller worah lin palaniswami lynch shevchenko i 1996 ) supporting state - wide immunisation tracking using multi - paradigm workflow technology proc . vldb96 pp . 263273 | 0.518473 | 3,384 | 1,598 | , Saving $0.1 in GPT-4. |
177,297 | we investigate scaling and memory effects in return intervals between price volatilities above a certain threshold @xmath0 for the japanese stock market using daily and intraday data sets .
we find that the distribution of return intervals can be approximated by a scaling function that depends only on the ratio between the return interval @xmath1 and its mean @xmath2 .
we also find memory effects such that a large ( or small ) return interval follows a large ( or small ) interval by investigating the conditional distribution and mean return interval .
the results are similar to previous studies of other markets and indicate that similar statistical features appear in different financial markets .
we also compare our results between the period before and after the big crash at the end of 1989 .
we find that scaling and memory effects of the return intervals show similar features although the statistical properties of the returns are different . | in recent years , financial markets have been studied using statistical physics approaches @xcite and some stylized facts have been observed including ( i ) the probability density function ( pdf ) of the logarithmic stock price changes ( log - returns ) has a power - law tail @xcite , ( ii ) the absolute value of log - returns are long - term power - law correlated @xcite .
statistical properties of price fluctuations are important to understand market dynamics , and are related to practical applications @xcite . in particular , the volatility of stocks attracted much attention because it is a key input of option pricing models such as the black - scholes @xcite .
yamasaki _ et al . _
@xcite investigated the return intervals between volatility above a certain threshold in the us stock and foreign exchange markets .
they analyzed _ daily _ data and found scaling and memory effects in return intervals .
@xcite studied the return intervals in _ intraday _ data of the us market , and found similar scaling and memory effects .
@xcite analyzed the memory in volatility return intervals . in this manuscript
, we further test the generality of the above findings in the japanese stock market data where we include both _
daily _ and _ intraday _ data sets .
we find that also in this case the pdf of return intervals mainly depends on the scaled parameter ; the ratio between the return intervals and their mean .
memory effects also exist in the return intervals sequences .
in addition , we study scaling and memory effects considering different type of market dynamic periods .
the japanese market in recent decades ( 1977 - 2004 ) can be divied into two periods , the inflationary ( before 1989 ) and the deflationary ( after 1989 ) .
kaizoji @xcite showed that some statistical properties of the returns are different in the two periods .
the absolute return distribution in the inflationary period behave as a _ power - law _ distribution , while the return distribution in the deflationary period obeys an _
exponential law_. here , we find that scaling and memory effects of the return intervals show similar features in both periods .
in this section , we analyze the statistical properties of the japanese stock market using _ daily _ and _ intraday _ return intervals .
we investigate the _ daily _ data of three representative companies , nippon steel , sony and toyota motor listed on the tokyo stock exchange ( tse ) for the 28-year period from 1977 to 2004 , a total of 7288 trading days . also , we study the _ intraday _ data of 1817 companies listed on the tse from january 1997 to december 1997 .
the sampling time is 1 minute and the data size is about 9 million .
the logarithmic return @xmath3 is written as @xmath4 where @xmath5 is the stock price at time @xmath6 , and the normalized volatility @xmath7 is defined as : @xmath8 where @xmath9 means time average .
we pick every _ event _ of volatility @xmath7 above a certain thresold @xmath0 .
the series of the time intervals between those events , depending on the threshold @xmath0 , @xmath10 , are generated .
we investigate the pdf @xmath11 to better understand its behavior and how it depends on the threshold @xmath1 ( the left panels of fig .
[ fig : pdf_daily ] ) .
the scaled pdf , @xmath12 , as a function of the scaled return intervals @xmath13 is shown in the right panels of fig .
[ fig : pdf_daily ] .
previous study @xcite showed the distributions of @xmath11 are different with different threshold @xmath0 , and we find the same result .
however , when plotting @xmath12 as a function of @xmath13 , we obtain an approximate collapse onto a single curve .
the collapse means that the distributions can be well approximated by the scaling relation @xmath14 the scaling function @xmath15 of eq .
( [ eq : p ] ) does not depend directly on the threshold @xmath0 but only through @xmath16 .
therefore , if @xmath11 is known for one value of @xmath0 , the distribution for other @xmath0 values can be predicted using eq .
( [ eq : p ] ) . figs .
[ fig : pdf_daily]g and [ fig : pdf_daily]h show that the same features , distribution and scaling of return intervals exist for _ intraday _ data after removing the intraday trends @xcite .
the size of the intraday data set is basically larger than that of daily data , and consists of 1817 companies .
therefore , we are able to extend our study to larger values of @xmath0 and get better statistics ( less scattering ) compared to those in fig . [ fig : pdf_daily ] ( a)-(f ) .
( color online ) distribution and scaling of return intervals using for ( a ) and ( b ) nippon steel , ( c ) and ( d ) sony , ( e ) and ( f ) toyota motor , and ( g ) and ( h ) mixture of 1817 japanese companies .
daily data is used for ( a)-(f ) , and intraday for ( g ) and ( h ) .
the sampling time for intraday data is 1 min .
symbols represent different threshold @xmath0 varying from 1 to 2 ( for ( a)-(f ) ) and 1 to 5 ( for ( g ) and ( h ) ) , respectively.,scaledwidth=100.0% ] previous similar studies on the us stock and foreign exchange market @xcite suggested that there might be a universal scaling function for the return time intervals of different financial markets .
we observe the same result also for both daily and intraday data of the japanese market , and it raises the possibility that the scaling function is universal .
we also test whether the sequence of the return intervals is fully characterized by the distribution @xmath11 .
if the sequence of @xmath1 are uncorrelated , the return intervals are independent of each other and chosen from the probability distribution @xmath11 .
however if they are correlated , the memory also affects the order in the @xmath17 time sequence .
@xmath18 represents the conditional pdf which is the probability of finding a return interval @xmath1 following a return interval @xmath19 .
if memory does not exist , we expect that the conditional pdf will be independent of @xmath19 and identical to @xmath11 .
we study @xmath18 for a range of @xmath19 values .
the full data set of @xmath17 is divided into eight subsets with return intervals in increasing order .
we show @xmath20 for @xmath19 being in the lowest subset ( full symbols ) and , in the largest subset ( open symbols ) in fig .
[ fig : memory_daily ] .
the results show that for @xmath19 in the lowest subset , the probability of finding @xmath1 below @xmath2 is enhanced compared to @xmath11 , while the opposite occurs for @xmath19 in the largest subset .
the pdfs , @xmath18 , for all thresholds collapse onto a single scaling function for each @xmath19 .
this suggests that @xmath11 does not characterize the sequence of @xmath1 and memory exists in the sequence .
( color online ) scaled conditional distribution @xmath21 as a funtion of @xmath13 using daily data for ( a ) nippon steel , ( b ) sony , ( c ) toyota motor , and ( d ) mixture of 1817 companies .
symbols represent different threshold @xmath0.,scaledwidth=100.0% ] the memory effects are also observed in the mean conditional return interval @xmath22 which is the first moment of @xmath23 shown in fig .
[ fig : mean ] , where we plot @xmath24 as a function of @xmath25 .
it is seen clearly that large ( or small ) @xmath1 tend to follow large ( or small ) @xmath19 .
note that the shuffled data ( open symbols ) exhibit a flat shape , which means @xmath1 is independent on @xmath19 .
the above results show that the return intervals @xmath1 strongly depend on the previous return interval @xmath19 .
( color online ) scaled mean conditional return interval @xmath24 as a function of @xmath25 for ( a ) nippon steel , ( b ) sony , ( c ) toyota motor , and ( d ) mixture of 1817 companies .
open symbols correspond to shuffled data .
the lines serve only as a guide to the eyes.,scaledwidth=100.0% ] we also analyze clusters of short and long return intervals in order to investigate clustering phenomena , which represent further and longer term correlations compared to @xmath18 and @xmath22 . the sequence of return intervals is devided into two bins by the median of the entire database .
the two bins consist of intervals which are above " and below " the median respectively .
a cluster is formed by @xmath26 consecutive return intervals that are above " or below " the median .
[ fig : cum_histo ] shows the cumulative distribution of clusters of size @xmath26 for three japanese companies and mixture of 1817 companies . both above " and below " clusters have long tails compared to the surrogate volatility shuffled case .
( color online ) cumulative distribution of size for return intervals clusters for ( a ) nippon steel , ( b ) sony , ( c ) toyota motor , and ( d ) mixture of 1817 companies .
the distributions consist of consecutive return intervals that are all above ( closed symbols ) or below ( open symbols ) the median of all the interval records .
the straight lines show the shuffled volatility case ( q=1 , above the median ) where memory is removed.,scaledwidth=100.0% ]
in this section , we investigate the nikkei 225 index data to answer one question : even the return distributions have different features , do the return time intervals show similar features ?
the nikkei 225 index reached the highest position on the last trading day of 1989 , but declined from the first trading day of 1990 .
it has dropped about 63 percent from 1990 to august of 1992 .
this is a famous japanese market bubble and crash .
therefore , the japanese market between 1977 and 2004 can be divided into two parts : the period of inflation , before december 1989 , and the period of deflation , after january of 1990 ( fig .
[ fig : nikkei ] ) .
kaizoji @xcite showed that the return statistics of those two periods are clearly different .
the return distribution in the inflationary period is approximated by an asymptotic power law , while the return distribution in the deflationary period seem to obey an exponential law .
the time series of nikkei 225 ( a ) index and ( b ) log return from january 1984 to december 2004.,scaledwidth=100.0% ] fig .
[ fig : inde](a ) represents the scaled pdf , @xmath12 , as a function of the scaled return intervals @xmath13 in the inflationary period ( full symbols ) and the deflationary period ( open symbols ) .
no significant differences is seen between these two periods . also , conditional mean return intervals of two periods show that @xmath1 depends on @xmath19 in a similar way in the two periods ( fig .
[ fig : inde](b ) ) .
it has been suggested that the pdf @xmath11 and the scaled pdf @xmath12 are universal functions for different financial markets .
here we observe that even though the return distributions of the periods are different , the return intervals show similar features .
( color online ) ( a ) distribution and scaling of return intervals and ( b ) scaled mean conditional return interval @xmath24 as a function of @xmath25 .
full symbols correspond to the inflationary period and open ones the deflationary period .
symbols represent different threshold @xmath0 .
the line is only a guide to the eyes.,scaledwidth=100.0% ]
we investigated scaling and memory effects in volatility return intervals for the japanese stock market using daily and intraday data sets .
for both data sets , we found that the distribution of return intervals are well approximated by a single scaling function that depends only on the ratio @xmath13 , and the scaling function is different from the poisson distribution expected for uncorrelated records . also , our results for the conditional distribution and mean return interval support the memory between subsequent return intervals , such that a large ( or small ) return interval is more likely to be followed by a large ( or small ) interval .
the clustering shown in fig .
[ fig : cum_histo ] shows that the memory exists even between nonsubsequent return intervals .
our results also support the possibility that the scaling and memory properties are similar functions for different financial markets .
in addition , we tested scaling and memory effects in the inflationary and deflationary periods of the japanese market .
while the return distributions show different features , the scaling and memory properties of the return intervals are similar .
it should be noted that similar scaling properties and memory in the return intervals have been found earlier also in climate @xcite and earthquakes @xcite .
f. wang , k. yamasaki , s. havlin , and h. e. stanley , phys .
e * 73 * , 026117 ( 2006 ) ; f. wang , p. weber , k. yamasaki , s. havlin , and h. e. stanley , eur .
j. b * 55 * , 123 ( 2007 ) ; f. wang , k. yamasaki , s. havlin , and h. e. stanley , arxiv:0707.4638 ( 2007 ) . | recent financial markets studied using statistical physics approaches stylized facts observed including probability density function of logarithmic stock price changes log - returns power - law tail absolute value of log - returns long - term power - law correlated statistical properties of price fluctuations important market dynamics related to practical applications volatility of stocks key input of option pricing models yamasaki investigated return intervals between volatility above threshold in us stock foreign exchange markets analyzed daily data found scaling memory effects in return intervals studied return intervals in intraday data market found similar scaling memory effects analyzed memory in volatility return intervals test generality findings in japanese stock market data daily intraday data pdf return intervals depends on scaled parameter ratio mean memory effects exist in return intervals sequences study scaling memory effects considering different market dynamic periods japanese market 1977 - 2004 divied into two periods inflationary deflationary after 1989 statistical properties of returns different in two periods absolute return distribution in inflationary period power - law distribution deflationary period exponential lawscaling memory effects return intervals show similar features periods analyze statistical properties japanese stock market using daily intraday return intervals investigate daily data three companies nippon steel sony toyota motor tokyo stock exchange 28-year period 1977 to 2004 7288 trading days study intraday data of 1817 companies january 1997 to december 1997 sampling time 1 minute data size 9 million logarithmic return @xmath3 written @xmath4 @xmath5 stock price @xmath6 normalized volatility @xmath7 @xmath8 @xmath9 time average pick every event volatility @xmath7 above thresold @xmath0 time intervals between events depending threshold @xmath0 generated investigate pdf @xmath11 understand behavior threshold @xmath1 scaled pdf @xmath12 function of scaled return intervals @xmath13 right panels study showed distributions of @xmath11 different with threshold @xmath0 find same result plotting @xmath12 as function of @xmath13 obtain approximate collapse single curve distributions approximated by scaling relation @xmath14 scaling function @xmath15 ofeq : p ] ) depend on threshold @xmath0 only through @xmath16 if @xmath11 known for one value @xmath0 distribution for other values predicted using eq figs show same features distribution scaling of return intervals for intraday data after removing intraday trends size intraday data set larger than daily data consists of 1817 companies extend study to larger values of @xmath0 better statistics less scattering)-(f distribution scaling of return intervals using nippon steel sony toyota motor of 1817 japanese companies daily data used for ( a)-(f ) intraday for ( g ) and ( h ) sampling time for intraday data 1 min symbols represent threshold @xmath0 from 1 to 2 for-(f 1 to 5 for g h,scaledwidth=100.0% previous studies on stock suggested universal scaling function for return time intervals of different financial markets same result for daily intraday data of japanese market possibility scaling function universal test sequence of return intervals characterized by distribution @xmath11 sequence @xmath1 uncorrelated return intervals independent chosen from probability distribution @xmath11 if correlated memory affects order @xmath17 time sequence @xmath18 represents conditional pdf probability finding return interval @xmath1 following @xmath19 if memory conditional pdf independent of @xmath19 identical to @xmath11 study @xmath18 for range @xmath19 values data set @xmath17 divided into eight subsets return intervals increasing order @xmath20 @xmath19 lowest subset largest subset fig @xmath19 lowest subset probability finding @xmath1 below @xmath2 enhanced opposite @xmath19 largest subset pdfs @xmath18 thresholds collapse single scaling function @xmath19 suggests @xmath11 characterize sequence @xmath1 memory exists sequence scaled conditional distribution @xmath21 funtion of @xmath13 using daily data for nippon steel sony toyota motor 1817 companies symbols represent different threshold memory effects observed in mean conditional return interval @xmath22 first moment of @xmath23 fig @xmath24 function of @xmath25large @xmath1 follow @xmath19 shuffled data flat shape @xmath1 independent on @xmath19 return intervals @xmath1 depend on previous @xmath19 scaled mean conditional return interval @xmath24 function of @xmath25 for nippon steel sony toyota motor 1817 companies open symbols correspond to shuffled data lines guide analyze clusters of short long return intervals investigate clustering phenomena correlations compared to @xmath18 @xmath22 sequence of return intervals devided into two bins by median database intervals above below median cluster formed by @xmath26 return intervals above or below median fig : cum_histo shows cumulative distribution of clusters of size @xmath26 for three japanese companies 1817 companies above below clusters have long tails compared to surrogate volatility shuffled case cumulative distribution of size for return intervals clusters for nippon steel sony toyota motor 1817 companies distributions of consecutive return intervals above or below median of interval records straight lines show shuffled volatility case ( q=1 above median where memory removedinvestigate nikkei 225 index data question return distributions different return time intervals similar features ? nikkei 225 index highest position last trading day 1989 declined from first trading day 1990 dropped 63 percent from 1990 to august 1992 famous japanese market bubble and crash japanese market between 1977 and 2004 divided into two parts inflation before december 1989 deflation after january 1990 return statistics of two periods different return distribution inflationary period approximated by asymptotic power law deflationary period obey exponential law time series of nikkei 225 index log return from january 1984 to december 2004. fig represents scaled pdf @xmath12 function of scaled return intervals @xmath13 in inflationary deflationary no significant differences between periods conditional mean return intervals periods show @xmath1 depends on @xmath19 similar suggested pdf @xmath11 and scaled pdf @xmath12 universal functions for different financial markets return distributions periods different return intervals show similar features distribution scaling return intervals scaled mean conditional return interval @xmath24 function of @xmath25 correspond to inflationary period deflationary periodsymbols represent threshold @xmath0 line guide to eyes,scaledwidth=100.0% investigated scaling memory effects in volatility return intervals japanese stock market daily intraday data sets distribution return intervals approximated by single scaling function on ratio @xmath13 different from poisson distribution for uncorrelated records results for conditional distribution mean return interval support memory between return intervals large or small return interval likely followed by large interval clustering in fig . shows memory exists between nonsubsequent return intervals results support scaling memory properties similar for different financial markets tested scaling memory effects in inflationary deflationary periods japanese market return distributions show different features scaling memory properties intervals similar similar scaling properties memory found in climate earthquakes f. wang k. yamasaki s. havlin h. e. stanley phys . e * 73 * , 026117 ( 2006 ). b * 55 * , 123 ( 2007 ) arxiv:0707.4638 ( 2007 ) | 0.442357 | 3,000 | 1,508 | , Saving $0.1 in GPT-4. |
177,298 | we present an estimation of the contribution of double parton scattering ( dps ) for jets widely separated in rapidity and for four - jet sample . in the case of four - jet production
we calculate cross section for both single - parton scattering ( sps ) using the code alpgen as well as for dps in lo collinear approach . the dps contribution is calculated within the so - called factorized ansatz and each step of dps
is calculated in the lo collinear approximation .
we show that the relative ( with respect to sps dijets and to the bfkl mueller - navelet ( mn ) jets ) contribution of dps is growing at large rapidity distance between jets .
the calculated differential cross sections as a function of rapidity distance between the most remote in rapidity jets are compared with recent results of ll and nll bfkl calculations for the mueller - navelet jet production at @xmath0 tev .
our results for four - jet sample are compared with experimental data obtained recently by the cms collaboration and a rather good agreement is achieved .
we propose to impose different cuts in order to enhance the relative dps contribution .
the relative dps contribution increases when decreasing the lower cut on the jet transverse momenta as well as when a low lower cut on the rapidity distance between the most remote jets is imposed . | many years ago mueller and navelet predicted strong decorrelation in relative azimuthal angle @xcite of jets with large rapidity separation due to exchange of the bfkl ladder between quarks ( see left panel of fig.[fig : diagrams ] ) . since then both leading - logarithmic and higher - order bfkl effects were calculated and discussed .
the effect of the nll correction is large and leads to significant lowering of the cross section .
the lhc opens a new possibility to study the decorrelation in azimuthal angle .
first experimental data measured at @xmath1 = 7 tev are expected soon @xcite .
also double parton scattering ( dps ) can be important in this context ( for diagrammatic representation of dps see right panel of fig.[fig : diagrams ] ) .
we discussed recently how important is the contribution of dps in the case of the jets widely separated in rapidity @xcite and for four - jet sample @xcite .
four - jet production was already discussed in the context of double parton scattering .
actually it was a first process where the dps was claimed to be observed experimentally @xcite .
however , in most of the past as well as current analyses the dps contribution to four - jet production is relatively small and single parton scattering ( sps ) driven by the 2 @xmath2 4 partonic processes dominates . on the theoretical side
the dps effects in four - jet production were discussed in refs .
a first theoretical estimate of sps four - jet production , including only some partonic subprocesses , and its comparison to dps contribution was presented in ref .
@xcite for tevatron .
some new kinematical variables useful for identification of dps were proposed in ref .
presence of perturbative parton splitting mechanism was discussed in ref .
@xcite . in our recent studies
we have shown how big can be the contribution of dps for jets widely separated in rapidity @xcite .
understanding of this contribution is important in the context of searching for bfkl effects or in general qcd higher - order effects @xcite . in the present letter
we wish to discuss also exclusive four - jet sample where the situation in the context of searching for dps is even better @xcite . in ref.@xcite
we showed how to maximize the dps contribution by selecting relevant kinematical cuts . here
we shall show only some examples .
-0.3 cm partonic cross sections used to calculate dps are calculated only in leading order .
then the cross section for dijet production can be written as : @xmath3 in our calculations we include all leading - order @xmath4 partonic subprocesses .
the @xmath5-factor for dijet production is rather small , of the order of @xmath6 .
it was shown in ref.@xcite that already the leading - order approach gives results in sufficiently reasonable agreement with recent atlas and cms inclusive jet data .
this simplified leading - order approach can be however used easily in calculating dps differential cross sections .
the multi - dimensional differential cross section can be written as : @xmath7 where @xmath8 and partons @xmath9 .
the combinatorial factors include identity of the two subprocesses .
each step of dps is calculated in the leading - order approach ( see eq.([lo_sps ] ) ) . in the calculations we have taken in most cases
@xmath10 = 15 mb .
phenomenological studies of @xmath10 summarized e.g. in @xcite give a similar value .
in fig . [ fig : deltay1 ] we show distribution in the rapidity distance between two jets in lo collinear calculation and between the most distant jets in rapidity in the case of four dps jets . in this calculation
we have included cuts for the cms expriment @xcite : @xmath11 ( -4.7,4.7 ) , @xmath12 ( 35 gev , 60 gev ) . for comparison
we show also results for the bfkl calculation from ref . @xcite . for this
kinematics the dps jets give relatively sizeable contribution only at large rapidity distance . the nll bfkl cross section ( long - dashed line ) is smaller than that for the lo collinear approach ( short - dashed line ) .
the collinear pqcd result is shown by the short - dashed line and the dps result by the solid line for @xmath1 = 7 tev ( left panel ) and @xmath1 = 14 tev ( right panel ) .
for comparison we show also results for the bfkl mueller - navelet jets in leading - logarithm and next - to - leading - order logarithm approaches from ref .
, title="fig:",width=188 ] 60 gev ) .
the collinear pqcd result is shown by the short - dashed line and the dps result by the solid line for @xmath1 = 7 tev ( left panel ) and @xmath1 = 14 tev ( right panel ) .
for comparison we show also results for the bfkl mueller - navelet jets in leading - logarithm and next - to - leading - order logarithm approaches from ref .
, title="fig:",width=188 ] -0.3 cm in fig .
[ fig : deltay-2 ] we show rapidity - difference distribution for even smaller lowest transverse momenta of the jet .
a measurement of such jets may be , however , difficult .
now the dps contribution may even exceed the standard sps dijet contribution , especially at the nominal lhc energy .
one could also try to measure correlations of semihard ( @xmath13 10 gev ) neutral pions with the help of so - called zero - degree calorimeters ( zdc ) @xcite .
+ -0.3 cm
first we wish to demonstrate how reliable our sps four jet calculation is . in fig .
[ fig : cms - data-1 ] we compare the results of calculation with the leading - order code alpgen @xcite with recent cms experimental data @xcite . in this analysis
the cms collaboration imposed different transverse momentum cuts on the leading , subleading , 3@xmath14 and 4@xmath15 jets . in this calculation
we have used an extra @xmath5-factor to effectively include higher - order effects @xcite .
we get relatively good description of both transverse moentum and pseudorapidity distributions of each of the four ( ordered in transverse momentum ) jets .
therefore we conclude that the calculation with the alpgen generator can be a reliable sps reference point for an exploration of the dps effects . having shown that our approach is consistent with existing lhc four - jet data
we wish to discuss how to find optimal conditions for `` observing '' the dps effects .
as shown in our previous paper @xcite on dijets widely separated in rapidity the distribution in rapidity separation of such jets seems a very good observable for observing the dps . in fig .
[ fig : deltay - dps-1 ] we show some examples of such distributions for different cuts on the jet transverse momenta for two collision energies @xmath1 = 7 tev and @xmath1 = 14 tev obtained with the condition of the four - jet observation .
we focus only on the distance between the most remote jets and do not check what happens in between . the higher collision energy or the smaller the lower transverse momentum cut the bigger the relative dps contribution . here
the relative dps contribution is much bigger than for jets widely separated in rapidity ( compare with fig .
[ fig : deltay1 ] and [ fig : deltay-2 ] ) .
in such a case one can therefore expect a considerable deficit when only sps four jets are included .
such cases would be therefore useful to `` extract '' the @xmath10 parameter .
any deviation from the `` canonical '' value of 15 mb would therefore shed new light on the underlying dynamics .
for example , a two - component model discussed in refs .
@xcite strongly suggests such dependences .
= 7 tev ( left column ) and @xmath1 = 14 tev ( right column ) for different cuts on jet transverse momenta ( identical for all four jets ) . , scaledwidth=100.0% ] = 7 tev ( left column ) and @xmath1 = 14 tev ( right column ) for different cuts on jet transverse momenta ( identical for all four jets ) .
, scaledwidth=100.0% ] = 7 tev ( left column ) and @xmath1 = 14 tev ( right column ) for different cuts on jet transverse momenta ( identical for all four jets ) .
, scaledwidth=100.0% ] = 7 tev ( left column ) and @xmath1 = 14 tev ( right column ) for different cuts on jet transverse momenta ( identical for all four jets ) .
, scaledwidth=100.0% ]
we have discussed how the double - parton scattering effects may contribute to large - rapidity - distance dijet correlations . as an example we have shown distributions in rapidity distance between the most - distant jets in rapidity .
the relative contribution of the dps mechanism increases with increasing distance in rapidity between jets .
we have also shown some predictions of the mueller - navelet jets in the ll and nll bfkl framework .
for the cms configuration our dps contribution is smaller than the dijet sps contribution , but only slightly smaller than that for the nll bfkl calculation .
we have shown that the relative effect of dps can be increased by lowering the jet transverse momenta . in this presentation
we have also discussed how to enhance the relative contribution of double - parton scattering for four - jet production .
first we have confronted results of our calculations with those obtained at the lhc by the cms collaborations .
the comparison indicates some evidence of dps at large pseudorapidities of the leading jet .
we have shown that imposing a lower cut on transverse momenta and rapidity distance between the most remote jets improves the situation considerably enhancing the relative contribution of dps .
a dedicated analysis of the dps effect is possible already with the existing data sample at @xmath1 = 7 tev .
the situation at larger energies , relevant for lhc run 2 should be even better . as a consequence we predict that azimuthal correlation between jets widely separated in rapidity should dissapear in the considered kinematical domain @xcite .
we have found that in some corners of the phase space the dps contribution can go even above 80 % @xcite . in this presentation
we have presented the detailed predictions .
once such cross sections are measured , one could try to extract the @xmath10 parameter from the four - jet sample and try to obtain its dependence on kinematical variables .
such dependence can be expected due to several reasons such as parton - parton correlations , hot spots , perturbative parton splitting .
f. abe _ et al .
_ [ cdf collaboration ] , phys . rev .
d * 47 * , 4857 ( 1993 ) .
b. ducloue , l. szymanowski and s. wallon , j. high energy phys . * 05 * , 096 ( 2013 ) ; arxiv:1302.7012 [ hep - ph ] . n. paver , and d. treleani , phys . lett . *
b146 * , 252 ( 1984 ) .
m. l. mangano , z. phys .
c * 42 * , 331 ( 1989 ) .
s. domdey , h. j. pirner and u. a. wiedemann , eur .
j. c * 65 * , 153 ( 2010 ) [ arxiv:0906.4335 [ hep - ph ] ] .
e. l. berger , c. b. jackson and g. shaughnessy , phys .
d * 81 * , 014014 ( 2010 ) [ arxiv:0911.5348 [ hep - ph ] ]
. b. blok , y. dokshitzer , l. frankfurt and m. strikman , phys .
d * 83 * , 071501 ( 2011 ) [ arxiv:1009.2714 [ hep - ph ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
d * 89 * , 092010 ( 2014 ) .
cms collaboration [ cms collaboration ] , cms - pas - fsq-12 - 002 .
f. abe et al .
( cdf collaboration ) , phys . rev . *
d56 * ( 1997 ) 3811 ; + phys .
* 79 * ( 1997 ) 584 .
m. l. mangano , m. moretti , f. piccinini , r. pittau and a. d. polosa , jhep * 0307 * , 001 ( 2003 ) [ hep - ph/0206293 ] .
z. bern , g. diana , l. j. dixon , f. febres cordero , s. hoeche , d. a. kosower , h. ita and d. maitre _ et al .
_ , phys .
lett . * 109 * , 042001 ( 2012 ) .
j. r. gaunt , j. high energy phys . * 01 * ( 2013 ) 042 . | ago mueller navelet predicted decorrelation azimuthal angle @xcite of jets large rapidity separation due to exchange bfkl ladder between quarks left panel leading - logarithmic higher - order bfkl effects calculated discussed effect nll correction large leads lowering cross section lhc opens possibility study decorrelation azimuthal angle experimental data at @xmath1 = 7 tev expected soon double parton scattering dps important representation right panel discussed contribution dps jets widely separated rapidity four - jet sample production discussed double parton scattering first process dps observed experimentally analyses dps contribution four - jet production small single parton scattering partonic processes dominates dps effects four - jet production discussed in first theoretical estimate of sps four - jet production comparison to dps contribution presented in ref . @xcite new kinematical variables for identification dps proposed ref perturbative parton splitting mechanism discussed in ref . @xcite recent studies shown contribution dps for jets widely separated rapidity @xciteunderstanding contribution important searching bfkl effects qcd higher - order effects @xcite present letter discuss four - jet sample situation searching dps better dps contribution selecting relevant kinematical cuts show examples -0.3 cm partonic cross sections dps calculated leading order cross section dijet production written as @xmath3 calculations include leading - order @xmath4 partonic subprocesses @xmath5-factor for dijet production small @xmath6 leading - order approach results reasonable agreement with atlas cms jet data simplified approach used calculating dps differential cross sections multi - dimensional differential cross section written @xmath7 @xmath8 @xmath9 combinatorial factors include identity two subprocesses each step dps calculated leading - order approach @xmath10 = 15 mb phenomenological studies of @xmath10 give similar value fig . distribution rapidity distance between two jets distant jets four dps jets included cuts for cms expriment @xcite @xmath11 ( -4.7,4.7 ) @xmath12 ( 35 60 gev ) comparison results bfkl calculation from @xcitedps jets give contribution at large rapidity distance nll bfkl cross section ( long - dashed line smaller lo collinear approach short - dashed line collinear pqcd result shown by short - dashed line dps result solid line for @xmath1 = 7 tev = 14 tev comparison results bfkl mueller - navelet jets in leading - next - to - leading - order logarithm approaches collinear pqcd result shown by short - dashed line dps result solid line for @xmath1 = 7 tev = 14 tev results mueller - navelet jets approaches -0.3 cm fig rapidity - difference distribution for smaller lowest transverse momenta jet measurement jets difficult dps contribution may exceed standard sps dijet contribution at nominal lhc energy measure correlations of semihard ( @xmath13 10 gev ) neutral pions with zero - degree calorimeters + -0.3 cm sps four jet calculation results with leading - order code alpgen @xcite recent cms experimental dataanalysis cms collaboration imposed transverse momentum cuts on leading subleading 3@xmath14 4@xmath15 jets calculation used extra @xmath5-factor include higher - order effects @xcite good description of transverse moentum pseudorapidity distributions of four jets calculation with alpgen generator reliable reference point for dps effects approach consistent with lhc four - jet data discuss optimal conditions for observing dps effects @xcite on dijets widely separated in rapidity distribution in rapidity separation good observable for observing dps fig . examples of distributions for different cuts jet transverse momenta for two collision energies @xmath1 = 7 tev = 14 tev four - jet observation focus on distance between remote jets check in between higher collision energy lower transverse momentum cut bigger relative dps contribution dps contribution bigger than for jets widely separated in rapidity expect considerable deficit when only sps four jets included useful to extract @xmath10 parameter deviation from value 15 mb light on underlying dynamics two - component model @xcite suggests dependences7 tev left column @xmath1 14 tev right column cuts jet transverse momenta identical four jets scaledwidth=100.0% 7 tev left column @xmath1 14 tev right column cuts transverse momenta identical four jets scaledwidth.0% 7 tev left column @xmath1 14 tev right column cuts momenta identical four jets scaledwidth.0% 7 tev left column @xmath1 14 tev right column cuts transverse momenta identical jets discussed double - parton scattering large - rapidity - distance dijet correlations shown distributions rapidity distance distant jets relative contribution dps increases distance rapidity jets predictions mueller - navelet jets ll nll bfkl framework cms configuration dps contribution smaller dijet sps contribution slightly smaller nll bfkl calculation relative effect dps increased lowering jet transverse momenta discussed enhance relative contribution double - parton scattering four - jet production confronted results calculations cms collaborations comparison indicates evidence dps large pseudorapidities leading jet lower cut transverse momenta rapidity distance remote jets improves situation relative contribution dpsanalysis dps effect possible with data sample at @xmath1 = 7 tev larger energies lhc run 2 better predict azimuthal correlation between jets separated rapidity dissapear in kinematical domain @xcite phase space dps contribution above 80 % @xcite detailed predictions cross sections measured extract @xmath10 parameter from four - jet sample obtain dependence on kinematical variables expected due to parton - parton correlations hot spots perturbative parton splitting f. abe et al . collaboration phys . 47 *, 4857 ( 1993 ) ducloue szymanowski wallon high energy phys . 05 *, 096 ( 2013 ) arxiv:1302.7012. paver. treleani phys. b146, 252 ( 1984 ) mangano phys 42, 331 ( 1989 ) domdey pirner wiedemann 65 *, 153 ( 2010 ) arxiv:0906.4335 hep - ph berger jackson shaughnessy phys 81 * 014014 ( 2010 ) arxiv:0911.5348 blok dokshitzer frankfurt strikman phys83 071501 2011 ) arxiv:1009.2714 hep - ph chatrchyan al cms collaboration phys 89 092010 2014 ) cms collaboration - pas - fsq-12 - 002 abe cdf collaboration phys . rev d56 1997 ) 3811 . 79 1997 ) 584. mangano moretti piccinini pittau polosa 0307 001 ( 2003 ) hep - ph/0206293 z. bern. diana dixon febres cordero hoeche kosower h. ita d. maitre al. phys . 109 042001 ( 2012 ) gaunt high energy phys . 01 2013 ) 042 | 0.438449 | 2,966 | 1,464 | , Saving $0.1 in GPT-4. |
177,311 | we employ the relativistic constituent quark model to give a unified description of the leptonic and semileptonic decays of pseudoscalar mesons ( @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 ) .
the calculated leptonic decay constants and form factors are found to be in good agreement with available experimental data and other approaches .
we reproduce the results of spin - flavor symmetry in the heavy quark limit . @xmath6 | semileptonic decays of pseudoscalar mesons allow to evaluate the elements of the cabibbo - kobayashi - maskawa ( ckm ) matrix , which are fundamental parameters of the standard model .
the decay @xmath7 provides the most accurate determination of @xmath8 , the semileptonic decays of d and b mesons , @xmath9 , @xmath10 and @xmath11 , can be used to determine @xmath12 , @xmath13 and @xmath14 , respectively .
the effects of strong interactions in these processes can be expressed in terms of form factors , which depend on @xmath15 , the squared momentum transferred to the leptonic pair .
information on the form factors are obtained by measuring the distributions of @xmath15 and decay angles .
the decays of heavy d and b mesons are of particular interest due to the spin - flavor symmetry observed for infinite quark masses @xcite . this symmetry allows to reduce the number of form factors and express them in terms of the universal isgur - wise function @xcite . also the scaling laws derived for some physical observables can be , in principle , tested experimentally . since the isgur - wise function can not be calculated from first principles , many models and nonperturbative approaches , which exhibit the heavy quark symmetry , have been employed to describe relevant phenomena .
however , it was found out , that the finite mass corrections are very important , especially , in the charm sector .
it appears that in some sense a step back should be done from using the heavy quark symmetry as a guide under model building to the straightforward calculations with full quark propagators .
then one has to check the consistency of the results with the spin - flavor symmetry in the heavy quark limit . in this paper
we employ the relativistic constituent quark model ( rcqm ) @xcite for the simultaneous description of both light and heavy flavored meson leptonic and semileptonic decays .
this model is based on the effective lagrangian describing the coupling of mesons with their quark constituents , and the compositeness condition .
the physical processes are described by the one - loop quark diagrams with free constituent propagators and meson - quark vertices related to the bethe - salpeter amplitudes .
the masses of lower - lying pseudoscalar ( ps ) mesons should be less than the sum of quark constituent masses to provide the absence of imaginary parts corresponding to quark production .
the adjustable parameters , the widths of bethe - salpeter amplitudes in momentum space and constituent quark masses , are determined from the best fit of available experimental data and some lattice determinations .
we found that our results are in good agreement with experimental data and other approaches . also we reproduce the results of spin - flavor symmetry for leptonic decay constants and semileptonic form factors in the heavy quark limit .
the shapes of vertex functions and quark propagators should be found from the bethe - salpeter and dyson - schwinger equations , respectively .
this is provided by the dyson - schwinger equation ( dse ) @xcite studies .
a dse - approach has been employed to provide a unified and uniformly accurate description of light- and heavy - meson observables @xcite .
a similar approach , based on the effective heavy meson lagrangian , has been described in @xcite in terms of a model based on meson - quark interactions , where mesonic transition amplitudes are represented by diagrams with heavy mesons attached to quark loops .
the free propagator has been used for light quarks .
however , the quark propagator obtained in the heavy quark limit has been employed for heavy quarks .
we employ an approach @xcite based on the effective interaction lagrangian which describes the transition of hadron into quarks . for example
, the transition of the meson @xmath16 into its constituents @xmath17 and @xmath18 is given by the lagrangian @xmath19 here , @xmath20 and @xmath21 are the gell - mann and dirac matrices , respectively , which provide the flavor and spin numbers of the meson @xmath16 .
the function @xmath22 is related to the scalar part of bethe - salpeter amplitude .
for instance , the separable form @xmath23 has been used in @xcite for pions .
the coupling constants @xmath24 is given by the so called _ compositeness condition _ proposed in @xcite and extensively used in @xcite .
that condition means that the renormalization constant of the meson field is equal to zero : @xmath25 where @xmath26 is the derivative of the meson mass operator defined by @xmath27.\ ] ] the invariant amplitudes describing the leptonic @xmath28 and semileptonic @xmath29 decays are written down @xmath30 where @xmath31 is the fermi weak - decay constant , @xmath32 is the appropriate element of the ckm matrix .
the matrix elements of the hadronic currents are given by @xmath33= f_h p^\mu \label{curlep}\\ & & \nonumber\\ m^{\mu}_{hh'}(p , p^\prime)&= & { 3\over 4\pi^2}g_hg_{h'}\!\!\int\!\!{d^4k\over 4\pi^2i } \phi_h(-k^2)\phi_{h'}(-k^2 ) \label{cursem}\\ & & \times { \rm tr}\biggl[\gamma^5 s_3(\not\ ! k)\gamma^5 s_2(\not\ ! k+\not\ !
p^\prime)o^\mu s_1(\not\ !
p ) \biggr ] \nonumber\\ & & \nonumber\\ & & = f_+(q^2)(p+p^\prime)^\mu + f_-(q^2)(p - p^\prime)^\mu\\ \nonumber\end{aligned}\ ] ] where @xmath34 is related to the bs - amplitude in momentum space , and @xmath35 is the propagator of the constituent quark with mass @xmath36 . as discussed before , to avoid the appearance of imaginary parts in eqs .
( [ curlep ] ) and ( [ cursem ] ) , we assume that @xmath37 which is a reliable approximation for the lower - lying mesons considered here . to evaluate the integral in eq .
( [ cursem ] ) @xmath38 where @xmath39 , we need to calculate the following integrals : @xmath40[m_2 ^ 2-(k+p^\prime)^2][m_3 ^ 2-k^2]}\,.\ ] ] using the cauchy representation for the function @xmath41 and then the standard techniques of the feynman @xmath42parametrization one finds ( @xmath43 ) @xmath44 \frac{t}{(1+t)^2}{\cal f}(z_i ) \nonumber\\ & & \nonumber\\ & & + p_\alpha^\mu p_\alpha^\nu p_\alpha^\delta \biggl(\frac{t}{1+t}\biggr)^3 { \cal f'}(z_i ) \biggr\ } \nonumber\end{aligned}\ ] ] where @xmath45 , @xmath46 , @xmath47 , and @xmath48-p_\alpha^2t/(1+t)$ ] .
finally , eq . ( [ int ] ) becomes @xmath49 with @xmath50 \nonumber\\ & & \nonumber\\ & & -{\cal f'}(z_i ) \biggl[(m_1+m_2)m_3 \nonumber\\ & & \nonumber\\ & & + \frac{t}{1+t}\biggl(-(\alpha_1+\alpha_2)(m_1m_3+m_2m_3-m_1m_2 ) \nonumber\\ & & \nonumber\\ & & + \alpha_1 p^2+\alpha_2 p^{\prime 2}\biggr ) -p_\alpha^2 \biggl(\frac{t}{1+t}\biggr)^2 \biggl(2-(\alpha_1+\alpha_2)\frac{t}{1+t}\biggr ) \biggr ] \biggr\}\ , .
\nonumber\end{aligned}\ ] ] the normalization condition is written in the form @xmath51 with @xmath52 .
the integrals corresponding to the matrix element of the leptonic decay @xmath28 and radiative decay of neutral meson @xmath53 are calculated following the same procedure .
we have @xmath54\phi(z_y ) \label{intlep}\\ & & \nonumber\\ & & \nonumber\\ k^{\mu\nu}(q_1,q_2)&= & \int \frac{d^4k}{4\pi^2i } \phi(-k^2 ) { \rm tr } \biggl\{\gamma^5 s(\not\ ! k-\not\ ! q_2)\gamma^\mu s(\not\ !
k ) \gamma^\nu s(\not\ ! k+\not\ !
q_1)\biggr\ } \nonumber\\ & & \nonumber\\ & = & i\varepsilon^{\mu\nu\alpha\beta}q_1^\alpha q_2^\beta k(p^2 ) \nonumber\\ & & \nonumber\\ k(p^2)&= & m\int\limits_0^\infty dt \biggl(\frac{t}{1+t}\biggr)^2 \int\limits_0 ^ 1 d\alpha_1 \int\limits_0^{1-\alpha_1 } d\alpha_2 \biggl(-\phi'(z_k)\biggr ) \label{intrad}\end{aligned}\ ] ] where @xmath55 $ ] and @xmath56+\alpha_1\alpha_2 p^2 t/(1+t)$ ] .
the physical observables are expressed in terms of the structural integrals written in eqs .
( [ intfin ] ) , ( [ intlep ] ) and ( [ intrad ] ) : @xmath57 ^ 2 , \label{lep}\\ & & \nonumber\\ & & \nonumber\\ f_+(q^2)&=&\frac{3}{4\pi^2}\ g_pg_{p'}\ i_+(m^2_p , m^2_{p'},q^2 ) , \label{sem}\\ & & \nonumber\\ \gamma(p\to p^\prime l\nu)&=&|v_{qq'}|^2\frac{g_f^2}{192\pi^3 m_p^3 } \int\limits_0^{t_- } dt |f_+(t)|^2 \biggl[(t_+-t)(t_--t)\biggr]^{3/2 } , \nonumber\end{aligned}\ ] ] with @xmath58 ( the extra factor 1/2 appears for @xmath59 in the final state ) . the leptonic heavy decay constants and semileptonic heavy to heavy form factors acquire a simple form in the heavy quark limit ,
_ i. e. _ when @xmath60 , @xmath61 and @xmath62 , @xmath63 with @xmath64 being a constant value . from eq .
( [ intfin ] ) by replacing the variables @xmath65 and @xmath66 , one obtains @xmath67 \nonumber\\ & & \nonumber\\ & = & \frac{m+m'}{2mm'}\cdot \frac{1}{2 } \int\limits_0 ^ 1 \frac{d\tau}{w } \int\limits_0^\infty du { \cal f}(\tilde z)\frac{m+\sqrt{u}}{m^2+\tilde z } \label{inthql}\end{aligned}\ ] ] where @xmath68 , @xmath69 and @xmath70
. the normalization condition can be obtained from eq .
( [ inthql ] ) by putting @xmath71 and @xmath72 .
we have @xmath73 where @xmath74 .
then the leptonic decay constant and semileptonic form factors are written as @xmath75\phi_h(\tilde z_0 ) \frac{m+\sqrt{u}/2}{m^2+\tilde z_0 } \label{lephql}\\ & & \nonumber\\ f_{\pm } & \rightarrow & \frac{m'\pm m}{2\sqrt{mm'}}\xi(w ) \hspace{1 cm } \xi(w)=\frac{1}{i_n}\int\limits_0 ^ 1 \frac{d\tau}{w } \int\limits_0^\infty du \phi_h^2(\tilde z ) \frac{m+\sqrt{u}}{m^2+\tilde z}\ , .
\label{semhql}\end{aligned}\ ] ] it is readily seen that we reproduce the scaling law for both leptonic decay constants and form factors , and obtain the explicit expression for the isgur - wise function @xcite .
the expressions obtained in the previous section for physical observables are valid for any vertex function @xmath34 . here
, we choose a gaussian form @xmath76 in minkowski space .
the magnitude of @xmath77 characterizes the size of the bs - amplitude and is an adjustable parameter in our approach .
thus , we have six @xmath78-parameters plus the four quark masses , all of which are fixed via the least - squares fit to the observables measured experimentally or taken from lattice simulations ( see , table [ t1 ] ) .
.calculated values of a range of observables ( @xmath79 in gev@xmath80 , leptonic decay constants in gev , form factors and ratios are dimensionless ) .
the obs . "
are extracted from refs .
the quantities used in fitting our parameters are marked by
@xmath81 " .
[ t1 ] [ cols="^,<,<,<,^,<,<,<",options="header " , ] | semileptonic decays of pseudoscalar mesons elements cabibbo - kobayashi - maskawa matrix fundamental parameters standard model decay @xmath7 determination @xmath8 semileptonic decays of d and b mesons @xmath9 determine @xmath12 @xmath13 @xmath14 effects strong interactions expressed form factors depend on @xmath15 momentum to leptonic pair form factors measuring distributions @xmath15 decay angles decays of heavy d and b mesons due to spin - flavor symmetry for infinite quark masses symmetry form factors universal isgur - wise function scaling laws for physical observables tested experimentally isgur - wise function calculated from first principles models nonperturbative approaches heavy quark symmetry employed phenomena finite mass corrections important in charm sector step back from heavy quark symmetry to straightforward calculations with full quark propagators check consistency results with spin - flavor symmetry heavy quark limit relativistic constituent quark model rcqm ) @xcite for description light heavy flavored meson leptonic semileptonic decaysmodel based on lagrangian coupling mesons quark constituents compositeness condition physical processes by one - loop quark diagrams free constituent propagators meson - quark vertices bethe - salpeter amplitudes lower - lying pseudoscalar mesons less than quark masses imaginary parts quark production adjustable parameters bethe - salpeter amplitudes quark masses determined from experimental data lattice determinations results with experimental data reproduce spin - flavor symmetry for leptonic decay constants semileptonic form factors in heavy quark limit shapes of vertex functions quark propagators from bethe - salpeter dyson - schwinger equations studies - approach unified description light- heavy - meson observables similar approach heavy meson lagrangian model meson - quark interactions mesonic transition amplitudes represented by diagrams heavy mesons quark loops free propagator used for light quarks quark propagator heavy quark limit for heavy quarks approach effective interaction lagrangian transition of hadron into quarkstransition meson @xmath16 into constituents by lagrangian @xmath19 @xmath20 gell - mann dirac matrices provide flavor spin numbers meson @xmath16 function @xmath22 related to scalar bethe - salpeter amplitude separable form @xmath23 used in @xcite for pions coupling constants @xmath24 compositeness condition renormalization constant meson field equal to zero @xmath26 derivative of meson mass operator @xmath27 invariant amplitudes leptonic semileptonic @xmath29 decays @xmath30 @xmath31 fermi weak - decay constant @xmath32 element ckm matrixmatrix elements hadronic currents given by @xmath33= f_h p^\mu\nonumber\\^{\mu{hh'}(p p^\prime) 3\over 4\pi^2}g_hg_{h'}^4k 4\pi^2i_h(-k^2)_{h'}(-k^2{cursem\gamma^5 s_3(\not k)^5 s_2( k+ p^\prime)o^\mu s_1( p \nonumber\\ f_+(q^2)(p+p^\prime)^\mu + f_-(q^2)(p - p^\prime)^\mu\\{aligned @xmath34 related bs - amplitude momentum space @xmath35 propagator constituent quark mass @xmath36 avoid imaginary parts @xmath37 reliable approximation lower - lying mesons evaluate integral @xmath38 @xmath39 calculate integrals @xmath40[m_2 ^ 2-(k+p^\prime)^2][m_3 ^ 2-k^2]cauchy representation @xmath41 standard techniques feynman @xmath42parametrization finds @xmath43 @xmath44 \frac{t}{(1+t)^2}{\cal f}(z_i ) \nonumber\\ p\alpha^^\delta (\frac{t}{1+t}\biggr)^3 f'}(z_i\nonumber\end{aligned} @xmath45 @xmath46 @xmath47 @xmath48-p_\alpha^2t/(1+t)$ becomes @xmath49 @xmath50 \cal f'}(z_i \biggl(m_1+m_2)m_3 \frac{t}{1+t}\biggl\alpha_1+\alpha_2)(m_1m_3+m_2m_2\alpha_1^2+\alpha_2 2} -p_\alpha^2 (\frac{t}{1+t}\biggr)^2(\alpha_1+\alpha_2)\frac{t}{1+t}\biggr\nonumber\end{aligned} normalization condition @xmath51 @xmath52integrals leptonic decay @xmath28 radiative decay neutral meson @xmath53 calculated same procedure @xmath54\phi(z_y ) \label{intlep}\\ \nonumber\\\mu\nu}(q_1,q_2)\frac{d^4k}{4\pi^2i \phi(-k^2 )\biggl\gamma^5\not- q_2)\gamma^\mu\nu+ q_1)\biggr \nonumber\\\varepsilon^{\mu\nu\alpha\beta}q_1^\alpha q_2^\beta k(p^2\nonumber\\ k(p^2)\limits_0^\infty\biggl{t{1+t}\biggr^2\limits\alpha_1\alpha_1\alpha_2\phi'(z_k)\biggr{intrad}\end{aligned} @xmath55 $ @xmath56+\alpha_1\alpha_2 p^2/(1+t)$ physical observables expressed structural integrals eqs .intfin intlep intrad @xmath57\label{lep \nonumber\\ f_+(q^2)\frac{3}{4\pi^2} g_pg_{p'}+(m^2_p{p'},q^2{sem\nonumber\\\gamma(p^\nu)|v_{qq'}\frac{g_f^2}{192\pi^3 m_p^3\limits_0^{t_-_+(t)|^2\biggl[(t_+\biggr]^{3/2\nonumber\end{aligned @xmath58 extra factor 1/2 @xmath59 final state leptonic heavy decay constants semileptonic heavy form factors simple form heavy quark limit @xmath60 @xmath61 @xmath62 @xmath63 @xmath64 constant valuereplacing variables @xmath65 obtains @xmath67\frac{m+m'}{2mm'}{1}{2\limits_0{d\tau}{w(\tilde z){m+\sqrt{u}}{m^2+\tilde z }}{aligned @xmath68 @xmath70 normalization condition from eq @xmath71 @xmath72 @xmath73 @xmath74 leptonic decay constant semileptonic form factors written as @xmath75(\tilde z_0 ){m+\sqrt{u}{m^2+\tilde z_0{lephql\nonumber\\ f_{\frac{m' m}{2\sqrt{mm'}}\xi(w ){1\frac{1}{i_n}\limits_0 1{d\tau}{w }_h^2(\tilde z ){m+\sqrt{u}}{m^2+\tilde z}{semhql}{aligned} reproduce scaling law for leptonic decay constants form factors obtain explicit expression isgur - wise function @xcite expressions previous section physical observables valid vertex function @xmath34 choose gaussian form @xmath76 minkowski space magnitude @xmath77 characterizes size bs - amplitude adjustable parameter six @xmath78-parameters plus four quark masses fixed via least - observables measured experimentally lattice simulations table t1.calculated values range observables @xmath79 gev@xmath80 leptonic decay constants gev form factors ratios dimensionless obs extracted from refs quantities fitting parameters marked by @xmath81 t1 cols=" " | 0.552747 | 3,140 | 1,744 | , Saving $0.1 in GPT-4. |
177,319 | a model of a thick brane in 5d bulk supported by two phantom scalar fields is considered . the comparison with a thick brane supported by two usual scalar fields
is carried out .
the distinctions between a thick brane supported by one usual scalar field and our model have been pointed out . :
thick branes ; phantom scalar fields . | a discovery of the accelerated expansion of the universe @xcite and a great deal of work attempting to model this phenomenon allows to suppose that existence of a special class of scalar fields phantom fields is possible in the universe @xcite .
the phantom fields are some type of matter with the violated weak or null energy conditions .
one of the ways of introduction of such fields is consideration of theories of a scalar field with a negative sign before a kinetic term ( ghost fields ) @xcite . in @xcite , using the ghost fields , we have obtained regular wormhole , particle - like and brane - like solutions .
another way consists in a consideration of braneworld models of dark energy which also allow violation of the weak or null energy conditions @xcite .
the accelerated expansion of the universe is being supported in phantom models by the phantom fields filling all the universe . in this paper
we suppose that the phantom field really fills the universe , and we use this assumption for obtaining a thick - brane solution in a 5d spacetime with two interacting phantom scalar fields . previously , one of us has already obtained a similar regular solution for two usual interacting scalar fields @xcite . here
we want to compare the solutions from @xcite with a solution to be found below , and to clarify the difference between thick - brane solutions with phantom and usual scalar fields .
also we want to compare our results with the results from @xcite where thick branes with one usual scalar field are under consideration ( see also earlier work on thick branes @xcite ) .
it was shown there that corresponding solutions have ads asymptotics , the potential @xmath0 must be an alternating function , and also some fine tuning condition should be satisfied .
we start from the lagrangian @xmath1,\ ] ] where capital latin indices run over @xmath2 , @xmath3 for the usual scalar fields , and @xmath4 for the phantom ones ; @xmath5 are two interacting scalar fields with the potential @xmath6 where @xmath7 is some constant .
the corresponding energy - momentum tensor is @xmath8 \right\}.\ ] ] we consider a problem when all variables depend only on the extra coordinate @xmath9 .
then metric for a 4d - brane embedded in an external five - dimensional spacetime can be written in the form @xmath10 using this metric , the einstein equations will be @xmath11 , \label{einst_phantom2}\end{aligned}\ ] ] where the first equation was obtained by summation of @xmath12 and @xmath13 components .
the equations for the scalar fields are @xmath14,\\ \chi^{\prime\prime}+4\frac{a^\prime}{a}\chi^{\prime}&=&\chi\left[2\varphi^2+\lambda_2\left(\chi^2-m_2 ^ 2\right)\right ] .
\label{field_phantom2}\end{aligned}\ ] ] let us examine the system of equations - with boundary conditions @xmath15 the conditions - and equation give the following value of the constant @xmath7 : @xmath16 let us choose the following numeric parameters for numerical analysis : @xmath17 ( solid lines ) and @xmath3 ( dashed lines).,width=453 ] for the cases @xmath4 ( solid line ) and @xmath3 ( dashed line).,width=453 ] for the cases @xmath4 ( solid line ) and @xmath3 ( dashed line).,width=453 ] the model with @xmath3 was considered in @xcite . here
we examine a case when @xmath4 , and compare obtained results with the results from @xcite . at numerical analysis of the system - one should solve a nonlinear problem for calculation of eigenvalues of the parameters @xmath18 and @xmath19 .
this problem is being solved by the shooting method which is described in detail in @xcite .
the obtained results are presented in figs .
[ fig : fields_phantom]-[fig : energy_phantom ] . the solutions for the case @xmath4
are shown by solid lines , and for the case @xmath3 by dashed lines .
the following values of the masses were found : @xmath20 and @xmath21 for the case @xmath4 , @xmath22 and @xmath23 for @xmath3 .
let us estimate asymptotic behavior of the solutions .
one can see from that asymptotically @xmath24 where @xmath25 is an asymptotic value of the metric function , and subscripts ( -1 ) and ( + 1 ) refer to cases with phantom and usual fields , respectively .
then , using these expressions and looking for a solution of equations - in the form @xmath26 we have the following asymptotic form for the scalar fields : @xmath27 } , \\ \delta \chi_{(-1)}&=&\chi_\infty \exp{\left[\left(2k_{(-1)}- \sqrt{4k_{(-1)}^2 + 2m_1 ^ 2-\lambda_2 m_2 ^ 2}\right)r\right ] } , \\ \delta \varphi_{(+1)}&=&\varphi_\infty \exp{\left[\left(2k_{(+1)}- \sqrt{4k_{(+1)}^2 + 2 \lambda_1m_1 ^ 2}\right)r\right ] } , \\ \delta \chi_{(+1)}&=&\chi_\infty \exp{\left[-\left(2k_{(+1)}+ \sqrt{4k_{(+1)}^2 + 2m_1 ^ 2-\lambda_2 m_2 ^ 2}\right)r\right]},\end{aligned}\ ] ] where @xmath28 are asymptotical values of the functions .
one can see from the obtained expressions that asymptotically @xmath29 and @xmath30 , i.e. the solutions are finite . in fig .
[ fig : energy_phantom ] the profiles of the energy density @xmath31 \label{energy - density}\ ] ] for the phantom and usual scalar fields are shown .
let us now study the dynamical stability of the above solutions against linear perturbations . to do this we have to write field equations following from the lagrangian keeping time dependence of metric and scalar fields functions .
then @xmath32 and @xmath33 components of the einstein equations will be : @xmath34&= & \epsilon\left[\frac{1}{2}a^2\left(\dot{\varphi}^2+\dot{\chi}^2\right)+\frac{1}{2}\left(\varphi^{\prime 2}+\chi^{\prime 2}\right)+v\right],\\ -3\left[-\frac{\ddot{a}}{a^3}+2\left(\frac{a^\prime}{a}\right)^2\right]&= & \epsilon\left[-\frac{1}{2}a^2\left(\dot{\varphi}^2+\dot{\chi}^2\right)-\frac{1}{2}\left(\varphi^{\prime 2}+\chi^{\prime 2}\right)+v\right],\\ 3\left[-\frac{\dot{a}a^\prime}{a^2}+\frac{\dot{a}^\prime}{a}\right]&= & \epsilon\left(-\dot{\varphi}\varphi^\prime-\dot{\chi}\chi^\prime\right),\end{aligned}\ ] ] where differentiation with respect to @xmath35 and @xmath9 is denoted by a @xmath36 and a @xmath37 correspondingly . introducing a new metric function @xmath38 via @xmath39 the component @xmath33 takes the form @xmath40 the corresponding equations for the scalar fields are @xmath41-\varphi^{\prime\prime}-4\alpha^{\prime}\varphi^{\prime}= -\varphi\left[2\chi^2+\lambda_1\left(\varphi^2-m_1 ^ 2\right)\right],\\ \label{sfe_chi } e^{-2\alpha}\left[\ddot{\chi}+2\dot{\alpha}\dot{\varphi}\right]-\chi^{\prime\prime}-4\alpha^{\prime}\chi^{\prime}= -\chi\left[2\varphi^2+\lambda_2\left(\chi^2-m_2 ^ 2\right)\right].\end{aligned}\ ] ] we perturb the solutions of the system - by expanding the metric function and scalar fields functions to first order as follows @xmath42 where the index @xmath43 indicates the static background solutions of equations - , and the index @xmath44 refers to perturbations .
then one has from : @xmath45 using the last expression , one can rewrite equations , as @xmath46\varphi_1- \left(\frac{4\epsilon}{3}\varphi_0^\prime\chi_0^\prime+4\varphi_0\chi_0\right)\chi_1+\omega^2 e^{-2\alpha_0}\varphi_1&=&0,\\ \chi_1^{\prime\prime}+4\alpha_0^\prime\chi_1^\prime- \left[\frac{4\epsilon}{3}\chi_0^{\prime
2}+\lambda_2\left(2\chi_0 ^ 2-m_2 ^ 2\right)\right]\chi_1- \left(\frac{4\epsilon}{3}\varphi_0^\prime\chi_0^\prime+4\varphi_0\chi_0\right)\varphi_1+\omega^2 e^{-2\alpha_0}\chi_1&=&0 . \label{sfe_chi_pert}\end{aligned}\ ] ] for existence of stable solutions it is necessary to provide positiveness of an eigenvalue @xmath47 .
if there is a negative eigenvalue @xmath47 then the solution will be unstable since then @xmath48 will grow exponentially . in order to make clear this question , we will search for numerical solutions of equations , with boundary conditions @xmath49 for a case
@xmath4 one can find that at @xmath50 the following regular solutions for @xmath51 exist ( see fig .
[ fig : perturb_phant ] ) .
( solid line ) and @xmath52 ( dashed line ) from equations , with the boundary conditions and @xmath50.,width=453 ] it indicates that the solutions with phantom scalar fields are stable for the case under consideration .
but for a case @xmath53 we could not find any regular solutions of equations , .
it seems that there is no stable solutions for usual scalar fields ( at least for the model s parameters we used here ) .
in our opinion , the main distinction of the phantom thick brane from a brane created by usual scalar fields consists in asymptotical behavior of the metric function @xmath54 : for the phantom thick brane @xmath55 , and for the usual brane @xmath56 .
we think that it happens due to the following reasons : let a solution has the following form near @xmath57 : @xmath58 their substitution in gives the following result @xmath59 it implies that at @xmath60 , @xmath61 and one has ( at least near @xmath57 ) a steadily increasing function which was confirmed by the numerical calculations presented in fig .
[ fig : metric_phantom ] . at @xmath62 , @xmath63 and one has a decreasing function .
another interesting feature of the obtained solution is that the energy density of the phantom brane at the origin of coordinates @xmath57 has an absolute maximum . at the same time
, the usual brane has a local minimum at @xmath57 .
physically , it implies that in the phantom case the brane looks like a hill on the background of phantom matter with constant energy density . in case of usual brane a maximum of energy is concentrated near the brane on both sides . in our opinion
, it may results in different scenarios of trapping of 4d matter on the phantom and usual branes . at comparison of our solutions with the results from @xcite
one might note the following differences : ( a ) the metric function @xmath54 increases exponentially ; ( b ) the potential @xmath64 does not change a sign ( see fig .
[ fig : potential ] ) ; ( c ) in consequence of the condition ( b ) , there is no a fine tuning condition . in ref .
@xcite such fine tuning condition is necessary for obtaining of a thick brane supported by one scalar field .
this fine tuning condition consists in the following : using the metric @xmath65 with the harmonic coordinate @xmath66 , such that @xmath67 , one should has the following fine tuning condition @xmath68 where @xmath69 in our case the corresponding condition could be obtained as follows : from equations , one has @xmath70 and because of @xmath71 , we have @xmath72 this indicates that the nontrivial potential @xmath73 does not change a sign as it can be seen from fig .
[ fig : potential ] .
so we have obtained the solutions describing a thick phantom brane in 5d bulk .
there are two main distinctive features of these solutions : a ) they have different asymptotic behavior ; b ) the solutions with phantom scalar fields are stable , and the solutions with usual scalar fields are not ( at least for the parameters @xmath74 we used here ) .
m. szydlowski , w. czaja and a. krawiec , phys . rev . *
e 72 * , 036221 ( 2005 ) , astro - ph/0401293 .
e. elizalde , s. nojiri and s.d .
odintsov , phys . rev .
* d 70 * , 043539 ( 2004 ) , hep - th/0405034 .
s. capozziello , s. nojiri , and s.d .
odintsov , `` unified phantom cosmology : inflation , dark energy and dark matter under the same standard '' , hep - th/0507182 . | discovery accelerated expansion universe @xcite work special class scalar fields phantom fields possible phantom fields are matter violated weak null energy conditions introduction fields theories scalar field negative sign before kinetic term ghost fields ) @xcite using ghost fields obtained regular wormhole particle - like brane - like solutions braneworld models dark energy allow violation weak null energy conditions accelerated expansion supported phantom models by phantom fields filling universe paper suppose phantom field fills universe use assumption for obtaining thick - brane solution 5d spacetime with two interacting phantom scalar fields obtained similar regular solution for two usual interacting scalar fields @xcite compare solutions @xcite with solution below clarify difference between thick - brane solutions with phantom usual scalar fields compare results with results @xcite thick branes with one usual scalar field corresponding solutions have ads asymptotics potential @xmath0 alternating function fine tuning condition satisfiedstart lagrangian @xmath1 capital latin indices run over @xmath2 @xmath3 usual scalar fields @xmath4 phantom ones @xmath5 two interacting scalar fields potential @xmath6 @xmath7 constant energy - momentum tensor @xmath8 consider problem variables depend on extra coordinate @xmath9 metric 4d - brane external five - dimensional spacetime written form @xmath10 metric einstein equations @xmath11_phantom2 first equation obtained summation @xmath12 @xmath13 components equations scalar fields @xmath14\chi{\prime}+4\frac{a\prime\left[2\varphi^2+\lambda_2\left_phantom2 examine system of equations boundary conditions @xmath15 conditions equation value constant @xmath7 @xmath16 choose numeric parameters numerical analysis @xmath17 solid lines ) @xmath3 ( dashed lines). cases @xmath4 solid line dashed model with @xmath3 considered in @xcite examine case @xmath4 compare results with results @xcitenumerical analysis solve nonlinear problem calculation eigenvalues parameters @xmath18 @xmath19 solved by shooting method described @xcite results presented in figs fields_phantom energy_phantom solutions @xmath4 shown by solid lines @xmath3 dashed lines values masses found @xmath20 @xmath21 @xmath4 @xmath22 @xmath23 @xmath3 estimate asymptotic behavior solutions @xmath24 @xmath25 asymptotic value metric function subscripts ( -1 ) + 1 ) refer to cases phantom usual fieldsusing expressions solution equations @xmath26 asymptotic form scalar fields @xmath27\delta \chi_{(-1)}\chi_\infty\left_(-1)} \sqrt{4k_{(-1)}^2 + 2m_1 2-\lambda_2 m_2 2\delta \varphi_{(+1)}\varphi_\infty_{(+1)} \sqrt{4k_{(+1)}^2 + 2 \lambda_1m_1 2\delta \chi_{(+1)}\chi_\infty{\left_{(+1)} \sqrt{4k_{(+1)}^2 + 2m_1 2-\lambda_2 m_2 2}{aligned @xmath28 asymptotical values functions expressions @xmath29 @xmath30 solutions finite fig . energy_phantom profiles energy density @xmath31 phantom scalar fields study dynamical stability solutions linear perturbations write field equations lagrangian time dependence metric scalar fields functions@xmath32 @xmath33 components einstein equations @xmath34\left\frac{1}\varphi}^2+\chi}^2{1}{2} 2}+v\right -3\left\frac{{a}}^3}+2\prime}}^2\right\frac{1}{2}a^2\left\varphi}^2+\chi}^2\right{1}{2} 2}{ 2}+v\right 3\left\frac{{a}a^\prime^2}{a}^\prime}{a}\right\left(-\dot{\varphi}\prime-{\chi}\chi^\prime{aligned differentiation @xmath35 @xmath9 denoted @xmath36 @xmath37new metric function @xmath38 @xmath39 @xmath33 @xmath40 equations scalar fields @xmath41-\varphi\prime\alpha\prime -\varphi\left[2\chi^2+\lambda_1\left(\varphi^2-m_1 2\right)\alpha}\left\alpha\varphi]-\chi\prime}-4\alpha^{\prime -\chi\left[2\varphi^2+\lambda_2\left(\chi^2-m_2 2\right) solutions system expanding metric function scalar fields first order @xmath42 index @xmath43 static solutions @xmath44 perturbations@xmath45 last expression rewrite equations @xmath46\varphi_1-(\frac{4\epsilon}{3}\varphi_0^\prime\chi_0^\prime+4\varphi_0\chi_0\right)+\omega^2 e^{-2\alpha_0}\varphi_1&=&0\chi_1\prime}+4\alpha_0^\prime\chi_1^-\frac{4\epsilon}{3}\chi_0^\prime 2}+\lambda_2\left(2\chi_0 2-m_2 2\right)\chi_1-\frac{4\epsilon}{3}\varphi_0^\prime\chi_0^\prime+4\varphi_0_0\right+\omega^2 e^{-2\alpha_0}\chi_1&=&0{sfe_chi_pert}\end{aligned} stable solutions positiveness eigenvalue @xmath47 negative eigenvalue @xmath47 solution unstable @xmath48 grow search numerical solutions equations conditions @xmath49 @xmath4 @xmath50 regular solutions @xmath51 exist fig solid line @xmath52 dashed line solutions phantom scalar fields stable @xmath53 regular solutionsno stable solutions for scalar fields model parameters main distinction of phantom thick brane from usual scalar fields consists in asymptotical behavior of metric function @xmath54 phantom thick brane @xmath55 usual brane @xmath56 due to reasons solution form near @xmath57 : @xmath58 substitution gives result @xmath59 implies at @xmath60 @xmath61 one steadily increasing function confirmed by numerical calculations in fig . at @xmath62 @xmath63 one has decreasing function energy density of phantom brane at origin coordinates @xmath57 has absolute maximum usual brane has local minimum at @xmath57 phantom case brane looks like hill on with constant energy density usual brane maximum energy concentrated near brane on both sides may results in different scenarios of trapping of 4d matter on phantom and usual branes comparison of solutions with results from @xcite differences : metric function @xmath54 increases exponentially potential @xmath64 change no fine tuning condition in @xcite fine tuning necessary for thick brane supported by one scalar fieldfine tuning condition metric @xmath65 harmonic coordinate @xmath66 @xmath67 fine tuning condition @xmath68 @xmath69 corresponding condition obtained equations @xmath70 @xmath71 @xmath72 nontrivial potential @xmath73 change sign fig . obtained solutions describing thick phantom brane in 5d bulk different asymptotic behavior phantom scalar fields stable usual scalar fields not parameters @xmath74. szydlowski w. czaja a. krawiec phys . rev . * e 72, 036221 ( 2005 ) astro - ph/0401293 elizalde s. nojiri odintsov phys . rev . * d 70 * 043539 ( 2004 ) hep - th/0405034 capozziello nojiri odintsov unified phantom cosmology inflation dark energy dark matter hep - th/0507182 | 0.554636 | 3,133 | 1,692 | , Saving $0.1 in GPT-4. |
177,321 | the apparent attenuation of muon neutrinos relative to electron neutrinos is a bit too low to be compatible with the most popular values of @xmath0
. fits to @xmath1 favor values of @xmath2 ev@xmath3 .
the fit minimized by the super kamioka group in estimating neutrino oscillation parameters neglects systematic errors .
the fit is dominated by systematic effects .
the data being combined in recent fits may not be compatible since there appear to be significant variations in the properties of the data with time .
a simple two component neutrino oscillation with @xmath0 in the range of 10@xmath4 to 10@xmath5 ev@xmath3 seems unable to account for the observations . | atmospheric neutrinos originate from the decay of unstable particles produced by cosmic ray interactions in the upper atmosphere .
the useful spectrum runs from about 200 mev to about 1330 mev .
but flux extends to much higher energies .
the super kamioka group has studied these out to about 10 gev .
the higher energy sample is only about 21% of the size of the sample below 1330 mev .
atmospheric neutrinos approximate the ideal of a two distance neutrino oscillation experiment .
figure [ dist ] shows the log of the flight length for neutrinos as a function of the cosine of the zenith angle .
events from below travel on the order of 8000 km .
those coming from above travel on the order of 10 km .
there is rather little solid angle at intermediate distances .
atmospheric neutrinos provide a good mixture of electron and muon neutrinos since a significant number of the muons at these energies decay before reaching the ground .
simple arguments suggest that the ratio of muon to electron neutrinos should be about 2 .
to summarize , atmospheric neutrinos are a mixture of muon and electron neutrinos coming from hadronic and muon decays .
the source is approximately at two distances that differ by three orders of magnitude .
most of the flux spans about a factor of 7 in energy .
the atmospheric neutrino anomaly@xcite is the observation of an apparent deficiency of muon type atmospheric neutrinos .
the anomaly is believed to be evidence for neutrino oscillations@xcite . but some of the details and assumptions used in the analysis cast some doubt on the oscillation hypothesis .
the neutrino sample is being drawn from natural sources which adds a degree of uncertainty .
almost all analyses have assumed that the anomaly is restricted to the muon neutrinos and that the electron neutrino flux is a good normalizer .
the ratio of observed to expected neutrino interactions is a measure of the possible reduction in the neutrino flux that is indicative of neutrino oscillations . due to a significant uncertainty in the overall flux normalization , as much as 25% ,
the quality of the attenuation statistic is improved by normalizing to something which is better known .
the value of @xmath7@xcite which attempts to normalize the muon neutrino flux to the electron neutrino flux is thought to have significantly smaller systematic errors , on the order of 5% .
an advantage that @xmath8 has over other observables is that it maximizes the use of the data .
it is insensitive to details of the angular correlation between neutrino direction and reconstructed lepton direction from the interaction .
a disadvantage is that one is averaging over two different baselines so that if only the long distance sample has oscillated significantly the result will be diluted by the unoscillated sample from the shorter baseline .
figure [ rnu ] illustrates the super kamioka measurement of @xmath8 .
the vertical scale is @xmath9 which combines the observed data , @xmath8 with a fit parameter @xmath10 .
the horizontal axis of figure [ rnu ] is the log@xmath11 of @xmath12 with @xmath0 in ev@xmath3 and @xmath13 measured in ev .
the points at the center of the plot are taken from references @xcite and @xcite and are plotted with @xmath14 which is obtained from the super kamioka fit described in the next section . since , for these point @xmath14 the value of @xmath15 can be read off the graph .
reference @xcite quotes @xmath16 for energies below 1330 mev .
reference @xcite quotes @xmath17 for the higher energy sample .
the diamond points in figure [ rnu ] are the same data but the values of @xmath10 and @xmath0 are a fit to these data points .
the fit yields @xmath18 and @xmath19 ev@xmath3 .
the points near the center , the super kamioka fit , would agree with the data much better if the points could be lowered .
but to do this would require @xmath20 which is unphysical .
it is difficult to assign a confidence level to our fit .
the points share many common systematic errors .
one would need to quantify these correlated errors in order to achieve a legitimate confidence interval .
since we have plotted @xmath12 the higher energy data points are to the left .
the @xmath0 mass scale is driven by the low value of @xmath8 measured for the higher energy data points . a reasonable fit can be obtained for @xmath2 ev@xmath3 . more details of the fit can be found in @xcite .
the curve plotted in figure [ rnu ] is an integral over the distances of figure [ dist ] for @xmath15 . for single component neutrino oscillations
the result is a function of @xmath12 and proportional to @xmath10 .
the curve illustrates the structure expected from the two distance scales involved . at
very small @xmath0 all path lengths are small compared to the oscillation length and so the value of @xmath8 is 1 . at very large @xmath0
the oscillation length is small compared to all path lengths in the problem so that @xmath15 will be @xmath21 , for a range of intermediate values of @xmath0 one path length is large compared to the oscillation length and the other is small so that @xmath22 , figure [ rofe ] illustrates the energy dependence of @xmath8 for various values of @xmath0 .
the figure was obtained by numerically integrating the honda 97 muon neutrino flux@xcite . for modest values of @xmath0
the value of @xmath8 is energy independent over the range indicated .
but it can never drop much below 0.75 . as @xmath0 rises portions of @xmath8
drop to below 0.75 and start to approach 0.5 . only at values of @xmath0 of about 0.1
ev@xmath3 is energy independence restored with @xmath23 .
figure [ rofe ] was generated for @xmath14 , its maximum value .
for smaller values of @xmath10 the allowed values of @xmath8 are greater .
this plot supports the fit of figure [ rnu ] in that the only curve that gives an energy independent value for @xmath1 at a value of @xmath8 below @xmath240.75 is @xmath25 ev@xmath3 the curve of figure [ rnu ] was done assuming an isotropic flux .
a check using the honda 97 flux reproduces the results to about 1% , except at the lowest energy point , e=250 mev , where the difference between isotropy and the calculated flux is less than about 3% for @xmath26 ev@xmath3 .
the physical significance of a low value of @xmath8 , if interpreted in terms of only a muon deficiency due to muon neutrino oscillations integrated over energy and distance , is that both distance scales must participate .
since the short distance scale is about 3 orders of magnitude shorter than the large one the favored value for @xmath0 must be comparably larger .
[ cols= " < , < , > , > , > , > , > , > " , ] the change in the measured electron rate means that @xmath8 must have changed .
@xmath8 rose from 0.61@xmath270.03 to 0.76@xmath270.04 .
this rise in @xmath8 brings it from a region which is not consistent with the hypothesis of neutrino oscillations over approximately half of the solid angle to a value consistent with such a hypothesis .
unfortunately there is no good reason to doubt the credibility of either data sample .
super kamioka has combined these samples in their analyses .
the oscillation analysis utilizes the electron events , which are not consistent through this period .
@xmath8 depends on the ratio of observations to simulations .
the simulations did not change in any significant way between these two periods@xcite so there is clear evidence that the electron sample does not provide a reliable normalization for the analysis of atmospheric neutrino data .
the assumption that the ratio of atmospheric electron to muon neutrino flux can be calculated to an accuracy of 5% is in error , unless some portion of the observed signal is not of atmospheric origin or is not neutrinos.@xcite the original super kamioka paper on the atmospheric neutrino flux@xcite notes , `` using a detailed monte carlo simulation , the ratio @xmath28 was measured to be 0.61@xmath270.03(_stat_)@xmath270.05(_sys _ ) , consistent with previous results from the kaimiokande , imb and soudan-2 experiments , and smaller than expected from theoretical models of atmospheric neutrinos '' .
this is no longer true .
the 1998 data is not consistent with previous results of the super kamioka detector or any of the earlier experiments cited .
an @xmath8 value of 0.76@xmath270.04 seems to be unique to this period .
the multi - gev data sample has no evidence of a rate variation in 1998 .
independent confirmation of the drop in event rate would help ally fears that it is due to instrumental or systematic effects .
the super kamioka group has maintained very good monitoring of the detector performance @xcite so such an explanation seems unlikely .
several factors indicate that the atmospheric neutrino anomaly may be more complicated than generally thought . in such cases , it is important to seek corroboration of the results from other experiments .
it is also important to be aware of simplifying assumptions that may have gone into the interpretation or the analysis of the data
. one must be careful not to ignore information simply because it is in conflict with a preferred solution .
it would appear that the electron neutrinos can no longer be used for relative normalization , at least until the variation is understood . in the early days
the anomaly was considered to be either a deficiency of muon like events or an excess of electron type events .
in fact , the results looked a bit like a systematic error in the classification method used for electron and muon events . since a random error would move events from the larger population , the muons , to the smaller population , the electrons .
the observed event rate for all events was much closer to most flux models than might be expected based the errors quoted for the estimates .
studies using the independent muon decay signature corroborate the morphology based results for muon and electron classification .
the `` neutrino '' events observed in underground detectors do not agree with expectations based on a simple source of atmospherically produced hadron and muon decays .
uncertainties in these expectations does not provide much guidance as to what might be wrong .
simple one component muon neutrino oscillations are unable to provide a convincing explanation for the small value of @xmath8 nor for recent apparent drop in the interaction rate .
i would like to thank david casper for useful comments on an earlier version of this manuscript . 9 t.j .
et al . _ ,
lett . * 57 * , 1986 ( 1986 ) .
particle data group , review of particle properties 2000 ( to be published ) . j.m .
losecco , hep - ph/9807432 , ( 1998 ) .
hirata _ et al .
_ , phys .
lett . * b205 * 416 ( 1988 ) . y. fukuda _ et al .
_ , phys .
lett . * b433 * 9 ( 1998 ) . y. fukuda _ et al .
_ , phys .
lett . * b436 * 3 ( 1998 ) . y. fukuda _ et al .
_ , phys .
lett . * 81 * 1562 ( 1998 ) .
the honda 97 flux tables were kindly provided by david casper of uc irvine , a member of the super kamioka collaboration .
m. messier , ph.d .
thesis , boston university ( 1999 ) + k. scholberg , proceedings of the 8th workshop on neutrino telescopes , ( m. baldo - ceolin editor ) * 1 * , 183 - 201 ( 1999 ) .
losecco , hep - ph/9903310 , ( 1999 ) .
s. kasuga , ph.d .
thesis , university of tokyo ( 1998 ) . | atmospheric neutrinos originate from decay unstable particles cosmic ray interactions spectrum 200 mev to 1330 mev flux extends to higher energies super kamioka group studied to 10 gev higher energy sample 21% below 1330 mev neutrinos approximate two distance neutrino oscillation experiment figure shows flight length neutrinos function zenith angle events from below travel 8000 km from above travel 10 km little solid angle at intermediate distances neutrinos provide mixture of electron and muon neutrinos muons decay before reaching ground ratio muon to electron neutrinos 2 neutrinos mixture of muon and electron neutrinos from hadronic muon decays source at two distances three orders of magnitude flux spans factor 7 in energy atmospheric neutrino anomaly@xcite deficiency of muon type neutrinos for neutrino oscillations doubt on oscillation hypothesis neutrino sample drawn from natural sources adds uncertainty analyses anomaly restricted to muon neutrinos electron neutrino flux good normalizerratio of observed to expected neutrino interactions reduction neutrino flux oscillations uncertainty in flux normalization 25% attenuation statistic improved by normalizing to better known value of @xmath7@xcite muon to electron smaller errors 5% @xmath8 maximizes use data insensitive to angular correlation between neutrino direction lepton direction disadvantage over two baselines long distance sample result diluted by sample figure super kamioka measurement of @xmath8 vertical scale is @xmath9 combines observed data @xmath8 fit parameter @xmath10 horizontal axis is log@xmath11 of @xmath12 with @xmath0 ev@xmath3 and @xmath13 measured in ev points at center plot taken from @xcite and plotted with @xmath14 obtained from super kamioka fit value of @xmath15 be read off graph reference @xcite @xmath16 for energies below 1330 mev @xmath17 for higher energy sample diamond points in same data values of @xmath10 and @xmath0 fit to data points fit yields @xmath18 and @xmath19 ev@xmath3points near center super kamioka fit agree data if lowered @xmath20 unphysical difficult to assign confidence level points share systematic errors quantify errors confidence interval plotted @xmath12 higher energy points left @xmath0 mass scale driven by low value @xmath8 higher energy reasonable fit for @xmath2 ev@xmath3 details in @xcite curve in figure integral over distances for @xmath15 single component neutrino oscillations result function of @xmath12 proportional to @xmath10 curve illustrates structure from two distance scales at small @xmath0 path lengths small oscillation length value @xmath8 1 large @xmath0 oscillation length small @xmath15 @xmath21 intermediate values one path length large other small @xmath22 figure illustrates energy dependence of @xmath8 for values @xmath0 obtained by numerically integrating honda 97 muon neutrino flux@xcite for modest values @xmath0 @xmath8 energy independent over range drop below 0.75 as @xmath0 rises @xmath8 below 0.75 approach 0.5 at values @xmath0 0.1 ev@xmath3 energy independence restored with @xmath23figure [ rofe ] generated for @xmath14 maximum value smaller values @xmath10 allowed values @xmath8 greater plot supports fit figure rnu ] only curve energy independent value @xmath1 at @xmath8 below @xmath240.75 is @xmath25 ev@xmath3 curve isotropic flux check honda 97 flux reproduces results to 1% except lowest energy point e=250 mev difference isotropy calculated flux less than 3% for @xmath26 ev@xmath3 physical significance low value @xmath8 muon deficiency due neutrino oscillations both distance scales must participate short distance scale 3 orders magnitude shorter large favored value @xmath0 must be larger change in measured electron rate means @xmath8 must changed @xmath8 rose from 0.61@xmath270.03 to 0.76@xmath270.04 rise from to consistent no reason to doubt credibility data sample super kamioka combined samples analyses oscillation analysis utilizes electron events not consistent @xmath8 depends on ratio observations to simulationssimulations change between periods clear evidence electron sample reliable normalization for atmospheric neutrino data assumption ratio atmospheric electron to muon neutrino flux 5% unless signal not atmospheric origin or not neutrinos original super kamioka paper on neutrino flux notes ratio @xmath28 measured 0.61@xmath270.03 consistent with results kaimiokande soudan-2 experiments smaller than expected from theoretical models of atmospheric neutrinos no longer true 1998 data not consistent with previous results super kamioka detector or earlier experiments @xmath8 value of 0.76@xmath270.04 unique to period multi - gev data sample no evidence of rate variation in 1998 independent confirmation of drop in event rate fears due to instrumental or systematic effects super kamioka group good monitoring detector performance explanation unlikely atmospheric neutrino anomaly may be complicated important to seek corroboration from other experiments aware of simplifying assumptions not ignore information conflict with preferred solution electron neutrinos can no longer be used for relative normalization until variation understoodearly days anomaly considered deficiency muon or excess electron results looked like systematic error in classification method electron muon events random error events from larger muons to smaller electrons observed event rate all events closer to flux models expected errors estimates studies using independent muon decay signature corroborate morphology results muon electron classification neutrino events in underground detectors agree with expectations simple source atmospherically produced hadron muon decays uncertainties expectations provide guidance one component muon neutrino oscillations provide convincing explanation for small value @xmath8 recent drop in interaction rate david casper for comments earlier version manuscript t.j . et al . lett 57 * 1986 particle data group review particle properties 2000 j.m . losecco hep - ph/9807432 1998 ) hirata _ et al . phys . lett b205 * 416 1988 ) fukuda phys b433 1998 b436 1998 81 1562 1998 honda 97 flux tables provided by david casper uc irvine member super kamioka collaboration m. messier , ph.d .boston university 1999 scholberg proceedings 8th workshop neutrino telescopes m. baldo ceolin editor 183 - 201 1999 losecco hep - ph/9903310 kasuga ph.d university tokyo 1998 | 0.434261 | 2,705 | 1,387 | , Saving $0.1 in GPT-4. |
177,331 | the radiative strength functions of @xmath0mo have been reanalyzed .
the enhanced @xmath1 strength for @xmath2 mev is confirmed . | the oslo group has reported unexpected enhancements in the radiative strength functions ( rsf ) of low energy @xmath1-rays for the @xmath3sc , @xmath4v , @xmath5fe and @xmath6mo nuclei @xcite .
these results , which have been obtained with the oslo method @xcite , have gained broad interest in the nuclear physics community .
the @xmath0mo nucleus has become a benchmark for other experimental groups trying to verify or falsify the oslo results .
unfortunately , @xmath0mo was not analyzed in an optimal way , and we have therefore decided to reanalyze these data sets .
the two data sets were recorded from the @xmath0mo(@xmath7he,@xmath7he@xmath8)@xmath0mo and @xmath9mo(@xmath7he,@xmath10)@xmath0mo reactions . in this work
we use the same key parameters for the normalizations as previously @xcite , namely a level density of @xmath11 mev@xmath12 and an average total radiative width of @xmath13 mev ( from @xmath14-wave resonances ) at the neutron binding energy @xmath15 .
experimental details and further references are found in refs .
@xcite .
the new analyzes concern two major points . in the old extraction of rsf
, we included the @xmath1-ray energies close or below the strong 778 kev @xmath16 ground band transition .
this transitional region in the experimental @xmath17 matrix is not properly subtracted in the first - generation procedure @xcite and is now excluded .
the second point concerns the estimate of the @xmath1-ray multiplicity as function of excitation energy , which is an important quantity both for normalizing the @xmath1 spectra with respect to each other and weighting the higher - order generation spectra in the subtraction procedure @xcite . in the previous analyses , we estimated the statistical multiplicity at excitation energy @xmath18 by introducing a lower @xmath1-ray threshold @xmath19 and an effective excitation energy @xmath20 giving @xmath21 where @xmath22 is the average energy of the @xmath1 spectrum for @xmath23 .
the @xmath24 parameter mimics the excitation energy at which the statistical @xmath1 transitions enter the ground band .
this treatment is applicable to rare earth nuclei , where the detector system efficiency for the lowest ground state band transitions , typically the @xmath25 and the @xmath16 transitions , is low .
however , for @xmath0mo the energy of the lowest ground band transitions are detected easily .
therefore , in the present analysis we use the straightforward expression for the total @xmath1-ray multiplicity @xmath26 where we simply divide the excitation energy by the average energy of the @xmath1 spectrum .
he,@xmath27 ) ( filled circles ) and the ( @xmath7he,@xmath7he ) ( open circles ) reaction .
the data from the new analysis is compared with previously published data @xcite . ]
he,@xmath27 ) ( filled circles ) and the ( @xmath7he,@xmath7he ) ( open circles ) reaction .
the data from the new analysis is compared with previously published data @xcite . ]
both the level density and the radiative strength function will be slightly modified by the new multiplicity expression .
figures [ fig : fig1 ] and [ fig : fig2 ] compare the new level densities and radiative strength functions with previous data @xcite . we see a very good resemblance between the pick - up reaction and the inelastic scattering reaction .
the error bars include statistical errors only .
the new level densities are very similar to the previous ones .
the same is true for the rsfs , except that the upbend is less pronounced due to the exclusion of the 778 kev transition .
the values of the data points can be found at http://ocl.uio.no / compilation/. there have been some misunderstandings concerning the description of the rsf upbend for low @xmath1 energies given by @xcite @xmath28 where @xmath29 and @xmath30 are fit parameters , and @xmath31 is given in mev .
the formula was chosen in order to fit the low - energy data by only two parameters .
we would like to stress that this description should not be used for @xmath1 energies lower than the experimental data points . in the extreme case
when @xmath32 , it is clear that the description is totally unrealistic as it gives wrong @xmath1 multiplicity .
the radiative strength function of @xmath0mo has been reanalyzed giving a slightly less pronounced enhancement at lower @xmath1-ray energies .
the data points at and below the 778 kev @xmath16 transition have been omitted . since extraction of level density is coupled to the radiative strength function ,
new level densities have also been presented . financial support from the norwegian research council ( nfr ) is gratefully acknowledged .
part of this work was performed under the auspices of the u.s .
department of energy by the university of california , lawrence livermore national laboratory under contract w-7405-eng-48 .
e.a , u.a , and g.e.m acknowledge support from the national nuclear security administration under the stewardship science academic alliances program through doe research grants no .
de - fg03 - 03-na00074 , no .
de - fg03 - 03-na00076 and u.s .
department of energy grant no .
de - fg02 - 97-er41042 .
a. c. larsen , m. guttormsen , r. chankova , t. lnnroth , s. messelt , f. ingebretsen , j. rekstad , a. schiller , s. siem , n. u. h. syed , and a. voinov , phys .
c * 76 , 044303 ( 2007 ) .
a. c. larsen , r. chankova , m. guttormsen , f. ingebretsen , t. lnnroth , s. messelt , j. rekstad , a. schiller , s. siem , n. u. h. syed , a. voinov , and s. w. degrd , phys . rev .
c * 73 , 064301 ( 2006 ) .
a. voinov , e. algin , u. agvaanluvsan , t. belgya , r. chankova , m. guttormsen , g.e .
mitchell , j. rekstad , a. schiller and s. siem , phys .
lett * 93 * , 142504 ( 2004 ) .
m. guttormsen , _ et al .
_ , phys .
c * 71 * , 044307 ( 2005 ) .
a. schiller , l. bergholt , m. guttormsen , e. melby , j. rekstad , and s. siem , nucl .
instrum .
methods phys .
a * 447 , 498 ( 2000 ) .
r. chankova , _ et al .
_ , phys .
c * 73 * , 034311 ( 2006 ) .
m. guttormsen , t. ramsy , and j. rekstad , nucl .
instrum .
methods a*255 * , 518 ( 1987 ) . * * * | oslo group reported enhancements in radiative strength functions of low energy @xmath1-rays for @xmath3sc nuclei results obtained with oslo method gained interest in nuclear physics @xmath0mo nucleus benchmark for groups @xmath0mo not analyzed decided to data sets two data sets recorded from @xmath0mo reactions use same parameters for normalizations level density of @xmath11 mev@xmath12 average total radiative width of @xmath13 mev from @xmath14-wave resonances at neutron binding energy @xmath15 details references in refs @xcite new analyzes concern two points old extraction included @xmath1-ray energies close below 778 kev @xmath16 ground band transition transitional region not subtracted in first procedure now excluded second point concerns estimate @xmath1-ray multiplicity as function of excitation energy important for normalizing @xmath1 spectra weighting higher - order generation spectra in subtraction procedureprevious analyses estimated statistical multiplicity at excitation energy @xmath18 introducing lower @xmath1-ray threshold @xmath19 effective excitation energy @xmath20 @xmath21 @xmath22 average energy @xmath1 spectrum for @xmath23 @xmath24 parameter mimics excitation energy statistical @xmath1 transitions enter ground band treatment applicable to rare earth nuclei detector system efficiency for lowest ground state band transitions @xmath25 @xmath16 low for @xmath0mo lowest ground band transitions detected easily present analysis straightforward expression for total @xmath1-ray multiplicity @xmath26 divide excitation energy by average energy @xmath1 spectrum@xmath27 open circles reaction data compared with previously published data level density radiative strength function slightly modified by new multiplicity expression figures compare new level densities radiative strength functions with previous data good resemblance between pick - up reaction inelastic scattering reaction error bars include statistical errors new level densities similar to previous rsfs upbend less pronounced exclusion 778 kev transition values data points found at http://ocl.uio.no / compilation/misunderstandings rsf upbend for low @xmath1 energies @xmath28 @xmath29 fit parameters @xmath31 in mev formula chosen fit low - energy data two parameters description used for @xmath1 energies lower than experimental data points extreme case @xmath32 description unrealistic wrong @xmath1 multiplicity radiative strength function @xmath0mo reanalyzed less enhancement at lower @xmath1-ray energies data points at below 778 kev @xmath16 transition omitted extraction level density coupled radiative strength function new level densities presented financial support from norwegian research council acknowledged work performed under u.s . department of energy university of california lawrence livermore national laboratory contract w-7405-eng-48 support national nuclear security administration stewardship science academic alliances program doe research grants-na00074 u.s . department of energy grant - 97-er41042. larsen guttormsen chankova lnnroth messelt ingebretsen rekstad schiller siem syed voinov phys . c * 76 , 044303 ( 2007 )chankova guttormsen ingebretsen lnnroth messelt rekstad schiller siem syed voinov degrd phys . rev c 73 064301 ( 2006 ) voinov algin agvaanluvsan belgya chankova guttormsen mitchell rekstad schiller s siem . 93 142504 ( 2004 ) guttormsen . c * 71 044307 ( 2005 ) schiller l. bergholt guttormsen. melby j. rekstad s. siem instrum methods phys . 447 , 498 ( 2000 ) chankova . c * 73 034311 ( 2006 ) guttormsen t. ramsy j. rekstad instrum. methods*255 518 ( 1987 ) | 0.266602 | 1,712 | 835 | , Saving $0.1 in GPT-4. |
177,332 | a classical result attributed to joachimsthal in 1846 states that if two surfaces intersect with constant angle along a line of curvature of one surface , then the curve of intersection is also a line of curvature of the other surface . in this note
we prove a global analogue of this result , as follows .
suppose that two closed convex surfaces intersect with constant angle along a curve that is not umbilic in either surface .
we prove that the principal foliations of the two surfaces along the curve are either both orientable , or both non - orientable .
we prove this by characterizing the constant angle intersection of two surfaces in euclidean 3-space as the intersection of a surface and a hypersurface in the space of oriented lines .
the surface is lagrangian , while the hypersurface is null , with respect to the canonical neutral kaehler structure .
we establish a relationship between the principal directions of the two surfaces along the intersection curve in euclidean space , which yields the result .
this method of proof is motivated by topology and , in particular , the slice problem for curves in the boundary of a 4-manifold .
= 0 | a classical result of differential geometry states that if two surfaces , @xmath0 and @xmath1 , in @xmath2 intersect with constant angle along a line of curvature of @xmath0 , then the curve of intersection is also a line of curvature of @xmath1 . in 1846
joachimsthal proved a special case of this theorem @xcite , while the general case was presented by bonnet in 1853 @xcite . in this note
we prove a global analogue of this result .
that is , consider two closed convex surfaces @xmath0 and @xmath1 intersecting transversely along a smooth simple closed curve @xmath3 .
decompose @xmath4 , @xmath5 , where @xmath6 are closed discs , with @xmath7 defined to be the disc with outward pointing normal given by the normal to @xmath1 projected onto @xmath0 , and similarly for @xmath8 .
we prove _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * main theorem . * _ if @xmath0 and @xmath1 intersect with constant angle along a curve that is not umbilic in either @xmath0 or @xmath1 , then the principal foliations of the two surfaces along the curve are either both orientable , or both non - orientable .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the main theorem is proven by characterizing the constant angle intersection of surfaces in @xmath2 as the intersection of a surface @xmath9 and a hypersurface @xmath10 in the space @xmath11 of oriented lines of @xmath2 . here
@xmath12 , where @xmath13 is the angle of intersection .
in this characterization , the surface @xmath9 is a lagrangian section and @xmath14 is a null hypersurface with respect to the neutral khler structure on the space of oriented lines . from these geometric properties we relate the principal directions of the surfaces along the intersection ( equation ( [ e : final ] ) ) and the result follows .
our methods are motivated by topology : the ultimate goal is to investigate whether a knot in the boundary of a 4-manifold is the boundary of a properly embedded disc - the slice problem . as a new geometric model , we propose subsets of @xmath11 and invariants derived from the neutral khler structure , which we claim can be utilized to detect obstructions to sliceness . for surfaces in @xmath2 ,
umbilic points are generically isolated and have a half - integer index associated with them given by the winding number of the principal foliation about the umbilic point .
let @xmath15 be a disc with non - umbilic boundary and only isolated umbilic points in the interior .
the total umbilic index @xmath16 is defined to be the sum of the umbilic indices on @xmath15 and when the boundary of @xmath15 is non - umbilic , it is equal to the winding number of the principal foliation along @xmath17 . from our perspective , umbilic points on @xmath18 correspond to complex points on @xmath9 and the total complex index @xmath19 of the corresponding disc in @xmath11 with totally real boundary is a smooth topological invariant @xcite .
thus our main theorem can be restated as : @xmath20 alternatively , our result provides a situation where one can bound this invariant by geometric data at the boundary hypersurface .
further results on this topic and its relation to the slice problem will be reported in a forthcoming paper .
let @xmath0 and @xmath1 be smooth convex surfaces which intersect along a curve @xmath3 at a constant angle @xmath21
. split the convex surfaces so that @xmath4 , @xmath5 , where @xmath6 are closed discs , with @xmath7 defined to be the disc with outward pointing normal given by the normal to @xmath1 projected onto @xmath0 , and similarly for @xmath8 , as in the diagram below .
now transfer this geometric data to the space @xmath11 of oriented lines of euclidean @xmath2 .
@xmath11 admits a canonical neutral khler structure @xmath22 that is invariant under the induced euclidean action @xcite , and , since @xmath11 is diffeomorphic to @xmath23 , write @xmath24 for @xmath11 endowed with this structure .
the surfaces @xmath0 and @xmath1 in @xmath2 will be treated differently : define @xmath25 to be the set oriented lines that intersect @xmath0 , forming an angle @xmath13 with the outward normal , and define @xmath26 be the set of oriented lines normal to @xmath1 .
clearly these are 3- and 2-dimensional submanifolds of @xmath24 , the latter being diffeomorphic to a 2-sphere while the former is diffeomorphic to a circle bundle over the 2-sphere . the submanifolds @xmath14 and @xmath9 intersect transversely along the curve @xmath27 which consists of the oriented normal lines to @xmath1 along the curve @xmath28 .
the constant angle condition in @xmath2 has been translated to an intersection condition for certain submanifolds in @xmath24 .
we now describe the geometric properties of these submanifolds . starting with @xmath9 , the set of oriented normals to the convex surface @xmath1 .
this is a lagrangian section of the bundle @xmath29 and is locally described by the gradient of the support function @xmath30 .
we now utilize local holomorphic coordinates @xmath31 on @xmath24 to describe this - see @xcite and references therein for further details .
the complex coordinate @xmath32 is the standard holomorphic coordinate on @xmath33 giving the direction of the oriented line , while the complex coordinate @xmath34 gives the perpendicular distance of the line to the origin .
together they yield @xmath35 the surface @xmath9 is the section @xmath36 where @xmath37 and this is lagrangian as @xmath38 define @xmath39 if @xmath40 and @xmath41 are the radii of curvature of @xmath1 , then @xmath42 thus umbilic points on @xmath1 are precisely the complex points on @xmath9 , while the half of the argument of @xmath43 determines the principal directions of the surface @xmath1 .
now suppose we have an oriented line @xmath61 and a point @xmath62 on this line .
thus , for some @xmath63 @xmath64 suppose further that @xmath31 is another oriented line that passes through @xmath62 and forms an angle @xmath13 with @xmath61 .
the former fact means that : @xmath65 substituting the expressions for @xmath66 and @xmath67 , and for @xmath68 from lemma [ l:1 ] , we conclude that the set of oriented lines that intersect @xmath61 at a parameter @xmath69 , forming an angle @xmath13 , is a circle in @xmath24 parameterized by @xmath53 as follows @xmath70 where @xmath12 . by assumption @xmath9 and @xmath10 intersect in the curve @xmath71 , consisting of the oriented normal lines to @xmath1 along the curve of intersection @xmath72 .
parameterize the curve so that @xmath3 is given by equations ( [ e : coordu ] ) where everything now depends upon a parameter @xmath73 .
note that @xmath74 and @xmath75 where we have used equation ( [ e : pot1 ] ) and the definitions of @xmath46 and @xmath47 .
identical expressions hold for the derivatives of @xmath76 and @xmath77 with all subscripts changed from 1 to 2 . to fix unit parameterization of the curve @xmath3 set @xmath78,\ ] ] for @xmath79 .
the curve @xmath71 of oriented lines are tangent to @xmath3 iff @xmath80 or @xmath81 , while , the oriented lines are normal to @xmath3 iff @xmath82 or @xmath83 . in summary , the intersection curve @xmath71 in @xmath24 is the solution to the two complex equations @xmath84 where all quantities depend on @xmath73 . in the second equation we use the given graph function @xmath85 , and similarly for @xmath76 .
thus we have five real unknowns encoded in @xmath86 and four real equations , yielding a one dimensional solution set parameterized by @xmath73 .
differentiate the second equation of ( [ e : int ] ) with respect to @xmath73 and use equations ( [ e : dr1 ] ) , ( [ e : unitpara ] ) and ( [ e : xi2dot ] ) , along with ( [ e : deta1 ] ) and the analogous equation for @xmath76 .
the result , after some computation and rearrangement , is @xmath89\nonumber\\ & + ( e^{ia_1}-\epsilon\xi_1)^2\psi_2\bigg[2\epsilon\kappa_1\frac{d a_1}{du } + \epsilon(\epsilon e^{2ia_1}+e^{ia_1}\xi_1)\sigma_1\nonumber\\ & \qquad\qquad\qquad+(e^{-ia_1}\epsilon\bar{\xi}_1- e^{-2ia_1})\bar{\sigma}_1 -(\epsilon(e^{ia_1}\bar{\xi}_1+e^{-ia_1}\xi_1)+\epsilon^2 - 1)\psi_1\bigg]\nonumber\\ & \qquad\qquad\qquad\qquad-(1+\epsilon^2)\kappa_1(e^{ia_1}-\epsilon\xi_1)^2=0.\nonumber\end{aligned}\ ] ] to transfer this to back to euclidean 3-space , denote the principal curvatures of @xmath0 by @xmath92 and those of @xmath1 by @xmath93 .
let @xmath94 be the angle between the principal direction of @xmath0 and the intersection curve and @xmath95 the corresponding angle on @xmath1 .
the last equation can now be succinctly written @xmath96 thus we obtain joachimsthal s theorem : @xmath3 is a line of curvature on @xmath0 iff it is a line of curvature on @xmath1 .
a further result of joachimsthal s also follows : if @xmath0 is a sphere ( so that @xmath97 ) and the surface @xmath1 intersects it at constant angle , then @xmath98 and so @xmath3 is a line of curvature on @xmath1 .
note that this equation states that the geodesic torsion of the intersection curve on the two surfaces are equal @xcite .
moreover , if @xmath3 contains no umbilic points on @xmath0 or @xmath1 , the winding numbers of the maps @xmath99 count the total umbilic index inside the curve on each surface , where the inside is determined by the outward pointing normals - see theorem 4 of @xcite .
for simplicity , write @xmath100 and @xmath101 so that the derivative of equation ( [ e : final ] ) with respect to @xmath73 is @xmath102 if @xmath103 everywhere along the curve then the winding numbers of the principal foliations are equal and the theorem holds .
otherwise , choose a point @xmath104 such that @xmath105 . by a deformation of the surfaces preserving constant angle we can also assure that @xmath106 ( an open condition ) . by a choice of parameterization of the intersection curve
@xmath3 we can set @xmath107 to be @xmath108 and @xmath109 .
furthermore , by a rotation of the complex coordinates @xmath32 we can set @xmath110 . then @xmath111 and @xmath112 for some @xmath113 . each principal foliation is orientable or non - orientable along @xmath3 according to whether the winding numbers @xmath114 or @xmath115 is even or odd , respectively . now evaluating equation ( [ e : las2 ] ) at @xmath108 and @xmath109 we find @xmath116 since none of @xmath117 or @xmath118 vanish at @xmath107 , both of these equations hold only if @xmath119 for some @xmath120 .
thus the foliations are both orientable or both non - orientable , as claimed . | classical result differential geometry states two surfaces @xmath0 intersect angle line curvature @xmath0 curve intersection line curvature @xmath1 1846 joachimsthal proved special case general case bonnet 1853 prove global analogue consider two closed convex surfaces @xmath0 @xmath1 intersecting transversely closed curve @xmath3 decompose @xmath4 @xmath5 @xmath6 closed discs @xmath7 disc outward pointing normal @xmath1 projected @xmath0 similarly @xmath8we prove _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * main theorem . * _ if @xmath0 and @xmath1 intersect with constant angle along curve not umbilic in or, principal foliations of two surfaces along curve are either both orientable , or both non - orientable ._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ main theorem is proven by characterizing constant angle intersection of surfaces in @xmath2 as intersection of a surface @xmath9 and hypersurface @xmath10 in space @xmath11 of oriented lines of @xmath2 . here @xmath12 , where @xmath13 is angle of intersection . in , surface @xmath9 is a lagrangian section and @xmath14 is a null hypersurface with to neutral khler structure on space of oriented lines . from these geometric properties we relate the principal directions of surfaces along intersection ( equation ( [ e : final ] ) ) and result follows .methods motivated by topology goal investigate knot in boundary 4-manifold embedded disc slice problem new model propose subsets of @xmath11 invariants from neutral khler structure detect obstructions sliceness surfaces @xmath2 umbilic points isolated half - integer index winding number principal foliation umbilic @xmath15 disc with non - umbilic boundary isolated umbilic points interior total umbilic index @xmath16 sum of umbilic indices @xmath15 - equal to winding number principal foliation along @xmath17 umbilic points on @xmath18 correspond to complex points on @xmath9 total complex index @xmath19 of disc @xmath11 boundary is smooth topological invariant @xcite main theorem restated as @xmath20 result invariant by geometric data at boundary hypersurface further results slice problem in forthcoming paper @xmath0 @xmath1 smooth convex surfaces intersect along curve constant angle split convex surfaces @xmath4 closed discs @xmath7 disc with outward pointing normal @xmath1 onto @xmath0 similarly for @xmath8transfer geometric data space @xmath11 oriented lines euclidean @xmath2 @xmath11 admits neutral khler structure @xmath22 invariant under euclidean action diffeomorphic @xmath23 write @xmath24 @xmath11 structure surfaces @xmath0 @xmath1 @xmath2 treated differently define @xmath25 oriented lines intersect @xmath0 angle outward normal @xmath26 oriented lines @xmath1 3- 2-dimensional submanifolds @xmath24 latter diffeomorphic 2-sphere former circle bundle 2-sphere submanifolds @xmath14 @xmath9 intersect transversely curve @xmath27 oriented normal lines @xmath1 curve @xmath28 constant angle condition @xmath2 translated intersection condition submanifolds @xmath24 describe geometric properties submanifolds @xmath9 oriented normals convex surface @xmath1 lagrangian section bundle @xmath29 described gradient support function @xmath30 utilize local holomorphic coordinates @xmath31 @xmath24 complex coordinate @xmath32 standard @xmath33 direction oriented line @xmath34 perpendicular distance line originyield @xmath35 surface @xmath9 section @xmath36 @xmath37 lagrangian @xmath38 define @xmath39 @xmath40 @xmath41 radii curvature @xmath1 @xmath42 umbilic points @xmath1 complex points @xmath9 argument @xmath43 determines directions surface @xmath1 oriented line @xmath61 point @xmath62 @xmath63 @xmath64 @xmath31 oriented line passes @xmath62 forms angle @xmath13 @xmath61 @xmath65 substituting expressions @xmath66 @xmath67 @xmath68 oriented lines intersect @xmath61 parameter @xmath69 angle @xmath13 circle @xmath24 parameterized @xmath53 @xmath70 @xmath12 @xmath9 @xmath10 intersect curve @xmath71 oriented normal lines @xmath1 curve intersection @xmath72 parameterize curve @xmath3 given equations depends parameter @xmath73 @xmath74 @xmath75 equation definitions @xmath46 @xmath47 identical expressions derivatives @xmath76 @xmath77 subscripts changed 1 to 2 fix unit parameterization curve @xmath3 set @xmath78 @xmath79curve @xmath71 oriented lines tangent @xmath3 oriented lines normal @xmath3 intersection curve @xmath71 @xmath24 solution complex equations @xmath84 quantities depend @xmath73 second equation use graph function @xmath85 @xmath76 five unknowns @xmath86 four equations one solution parameterized @xmath73 differentiate second equation [ e : int ] ) @xmath73 use equations e : dr1 ] e : unitpara ] : xi2dot ] e : deta1 ] ) analogous equation @xmath76result computation @xmath89\nonumber\\ e^{ia_1}-\epsilon\xi_1)^2\psi_2\epsilon\kappa_1 +\epsilon e{2ia_1}+e{ia_1\xi_1)\sigma_1\qquad+(e^{-ia_1}\epsilon\xi}_1}_1 -\epsilon\xi_1)+\epsilon^2 - 1)\psi_1\qquad-(1+\epsilon^2)\kappa_1(e^{ia_1}-\epsilon\xi_1)^2=0.\nonumber{aligned} transfer euclidean 3-space denote curvatures @xmath0 by @xmath92 @xmath1 by @xmath93 @xmath94 angle between direction @xmath0 intersection curve @xmath95 angle @xmath1 last equation written @xmath96 joachimsthal s theorem : @xmath3 line of curvature on @xmath0 @xmath1 @xmath0 sphere @xmath1 intersects constant angle @xmath98 @xmath3 line of curvature on @xmath1equation states geodesic torsion intersection curve surfaces equal @xcite if @xmath3 no umbilic points @xmath0 winding numbers @xmath99 count total umbilic index inside curve determined by outward normals theorem 4 @xcite write @xmath100 @xmath101 derivative equation @xmath73 is @xmath102 if @xmath103 curve winding numbers principal foliations equal theorem holds choose point @xmath104 @xmath105 deformation surfaces constant angle @xmath106 parameterization intersection curve @xmath3 set @xmath107 @xmath108 @xmath109 rotation coordinates @xmath32 set @xmath110 @xmath113 each principal foliation orientable or non - orientable along @xmath3 according winding numbers @xmath114 @xmath115 even or odd evaluating equation at @xmath108 @xmath109 @xmath116 none @xmath117 or @xmath118 vanish at @xmath107 equations hold if @xmath119 @xmath120 foliations both orientable or non - orientable | 0.557284 | 3,347 | 1,822 | , Saving $0.1 in GPT-4. |
177,341 | this contribution presents the optical spectral properties of fir - luminous galaxies , whose distinctive feature is often the simultaneous presence in the spectra of a strong @xmath0 line in absorption and of emission lines ( e(a ) spectra ) .
a discrepancy between the star formation rate estimated from the fir luminosity and that derived from the @xmath1 luminosity persists even after having corrected the @xmath1 flux for dust according to the observed balmer decrement .
it is shown that the e(a ) spectrum can be reproduced assuming a current starburst and a dust extinction affecting the youngest stellar populations much more than the older stars . | the most distinctive feature in the optical spectra of fir - luminous galaxies is not the equivalent width of the emission lines , but the equivalent width of the @xmath0 line in absorption , which is generally stronger than in the spectra of optically selected galaxies .
in fact , emission line spectra with ew(@xmath2 @xmath3 ( e(a ) spectra ) are frequent among fir galaxies , while the great majority of nearby spirals in optical samples have ew(@xmath4 @xmath3 ( e(c ) spectra ) ( poggianti et al . 1999 [ p99 ] ; poggianti & wu 2000 [ pw00 ] ) . the e(a ) spectral class differs from the so - called `` k+a '' ( or `` e+a '' ) galaxies which only display a strong @xmath0 line in absorption , but no emission lines .
what is the origin of the difference in the @xmath0 strength between dusty starburst galaxies and quiescent spirals ?
what star formation and dust properties generate the peculiar spectral combination found in e(a ) galaxies ? in the following i will address these issues presenting the analysis of a spectroscopic survey of luminous iras galaxies ( pw00 ) and the results of an effort to model the e(a ) spectrum in detail ( poggianti , bressan & franceschini 2001 ) .
in pw00 we analyzed the spectral characteristics of a complete sample of iras galaxies ( wu et al .
1998a , b ) comprising 73 very luminous infrared galaxies ( vlirgs , @xmath5 ) with a median @xmath6 and 40 companion galaxies , selected from the 2 jy redshift survey of strauss et al .
the great majority of these galaxies show evidence for a strong interaction or a merger ( wu et al .
the spectra typically cover the central 2 kpc of the galaxies at a @xmath7 @xmath3 resolution .
table 1 shows the fraction of vlirgs as a function of the spectral class ; a detailed description of each class can be found in pw00 .
more than half of these galaxies have an e(a ) spectrum , about 1 out of 4 has an e(c ) spectrum , and only 1 out of 10 has very strong [ oii ] emission ( e(b ) type ) .
none of the e(a ) spectra would be classified as a seyfert1 or 2 according to the standard diagnostic diagrams of line intensity ratios .
notably no k+a spectrum is found among the fir luminous galaxies , neither among their companions : all the vlirgs display emission lines in their spectra . )
= 0.4 ew(@xmath1 + nii ) ) . ] a general characteristic of the e(a ) spectra in all environments and at all redshifts appears to be the low [ oii]/@xmath1 ratio ( see fig .
1 ) : both the equivalent width and the flux ratios of these lines are a factor of 2 lower than the median ratios observed in nearby spirals . though such low ratios may be caused by various reasons , the most likely explanation is reddening by dust : the color excess e(b - v ) derived from the observed balmer decrement ( see table 1 ) is consistent with and fully accounts for the observed [ oii]/@xmath1 values .
it is noteworthy that the difference in the _ equivalent width _ ratio between this sample and optically selected nearby spirals is entirely due to the difference in the average _ flux _ ratio of these two lines .
this can be interpreted as an effect of _ selective dust extinction _ which affects the youngest stellar generations ( still deeply embedded in large amounts of dust and responsible for the ionizing photons producing the emission lines ) much more than older , less extincted stellar populations producing the continuum underlying the lines : the net result is a low ew ratio .
additional evidence for a selective extinction will be discussed in sec .
luminosities in solar units . no dust correction has been applied .
the best fit to the datapoints is shown as a solid line .
the relation found by devereux & young ( 1990 ) for a sample of field spiral galaxies in the local universe is extrapolated to the fir luminosities of the present dataset and is shown as a dotted line .
`` e '' spectra are those with at least one detected emission line , but @xmath0 unmeasurable . ] as in all fir luminous samples , there is a deficiency of @xmath1 luminosity at a given fir luminosity as compared to optically selected spirals ( fig . 2 ) and this scarcity of @xmath1 flux translates directly into a difference in the sfr estimate : even applying to the @xmath1 flux the dust correction derived from the balmer decrement is not sufficient to reconcile the two estimates of current star formation obtained from the fir and @xmath1 luminosities ( fig .
3 ) . after dust correction
, the @xmath8 is still a factor of @xmath9 lower than the @xmath10 ; it is unlikely this discrepancy is entirely due to the limited aperture of the spectroscopic slit and i will come back to this point in sec .
flux versus the fir based estimate .
the dotted line is the fit to the e(a ) population .
the solid line shows the relation @xmath11 , and the dashed lines are found for 1 mag extinction at @xmath1 ( average extinction in nearby spirals ) and for the average extinction in e(a ) s in w98 sample as determined by the balmer decrement ( e(b - v)=1.1 , a(@xmath12 mag ) ] fraction of galaxies and e(b - v ) as a function of the spectral class .
lccl class & % & median e(b - v ) & comments ( see pw00 for details ) + e(a ) & 0.56@xmath130.10 & 1.11 & strong balmer absorption plus emission + e(c ) & 0.25@xmath130.07 & 0.68 & weak / moderate balmer absorption plus emission + e(b ) & 0.10@xmath130.04 & 0.62 & strong emission ( ew([oii])@xmath14 ) + sey1 & 0.10@xmath130.04 & & seyfert1 from broad hydrogen lines in emission + ( thin line ) .
the latter is taken from fig .
12 in flores et al .
( 1999 ) and has been normalized to the vlirgs spectrum over the wavelength range in common .
this is the average spectrum of 5 iso - detected galaxies at @xmath15 whose spectral energy distributions at visible , near - ir , mir and radio wavelengths resemble those of highly reddened starbursts in the local universe .
the spectral resolution is 10 @xmath3 ( wu ) and 40 @xmath3 ( flores ) .
] the analysis presented so far refers to a well defined sample of iras luminous galaxies , but the e(a ) phenomenon appears to be widespread among ir luminous galaxies at any redshift ( see pw00 for a census of the e(a ) occurrence ) . as an example , fig .
4 shows that the average e(a ) spectrum of wu s sample is very similar to the average spectrum of distant starburst galaxies detected at 15 micron by iso , both in the continuum shape and in the strength of the [ oii ] , @xmath0 and @xmath16 lines .
a selective dust extinction has been proposed as the physical origin of the e(a ) spectra ( p99 , pw00 ) : dust obscuration affects the youngest stellar generations more than the older stars .
this is expected to explain both the observed [ oii]/@xmath1 ratio , as discussed in the previous section , and the @xmath0 strength because the stellar populations responsible for this line ( with ages a few times @xmath17 1.5 @xmath18 yr ) have had time to drift away from or disperse the dusty molecular clouds where they were born and their emission can dominate the integrated spectrum at 4000 @xmath3 if younger stars are more heavily obscured . furthermore , a selective extinction appears to be the most plausible explanation for the fact that , even within the same spectrum , different values of extinction are measured depending on the spectral region / feature used to estimate it : for example , the dust attenuation of the uv / optical stellar continuum is often measured to be lower than the obscuration of the emission lines ( see pw00 for a reference list ) . in order to verify whether the hypothesis of selective extinction in a starburst galaxy can indeed account for the e(a ) spectrum , a simplified spectrophotometric model , including only 10 stellar populations ,
has been developed ( poggianti , bressan & franceschini 2001 ) .
this represents the minimum set of stellar populations that are known to be essential because they affect the spectral features that we wish to reproduce : four young generations ( @xmath19 yr ) responsible for the ionizing photons that produce the emission lines ; five intermediate populations ( @xmath20 yr ) with the strongest balmer lines in absorption , and older stars modelled as a constant star formation rate between 1 and 12 gyr before the moment of the observation which can give a significant contribution to the continuum .
the spectrum of each stellar generation is found from a spectrophotometric model that includes both the stellar and the nebular contribution ( barbaro & poggianti 1997 ) and it is extincted with its own extinction value that is allowed to vary from a stellar population to another assuming a dust screen with a standard galactic extinction law and a salpeter imf between 0.1 and 100 @xmath21 .
the results of this model are compared with the average e(a ) spectrum of the vlirgs discussed in sec . 2 ;
the quality of the fit is assessed considering the differences between the model and the observed spectrum in the equivalent widths of four lines ( [ oii]@xmath223727,h@xmath23 , h@xmath24 and h@xmath25 ) and the continuum flux in eight spectral windows in the range 37706460 .
the difference between the model and the observed spectrum is also shown ( at flux@xmath26 ) .
the vertical segment on the right side represents the @xmath27 error in @xmath1 .
_ right : _ sfr ( histogram , normalized to = 1 in the old population ) and e(b - v ) ( dots ) of the models whose spectrum is shown in the left panel .
the observed fir / v ratio is = 88.0.,scaledwidth=110.0% ] \1 ) the e(a ) spectrum is consistent with the starburst / selective extinction scenario .
the upper left panel of fig . 5 presents the comparison between the average e(a ) spectrum and a model of a burst that began @xmath28 yr ago ; the extinction of the starburst populations is significantly higher than that of the older generations ( see the right panel in fig .
the fit is remarkable , both as far as the line equivalent widths and the continuum are concerned .
but this is only at the 4% level . ]
\a ) slit aperture effects i.e. a mismatch in the galactic area sampled by the optical spectrum ( central 2 kpc ) and by iras ( integrated ) .
given that the ir emissivity is usually concentrated in the central regions of luminous infrared galaxies , it is hard to envisage how slit effects can account for the discrepancy between the modelled and the observed ir flux .
\b ) dusty starburst galaxies often have star forming regions which are completely obscured at optical wavelengths , hence give no contribution to the spectrum but produce a significant fraction of the fir luminosity ( e.g. mirabel et al .
the observed fir / v ratio can be reproduced by starburst models with regions that are highly obscured by dust with an e(b - v ) even greater than in the models of fig .
\3 ) no strong constraint can be placed on the burst duration .
models with a starburst as short as a few times @xmath29 yr and as long as @xmath18 yr are able to fit the observed e(a ) spectrum as long as : a ) the youngest populations are highly extincted and b ) the old population provides a contribution but does not overwhelm the intermediate age contribution at 4000 .
as an example , the lower left panel in fig .
5 shows the excellent fit obtained with a current burst that began @xmath30 yr ago . in the case of short bursts ( @xmath31 yr ) ,
the strong @xmath0 line is not produced by the stars born during the starburst event , but by the previous stellar populations .
\4 ) we also considered a family of `` post - starburst models '' with extinction , assuming a galaxy is seen _
after _ a strong starburst phase when a small amount of residual star formation activity is still ongoing . this type of models besides accounting for not more than @xmath32 of the observed fir luminosity fail to reproduce simultaneously the @xmath1 and the @xmath16 lines , either underestimating the former or overestimating the latter .
g. barbaro & b.m .
poggianti : a&a * 490 * , 504 ( 1997 ) n.a .
devereux , j.s .
young : apj * 350 * , l25 ( 1990 ) h. flores , f. hammer , t.x .
thuan , c. cesarsky , f.x .
desert , a. omont , s.j .
lilly , s. eales , d. crampton , o. le fevre : apj * 517 * , 148 ( 1999 ) i.f .
mirabel , l. vigroux , v. charmandaris , m. sauvage , p. gallais , d. tran , c. cesarsky , s.c .
madden , p.a .
duc : a&a * 333 * , l1 ( 1998 ) b.m .
poggianti , i. smail , a. dressler , w.j .
couch , a.j .
barger , h. butcher , r.s .
ellis , a. oemler : apj * 518 * , 576 ( 1999 , p99 ) b.m .
poggianti & h. wu : apj * 529 * , 157 ( 2000 , pw00 ) b.m .
poggianti , a. bressan , a. franceschini : apj , in press ( 2001 ) astro - ph 0011160 m.a .
strauss et al . :
apjs * 83 * , 29 ( 1992 ) h. wu , z.l .
zou , x.y .
xia , z.g .
deng : a&as * 127 * , 521 ( 1998a , w98a ) h. wu , z.l .
zou , x.y .
xia , z.g .
deng : a&as * 132 * , 181 ( 1998b , w98b ) | distinctive feature in spectra fir - luminous galaxies not emission lines @xmath0 line in absorption stronger than selected galaxies emission line spectra with ew(@xmath2 @xmath3 e spectra frequent among fir galaxies majority of nearby spirals have ew(@xmath4 @xmath3 e spectra poggianti 1999 wu 2000 e(a ) spectral class differs from `` k+a galaxies strong @xmath0 line in absorption no emission lines origin of difference in @xmath0 strength between dusty starburst galaxies quiescent spirals ? star formation dust properties generate peculiar spectral combination in e(a ) galaxies ? analysis spectroscopic survey of luminous iras galaxies e(a ) spectrum analyzed spectral characteristics of sample iras galaxies wu 73 luminous infrared galaxies with median @xmath6 40 companion galaxies selected from 2 jy redshift survey of strauss et al majority show strong interaction or merger spectra cover central 2 kpc galaxies at @xmath7 @xmath3 resolution table 1 shows fraction vlirgs spectral class detailed description in pw00 half galaxies have e(a ) spectrum 1 out of 4 has e(c ) spectrum 1 out of 10 has strong oii ] emission ( e(b ) type none e(a ) spectra seyfert1 or 2 according standard diagnostic diagrams line intensity ratios no k+a spectrum among fir luminous galaxies neither companions vlirgs display emission lines = 0.4 ew(@xmath1 + nii ) ) characteristic e(a ) spectra low [ oii]/@xmath1 ratio fig . 1 equivalent width flux ratios factor 2 lower than median ratios nearby spirals low ratios likely explanation reddening by dust color excess e(b - v ) from balmer decrement table 1 consistent with accounts for [ oii]/@xmath1 values difference in equivalent width _ ratio between sample nearby spirals due to difference average flux _ ratio effect selective dust extinction affects youngest stellar generations than older populations net result low ew ratio additional evidence for selective extinction discussed in sec in solar units no dust correction appliedbest fit to datapoints solid line relation by devereux & young 1990 for field spiral galaxies extrapolated to fir luminosities dataset as dotted line e spectra with detected emission line @xmath0 unmeasurable deficiency of @xmath1 luminosity at fir compared to spirals scarcity translates into difference in sfr estimate dust correction from balmer decrement estimates star formation from fir @xmath1 luminosities after dust correction @xmath8 @xmath9 lower than @xmath10 unlikely discrepancy due to limited aperture spectroscopic slit dotted line fit to e(a ) population solid line shows relation @xmath11 dashed lines for 1 mag extinction at @xmath1 average extinction in e(a ) w98 sample by balmer decrement ( e(b - v)=1.1 @xmath12 mag fraction of galaxies e(b - v function of spectral class median e(b - v ) + e(a ) & 0.56@xmath130.10 & 1.11 strong balmer absorption plus emission + e(c ) & 0.25@xmath130.07 & 0.68 weak moderate balmer absorption emission + e(b ) 0.10@xmath130.04 0.62 strong emission@xmath14 sey1 0.10@xmath130.04 seyfert1 hydrogen lines emission thin line latter taken from fig . 12 flores et al. 1999 normalized vlirgs spectrum average spectrum 5 iso - detected galaxies @xmath15 resemble reddened starbursts spectral resolution 10 @xmath3 ( wu ) 40 @xmath3 ( flores ) analysis refers sample iras luminous galaxies e(a ) phenomenon widespread ir luminous galaxies redshift pw00 census e(a ) occurrence fig . 4 average e(a ) spectrum wu s sample similar distant starburst galaxies 15 micron iso continuum shape strength oii @xmath0 @xmath16 lines selective dust extinction proposed origin e(a ) spectra dust obscuration affects youngest stellar generations olderexpected explain observed [ oii]/@xmath1 ratio @xmath0 strength stellar populations responsible line ages @xmath17 1.5 @xmath18 yr ) dusty molecular clouds emission dominate spectrum at 4000 @xmath3 if younger stars obscured selective extinction plausible explanation different values extinction measured depending spectral region / feature dust attenuation of uv / optical stellar continuum lower than obscuration emission lines see pw00 reference list verify selective extinction in starburst galaxy for e(a ) spectrum simplified spectrophotometric model 10 stellar populations developed ( poggianti 2001 ) minimum stellar populations essential spectral features four young generations ( @xmath19 yr ) for ionizing photons five intermediate populations ( @xmath20 yr ) strongest balmer lines older stars constant star formation rate between 1 and 12 gyr before observation significant contribution to continuum spectrum of each stellar generation found from spectrophotometric model stellar nebular contribution 1997 extincted with own extinction value from stellar population assuming dust screen standard galactic extinction law salpeter imf between 0.1 and 100 @xmath21 .results model compared with average e(a ) spectrum vlirgs 2 quality fit assessed considering differences model observed spectrum in equivalent widths of four lines@xmath223727 continuum flux in eight spectral windows range 37706460 difference between model observed spectrum shown at flux@xmath26 vertical segment represents @xmath27 error in @xmath1 sfr histogram 1 e(b - v ) dots models spectrum left observed fir / v ratio = 88.0.,scaledwidth=110.0% e(a ) spectrum consistent with starburst / selective extinction scenario panel fig . 5 comparison average e(a ) spectrum model burst @xmath28 yr ago extinction starburst populations higher than older generations fit remarkable line equivalent widths continuum only at 4% level slit aperture effects mismatch in galactic area optical spectrum iras ir emissivity concentrated in central regions luminous infrared galaxies hard to envisage slit effects for discrepancy between modelled observed ir flux dusty starburst galaxies have star forming regions obscured at optical wavelengths no contribution spectrum produce significant fraction fir luminositymirabel et al observed fir / v ratio reproduced by starburst models regions obscured by dust e(b - v ) greater than fig no constraint on burst duration models starburst short @xmath29 yr long @xmath18 yr fit e(a ) spectrum youngest populations extincted old population overwhelm intermediate age contribution at 4000 lower left panel . 5 shows excellent fit with current burst @xmath30 yr ago short bursts @xmath31 yr ) strong @xmath0 line not produced by stars born previous stellar populations family post - starburst models with extinction galaxy seen after strong starburst residual star formation activity ongoing models not more than @xmath32 fir luminosity fail to reproduce @xmath1 @xmath16 lines underestimating or overestimating barbaro poggianti : a * 490 *, 504 ( 1997 ) devereux young * 350 * l25 ( 1990 ) flores hammer thuan cesarsky desert omont lilly crampton le fevre * 517 * 148 ( 1999 ) mirabel vigrouxcharmandaris sauvage gallais tran cesarsky madden duc 333 l1 ( 1998 ) poggianti smail dressler couch barger butcher ellis oemler apj 518 576 ( 1999 p99 ) poggianti h. wu apj 529 157 ( 2000 pw00 ) poggianti bressan franceschini apj press ( 2001 ) astro - ph 0011160 strauss apjs * 83 * 29 ( 1992 ) h. wu z zou xia deng 127 * 521 ( 1998a w98a wu zou deng * 132 * 181 ( 1998b w98b | 0.507551 | 3,279 | 1,669 | , Saving $0.1 in GPT-4. |
177,345 | we revisit the classical problem of nucleated polymerisation and derive a range of exact results describing polymerisation in systems intermediate between the well - known limiting cases of a reaction starting from purely soluble material and for a reaction where no new growth nuclei are formed . | the classical theory of nucleated polymerisation@xcite describes the growth of filamentous structures formed through homogeneous nucleation@xcite .
this framework was initially developed by oosawa and coworkers in the 1960s@xcite to describe the formation of biofilaments , including actin and tubulin .
this theory has been generalised to include secondary nucleation processes by eaton and ferrone@xcite in the context of their pioneering work elucidating the polymerisation of sickle haemoglobin , and by wegner@xcite in order to include fragmentation processes into the growth model for actin filaments . for irreversible growth in the absence of pre - formed seed material and secondary nucleation pathways , in 1962
oosawa presented solutions to the kinetic equations which were very successful in describing a variety of characteristics of the polymerisation of actin and tubulin .
the other limiting case , namely where seed material is added at the beginning of the reaction and where no new growth nuclei are formed during the reaction , is also well known . in this paper
, we present exact results which encompass all cases between these limiting scenarios , extending the results of oosawa for a system dominated by primary nucleation to the case where an arbitrary concentration of pre - formed seed material is present .
we also discuss a range of general closed form results from the oosawa theory for the behaviour of a system of biofilaments growing through primary nucleation and elongation .
we then compare the behaviour of systems dominated by primary nucleation to results derived recently for systems dominated by secondary nucleation .
the theoretical description of the polymerisation of proteins such as actin and tubulin to yield functional biostructures was considered in the 1960s by oosawa@xcite . for a system that evolves through primary nucleation of new filaments , elongation of existing filaments , and depolymerisation from the filament ends , the change in concentration of filaments of size @xmath0 ,
denoted @xmath1 , is given by the master equation@xcite : @xmath2 where @xmath3 , @xmath4 , @xmath5 are rate constants describing the elongation , depolymerisation and nucleation steps and @xmath6 is the concentration of free monomeric protein in solution .
the factor of 2 in eq . originates from the assumption of growth from both ends . for the case of irreversible biofilament growth
, the polymerisation rate dominates over the depolymerisation rate ; from eq .
, the rate of change of the number of filaments , @xmath7 , and the free monomer concentration , @xmath6 , were shown by oosawa under these conditions @xcite to obey : @xmath8 combining eqs .
and yields a differential equation for the free monomer concentration@xcite : @xmath9 . here ,
we integrate these equations in the general case where the initial state of the system can consist of any proportion of monomeric and fibrillar material ; this calculation generalises the results presented by oosawa to include a finite concentration of seed material present at the start of the reaction . beginning with eqs . and , the substitution @xmath10 followed by multiplication through by @xmath11 yields : @xmath12 = \frac{d}{dt}e^{n_c z}\ ] ] integrating both sides results in : @xmath13 we obtain a separable equation for @xmath11 , which can be solved to yield : @xmath14 integration and exponentiation yields the expression for @xmath6 : @xmath15^{1/n_c}\ ] ] inserting the appropriate boundary conditions in terms of @xmath16 and @xmath17 fixes the values of the constants @xmath18 and @xmath19 , resulting in the final exact result for the polymer mass concentration @xmath20 : @xmath21^{\beta } \label{eq : oosawaseededm}\ ] ] where the effective rate constant @xmath22 is given by @xmath23 and @xmath24 , @xmath25 , @xmath26 for @xmath27 .
we note that this expression only depends on two combinations of the microscopic rate constants , @xmath28 and @xmath22 .
the result reveals that @xmath22 controls the aggregation resulting from the newly formed aggregates , whereas @xmath29 defines growth from the pre - formed seed structures initially present in solution . in the special case of the aggregation reaction starting with purely soluble proteins , @xmath30 , @xmath31 ,
these expressions reduce to @xmath32 and @xmath33 , and eq .
yields the result presented by oosawa@xcite and the single relevant parameter in the rate equations is @xmath22 .
interestingly , generalisations of eq . which include secondary pathways ,
maintain the dependence on @xmath22 and @xmath29 but introduce an additional parameter analogous to @xmath22 for each active secondary pathway@xcite .
an expression for the evolution of the polymer number concentration , @xmath7 may be derived using eq . .
direct integration of eq . gives the result for @xmath7 : @xmath34 eqs . and give in closed form the time evolution of the biofilament number and mass concentration growing through primary nucleation and filament elongation . ; a lag - phase exists when the initial gradient is not the maximal gradient .
the numbers accompanying each curve are @xmath35 ; eq . predicts that a lag - phase only exists when this ratio is less than unity .
( a ) : polymerisation in the presence of an increasing quantity of seed material of a fixed average length ( 5000 monomers per seed ) added at the beginning of the reaction .
the seed concentrations given as a fraction of the total concentration of monomer present are right to left ) : 0 , 0.01 , 0.04 , 0.1 , 0.2 , 0.5 .
( b ) : nucleated polymerisation in the presence of a fixed quantity ( 1% of total monomer in the system ) of seed material of varying average length . the average number of monomer per seed are ( right to left ) : n / a ( unseeded ) , 5000 , 1000 , 500 , 200 , 50 . the other parameters for both panels
are : @xmath36 , @xmath37 , @xmath38 , @xmath39m@xmath40 s@xmath40 .
[ fig : seeded],scaledwidth=100.0% ] insight into the early time behaviour of the polymer mass concentration can be obtained by expanding eq .
for early times to yield : @xmath41t^2 /2
+ \mathcal{o}(t^3)\ ] ] this expression recovers the characteristic @xmath42 dependence of the oosawa theory and has an additional term linear in time relating to the growth of pre - formed aggregates . in many cases , eq .
describes a sigmoidal function with a lag phase .
the time of maximal growth rate , @xmath43 , can be found from the inflection point of the sigmoid from the condition @xmath44 : @xmath45 ( \mu \beta^{-\frac{1}{2}}\lambda_0)^{-1}\ ] ] such that a lag phase exists only for : @xmath46 using the composition @xmath47 reduces this to the simple condition : @xmath48 in other words , a point of inflection exists if the growth through elongation from the ends of pre - existing seeds , @xmath29 , is less effective that the effective growth through nucleation and elongation of new material , @xmath22 .
this result imples that an increased nucleation rate promotes the existence of an inflection point , whereas an increased elongation rate or an increased level of seeding tends to disfavour its existence . in particular
, we also note that in the absence of nucleation , an inflection point can not exist in the polymer mass concentration as a function of time .
interestingly , the result eq
. is analogous to the criterion applicable for fragmentation dominated growth where a lag phase only exists when the parameters controlling fragmentation - related secondary nucleation is larger than @xmath49 .
the maximal growth rate , @xmath50 , is given by : @xmath51 which occurs at a polymer mass concentration @xmath52 given from eq .
: @xmath53 the lag time , @xmath54 , is then given by : @xmath55 ( \mu \beta^{-\frac{1}{2 } } \lambda)^{-1 } \end{split } \label{eq : oo2}\ ] ] interestingly , from eq .
, we note that a point of inflection can never exist for @xmath7 for simple nucleated polymerisation .
by contrast , when secondary pathways are active , an inflection point can frequently be present@xcite .
> p5.0cm|>p3.0cm|>p3.0cm|>p4.7 cm & * primary nucleation*&*fragmentation*&*monomer - dependent secondary nucleationkinetic parameters*&@xmath22 , @xmath3&@xmath22 , @xmath56 , @xmath3&@xmath22 , @xmath57 , @xmath3**early time growth**&polynomial&exponential&exponential**scaling behaviour + ( lag time , max growth rate)**&yes + with @xmath22&yes + with @xmath56&yes + with @xmath57 * positive feedback*&no&yes&yes many systems that evolve through nucleated polymerisation display characteristic scaling behaviour@xcite .
this behaviour can be seen to be a consequence of the fact that under many conditions , the rate equations are dominated by a single parameters that corresponds to the dominant form of nucleation : @xmath22 for classical nucleated polymerisation and @xmath58 for polymerisation in the presence of secondary pathways .
these parameters have the general form @xmath59 where @xmath60 corresponds to the nucleation process and @xmath61 is related to the monomer dependence of this process : @xmath62 , where @xmath63 is the critical nucleus size for primary nucleation , @xmath64 for fragmentation driven growth and @xmath65 , the secondary nucleus size in cases where monomer - dependent secondary nucleation is dominant .
the dominance of a single combination of the rate constants implies that many of the macroscopic system observables will be correlated since they are dependent on the same parameter .
a striking examples of this behaviour is provided by the very general correlation between the lag - time and the maximal growth rate@xcite , which is manifested in the present case in eqs . and as @xmath66 and @xmath67 .
interestingly the rate equations describe sigmoidal curves both in the presence and in the absence of secondary nucleation processes . for more complex primary nucleation pathways@xcite the polynomial form for the early time solution
is maintained , but higher - order exponents are obtained . in the absence of secondary processes , however , the lag - phase is less marked since the early time rise is a slower polynomial relationship rather than the exponential onset characteristic of secondary pathways@xcite .
this observation implies that the difference between a high - order polynomial and an exponential may not be apparent in experimental data in the presence of noise , and therefore a global analysis of the system under different conditions is required in order to obtain robust mechanistic information@xcite .
in this paper , we have provided results for the time course of nucleated polymerisation for systems that are initially in a mixed state and contain both monomeric and fibrillar material .
these results generalise the classical oosawa theory that describes the formation of biofilaments to cases where an arbitrary amount of pre - formed seed material is present in the system . furthermore
, these results represent a reference to which polymerisation driven by secondary pathways can be compared .
we are grateful to the schiff foundation ( siac ) , and to the wellcome ( mv , cmd , tpjk ) and leverhulme trusts ( cmd ) for financial support .
cohen , s.i.a . ; vendruscolo , m. ; dobson , c.m . ; knowles , t.p.j .
nucleated polymerisation with secondary pathways ii .
determination of self - consistent solutions to growth processes described by non - linear master equations . .
knowles , t.p.j . ; waudby , c.a . ; devlin , g.l . ; cohen , s.i.a . ; aguzzi , a. ; vendruscolo , m. ; terentjev , e.m . ; welland , m.e . ;
dobson , c.m .
an analytical solution to the kinetics of breakable filament assembly .
, _ 326 _ , 15331537 . | classical theory of nucleated polymerisation describes growth filamentous structures through homogeneous nucleation framework developed by oosawa coworkers 1960s formation biofilaments including actin tubulin theory generalised to secondary nucleation by eaton ferrone wegner fragmentation growth model for actin filaments for irreversible growth pre - formed seed material secondary nucleation pathways 1962 oosawa presented solutions kinetic equations characteristics polymerisation of actin tubulin limiting case seed material added at no new growth nuclei formed known paper present results all cases between limiting scenarios extending results for primary to arbitrary concentration pre - formed seed material discuss general closed form results from theory for system biofilaments growing through primary nucleation compare behaviour of systems dominated by primary nucleation to results systems secondary nucleation theoretical description of polymerisation of proteins actin tubulin functional biostructures considered 1960s by oosawasystem evolves through nucleation new elongation existing depolymerisation ends change in concentration filaments size @xmath0 given by master equation@xcite : @xmath2 @xmath3 rate constants elongation depolymerisation nucleation @xmath6 concentration free monomeric protein in solution factor 2 from assumption growth both ends irreversible biofilament growth polymerisation rate dominates depolymerisation rate rate change number filaments @xmath7 free monomer concentration @xmath6 shown by oosawa under conditions @xcite @xmath8 yields differential equation for free monomer concentration@xcite : @xmath9 integrate equations case initial state system any proportion monomeric fibrillar material calculation generalises results include finite concentration seed material start reactionsubstitution @xmath10 multiplication @xmath11 yields @xmath12 = {d}{dt}e{n_c integrating sides results @xmath13 separable equation @xmath11 solved @xmath14 integration exponentiation yields expression @xmath6 : @xmath15^{1/n_c} inserting boundary conditions @xmath16 fixes values constants @xmath18 @xmath19 final result polymer mass concentration @xmath20 : @xmath21^{{eq oosawaseededm effective rate constant @xmath22 given by @xmath23 @xmath24 @xmath27 expression depends on microscopic rate constants @xmath28 @xmath22 @xmath22 controls aggregation formed aggregates @xmath29 defines growth pre - formed seed structures reaction soluble proteins expressions reduce to @xmath32 @xmath33 yields result oosawa@xcite single relevant parameter @xmath22 generalisations eq secondary pathways maintain dependence on @xmath22 @xmath29 introduce additional parameter @xmath22 each active secondary pathway expression evolution polymer number concentration @xmath7 derived using integrationresult @xmath7 @xmath34 time evolution biofilament number mass concentration through primary nucleation filament elongation lag - phase exists initial gradient not maximal numbers curve @xmath35 lag - phase ratio less than unity polymerisation increasing quantity seed material fixed average length ( 5000 monomers per seed beginning reaction seed concentrations 0 0.01 0.04 0.1 0.2 0.5 nucleated polymerisation fixed quantity 1% total monomer seed material varying average length average number monomer per seed n / a , 5000 1000 500 200 50 other parameters panels @xmath36 @xmath37 @xmath38 @xmath39m@xmath40 fig seeded,scaledwidth=100.0% early time behaviour polymer mass concentration expanding eq . @xmath41t^2 /2 + \mathcal{o}(t^3)\ expression recovers @xmath42 dependence oosawa theory additional term linear time growth pre - formed aggregates describes sigmoidal function with lag phasemaximal growth rate @xmath43 found from inflection point sigmoid condition @xmath44 : @xmath45 (\beta^{-\frac{1}{2}}\lambda^{-1} lag phase exists for @xmath46 composition @xmath47 reduces to condition @xmath48 point of inflection exists if growth through elongation from seeds @xmath29 less effective growth nucleation new material @xmath22 increased nucleation rate promotes inflection point increased elongation rate level seeding absence of nucleation inflection point exist in polymer mass concentration function time result analogous to fragmentation dominated growth lag phase exists when fragmentation secondary nucleation larger than @xmath49 maximal growth rate @xmath50 given by @xmath51 at polymer mass concentration @xmath52 lag time @xmath54 given by @xmath55 ( \mu \beta^{-\frac{1}{2\lambda)^{-1 } \end{split{eq oo2} point of inflection for @xmath7 for simple nucleated polymerisation when secondary pathways active inflection point present p5.0cm>p3.0cm7 cm primary nucleation*fragmentation*monomer - dependent secondary nucleationkinetic parameters@xmath22 @xmath3 time&polynomial&exponential behaviour lag time max growth rate @xmath22 @xmath56 @xmath57 positive feedback systems through nucleated polymerisation display scaling behaviour rate equations dominated by single parameters dominant form nucleation @xmath22 classical nucleated polymerisation @xmath58 polymerisation secondary pathways parameters general form @xmath59 @xmath60 nucleation process @xmath61 monomer dependence @xmath62 @xmath63 critical nucleus size for primary nucleation @xmath64 for fragmentation driven growth @xmath65 secondary nucleus size monomer - dependent secondary nucleation dominant dominance of single combination rate constants implies macroscopic system observables correlated dependent on same parameter general correlation between lag - time maximal growth rate manifested eqs @xmath66 @xmath67 rate equations describe sigmoidal curves presence absence secondary nucleation processescomplex primary nucleation pathways polynomial form early time solution maintained higher - order exponents obtained absence of secondary processes lag - phase less marked early time rise slower polynomial relationship exponential onset secondary pathways observation implies difference between high - order polynomial exponential not apparent in experimental data presence noise global analysis system under different conditions required obtain robust mechanistic information paper provided results for time course of nucleated polymerisation for systems initially mixed state monomeric and fibrillar material results generalise classical oosawa theory formation biofilaments to cases arbitrary amount pre - formed seed material present results represent reference polymerisation driven by secondary pathways compared grateful to schiff foundation wellcome leverhulme trusts for financial support cohen vendruscolo dobson knowles nucleated polymerisation with secondary pathways determination of self - consistent solutions to growth processes non - linear master equations devlin terentjev welland dobson analytical solution to kinetics of breakable filament assembly_ 326 15331537 | 0.498738 | 2,744 | 1,406 | , Saving $0.1 in GPT-4. |
177,349 | we posit that accreting compact objects , including stellar mass black holes and neutron stars as well as supermassive black holes , may undergo extended periods of accretion during which the angular momentum of the disk at large scales is misaligned with that of the compact object . in such a scenario , lense - thirring precession caused by the rotating compact object can dramatically affect the disk . in this presentation
we describe results from a three - dimensional relativistic magnetohydrodynamic simulation of an mri turbulent disk accreting onto a tilted rapidly rotating black hole . for this case
, the disk does not achieve the commonly described bardeen - petterson configuration ; rather , it remains nearly planar , undergoing a slow global precession .
accretion from the disk onto the hole occurs predominantly through two opposing plunging streams that start from high latitudes with respect to both the black - hole and disk midplanes .
this is a consequence of the non - sphericity of the gravitational spacetime of the black hole . | recently , our group has been working on three - dimensional , fully relativistic simulations of tilted thick disks , first in the hydrodynamic limit @xcite and here in full general relativistic mhd ( grmhd ) .
a general relativistic treatment ensures that all important relativistic features , such as the cusp in the potential and the einstein and lense - thirring precessions of the orbits , as well as any higher order couplings of these features , are treated properly .
the inclusion of magnetic fields is important because it is now widely believed that local `` viscous '' stresses are generated by turbulence that results from the magnetorotational instability ( mri ) @xcite .
this work is carried out using the cosmos++ astrophysical magnetohydrodynamics code @xcite . for this work ,
the key features of cosmos++ are the flexible mesh structure , adaptive gridding , and three - dimensional general relativistic mhd capabilities .
this simulation is carried out on a spherical - polar mesh with an equivalent resolution of @xmath0 zones . the full set up of this problem and higher
resolution simulations will be described in a forthcoming paper ( fragile et al .
, in preparation ) .
perhaps the most striking feature of the simulated disk at late times are the two opposing plunging streams that start from high latitudes both with respect to the black - hole symmetry plane and the disk midplane .
figure [ fig : plunge ] shows the structure of the full disk and a zoomed in view of the plunging region with the plunging streams .
note that one stream remains entirely above the black - hole symmetry plane , while the other remains below . as material passes through this plunging stream it undergoes strong differential precession .
the precession totals approximately @xmath1 ; thus , the material enters the black hole from the opposite azimuth from which it began its plunge .
orbits at the pressure maximum of the initial torus ) .
this figure is oriented such that the black - hole spin axis points straight up .
the @xmath2-axis of the triad gives the original orientation of the disk angular momentum vector .
the main body of the disk shows very little warping or realignment toward the symmetry plane of the black hole .
the plunging region ( shown in inset ) shows two opposing , high - latitude streams of material connecting the disk to the horizon ( indicated by arrows ) .
[ fig : plunge ] ] the interesting question from figure [ fig : plunge ] is _ why _ do these opposing accretion streams start from such high latitude with respect to the black - hole symmetry plane and disk midplane ? fundamentally this is due to the aspherical nature of the gravitational spacetime of the rotating black hole . a nice way to illustrate
this is to consider the dependence of @xmath3 , the coordinate radius of the innermost stable circular orbit , on inclination .
figure [ fig : isco ] illustrates this dependence for a few selected cases of @xmath4 .
the key point of the plot is that orbital stability around a rotating black hole is strongly dependent on the inclination of the orbit .
notice that the unstable region increases monotonically for increasing inclination .
we can make better use of the information in figure [ fig : isco ] by converting it to a polar plot and overlaying it on a plot of data from the simulation , as is done in figure [ fig : slice ] .
the polar plot creates an `` isco surface '' , which gives a clear indication of where the most unstable regions are .
note that the plunging orbits highlighted previously start near where the disk first encounters the isco surface .
more precisely , the streams start near the largest cylindrical radius ( @xmath5 ) of the isco surface , measured with respect to the angular momentum axis of the disk .
the increase in @xmath3 with inclination then explains why these plunging orbits start at high inclinations relative to the black - hole symmetry plane and the disk midplane , since this is where the gravitational potential is most unstable .
this work was supported under the following nsf programs : partnerships for advanced computational infrastructure , distributed terascale facility ( dtf ) and terascale extensions : enhancements to the extensible terascale facility . | group working on three - dimensional relativistic simulations of tilted thick disks first in hydrodynamic limit @xcite full general relativistic mhd general relativistic treatment ensures important features cusp potential einstein lense - thirring precessions higher order couplings inclusion of magnetic fields important local viscous stresses generated by turbulence from magnetorotational instability @xcite work using cosmos++ astrophysical magnetohydrodynamics code key features flexible mesh structure adaptive gridding three - dimensional general relativistic mhd capabilities simulation on spherical - polar mesh with equivalent resolution of @xmath0 zones full set up higher resolution simulations described in forthcoming paper striking feature simulated disk two opposing plunging streams from high latitudes black - hole symmetry disk midplane figure shows structure full disk zoomed view of plunging region with streams one stream above black - hole symmetry plane other below material passes through stream undergoes strong differential precession approximately @xmath1 material enters black hole from opposite azimuth orbits at pressure maximum initial black - hole spin axis points straight up@xmath2-axis triad gives original orientation disk angular momentum vector main body disk shows little warping realignment toward symmetry plane black hole plunging region shows two opposing high - latitude streams material connecting disk to horizon indicated by arrows fig : plunge question why opposing accretion streams start from high latitude black - hole symmetry plane disk midplane ? due to aspherical nature gravitational spacetime rotating black hole dependence of @xmath3 coordinate radius stable circular orbit on inclination figure isco illustrates dependence orbital stability around rotating black hole dependent on inclination orbit unstable region increases monotonically for increasing inclination information converting to polar plot overlaying on plot data simulation polar plot creates isco surface indication most unstable regions plunging orbits start near disk encounters isco surface streams start near largest cylindrical radius ( @xmath5 ) of isco surface angular momentum axis disk increase in @xmath3 with inclination explains why plunging orbits start at high inclinations relative to black - hole symmetry plane disk midplane where gravitational potential most unstablework supported under nsf programs partnerships advanced computational infrastructure distributed terascale facility terascale extensions facility | 0.188913 | 882 | 430 | , Saving $0.0 in GPT-4. |
177,350 | we examine the hi+@xmath0 content of blue - sequence e / s0s a recently identified population of galaxies that are morphologically early type , but reside alongside spiral galaxies in color vs. stellar mass space .
we test the idea that the majority of low - to - intermediate mass blue - sequence e / s0s may be settled products of past mergers evolving toward later - type morphology via disk regrowth .
we find that blue - sequence e / s0s with stellar masses @xmath1 have atomic gas - to - stellar mass ratios of 0.1 to @xmath21.0 , comparable to those of spiral galaxies .
preliminary co(1 - 0 ) maps reveal disk - like rotation of molecular gas in the inner regions of several of our blue - sequence e / s0s , which suggests that they may have gas disks suitable for stellar disk regrowth . at the current rate of star formation
, many of our blue - sequence e / s0s will exhaust their atomic gas reservoirs in @xmath33 gyr . over the same time period ,
most of these galaxies are capable of substantial growth in the stellar component .
star formation in blue - sequence e / s0s appears to be bursty , and likely involves inflow triggered by minor mergers and/or interactions . | blue - sequence e / s0s represent a unique population of galaxies morphologically classified as early types , but occupying the same blue locus as spirals in color vs. stellar mass space .
while high - mass blue - sequence e / s0s often appear to be young merger / interaction remnants likely to fade to the red sequence , blue - sequence e / s0s with lower stellar masses appear to be much less disturbed .
these galaxies typically occupy low - density field environments where fresh gas infall is possible , and may provide an evolutionary link between traditional early - type galaxies and spirals through disk regrowth @xcite .
we examine the disk growth potential of blue - sequence e / s0s by measuring cold gas reservoirs the raw material for star formation .
our sample includes all 14 blue - sequence e / s0s with stellar masses @xmath4 ( chosen with respect to where blue - sequence e / s0s tail off , figure [ fig.samplemass]a ) from the nearby field galaxy survey ( nfgs , * ? ? ?
we also include 11 red - sequence and two mid - sequence nfgs e / s0s with comparable stellar masses for comparison .
because the nfgs was designed to be a statistically representative sample of the local universe , spanning the natural variety of galaxy morphologies , masses , and environments , these 27 galaxies should correspond to a broad range of evolutionary stages .
we have complete hi data for our sample of e / s0s with @xmath5 .
all of our blue - sequence e / s0s and both mid - sequence e / s0s are detected in hi and have atomic gas masses ranging from @xmath6 to almost @xmath7 .
in contrast , four of our 11 red - sequence e / s0s are not detected .
normalized to stellar mass , the atomic gas masses for 12 of the 14 blue - sequence e / s0s range from 0.1 to @xmath21.0 ( figure [ fig.samplemass]b ) , demonstrating that morphological transformation is likely if the detected gas can be converted into stars .
these gas - to - stellar mass ratios are comparable to those of spiral and irregular galaxies and have a similar dependence on stellar mass .
high - resolution mapping of the molecular gas distribution in blue - sequence e / s0s is underway with the combined array for research in millimeter - wave astronomy ( carma ) .
the molecular gas mass derived from the co(1 - 0 ) maps boosts some of the galaxies further into the region occupied by spirals .
these maps also reveal disk - like rotation of molecular gas in the inner regions of some of our blue - sequence e / s0s ( figure [ fig.carma ] ) , which suggests that they may have gas disks suitable for stellar disk growth .
hi profiles and resolved ionized - gas rotation curves also confirm the presence of disks in many of our blue - sequence e / s0s @xcite .
the timescale that can be directly estimated from the atomic gas mass and star formation rate of a galaxy is the gas exhaustion time the amount of time it would take to convert all the gas into stars , assuming the current star formation rate remains constant .
we find that blue - sequence e / s0s will typically exhaust their gas reservoirs in @xmath33 gyr without fresh gas infall @xcite .
figure [ fig.starformation]a shows that over half ( 9 of 14 ) of our blue - sequence e / s0s can increase their stellar masses by @xmath8@xmath9 over 3 gyr in either of two limiting scenarios constant star formation rate with ongoing infall , or exponentially declining star formation without infall .
this growth is comparable to the amount of fractional stellar mass growth by spiral and irregular galaxies in the same time ( grey vertical lines ) and suggests that most of our blue - sequence e / s0s are capable of significant morphological transformation .
while high - mass blue - sequence e / s0s often show signs of major mergers or interactions , blue - sequence e / s0s with @xmath10 below @xmath11 appear more settled , with a minority experiencing major disturbances .
nonetheless , the similarity of their compact morphologies and relatively hot stellar dynamics to red - sequence e / s0s suggests that many of these galaxies are products of major mergers in the past @xcite .
the long gas exhaustion timescales for blue - sequence e / s0s allow time for minor mergers / interactions with small companions to trigger episodic inflows of gas , resulting in bursts of star formation .
in addition to having bluer inner and outer disk colors than red - sequence e / s0s , half of the blue - sequence e / s0s have centers that are bluer than their outer disks .
figure [ fig.starformation]b shows that all but one of our blue - sequence e / s0s and both mid - sequence e / s0s are on the more blue - centered end of the distribution for spiral / irregular galaxies . for galaxies of all morphologies , blue - centeredness reflects central star formation enhancements and correlates strongly with morphological peculiarities and the presence of nearby companions @xcite .
most of the more blue - centered blue - sequence e / s0s also have enhanced total specific star formation rates @xcite .
these results are broadly consistent with a picture of episodic outer - disk and inner - disk ( pseudobulge ) growth , allowing e / s0s to evolve toward typical spiral galaxy morphology @xcite .
jansen , r. a. , franx , m. , fabricant , d. , & caldwell , n. 2000 , apjs , 126 , 271 kannappan , s. j. , jansen , r. a. , & barton , e. j. 2004 , aj , 127 , 1371 kannappan , s. j. , guie , j. m. , & baker , a. j. 2009 , aj , 138 , 579 wei , l. h. , kannappan , s. j. , vogel , s. n. , & baker , a. j. 2009 , submitted | blue - sequence e / s0s unique population early types same blue locus as spirals high - mass young merger interaction remnants fade red sequence with lower stellar masses less disturbed galaxies occupy low - density field environments fresh gas infall possible evolutionary link between early - type galaxies spirals disk regrowth examine disk growth potential measuring cold gas reservoirs for star formation sample includes 14 blue - sequence e / s0s with stellar masses @xmath4 from field galaxy survey 11 red - sequence two mid - sequence nfgs e / s0s with comparable stellar masses comparison statistically representative sample local universe 27 galaxies correspond broad evolutionary stages complete hi data for sample e / s0s with @xmath5 blue - sequence mid - sequence e detected in hi atomic gas masses ranging from @xmath6 to almost @xmath7 four of 11 red - sequence e / s0s not detectednormalized to stellar mass atomic gas masses for 12 of 14 blue - sequence e / s0s range 0.1 to @xmath21.0 morphological transformation likely if gas converted into stars gas - to - stellar mass ratios comparable to spiral irregular galaxies dependence on stellar mass high - resolution mapping of molecular gas distribution in underway with astronomy molecular gas mass boosts galaxies into spirals maps reveal disk - like rotation of gas in inner regions gas disks for stellar disk growth ionized - gas rotation curves confirm presence of disks in e / s0s timescale estimated from atomic gas mass star formation rate is gas exhaustion time convert gas into stars blue - e s0s exhaust gas reservoirs in @xmath33 gyr without fresh gas infall over half ( 9 of 14 ) - sequence e / s0s can increase stellar masses by @xmath8@xmath9 over 3 gyr in constant star formation rate or declining star formation without infallgrowth comparable to fractional stellar mass growth spiral irregular galaxies suggests blue - sequence e / s0s significant morphological transformation high - mass blue sequence e / s0s show major mergers interactions with below @xmath11 settled minority experiencing major disturbances similarity compact morphologies hot stellar dynamics to red - sequence e / s0s suggests galaxies products of major mergers long gas exhaustion timescales allow minor mergers interactions trigger inflows gas star formation bluer inner outer disk colors half s0s have centers bluer figure shows blue - sequence e mid - sequence e / s0s on more blue - centered end of distribution for spiral irregular galaxies blue - centeredness reflects central star formation enhancements correlates with morphological peculiarities presence nearby companions more blue - centered e / s0s enhanced total specific star formation rates results consistent with episodic outer - inner - disk growth allowing e / s0s evolve toward typical spiral galaxy morphology jansen franx fabricant caldwell2000 126 271 kannappan s j jansen barton e j. 2004 127 1371 kannappan guie j. m. baker a. j. 2009 138 wei l h kannappan vogel s n. baker a. j. 2009 | 0.263232 | 1,319 | 635 | , Saving $0.0 in GPT-4. |
177,355 | surface waves , named here as dyakonov tamm waves , can exist at the planar interface of an isotropic dielectric material and a chiral sculptured thin film ( stf ) . due to the periodic nonhomogeneity of a chiral stf ,
the range of the refractive index of the isotropic material is smaller but the range of the propagation direction in the interface plane is much larger , in comparison to those for the existence of dyakonov waves at the planar interface of an isotropic dielectric material and a columnar thin film . | less than two decades ago , dyakonov @xcite theoretically predicted the propagation of a surface wave at the planar interface of an isotropic dielectric material and a positively uniaxial dielectric material with its optic axis wholly parallel to the interface plane .
if @xmath0 indicates the angle between the optic axis and the direction of surface wave propagation , and @xmath1 is the refractive index of the isotropic dielectric material , then the _ dyakonov wave
_ exists for rather narrow ranges of @xmath0 and @xmath1 .
the consequent significance of dyakonov waves for optical sensing and waveguiding was recognized thereafter @xcite . since then
, the concept of the dyakonov wave has been extended to the planar interfaces of isotropic and biaxial dielectric materials @xcite .
the possibility of the anisotropic material being artificially engineered , either as a photonic crystal with a short period in comparison to the wavelength @xcite or as a columnar thin film ( ctf ) @xcite , has also emerged .
let us note here that the dyakonov wave still remains to be experimentally observed , in part due to the narrow range of @xmath0 for its existence @xcite .
the anisotropic material is taken to be homogeneous in all of the foregoing and other reports on the dyakonov wave .
what if the anisotropic material were to be chosen as periodically nonhomogeneous in a direction normal to the bimaterial interface ?
this question initiated a research project , the first results of which are being communicated here .
being a natural extension of a ctf , a chiral sculptured thin film ( stf ) was chosen as the periodically nonhomogeneous anisotropic material @xcite .
a chiral stf is made by directing a vapor flux in vacuum at an oblique angle onto a rotating substrate .
under suitable conditions , an assembly of parallel nanohelixes of the evaporated species forms , with the helical axes perpendicular to the substrate .
an example of a single nanohelix is illustrated in figure [ fig : chistf ] . by adjusting deposition parameters ,
both the pitch @xmath2 and the angle of inclination @xmath3 $ ] can be controlled .
each nanohelix , composed of multimolecular clusters with @xmath4 nm diameter , is effectively a continuously bent column of @xmath5-nm cross sectional diameter .
therefore , at visible frequencies and lower , a chiral stf may be regarded as a linear , locally orthorhombic , unidirectionally nonhomogeneous continuum whose relative permittivity dyadic is akin to that of chiral smectic liquid crystals @xcite . [
cols="^ " , ] calculations were also made at @xmath6 and @xmath7 nm , with the assumption of the same constitutive and geometric parameters as at @xmath8 nm .
qualitatively similar conclusions on the width of the ranges of @xmath1 and @xmath0 for dyakonov
tamm waves were drawn from the numerical results obtained .
furthermore , the minimum value of @xmath9 was found to increase and the maximum values of @xmath10 $ ] , @xmath11 $ ] , and @xmath12 $ ] were found to decrease , as the ratio @xmath13 increases .
additional calculations , not shown here , were made for an identical chiral stf except with structural left
handedness ( @xmath14 ) .
the exact same results were obtained with the left handed chiral stf as with the right handed chiral stf presented in figures 24 .
it must borne in mind that , by changing @xmath15 from @xmath16 to @xmath17 in ( [ szdyadic ] ) , we invert not only the structural handedness of the chiral stf but also the sense of rotation brought about by @xmath18 .
to conclude , we examined the phenomenon of surface wave propagation at the planar interface of an isotropic dielectric material and a chiral sculptured thin film .
the boundary value problem was formulated by marrying the usual formalism for the dyakonov wave at the planar interface of an isotropic dielectric material and a columnar thin film with the methodology for tamm states in solid state physics .
the solution of the boundary value problem let us deduce the existence of dyakonov
tamm waves . in constitutive terms ,
the major difference between a ctf and a chiral stf is the periodic nonhomogeneity enshrined in @xmath19 ; in the limit @xmath20 , a nanohelix uncurls into a nanorod , and a chiral stf transmutes into a ctf .
however , the distinction between chiral stfs and ctfs is nontrivial , as may be deduced from the floquet
lyapunov theorem @xcite . in comparison to the dyakonov wave localized to the planar interface of an isotropic dielectric material and a ctf @xcite
, we found that the @xmath1range for the existence of a dyakonov
tamm wave at the planar interface of an isotropic dielectric material and a chiral stf is smaller .
however , the @xmath0range is much larger in width : in comparison to @xmath21 with ctfs @xcite and @xmath22 with effectively uniaxial , short
period photonic crystals @xcite the width of the @xmath0range is as high as @xmath23 in figure [ fig : vbarlambda633 ] with chiral stfs .
this implies that dyakonov
tamm waves could be detected much more easily than dyakonov waves .
hodgkinson , q.h .
wu and j. hazel , `` empirical equations for the principal refractive indices and column angle of obliquely deposited films of tantalum oxide , titanium oxide , and zirconium oxide '' appl
. opt . * 37 * , 26532659 ( 1998 ) .
a. lakhtakia and w.s .
weiglhofer , `` further results on light propagation in helicoidal bianisotropic mediums : oblique propagation '' proc . r. soc . lond .
a * 453 * , 93105 ( 1997 ) ; correction : * 454 * , 3275 ( 1998 ) . | two decades ago dyakonov @xcite predicted propagation surface wave at interface isotropic and uniaxial dielectric material optic axis parallel to interface plane @xmath0 angle between optic axis surface wave propagation @xmath1 refractive index of isotropic dielectric material dyakonov wave exists for narrow ranges of @xmath0 and @xmath1 significance of waves for optical sensing waveguiding recognized concept extended to planar interfaces of isotropic biaxial dielectric materials possibility of anisotropic material artificially engineered as photonic crystal short period or columnar thin film emerged dyakonov wave remains experimentally observed due to narrow range of @xmath0 anisotropic material homogeneous in reports on dyakonov wave if periodically nonhomogeneous bimaterial interface ? initiated research project first results communicated chiral sculptured thin film ) chosen as periodically nonhomogeneous anisotropic material @xcite stf made by directing vapor flux in vacuum oblique angle onto rotating substrateconditions assembly parallel nanohelixes evaporated species forms helical axes perpendicular to substrate example single nanohelix illustrated in figure adjusting deposition parameters pitch @xmath2 angle inclination @xmath3 controlled each nanohelix multimolecular clusters @xmath4 nm diameter continuously bent column of @xmath5-nm diameter at visible frequencies chiral stf linear locally orthorhombic nonhomogeneous continuum relative permittivity akin to chiral smectic liquid crystals calculations at @xmath6 and @xmath7 nm same constitutive geometric parameters as @xmath8 nm similar conclusions on width ranges of @xmath1 @xmath0 for dyakonov tamm waves drawn from numerical results minimum value of @xmath9 maximum values of @xmath10 decrease ratio @xmath13 increases additional calculations for identical chiral stf structural left handedness ( @xmath14 ) same results obtained with left handed right handed chiral stf in figures 24 changing @xmath15 from to @xmath17 structural handedness chiral stf sense of rotation @xmath18examined surface wave propagation at planar interface isotropic dielectric material chiral sculptured thin film boundary value problem formulated formalism dyakonov wave with methodology for tamm states in solid state physics solution existence of dyakonov tamm waves major difference between ctf chiral stf is periodic nonhomogeneity in @xmath19 @xmath20 nanohelix uncurls into nanorod chiral stf transmutes into ctf distinction between chiral stfs and ctfs nontrivial from floquet lyapunov theorem comparison to dyakonov wave interface isotropic dielectric material ctf @xmath1range for dyakonov tamm wave smaller @xmath0range larger in width comparison to @xmath21 @xmath22 width @xmath0range as high as @xmath23 in figure with chiral stfs implies dyakonov tamm waves detected easily hodgkinson , q.h . wu j. hazel `` empirical equations for refractive indices column angle of obliquely deposited films of tantalum oxide titanium oxide zirconium oxide ''37 26532659 1998 ) lakhtakia weiglhofer results light propagation helicoidal bianisotropic mediums oblique propagation soc lond 453 93105 1997 ) correction 3275 1998 ) | 0.259669 | 1,399 | 709 | , Saving $0.0 in GPT-4. |
177,364 | some past and ongoing explorations of the spectrum of qcd using monte carlo simulations on a space - time lattice are described .
glueball masses in the pure - gauge theory are reviewed , and the energies of gluonic excitations in the presence of a static quark - antiquark pair are discussed .
current efforts to compute the baryon spectrum using extended three - quark operators are also presented , emphasizing the need to use irreducible representations of the cubic point group to identify spin quantum numbers in the continuum limit . | spectroscopy is a powerful tool for distilling the key degrees of freedom in a given system . at the present time , the best way of extracting the spectrum of states from the qcd lagrangian is monte carlo computer calculations using a space - time lattice .
a spectrum determination requires the extraction of many excited - state energies , so a brief discussion on how excited - state energies can be determined from monte carlo estimates of correlation functions in euclidean field theory is warranted . in this talk ,
several past explorations of the qcd spectrum are outlined , in particular , glueball masses in the pure - gauge theory and the energies of gluonic excitations of the so - called static quark - antiquark potential .
also , ongoing efforts by the lattice hadron physis collaboration ( lhpc ) to determine the baryon spectrum using extended three - quark operators are described .
the monte carlo method can be applied to obtain estimates of the path integrals which yield a hermitian matrix of correlation functions @xmath0 , where @xmath1 creates the states of interest at time @xmath2 and @xmath3 annihilates such states at a later time @xmath4 .
the procedure for extracting the lowest energies @xmath5 from this matrix is well known@xcite .
let @xmath6 denote the eigenvalues of the hermitian matrix @xmath7 , where @xmath8 is some fixed reference time ( typically small ) and the eigenvalues , also known as the _ principal _ correlation functions , are ordered such that @xmath9 as @xmath4 becomes large .
then one can show that @xmath10 determinations of the principal correlators @xmath6 for large temporal separations @xmath4 yield the desired energies @xmath11 . since statistical fluctuations grow with increasing @xmath4 , it is crucial that contributions from higher - lying states be diminished so that the desired lowest - lying energies dominate the principal correlators well before the signal - to - noise ratio falls .
judiciously chosen quark - field and gluon - field smearings is one important ingredient for reducing couplings to the short wavelength modes of the theory .
the use of large sets of extended operators is another key ingredient .
our monte carlo calculations are carried out in a finite - sized box with periodic boundary conditions .
given the finite volume and the discrete nature of the allowed momenta in the box , the masses and widths of resonances ( unstable hadrons ) can not be calculated directly , but must be deduced from the discrete spectrum of finite - volume stationary states for a range of box sizes@xcite .
such applications require prohibitive computational resources , but our goal in the baryon project is to obtain a first exploratory scan of the spectrum , so simply obtaining the finite - volume spectrum for a few judiciously - chosen volumes should suffice for ferreting out the hadron resonances from the less interesting scattering states .
glueball masses in the pure - gauge theory , shown in fig .
[ fig : one ] , were the first exploration of the qcd spectrum@xcite in which i was involved .
the spectrum can be fairly well described by a bag model description of the gluons , whereas string models seem to fare less well .
inclusion of light - quark effects is an ongoing challenge .
the issue of string formation of the gauge - field in the presence of a static quark - antiquark pair was my next exploration@xcite .
for quark - antiquark separations @xmath13 greater than 2 fm , the spectrum of gluonic excitations agreed without exception with that expected from an effective string theory description of the gauge field ( see fig .
[ fig : one ] ) , and a fine structure provided tantalizing clues to the nature of such an effective string theory . a dramatic level reordering was observed as @xmath13 became smaller , suggesting a bag model picture or multipole expansion may be more relevant at such scales .
these energies were used@xcite as a starting point in a born - oppenheimer treatment of heavy - quark mesons , both conventional and hybrid ( bound by an excited gluon field ) .
agreement of level splittings from direct monte carlo calculations with those in the leading born - oppenheimer approximation validated such a treatment and provided a compelling physical picture of both conventional and hybrid heavy - quark mesons .
i am currently a member of the lattice hadron physics collaboration ( lhpc )
. one of our goals is to compute the spectrum of baryon resonances , with an eye towards later determining the meson spectrum .
we have designed large sets of extended , gauge - invariant three - quark operators to facilitate this task@xcite . the usual operator construction which mimics
the approach one would take in continuous space - time is very cumbersome , especially when tackling an entire spectrum .
our approach combines the physical characteristics of baryons with the symmetries of the lattice regularization of qcd used in simulations . for baryons at rest ,
our operators are formed using group - theoretical projections onto the irreducible representations ( irreps ) of the @xmath14 symmetry group of a three - dimensional cubic lattice .
there are four two - dimensional irreps @xmath15 , @xmath16 and two four - dimensional representations @xmath17 and @xmath18 . the continuum - limit spins @xmath19 of our states can be deduced by examining degeneracy patterns across the different @xmath14 irreps .
baryons are expected to be rather large objects , and hence , local operators will not suffice . our approach to constructing extended operators is to use covariant displacements of the quark fields .
displacements in different directions are used to build up the appropriate orbital structure , and displacements of different lengths can build up the needed radial structure .
there are six different spatial orientations that we use , shown in fig [ fig : two ] .
the singly - displaced operators are meant to mock up a diquark - quark coupling , and the doubly - displaced and triply - displaced operators are chosen since they favor the @xmath20-flux and @xmath21-flux configurations , respectively .
we have now finished optimizing the quark - field and gauge - field smearing parameters , and have begun low - statistics runs to prune out the ineffectual and overly - noisy operators .
some sample effective mass plots are shown in fig .
[ fig : three ] .
our goal is to find the smallest set of operators out of the several hundred we have constructed which are useful for extracting some number of lowest - lying states , and to push the technology to maximize the number of excited - state energies which can be reliably determined .
we shall report our findings in the near future .
this work was supported by the u.s . national science foundation through grant phy-0354982 . | spectroscopy freedom system best extracting spectrum from qcd lagrangian monte carlo computer calculations space - time lattice spectrum determination requires excited - state energies discussion monte carlo estimates correlation euclidean field theory past explorations qcd spectrum outlined glueball masses pure - gauge theory energies of gluonic excitations static quark - antiquark potential efforts lattice hadron physis collaboration determine baryon spectrum using extended three - quark operators described monte carlo method estimates path integrals hermitian matrix of correlation functions @xmath0 @xmath1 creates states interest annihilates states procedure extracting lowest energies @xmath5 known @xmath6 eigenvalues hermitian matrix @xmath7 @xmath8 fixed reference time eigenvalues principal correlation functions ordered @xmath9 @xmath4 large @xmath10 determinations of principal correlators @xmath6 for large temporal separations yield desired energies statistical fluctuations grow @xmath4 contributions from higher - lying states desired lowest - lying energies dominate principal correlators before signal - to - noise ratio fallschosen quark gluon - field smearings reducing couplings to short wavelength modes large sets extended operators key monte carlo calculations in finite - sized box with periodic boundary conditions finite volume discrete momenta masses widths of resonances calculated directly deduced from spectrum finite - volume stationary states for box sizes applications require computational resources goal baryon project first exploratory scan spectrum obtaining finite - volume spectrum for volumes for ferreting hadron resonances from less scattering states glueball masses in pure - gauge theory shown in fig . one first exploration of qcd spectrum@xcite spectrum described by bag model description gluons string models less well inclusion of light - quark effects challenge issue string formation gauge static quark - antiquark pair next exploration for quark - antiquark separations @xmath13 greater than 2 fm spectrum gluonic excitations agreed with effective string theory description fig fine structure clues dramatic level reordering observed as @xmath13 smaller bag model picture multipole expansion relevant scalesenergies used starting point born - oppenheimer treatment of heavy - quark mesons conventional and hybrid bound gluon field agreement of level splittings from monte carlo calculations born - oppenheimer approximation validated treatment provided compelling physical picture of hybrid mesons member of lattice hadron physics collaboration compute baryon resonances determining meson spectrum designed extended gauge - invariant three - quark operators usual operator construction cumbersome spectrum approach combines physical characteristics baryons with symmetries lattice regularization of qcd baryons at rest operators formed using group - theoretical projections irreducible representations @xmath14 symmetry group three - dimensional cubic lattice four two - dimensional irreps @xmath15 two four - dimensional @xmath17 continuum - limit spins @xmath19 deduced by examining degeneracy patterns across @xmath14 irreps baryons large objects local operators suffice approach extended operators covariant displacements of quark fields displacements different directions build orbital structure different lengths radial structure six spatial orientations shown in figsingly - displaced operators mock diquark - quark coupling doubly triply - displaced favor @xmath20-flux @xmath21-flux configurations finished optimizing quark - field gauge - field smearing parameters begun low - statistics runs prune ineffectual noisy operators effective mass plots shown in fig . goal find smallest set operators useful for extracting lowest - lying states push technology maximize excited - state energies determined report findings work supported by u.s . national science foundation grant phy-0354982 | 0.264214 | 1,444 | 716 | , Saving $0.0 in GPT-4. |
177,366 | we study the nonequlibrium transport through a quantum dot weakly coupled to luttinger liquids ( ll ) .
a general current expression is derived by using nonequilibrium green function method .
then a special case of the dot with only a single energy level is discussed . as a function of the dot s energy level , we find that the current as well as differential conductance is strongly renormalized by the interaction in the ll leads . in comparison with the system with fermi liquid ( fl ) leads , the current is suppressed , consistent with the suppression of the electron tunneling density of states of the ll ; and the outset of the resonant tunneling is shifted to higher bias voltages . besides , the linear conductance obtained by furusaki using master equation can be reproduced from our result .
pacs numbers:73.63.-b,71.10.pm , 73.40.gk , 73.23.-b , tunneling phenomena in mesoscopic systems have been intensively studied for years both experimentally and theoretically . since early 1990s , the tremendous progress in nanofabrication and the discovery of novel one - dimensional ( 1d ) materials , especially the carbon nanotubes,@xcite have led to a revived interest in studying the tunneling in systems including 1d
conductors.@xcite@xcite it is well - known that at low temperatures the two - dimensional ( 2d ) and three - dimensional ( 3d ) interacting electron systems can be described by landau s fermi liquid ( fl ) theory .
however , the interacting electron systems in one dimension ( 1d ) is quite different , it behaves as luttinger liquid ( ll),@xcite@xcite@xcite for that theory predicts a series of unique characters , e.g. , the absence of landau quasiparticles , spin - charge separation , suppression of the electron tunneling density of states ( dos ) , anomalous power laws for transport coefficients with interaction - dependent exponents , etc.@xcite@xcite naturally , one may ask whether the ( resonant ) tunneling between lls is different from that of fls .
recently , furusaki reported a theoretical study of the resonant tunneling through a quantum dot ( qd ) weakly coupled to two ll leads ( hereafter as ll - qd - ll).@xcite assuming that the tunneling is an incoherent sequential tunneling , the author derived the linear conductance using the master equation approach .
similar studies have been done by other authors earlier.@xcite@xcite@xcite in this paper , we present an investigation for a ll - qd - ll system .
differently , we consider the tunneling as a coherent process and study the nonequilibrium current through the system with a finite bias voltage applied on the two ll leads . by using nonequilibrium green function technique,@xcite a general current formula
is derived .
we find that both the current and differential conductance are strongly renormalized by the interaction in the ll leads . in current - voltage characteristics ,
the current is first suppressed , consistent with the suppression of the electron tunneling density of states ; and the outset of the resonant tunneling is found to be shifted to higher bias voltages , comparing to the system fl - qd - fl . finally , the linear conductance obtained by furusaki using master equation can be reproduced from our result .
the hamiltonian of the system can be split into three parts : @xmath0 , where @xmath1 denotes the left and right ll leads in its standard form ( see @xcite ) , @xmath2 is the hamiltonian of the qd , with @xmath3 the creation / annihilation operators of the n - th energy level in the dot ( for simplicity , the intradot coulomb interaction has been neglected ) .
@xmath4 is the tunneling hamiltonian and can be written as @xmath5 in which @xmath6 are the fermi operators at the end points of the left / right lead ; @xmath7 is the tunneling constant .
the spin indices have been suppressed since they are not important in the following discussion .
defining the mixed casual green function @xmath8 and green functions of all the other types in the standard form,@xcite the current from left ll lead into the qd under a bias voltage @xmath9 applied on the right lead can be evaluated as follows ( in units of @xmath10 ) : @xmath11 , \ ] ] where @xmath12 is the dyson equation for contour green function @xmath13 , @xmath14 is the full green function of the qd , @xmath15 is the local green function at the end point of the left ll lead without the coupling to the qd , and @xmath16 is the contour - ordering operator . after using langreth theorem of analytic continuation , and the fourier transformation , the current can then be expressed as : @xmath17 .\ ] ] the dyson equation for @xmath18 is @xmath19 , where @xmath20 is the irreducible self - energy and @xmath21 is the free green function of the dot without the coupling between the qd and leads .
using langreth theorem and noticing @xmath22 , the current expression , eq.(3 ) , can be reduced to @xmath23 ^{-1}\left [ g_l^{>}\sigma ^{<}-g_l^{<}\sigma ^{>}\right ] \left [ \left ( g^a\right ) ^{-1}-\sigma ^a\right ] ^{-1}\right\ } _ { n , m}t_{m , l}.\ ] ] neglecting the off - diagonal tunneling matrix elements and their energy - dependence , which is equivalent to take @xmath24,@xcite then eq.(3 ) reduces to @xmath25 which describes the nonequilibrium current in ll - qd - ll , and is the the central result of this work .
similarly , one can easily obtain the current from right ll lead to the qd , @xmath26 , just by changing the subscript in eq.(5 ) @xmath27 , and easily to see @xmath28 , the current conservation , as expected . as an important example ,
let us consider the quantum dot with only a single energy level , i.e. , @xmath29 .
after taking a contour integral , eq.(5 ) reduces to @xmath30 in which @xmath31 and @xmath32 are the lesser ( greater ) green functions of the left and right leads without the couplings to the dot , and can be directly quoted from:@xcite @xmath33 where @xmath34 is the temperature , and @xmath35 the bias voltage .
@xmath36 is defined as @xmath37 , here @xmath38 is the gamma function , @xmath39 are interaction parameters characterizing the left / right ll liquids , which should not be confused with the green functions with the same symbols , @xmath40 describes the effective level broadening of the dot , proportional to @xmath41 , and @xmath42 is the high - energy cutoff or a band width.@xcite in the lowest order approximation , eq.(6 ) reduces to @xmath43 eq.(8 ) describes the nonequilibrium current of ll - qd - ll under a finite bias voltage . in the following , we present the numerical studies based on eq.(8 ) for the symmetric case , i.e. , @xmath44 , @xmath45 and @xmath46 .
fig.1 shows the current vs bias voltage at a fixed value @xmath47 for different interaction parameters of the ll leads , @xmath48 .
the system with fl leads is also shown for comparison ( see the curve with @xmath49 in fig.1 ) .
we find the following features : ( 1 ) for fl - qd - fl , a steep increase occurs around @xmath50 , which is consistent with the suppression of the electron tunneling of states of the ll .
the suppression becomes stronger with the decrease of @xmath51 ( or the increase of the interactions in the leads ) .
( 2 ) the outset of the resonant tunneling varies with the change of @xmath51 , from @xmath52 for @xmath49 ( i.e. , the fl leads ) to @xmath53 for @xmath54 ( i.e. , the strongest interacting ll leads ) .
a current plateau appears again for @xmath55 around @xmath53 .
( 3 ) somehow , the curves with @xmath56 intersect approximately at a point ( @xmath57 , @xmath58 ) , which may indicate the peculiar behavior for strong interacting case .
the second feature above mentioned can also be seen from fig.2 , in which the differential conductance vs @xmath59 for different interaction parameters of ll , @xmath60 , is presented .
the location of the peak of conductance is shifted from @xmath61 to @xmath62 with the increase of @xmath51 .
the nonequilibrium current as a function of @xmath59 at a fixed bias is shown in fig.3 . comparing to the current of fl - qd - fl
, these curves are strongly renormalized by the interactions of the ll leads : instead a mesa for fl - qd - fl , a peak occurs at @xmath63 ; and the half - width of the peak becomes narrower with the increase of the interactions of electrons in the ll leads .
the dependence of the peak current on @xmath34 and @xmath64 can be obtained by estimating the current @xmath65 at @xmath63 . because the asymptotic limit of the gamma function is @xmath38 @xmath66 @xmath67,@xcite one finds that at low temperatures @xmath68 the current behaves as @xmath69 @xmath70 @xmath71 , whereas at high temperatures @xmath72 @xmath69 @xmath70 @xmath73 .
similar scaling forms have been derived by chamon and wen by considering the incoherent sequential tunneling.@xcite it should be pointed out that our calculation is valid only when the higher order contributions ( @xmath74 @xmath75 ) can be neglected .
the pole of the integrand in eq.(4 ) satisfies the equation @xmath76 .
neglecting the shift of the energy level of the qd , the lowest order solution is @xmath77 .
thus , the higher order correction can be neglected only if @xmath78 , or equivalently , @xmath79 . using the asymptotic limit of the gamma function ,
one finds that at low temperatures @xmath80 the condition reduces to @xmath81 and @xmath82 ; whereas at high temperatures @xmath83 @xmath84 . therefore , our evaluations validate only when @xmath85 and @xmath86.@xcite finally , as a quick check , we can obtain the linear conductance @xmath87 from eq.(7 ) by simply taking @xmath88 , which gives : @xmath89 } \frac{\gamma _ l\left ( \varepsilon \right ) \gamma _ r\left ( \varepsilon \right ) } { \gamma _ l\left ( \varepsilon \right ) + \gamma _ r\left ( \varepsilon \right ) } .\ ] ] this is exactly the linear conductance obtained by furusaki,@xcite and was confirmed by experiment recently.@xmath90 in conclusion , we derive a general formula for nonequilibrium dc current through a quantum dot weakly coupled to two luttinger liquids .
we consider the tunneling as a coherent process and use nonequilibrium green function technique , different from chamon and wen and furusaki . in symmetric case and for the dot with a single energy level ,
we find that both the current and the differential conductance are strongly renormalized by electron - electron interactions in the ll leads .
we hope that further experiments may provide evidence for these theoretical predictions .
the authors would like to thank guang - shan tian for many helpful discussions .
this work was supported by nsfc under grant no.10074001 .
one of the authors ( t.h.lin ) would also like to thank the support from the visiting scholar foundation of the state key laboratory for mesoscopic physics in peking university .
@xmath91 to whom correspondence should be addressed .
@xmath92 s. iijima , nature * 354 * , 56 ( 1991 ) ; a. thess _
et al_. , science * 273 * , 483 ( 1996 ) .
r. egger _
et al .
_ , cond - mat/0008008 .
a. f. morpurgo _
et al . _ , science * 286 * , 263 ( 1999 ) .
s. tomonaga , prog .
theor .
phys .
( kyoto ) * 5 * , 544 ( 1950 ) .
j. m. luttinger , j. math . phys .
* 4 * , 1154 ( 1963 ) .
f. d. m. haldane , prl * 47 * , 1840 ( 1981 ) . * * j. voit , rep .
. phys . * 57 * , 977 ( 1994 ) ; cond - mat/0005114 . s. rao and d.
sen , cond - mat/0005492 .
a. furusaki , phys . rev . * b 57 * , 7141 ( 1998-ii ) . c. l. kane and m. p. a. fisher , phys .
rev . *
b 46 * , 15 233 ( 1992 ) .
a. furusaki and n. nagaosa , phys . rev .
* b 47 * , 3827 ( 1993 ) . c. de c. chamon and x. g. wen , phys .
rev .
lett . * 70 * , 2605 ( 1993 ) .
h. haug and a. -p .
jauho , _ quantum kinetics in transport and optics of semiconductors _
( springer - verlag , berlin , 1998 ) .
y. meir et al . , phys .
rev .
lett . * 66 * , 3048 ( 1991 ) ; l. wang _
et al . _ ,
phys .
rev .
lett .
, * 73 * , 585 ( 1994 ) . i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series , and products _
( academic press , inc . , new york , 1980 ) .
these conditions are different from that of chamon and wen,@xcite where the current formula stands down to zero temperature for interacting parameter @xmath93 . in our calculation , the imaginary part of the self - energy has been neglected , which will play important role in the tunneling process when @xmath94 .
o. m. auslaender _
et al . _ ,
phys .
rev . lett . * 84 * , 1764 ( 2000 ) . | * current , scaled by its maximum at @xmath95 , vs bias voltage for @xmath96 .
@xmath97 is the energy level of the dot , @xmath51 is the interaction parameters of the leads .
curves for @xmath98 intersect approximately at a point @xmath99 .
the case with fl leads @xmath100 is also shown for comparison .. * differential conductance scaled by its maximum at @xmath101 as a function of @xmath59 for @xmath102 .
the location of the maximum is shifted away from @xmath103 for @xmath49 to @xmath63 for @xmath104 . *
current vs @xmath59 for @xmath105 the current is scaled by its maximum at bias voltage @xmath106 for @xmath107(luttinger liquid ) , a peak occurs at @xmath108 , and the half width of the peak becomes narrower with the decrease of @xmath109 . the curve with @xmath49 ( fermi liquid ) is also shown for comparison . | current scaled maximum @xmath95 vs bias voltage @xmath96 @xmath97 energy level dot @xmath51 interaction parameters leads curves @xmath98 intersect @xmath99 case leads @xmath100 shown comparison differential conductance scaled maximum @xmath101 function @xmath59 @xmath102 location maximum shifted from @xmath103 to @xmath63 current vs @xmath59 @xmath105 scaled maximum bias voltage @xmath107 liquid peak @xmath108 half width narrower decrease @xmath109 curve @xmath49 fermi liquid shown comparison | 0.068893 | 220 | 119 | , Saving $0.0 in GPT-4. |
177,373 | we have fabricated high - quality fese@xmath0 superconducting films with a bulk @xmath1 of @xmath2 k on different substrates , al@xmath3o@xmath4(0001 ) , srtio@xmath4(100 ) , mgo(100 ) , and laalo@xmath4(100 ) , by using a pulsed laser deposition technique .
all the films were grown at a high substrate temperature of 610 @xmath5c , and were preferentially oriented along the ( 101 ) direction , the latter being to be a key to fabricating of fese@xmath0 superconducting thin films with high @xmath1 . according to the energy dispersive spectroscopy data , the fe : se composition ratio was 1:0.90@xmath60.02 .
the fese@xmath0 film grown on a srtio@xmath4 substrate showed the best quality with a high upper critical magnetic field [ @xmath7(0 ) ] of 56 t. | a new iron age has occurred in the field of superconductivity and started when the discovery of the lafeaso@xmath0f@xmath8 superconductor in 2008 gave rise to a huge interest in depressed superconductivity division @xcite . over the past two years
, an enormous number of papers have been published and more than 70 new iron - based superconductors have been discovered @xcite . generally , iron - based superconductors are classified as refeaso ( 1111 family , re : rare earths ) , aefe@xmath3as@xmath3 ( 122 family , ae : alkaline earths ) , afeas ( 111 family , a : li , na ) , and fech ( 11 family , ch : chalcogens ) . among these families , a tetragonal phase fese is an attractive material for finding the origin of iron - based superconductors due to its simple crystal structure @xcite .
it has just superconducting layers without block layers , and its superconducting critical temperature ( @xmath1 ) is very sensitive to change in the strain .
although the @xmath1 of bulk fese is as low as 8.5 k , its @xmath1 can be raised to 37 k under pressure @xcite .
in addition , the @xmath1 of fese@xmath9te@xmath9 thin films can be enhanced to 21 k by controlling the biaxial strain @xcite .
however , compared with te- or s - doped superconductors @xcite , research on pure fese superconductors is rarely carried out .
synthesis of single - phase fese superconductors with a tetragonal structure is not easy because it has several types of phases , such as tetragonal ( pbo - type ) , hexagonal ( nias - type ) , and fe@xmath10se@xmath11 phases @xcite .
moreover , most of them , except for the tetragonal phase , are known to be ferromagnetic phases @xcite . according to recent reports @xcite , the possibility of ferromagnetism was presented in fese thin films with a tetragonal structure at temperature near room temperature , but this issue is still under debate .
nevertheless , the observation of superconductivity in the tetragonal phase of fese by hsu @xmath12 @xcite has offered a new member ( 11 family ) to the class of iron - based superconductors .
also , an expectation of a higher @xmath1 in fete than in fese , which is caused by strong paring @xcite , has brought about active research on the fe - te - se system @xcite .
if the intrinsic properties are to be investigated precisely , undoped fese or fete single crystals with large sizes or thin films are required because isovalent substitution induces lattice strain due to the different ionic sizes .
very recently , han @xmath12 @xcite and si @xmath12 @xcite reported the appearance of superconductivity in fete films due tensile stress effects and oxygen incorporation , respectively , while no superconductivity was observed in bulk samples @xcite . in the case of fese films ,
only a few papers have been published @xcite , and the experimental results are still unsatisfactory compared to those for bulk samples .
furthermore , there has been little progress in the growth of pure fese films whereas some research groups have fabricated high - quality te - doped fese thin films @xcite . in this work , we report the fabrication of fese@xmath0 films with a bulk @xmath1 of @xmath2 k on several types of substrates , al@xmath3o@xmath4(0001 ) [ ao ] , srtio@xmath4(100 ) [ sto ] , mgo(100 ) , and laalo@xmath4(100 ) [ lao ] , by using a pulsed laser deposition ( pld ) technique .
all samples were grown at a high substrate temperature of 610 @xmath5c and showed high upper critical fields with a ( 101 ) preferred orientation .
homemade fese targets having a stoichiometric ratio of fe : se = 1:1 were used in this study , and the laser beam was generated by using a lambda physik krf excimer laser ( @xmath13 = 248 nm ) .
our pld system is described in detail elsewhere @xcite . a laser energy density of 1.15 j/@xmath14
, a high repetition rate of 48 hz , and high substrate temperature ( @xmath15 ) of 610 @xmath5c were used for the deposition of fese@xmath0 films in a high vacuum state of @xmath16 10@xmath17 torr .
the high frequency of the laser is favorable for the fabrication of films containing highly - volatile materials @xcite .
high growth rates of @xmath18 nm / s were obtained under these growth conditions .
the compositional ratio of the deposited films was fe : se = 1:0.90@xmath60.02 .
the crystal phases and orientations of the fabricated fese@xmath0 films were investigated by using x - ray diffraction ( xrd ) .
the surface morphology and thickness were checked by using scanning electron microscopy ( sem ) .
the elemental composition ratio was analyzed by using energy dispersive spectroscopy ( eds ) based on the results for a standard stoichiometric sample . a vibrating sample measurement system ( squid - vsm , quantum design ) and a physical property measurement system ( ppms , quantum design ) were used for the study of their superconducting properties .
the temperature dependences of the resistivity ( @xmath19 ) of the fese@xmath0 films are shown in figure 1 .
a se deficiency of @xmath20 = 0.10@xmath60.02 , which was comparable with previously reported samples @xcite , was examined by using eds .
all of the deposited films showed a @xmath21 of @xmath2 k and a @xmath22 of @xmath23 k , and their transition widths ( @xmath24@xmath1 ) decreased with increasing residual resistivity ratio ( rrr ) ; these data are summarized in table 1 .
all these data are very close to the previous results for bulk samples @xcite and show a higher @xmath22 and a narrower @xmath24@xmath1 than previously reported @xcite . in figure 1(a ) , the @xmath19(@xmath25 ) of the film grown on a sto substrate is multiplied by 5 because it has small values compared with those of the other film .
the change of slope around 120 k seems to originate from the influence of a spin - density wave ( sdw ) @xcite .
films grown on several types of substrates . and
( b ) an enlarged view near the superconducting transition temperature , with the resistivity values normalized to those at 15 k. the inset of ( b ) shows the field - cooled ( fc ) and the zero - field - cooled ( zfc ) measurements for an fese@xmath0 film grown on a srtio@xmath26 substrate at @xmath27 = 20 oe . ]
figure 1(b ) is a magnified view near the superconducting transition temperature ; the resistivity values are normalized to those at 15 k. the inset shows the temperature dependence of the magnetization ( @xmath28 ) of the fese@xmath0 film fabricated on a sto substrate .
the clear diamagnetic signal indicates that it has good bulk superconductivity .
although the irreversible temperature of @xmath16 6 k in the @xmath29 curve is lower than the @xmath1 obtained from @xmath19(@xmath25 ) , obtaining the superconducting signal from magnetization measurements in fese superconductors is not easy due to its intrinsic magnetism @xcite or its magnetic impurities , such as hexagonal fese phases and fe impurities @xcite .
@*6l substrate&@xmath21(k)&@xmath22(k)&@xmath24@xmath1(k)&rrr&fe : seal@xmath3o@xmath4&10.5&8.2&2.3&4.5&1:0.91srtio@xmath4&11.4&8.3&3.1&4.2&1:0.92mgo&11.7&8.1&3.6&3.4&1:0.88laalo@xmath4&11.2&7.7&3.5&3.3&1:0.91 scan xrd patterns of the fese@xmath0 films grown on al@xmath30o@xmath26(0001 ) and srtio@xmath26(100 ) substrates .
the preferred orientation is seen to be along the ( 101 ) direction ( t : tetragonal , h : hexagonal ) . ]
films grown on several different substrates : ( a ) al@xmath30o@xmath26 , ( b ) srtio@xmath26 , ( c ) mgo , and ( d ) laalo@xmath26 .
all surface morphologies show well - linked states without cracks . ]
figure 2 shows the @xmath31 scan xrd patterns of fese@xmath0 films prepared on the al@xmath3o@xmath4(0001 ) and srtio@xmath4(100 ) substrates .
our samples are mainly oriented along the ( 101 ) direction of the tetragonal ( t ) structure , but other phases exist in tiny amounts , such as hexagonal ( h ) fese and fe@xmath10se@xmath11 .
the lattice constants for the @xmath32- and @xmath33-axis are 3.769 and 5.490 , respectively , which are close to the values observed in the bulk samples @xcite .
no peak shift is detected regardless of the substrate .
we believe that this is due to the growth direction of the film being along the ( 101 ) direction rather than along the @xmath33-axis or the @xmath34-axis .
wang @xmath12 @xcite reported that the orientation of the t(101 ) direction is the key to fabricating of fese@xmath0 films with strong superconductivity .
the surface morphologies of the fese@xmath0 films on the ao , sto , mgo , and lao substrates were checked by using scanning electron microscopy ( sem ) , as shown in figure 3 .
the sem images , ( a ) , ( b ) , ( c ) , and ( d ) , show that compared with previous reports @xcite , all films have well - connected morphologies without cracks , which is one factor in obtaining higher - quality samples with a sharper superconducting transition .
the white tiny grains in figure 3(b ) were shown to be fe particles by using eds .
films grown on lao had larger fe lumps than ones grown on sto .
these fe lumps , which were ferromagnetic , are a possible source of the proximity effect that accompanies a suppression of @xmath1 @xcite .
these locally weakened superconducting regions generate a broad superconducting transition .
the thicknesses determined from the sem cross - sectional images in ( a ) , ( b ) , ( c ) , and ( d ) are 1.5 , 1.6 , 1.6 , and 1.8 @xmath35 m , respectively .
( @xmath25 ) curves for fese@xmath0 films on srtio@xmath26 substrates under magnetic fields up to 7 t for a field parallel to the substrate plane .
the inset presents the linear fittings for @xmath7(@xmath25 ) at @xmath36 , @xmath37 , and @xmath38 . ] finally , we investigated the temperature dependence of the upper critical magnetic field ( @xmath7(@xmath25 ) ) for the fese@xmath0 films .
figure 4 displays the @xmath39 curves of the film prepared on a sto substrate for various magnetic fields from 0 to 7 t , and the inset shows the fitting of @xmath7(@xmath25 ) .
the data points were obtained from the @xmath39 curves for three defined criteria ; @xmath38 ( 90% of @xmath40 ) , @xmath37 ( 50% of @xmath40 ) , and @xmath36 ( 10% of @xmath40 ) are equivalent to critical temperatures of @xmath21 , @xmath41 , and @xmath22 , respectively .
the values of @xmath7(0 ) at @xmath38 are 58 , 56 , 50 , and 57 t for films grown on ao , sto , mgo , and lao substrates , respectively , which were estimated by using a linear extrapolation @xcite .
these high values of @xmath7(0 ) open up the possibility for practical applications , provided we can grow better films with higher values of @xmath1 .
in summary , we have deposited fese@xmath0 films with bulk superconducting properties on various types of substrates , al@xmath3o@xmath4(0001 ) , srtio@xmath4(100 ) , mgo(100 ) , and laalo@xmath4(100 ) , by using a pulsed laser deposition system .
all films were preferentially oriented along the ( 101 ) direction , and showed upper critical fields as high as 50 t. our results showed that ( 101)-oriented fese@xmath0 films with bulk superconductivity could be fabricated at high growth temperature regardless of type of substrate . | new iron age in superconductivity started discovery of lafeaso@xmath0f@xmath8 superconductor in 2008 interest in depressed superconductivity division past two years papers published 70 new iron - based superconductors discovered iron - based superconductors classified as refeaso rare earths aefe@xmath3as@xmath3 alkaline earths afeas 111 fech ( 11 chalcogens tetragonal phase fese attractive material for iron - based superconductors simple crystal structure superconducting layers without block layers superconducting critical temperature sensitive to change strain @xmath1 of bulk fese low 8.5 k can be raised to 37 k under pressure @xmath1 of fese@xmath9te@xmath9 thin films can be enhanced to 21 k by controlling biaxial strain research on pure fese superconductors rarely carried synthesis of single - phase fese superconductors with tetragonal structure not easy several types of phases tetragonal hexagonal fe@xmath10se@xmath11 phases most known be ferromagnetic phasesrecent reports possibility of ferromagnetism in fese thin films tetragonal structure at near room temperature under debate observation of superconductivity in tetragonal phase fese by hsu @xmath12 new member to iron - based superconductors expectation of higher @xmath1 in fete than fese strong paring active research on fe - te - se system undoped fese or fete single crystals with large sizes thin films required isovalent substitution induces lattice strain different ionic sizes han si reported superconductivity in fete films due tensile stress effects oxygen incorporation no superconductivity in bulk samples fese films few papers published experimental results unsatisfactory compared to bulk samples little progress in growth of pure fese films some groups fabricated high - quality te - doped fese thin films fabrication of fese@xmath0 films with bulk @xmath1 of @xmath2 k on substrates pulsed laser deposition technique samples grown at high substrate temperature 610 @xmath5c showed high upper critical fields ( 101 ) preferred orientationfese targets ratio fe : se = 1:1 used study laser beam generated lambda physik krf excimer laser @xmath13 = 248 nm ) pld system described laser energy density 1.15 j/@xmath14 high repetition rate 48 hz high substrate temperature 610 @xmath5c used for deposition fese@xmath0 films high vacuum state 10 torr high frequency laser favorable fabrication films highly - volatile materials high growth rates @xmath18 nm / s obtained compositional ratio films fe : se = 1:0.90@xmath60.02 crystal phases orientations films investigated x - ray diffraction surface thickness checked scanning electron microscopy elemental composition ratio analyzed energy dispersive spectroscopy vibrating sample measurement system physical property measurement system used superconducting properties temperature dependences resistivity fese films shown in figure 1 se deficiency @xmath20 = 0.10@xmath60.02 comparable samples examined eds deposited films showed @xmath21 k @xmath22 k transition widths decreased increasing residual resistivity ratio data summarized in table 1 close to previous results bulk samples higher @xmath22 narrower @xmath24@xmath1 figure 1(a @xmath19( ) film on sto substrate multiplied by 5 small values change slope around 120 k from spin - density wave films on substrates enlarged view near superconducting transition temperature resistivity values normalized at 15 k field - cooled zero - - cooled measurements fese@xmath0 film srtio@xmath26 substrate at @xmath27 = 20 oe figure 1(b magnified view near superconducting transition temperature resistivity values normalized to at 15 k temperature dependence magnetization @xmath28 ) fese@xmath0 film sto substrate clear diamagnetic signal indicates good bulk superconductivity irreversible temperature of @xmath16 6 k @xmath29 curve lower than @xmath1 @xmath19( ) obtaining superconducting signal from magnetization fese superconductors not easy due intrinsic magnetism magnetic impuritiessubstrate@xmath21@xmath3o@xmath4&10.5&8.2&2.3&4.5&1:0.91srtio@xmath4&11.4&8.3&3.1&4.2&1:0.92mgo&11.7&8.1&3.6&3.4&1:0.88laalo@xmath4&11.2&7.7&3.5&3.3&1:0.91 patterns fese@xmath0 films al@xmath30o@xmath26(0001 srtio@xmath26(100 ) substrates preferred ( 101 ) direction t tetragonal h hexagonal films substrates al@xmath30o@xmath26 srtio@xmath26 mgo laalo@xmath26 surface morphologies well - linked states without cracks figure 2 shows @xmath31 scan xrd patterns fese@xmath0 films al@xmath3o@xmath4(0001 srtio@xmath4(100 ) substrates samples oriented ( 101 ) direction tetragonal other phases hexagonal lattice constants @xmath32- @xmath33-axis 3.769 5.490 close values bulk samples no peak shift due growth direction ( 101 ) direction@xmath12 @xcite reported t(101 ) direction key to fese@xmath0 films strong superconductivity surface morphologies on ao sto mgo lao substrates checked scanning electron microscopy figure 3 sem images show all films well - connected morphologies without cracks higher - quality samples sharper superconducting transition white tiny grains in figure 3(b ) fe particles films lao had larger fe lumps sto ferromagnetic possible source proximity effect suppression weakened superconducting regions generate broad superconducting transition thicknesses sem cross - sectional images 1.5 1.6 1.8 @xmath35 m @xmath25 ) curves for fese@xmath0 films on srtio@xmath26 substrates under magnetic fields up to 7 t parallel substrate plane presents linear fittings for @xmath7(@xmath25 ) @xmath36 @xmath37 @xmath38 investigated temperature dependence of upper critical magnetic field @xmath7(@xmath25 ) for fese@xmath0 filmsfigure 4 displays @xmath39 curves sto substrate for magnetic fields 0 to 7 t shows fitting @xmath7( ) data points obtained from @xmath39 curves for three criteria @xmath38 90% 50% 10% equivalent critical temperatures @xmath21 @xmath22 values @xmath7(0 ) @xmath38 58 56 50 57 t for films ao sto mgo lao substrates estimated linear extrapolation @xcite high values @xmath7(0 ) open practical applications better films higher values @xmath1 deposited fese@xmath0 films bulk superconducting properties on substrates mgo laalo pulsed laser deposition system all films oriented ( 101 ) direction showed upper critical fields 50 t results showed ( 101)-oriented fese@xmath0 films bulk superconductivity fabricated at high growth temperature regardless substrate | 0.437644 | 2,999 | 1,584 | , Saving $0.1 in GPT-4. |
177,377 | experimental studies of power corrections with @xmath0 data are reviewed .
an overview of the available data for jet and event shape observables is given and recent analyses based on the dokshitzer - marchesini - webber ( dmw ) model of power corrections are summarised .
the studies involve both distributions of the observables and their mean values .
the agreement between perturbative qcd combined with dmw power corrections and the data is generally good , and the few exceptions are discussed . the use of low energy data sets highlights deficiencies in the existing calculations for some observables .
a study of the finiteness of the physical strong coupling at low energies using hadronic @xmath1 decays is shown . | data from hadron production in @xmath0 annihilation experiments has been one of the main sources of experimental information for the study of non - perturbative qcd of hadron jets , see e.g. @xcite .
compared to other processes where power corrections have been studied such as hadron production in deep inelastic lepton - nucleon scattering @xcite there are two main advantages : leptonic initial state : : the incoming electron and positron are leptons and thus there is no interference between initial state and the hadronic final state .
this makes the interpretation of of the final state in terms of qcd much more robust .
large range of energies : : the currently available data from @xmath0 annihilation to hadrons span a range of centre - of - mass ( cms ) energies of more than an order of magnitude . since many effects in qcd depend on the energy scale comparing data sets recorded at different cms energies yields strong tests of the theory . from reconstructed hadronic final states jet production or event shape observables
are calculated , see e.q . @xcite . as an example the observable thrust @xmath2
is defined by @xmath3 where @xmath4 is the 3-momentum of particle @xmath5 and @xmath6 .
a value of @xmath7 corresponds to an ideal 2-jet event , events with three final state objects have @xmath8 while completely spherical events with many final state particles can approach @xmath9 .
event shape observables @xmath10 are conventionally defined to have @xmath11 for ideal two - jet configurations and thus the thrust is often analysed in terms of @xmath12 .
the observables can be classified as 3-jet or 4-jet observables depending on whether @xmath13 occurs in the cms for final states with @xmath14 or @xmath15 objects .
the energy of the electron and electron beams in most @xmath0annihilation collider experiments is equal due to the simplified construction and operation of a circular accelerator with only one beam line . as a consequence the cms of hadronic final states produced in @xmath0 annihilation coincides with the laboratory system in the absence of hard initial state radiation ( isr ) .
a typical @xmath0 annihilation experiment is shown in figure [ fig_jade ] .
the jade detector ( for details see @xcite ) was operated at the petra accelerator @xcite from 1979 to 1986 at cms energies @xmath16 gev .
the design of the experiment is similar to the more recent lep experiments .
the interaction region , where the electron and positron beams collide , is surrounded by gaseous tracking detectors ( jet chamber ) inside a solenoidal magnetic field of 0.48 t. the tracking detectors are inside a calorimeter consisting of 2712 lead glass blocks to measure the energy of charged and neutral particles .
tracking and calorimeter efficiency is good in the region @xmath17 were @xmath18 is the angle w.r.t .
the beam axis .
therefore most @xmath0 annihilation events are fully contained in the detector and complete reconstruction with high efficiency and good precision is possible .
this reduces the size of experimental corrections for acceptance and resolution and thus the corresponding experimental uncertainties .
the experimental conditions vary significantly between the various cms energies of the data sets .
figure [ fig_data ] ( left ) @xcite shows the measured cross sections of several @xmath0 annihilation processes as a function of the cms energy @xmath19 .
the @xmath20 dependence as expected from qed at low @xmath19 is clearly visible . at the @xmath21 peak around @xmath22
gev the cross sections increase by a factor of about 100 compared to their lowest values at @xmath23 gev or @xmath24 gev . at @xmath25 background from two - photon interactions with hadronic final states ( @xmath26 ) , from production of @xmath1-lepton pairs decaying to hadrons and from hadronic final states with isr and
thus reduced cms is significant . requiring more than four tracks of charged particles ( events with exactly four tracks not in a 1- vs 3-prong configuration
may also be accepted @xcite ) reduces the backgrounds from @xmath1-pairs and two - photon interactions to negligible levels .
the data collected on or near the @xmath21 peak are essentially free of backgrounds due to the high energy and large cross section .
isr effects are also suppressed .
the high energy samples taken by the lep experiments during the lep 2 phase have again backgrounds from two - photon interactions , isr and for @xmath27 gev from production of four - fermion final states dominated by @xmath28 pairs decaying to hadrons . [ cols="^,^ " , ] the analysis @xcite defines an effective charge @xmath29 via @xmath30 .
the relation between @xmath29 and @xmath31 is known perturbatively .
figure [ fig_softfreeze ] ( right ) shows the variable @xmath29 as a function of @xmath32 extracted from the data ( shaded bands ) and using different predictions ( lines ) .
the band from analysis of the inclusive spectrum labelled `` @xmath33 '' stays fairly flat down to @xmath34 gev@xmath35 and then rises steeply due to the @xmath36 resonance .
the other bands are from analysis of spectra for even or odd number of pions in the final state and have strong non - perturbative effects .
the solid line labelled `` @xmath37 4 loop @xmath38 '' corresponds to the most complete qcd prediction for the running of the effective charge including an estimate of 4th order term .
this prediction describes the data well down to @xmath39 gev@xmath35 and stays finite at @xmath40 .
one also observes that the nnlo prediction labelled `` @xmath37 3 loop '' also describes the data at medium values of @xmath32 and is finite at @xmath40 .
it is thus possible to consistently describe data for hadronic @xmath1 decays down to @xmath41 gev@xmath35 with an effective coupling which stays finite at @xmath40 .
these observations are consistent with earlier work @xcite .
in this review several topics were not covered due to a lack of results .
one issue is the study of 4-jet event shape observables in the planar limit .
the theoretical aspects and preliminary experimental results are discussed in contributions b001 and b002 to this conference .
however , there are no published experimental studies which could be subject of this review , but hopefully there will be results on this important subject in the near future .
other models for the description of soft corrections to hard qcd processes such as shape functions , the single - dressed - gluon approximation ( sdg ) are subject of contributions to this conference but there are no detailed experimental studies . from an experimental point of view
the dmw model is attractive , because it makes strong predictions about relations between observables and has only universal free parameters .
also , predictions actually exist for several observables such that direct comparisons are possible . *
the values for @xmath42 from different observables are consistent within 20 to 30% .
however , it has to be understood if this is a variation one expects from the theoretical uncertainties in the calculations .
it is often claimed that the theoretical uncertainty on the milan factor @xmath43 of about 20% due to missing higher order terms could explain the scatter of @xmath42 results .
however , since @xmath43 is a universal factor it is not clear if it can explain differences between different observables . *
the values for @xmath44 from different observables are consistent with each other within the uncertainties of the perturbative qcd calculations as expected .
however , direct comparison of the results from power correction analyses and analyses with hadronisation corrections derived from monte carlo simulation show a systematic bias for @xmath44 of about 6% ( mean values ) or 9% ( distributions ) .
this bias is comparable to or larger than the total uncertainties quoted for determinations of @xmath44 from event shape or jet observables in @xmath0 annihilation .
the bias is significantly larger than the hadronisation model uncertainties of less than 2% quoted for @xmath44 determinations using lep data @xcite . in the future
the studies of power corrections with @xmath0 annihilation data would profit from further analyses using the jade data in the intermediate energy region @xmath45 gev . in particular the eec and 4-jet observables with or without requiring the planar limit would be attractive subjects
. the impact of nnlo perturbative qcd calculations for 3-jet observables could also be significant .
firstly , the power correction calculations will have to adapted to the improved predictions .
secondly , with there is potential for high precision measurements of @xmath44 with reduced theoretical uncertainties . with selected observables with
suppressed power corrections such as @xmath46 one could obtain the matching reduced hadronisation uncertainties .
other approaches to understand soft corrections in terms of qcd should be studied in more detail .
the connection with rs optimisation discussed in contributions t001 and t002 is also an interesting new area of research .
s. kluth : in : 2001 qcd and high energy hadronic interactions , j.trnthanh vn ( ed . ) , 115 .
the gioi publishers , vietnam , 2002 , contributed to xxxvith recontre de moriond , qcd and high energy hadronic interactions , les arcs , france , march 1724 , 2001 | hadron production in @xmath0 annihilation experiments main for non - perturbative qcd hadron jets lepton - nucleon scattering two advantages : leptonic initial state incoming electron positron are leptons no interference between hadronic final state interpretation final state qcd robust large range of energies data @xmath0 centre - of - mass energies comparing data cms energies tests theory from reconstructed hadronic final states jet production event shape observables calculated observable thrust @xmath2 defined by @xmath3 @xmath4 3-momentum of particle @xmath5 @xmath6 @xmath7 corresponds ideal 2-jet event events with three final state objects have @xmath8 spherical events many final particles approach @xmath9 event shape observables @xmath10 @xmath11 for ideal two - jet configurations thrust analysed in @xmath12 classified as 3-jet or 4-jet depending on @xmath13 @xmath14 or @xmath15 objects energy electron beams in @xmath0annihilation collider experiments equal due to simplified construction operation circular accelerator one beam linehadronic final states @xmath0 annihilation coincides with laboratory system hard initial state radiation typical @xmath0 annihilation experiment in figure_jade jade detector operated at petra accelerator 1979 to 1986 at energies @xmath16 gev design similar to lep experiments interaction region electron positron beams collide surrounded by gaseous tracking detectors solenoidal magnetic field 0.48 t detectors inside calorimeter 2712 lead glass blocks energy charged neutral particles tracking calorimeter efficiency good in region @xmath17 @xmath18 angle beam axis annihilation events contained in detector reconstruction with high efficiency precision possible reduces experimental corrections experimental uncertainties experimental conditions vary between cms energies data sets shows measured cross sections of @xmath0 annihilation processes cms energy @xmath19 @xmath20 dependence from qed at low @xmath19 visible @xmath21 peak around @xmath22 gev cross sections increase by 100 lowest values at @xmath23 gev @xmath25 background - photon interactions hadronic final states @xmath1-lepton pairs reduced cms significantthan four tracks charged particles accepted reduces backgrounds from @xmath1-pairs two - photon interactions data near @xmath21 peak free of backgrounds high energy large cross section isr effects suppressed high energy samples lep experiments 2 phase have backgrounds from two - photon interactions isr @xmath27 four - fermion final states @xmath28 pairs decaying hadrons analysis @xcite defines effective charge @xmath29 via @xmath30 relation between @xmath29 @xmath31 known perturbatively shows variable @xmath29 function of @xmath32 extracted data predictions band from @xmath33 flat to @xmath34@xmath35 rises due to @xmath36 resonance other bands from analysis even odd number pions final state strong non - perturbative effects solid line @xmath37 4 loop @xmath38 complete qcd prediction effective charge estimate 4th order term describes data to @xmath39@xmath35 finite at @xmath40 prediction @xmath37 3 loop describes data at medium values @xmath32 finite at @xmath40possible consistently describe data for hadronic @xmath1 decays to @xmath41@xmath35 effective coupling finite at @xmath40 observations consistent with earlier work @xcite review topics not covered lack of results issue study 4-jet event shape observables in planar limit theoretical aspects preliminary experimental results discussed in contributions b001 and b002 no published experimental studies results future other models soft corrections to hard qcd processes shape functions single - dressed - gluon approximation no detailed experimental studies dmw model attractive makes strong predictions relations observables universal free parameters predictions exist for several observables direct comparisons possible values for @xmath42 from observables consistent within 20 to 30% variation from theoretical uncertainties theoretical uncertainty on milan factor @xmath43 20% missing higher order terms could explain scatter @xmath42 results @xmath43 universal factor not clear explain differences observables values for @xmath44 consistent within uncertainties perturbative qcd calculations direct comparison results power correction analyses with hadronisation corrections simulation systematic bias for @xmath44 of 6% or 9%bias comparable to larger than uncertainties @xmath44 jet observables @xmath0 annihilation bias larger than hadronisation model uncertainties less than 2% @xmath44 determinations lep data studies power corrections @xmath0 annihilation profit from analyses jade data intermediate energy region @xmath45 gev eec 4-jet observables planar limit attractive impact nnlo perturbative qcd calculations for 3-jet observables significant power correction calculations improved predictions potential for high precision measurements @xmath44 reduced theoretical uncertainties observables suppressed power corrections @xmath46 reduced hadronisation uncertainties approaches soft corrections qcd connection with rs optimisation contributions t001 t002 interesting research s. kluth 2001 qcd high energy hadronic interactions.trnthanh gioi publishers vietnam xxxvith recontre de moriond qcd high energy hadronic interactions france | 0.318759 | 2,121 | 1,070 | , Saving $0.1 in GPT-4. |
177,385 | we analyze the dependence of cp - odd asymmetries on longitudinally polarized @xmath0 and @xmath1 beams in neutralino production and subsequent leptonic two - body decays .
we give numerical examples of the asymmetries and of the cross sections in the minimal supersymmetric standard model with complex parameters @xmath2 , @xmath3 and @xmath4 for a collider with @xmath5 gev .
we show that longitudinally polarized electron and positron beams can enhance considerably both the asymmetries and the cross sections .
-0 mm -10 mm 220 mm 160 mm wue - itp-03 - 016 + uwthph-2003 - 24 + hephy - pub 779/03 + hep - ph/0310011 + * impact of beam polarization on cp asymmetries in neutralino pair production * 2.5em + [ 1ex ] _
@xmath6 institut fr theoretische physik , universitt wien , boltzmanngasse 5 , a-1090 wien , austria _ + _ @xmath7 institut fr theoretische physik , universitt wrzburg , am hubland , d-97074 wrzburg , germany _ + _ @xmath8 institut fr hochenergiephysik , sterreichische akademie der wissenschaften , nikolsdorfergasse 18 , a-1050 wien , austria _ + 1em | in the neutralino sector of the minimal supersymmetric standard model ( mssm ) @xcite , the gaugino mass parameter @xmath3 , the higgsino mass parameter @xmath2 , and the trilinear coupling parameter @xmath4 in the stau sector , can be complex .
the physical phases @xmath9 , @xmath10 and @xmath11 of these parameters can cause large cp - violating effects already at tree level . in neutralino production @xmath12 and the subsequent leptonic two - body decay of one of the neutralinos @xmath13 and of the decay slepton @xmath14
the neutralino spin correlations lead to several cp - odd asymmetries . with the triple product @xmath15
, we define the _ t - odd _ asymmetry of the cross section @xmath16 for the processes ( [ production])-([decay_2 ] ) : @xmath17 if absorbtive phases are neglected , @xmath18 is cp - odd due to cpt invariance .
the dependence of @xmath18 on @xmath9 and @xmath10 was analyzed in @xcite . in case
the neutralino decays into a @xmath19-lepton , @xmath20 , @xmath21 , the t - odd transverse @xmath22 and @xmath23 polarizations @xmath24 and @xmath25 , respectively , give rise to the _ cp - odd _
observable @xmath26 which is also sensitive to @xmath11 . for various mssm scenarios ,
@xmath27 was discussed in @xcite . for measuring the asymmetries , it is crucial to have both large asymmetries and large cross sections . in this note
we study the impact of longitudinally polarized @xmath0 and @xmath1 beams of a future linear collider in the 500 gev range on the asymmetries @xmath18 , @xmath27 and on the cross sections @xmath16 .
we present numerical results for @xmath28 with the subsequent leptonic decay of @xmath29 for a linear collider with @xmath5 gev . for @xmath18 , eq .
( [ tasymmetry ] ) , we study the neutralino decay into the right selectron and right smuon , @xmath30 , @xmath31 and for @xmath27 , eq .
( [ acp ] ) , that into the lightest scalar tau , @xmath32 .
we study the dependence of the asymmetries and the cross sections on the beam polarizations @xmath33 and @xmath34 for fixed parameters @xmath35 , @xmath36 , @xmath37 , @xmath38 and @xmath39 .
we assume @xmath40 and use the renormalization group equations @xcite for the selectron and smuon masses , @xmath41 with @xmath42 gev . the interaction lagrangians and details on stau
mixing can be found in @xcite . in fig .
[ plot_1]a we show the dependence of @xmath18 on the beam polarization for @xmath43 and @xmath44 . a small value of @xmath10 is suggested by constraints on electron and neutron electric dipole moments ( edms ) @xcite for a typical susy scale of the order of a few 100 gev ( for a review see , e.g. , @xcite ) .
it is remarkable that in our scenario the asymmetry can be close to 10% even for the small value of @xmath43 and for @xmath45 . the cross section @xmath46 is shown in fig .
[ plot_1]b . for our scenario with @xmath47 gev and @xmath48 ,
the neutralino branching ratio is br@xmath49 = 0.63 ( summed over both signs of charge ) and br(@xmath50 ) = 1 .
note that the asymmetry @xmath18 and the cross section @xmath16 are both considerably enhanced for negative positron and positive electron beam polarization .
this choice of polarization enhances the contributions of the right slepton exchange in the neutralino production , eq .
( [ production ] ) , and reduces that of left slepton exchange @xcite . while the contributions of right and left slepton exchange enter @xmath16 with the same sign , they enter @xmath18 with opposite sign , which accounts for the sign change of @xmath18 .
( 10,8)(0,0 ) ( 0,0 ) ( 3.,7.5 ) ( 6.5,-.3)@xmath33 ( 0.2,7.3)@xmath34 ( 1.2,6.3)-4.5 ( 2.2,5.2)-3 ( 3.0,4.4)0 ( 3.6,3.8)3 ( 4.5,3)6 ( 5.2,2.1)8 ( 6.5,1.)9 ( 0.5,-.3)fig . [ plot_1]a ( 8,0 ) ( 9.5,7.5 ) ( 14.5,-.3)@xmath33 ( 8.2,7.3)@xmath34 ( 8.5,-.3)fig . [ plot_1]b ( 9.2,1.2)4 ( 9.5,2.1)12 ( 10.0,3.6)20 ( 10.2,5.2)24 ( 11.5,3.9)24 ( 9.3,6.1)28 ( 12.3,3.1)28 ( 13.2,2.3)36 ( 14.,1.5)48 ( 14.6,0.8)60 ( 11.8,5.3)20 ( 13.2,5.8)12 ( 14.3,6.3)4 in fig .
[ plot_2]a we show the contour lines of the @xmath19 polarization asymmetry @xmath27 , eq .
( [ acp ] ) , for @xmath51 and @xmath52 in the @xmath33-@xmath34 plane .
we have chosen a large value of @xmath53 gev because @xmath27 increases with increasing @xmath54 @xcite .
for unpolarized beams the asymmetry is 1% .
however , it reaches values of more than @xmath55 if the @xmath0 and @xmath1 beams are polarized with the opposite sign .
if at least one of the beams is polarized ( e.g. @xmath56 , @xmath57 ) , the asymmetries are somewhat smaller ( @xmath58 ) .
the reason for this dependence is again the enhancement of either the right or the left selectron exchange contributions in the production process .
the cross section @xmath59 is shown in fig . [ plot_2]b with @xmath60 .
also @xmath16 is very sensitive to variations of the beam polarization and varies between 1 fb and 30 fb .
( 10,8)(0,0 ) ( 0,0 ) ( 3.,7.5 ) ( 6.5,-.3)@xmath33 ( 0.2,7.3)@xmath34 ( 1.,6.4)-14.5 ( 2.,5.4)-12 ( 3.0,4.4)-6 ( 3.8,3.7)0 ( 4.5,3)6 ( 5.3,2.1)12 ( 6.3,.8)13.5 ( 0.5,-0.3)fig . [ plot_2]a ( 8,0 ) ( 9.5,7.5 ) ( 14.5,-.3)@xmath33 ( 8.2,7.3)@xmath34 ( 8.5,-.3)fig . [ plot_2]b ( 9.1,0.8)1 ( 9.6,1.2)5 ( 10.2,1.8)10 ( 11.0,2.6)15 ( 11.1,3.7)17 ( 11.,4.7)20 ( 10.2,5.5)25 ( 9.5,6.0)30 ( 13.,3.3)17 ( 12.8,4.3)15 ( 13.6,5.2)10 ( 14.3,5.9)5 ( 14.7,6.4)1 ( 13.3,2.1)20 ( 14.1,1.3)25 ( 14.6,0.7)30 since the asymmetry @xmath27 is also very sensitive to the phases @xmath9 and @xmath10 we show for @xmath61 and @xmath62 , the dependence of @xmath27 and @xmath63 on the beam polarization in figs .
[ plot_3]a , b , respectively .
the neutralino branching ratio is @xmath64 for our scenario . despite the small phases , @xmath27 reaches values up to @xmath65 for negative @xmath1 and positive @xmath0 beam polarizations .
( 10,8)(0,0 ) ( 0,0 ) ( 3.,7.5 ) ( 6.5,-.3)@xmath33 ( 0.2,7.3)@xmath34 ( 1.4,6.1)-12 ( 2.9,5.)-9 ( 3.4,4.3)-6 ( 3.9,3.7)-3 ( 4.5,3.2)0 ( 5.3,2.2)3 ( 6.4,.9)6.5 ( 0.5,-.3)fig . [ plot_3]a ( 8,0 ) ( 9.5,7.5 ) ( 14.5,-.3)@xmath33 ( 8.2,7.3)@xmath34 ( 8.5,-.3)fig . [ plot_3]b ( 9.6,1.2)5 ( 10.4,1.7)10 ( 11.4,2.1)15 ( 11.6,3.8)17 ( 10.9,4.6)20 ( 10.1,5.2)25 ( 9.5,5.7)30 ( 9.1,6.3)35 ( 12.9,4.3)15 ( 13.6,5.)10 ( 14.3,5.9)5 ( 13.4,2.1)17 ( 14.1,1.4)20 ( 14.7,0.7)25
within the mssm we have analyzed the dependence on the beam polarization of two cp - odd asymmetries in @xmath66 and the subsequent leptonic two - body decay of @xmath67 .
for the decay process @xmath68 , @xmath69 with @xmath70 , we have found that the asymmetry @xmath71 of the triple product @xmath72 , which is sensitive to @xmath9 and @xmath10 , can be twice as large if polarized beams are used , with e.g. @xmath56 and @xmath73 .
also for these polarizations the cross section can be enhanced up to a factor of 2 . for the neutralino decay , @xmath74
, we have given numerical examples for the beam polarization dependence of the cp - odd @xmath19 polarization asymmetry @xmath27 , which is also sensitive to @xmath11 .
for the scenarios considered , both @xmath27 and the cross section depend sensitively on the beam polarizations and can be enhanced by a factor between 2 and 3 . the dependence on the beam polarizations of the asymmetries @xmath18 , @xmath27 and of the cross sections is due to the contributions from right and left selectron exchange in the neutralino production process .
generally , negative ( positive ) @xmath1 and positive ( negative ) @xmath0 beam polarization enhances right ( left ) selectron exchange .
due to the fact that both asymmetries and cross sections can be enhanced significantly , we conclude that the option of having both beams polarized at an @xmath75-collider is advantageous for the determination of the cp - odd asymmetries .
this work was supported by the ` fonds zur frderung der wissenschaftlichen forschung ' ( fwf ) of austria , projects no .
p13139-phy and no .
p16592-n02 and by the european community s human potential programme under contract hprn - ct-2000 - 00149 . this work was also supported by the deutsche forschungsgemeinschaft ( dfg ) under contract fr 1064/5 - 1 .
i. s. altarev et al . , phys . lett . * b 276 * ( 1992 ) 242 ; i. s. altarev et al . ,
phys . atom . nucl.*59 * ( 1996 ) 1152 and yad .
* 59 n 7 * ; e. d. commins , s. b. ross , d. demille , b. c. regan , phys
. rev . * a 50 * ( 1994 ) 2960 . g. moortgat - pick , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , econf * c010630 * , e3008 ( 2001 ) [ arxiv : hep - ph/0202082 ] ; g. moortgat - pick , a. bartl , h. fraas and w. majerotto , eur .
j. c * 18 * ( 2000 ) 379 . | neutralino sector supersymmetric standard model @xcite gaugino mass @xmath3 higgsino mass parameter @xmath2 trilinear coupling parameter @xmath4 complex physical phases @xmath9 cause cp - violating effects tree level neutralino production @xmath12 leptonic two - body decay decay spin correlations lead cp - odd asymmetries triple product @xmath15 t - odd _ asymmetry cross section @xmath16 production @xmath17 absorbtive phases neglected @xmath18 cp - odd cpt invariance dependence @xmath18 @xmath9 analyzed @xcite neutralino decays @xmath19-lepton t - odd transverse @xmath22 @xmath23 polarizations @xmath24 cp - odd _ observable @xmath26 sensitive @xmath11 mssm scenarios @xmath27 discussed @xcite crucial large asymmetries cross sections impact longitudinally polarized @xmath0 @xmath1 beams future linear collider 500 gev range asymmetries @xmath18 @xmath27 cross sections @xmath16present numerical results @xmath28 leptonic decay @xmath29 linear collider @xmath5 gev @xmath18 tasymmetry study neutralino decay right selectron right smuon @xmath30 @xmath31 @xmath27 lightest scalar tau @xmath32 study asymmetries cross sections beam polarizations @xmath33 @xmath34 fixed parameters @xmath35 @xmath39 assume @xmath40 use renormalization group equations selectron smuon masses @xmath41 @xmath42 gev interaction lagrangians details stau mixing @xcite fig . plot_1 dependence @xmath18 beam polarization @xmath43 @xmath44 small value @xmath10 suggested constraints electron neutron electric dipole moments edms typical susy scale 100 gev scenario asymmetry close 10% small value @xmath43 @xmath45 cross section @xmath46 shown fig . plot_1]b scenario @xmath47 gev @xmath48 neutralino branching ratio br@xmath49 = 0.63 br(@xmath50 ) = 1 .asymmetry @xmath18 cross section @xmath16 enhanced negative positron positive electron beam polarization enhances contributions right slepton exchange production reduces left slepton exchange contributions enter @xmath16 same sign @xmath18 opposite sign change 10,8) 3.,7.5 6.5,-.3)@xmath33 ( 0.2,7.3)@xmath34 ( 1.2,6.3)-4.5 2.2,5.2 3.0,4.4 3.6,3.8)3 4.5,3)6 5.2,2.1)8 6.5,1.)9 0.5,- 9.5,7.5 ) 14.5,-.3)@xmath33 8.2,7.3)@xmath34 ( 8.5,-,1.2)4 9.5,2.1 10.0,3.6 11.5,3.9 9.3,6.1 12.3,3.1 13.2,2.3,1.5 11.8,5 14.3,6.3 contour lines @xmath19 polarization asymmetry @xmath27 @xmath51 @xmath52 @xmath33-@xmath34 plane large value @xmath53 gev @xmath27 increases @xmath54 @xcite unpolarized beams asymmetry 1%reaches values more than @xmath55 if beams polarized opposite if one polarized asymmetries smaller ( @xmath58 ) dependence enhancement right left selectron exchange contributions production cross section @xmath59 in with @xmath60 @xmath16 sensitive to beam polarization varies between 1 and 30 fb 10,8) 3.,7.5 ) 6.5,-.3)@xmath33 ( 0.2,7.3)@xmath34 ( 1.,6.4)-14.5.4)-12 3.0,4.4)-6 3.8,3.7)0,3)6 5.3,2.1)12 6.3,.8)13.5 ( 0.5,-0.3,7.5 ) 14.5,-.3)@xmath33 ( 8.2,7.3)@xmath34 ( 8.5,-.8)1,1.2)5 10.2,1.8,2.6)15 11.1,3.7)17 9.5,6.0)30,3.3 12.8,4.3 14.3,5.9)5 14.7,6.4)1,2.1 14.1,1.3)25 14.6,0.7)30 asymmetry @xmath27 sensitive to phases @xmath61 dependence of @xmath27 @xmath63 on beam polarization neutralino branching ratio is @xmath64 scenariosmall phases @xmath27 reaches negative positive beam polarizations 10,8)(0,0 3.,7.5 ) 6.5,-.3)@xmath33 0.2,7.3) 1.4,6.1 2.9,5. 3.4,4.3 3.9,3.7)-3 4.5,3.2 5.3,2.2)3 6.4,.9)6.5 0.5,- 9.5,7.5 ) 14.5,-.3)@xmath33 8.2,7.3),- 9.6,1.2)5 10.4,1.7)10 11.4,2.1)15 11.6,3.8)17 10.9,4.6 10.1,5.2)25 9.5,5.7)30 9.1,6.3)35 12.9,4.3)15 13.6,5. 14.3,5.9)5 13.4,2.1)17 14.1,1.4 14.7,0.7)25 analyzed dependence beam polarization cp - odd asymmetries @xmath66 leptonic two - body decay @xmath67 asymmetry @xmath71 sensitive twice large if polarized beams cross section enhanced factor 2 decay examples beam polarization dependence cp - odd polarization asymmetry @xmath27 sensitive to @xmath11scenarios considered @xmath27 cross section depend on beam polarizations enhanced by factor between 2 3 dependence on beam polarizations asymmetries @xmath18 cross sections due to contributions right left selectron exchange neutralino production process negative @xmath1 positive @xmath0 beam polarization enhances right left selectron exchange asymmetries cross sections enhanced both beams polarized at @xmath75-collider advantageous for cp - odd asymmetries work supported by fonds zur frderung wissenschaftlichen forschung austria projects p13139-phy p16592-n02 european community s human potential programme contract hprn - ct-2000 - 00149 supported by deutsche forschungsgemeinschaft dfg ) contract fr 1064/5 - 1 altarev et al . . lett 276 ( 1992 ) 242 1996 ) 1152 7 commins ross demille regan phys. rev 50 1994 ) 2960 moortgat - pick summer study on future of particle physics snowmass 2001 ) graf c010630 e3008 2001 arxiv hep - ph/0202082 moortgat - pick bartlfraas w. majerotto eur 18 2000 379 | 0.425867 | 3,135 | 1,739 | , Saving $0.1 in GPT-4. |
177,391 | motivated by the results of two - dimensional conformal field theory ( cft ) we investigate the finite - size scaling of the mass spectrum of an ising model on three - dimensional lattices with a spherical cross section .
using a cluster - update monte carlo technique we find a linear relation between the masses and the corresponding scaling dimensions , in complete analogy to the situation in two dimensions .
amplitude ratios as well as the amplitudes themselves appear to be universal in this case . | divergent correlation lengths of a statistical system approaching a critical point establish the symmetry of scale invariance on which finite - size scaling theory is based . in a continuum critical theory , however , there are additional symmetries present , namely translational , rotational and inversion invariance
. the implications of the group of these symmetries , as explored in conformal field theory @xcite , go far beyond simple scaling theory , especially for two - dimensional systems , where the whole content of scaling operators as well as for example higher correlation functions can be generically extracted from this symmetry .
as for a prominent result , conformal invariance suffices to determine the critical two - point function in the plane . using a conformal ( logarithmic ) transformation that wraps the plane around an infinite - length cylinder one thus arrives at an expression for the correlation function of a primary ( conformally covariant ) operator on the cylinder @xcite , a geometry one generically considers in transfer matrix calculations . the fact that this transformation does only affect the radial but not the angular part of the coordinates led cardy @xcite to a generalization for the higher - dimensional geometries @xmath0 .
thus , exactly as in two dimensions , the large distance exponential decay of correlations is determined by a correlation length that is given by the inverse of the scaling dimension @xmath1 of the operator under consideration , @xmath2 where @xmath3 is the radius of @xmath4 .
it has to be noted , however , that the meaning of a _ primary _ operator is a priori not well defined in three dimensions so that this result has to be considered a conjecture and a numerical analysis is more than an exercise in this case . a first attempt to establish such evidence using the hamiltonian formulation of the ising model and platonic solids as discretization of the sphere @xmath5 was inconclusive due to the restricted number and size of regular polyhedra @xcite .
using _ approximate _
quadrangulations we are able to check cardy s conjecture for systems sufficiently large to carry out a proper finite - size scaling ( fss ) analysis .
there are several straightforward choices for model lattices with spherical topology @xcite ; the more natural are variants of a rectangular mesh on a cube .
different refinements can be applied to reduce the concentration of the curvature around the corners of the cube such as filling in triangles instead of the cube corners or projecting the cube on the sphere by varying the link lengths , i.e. , applying an appropriate site - dependent weight function to the action . for those cube - like lattices
, however , differences in the scaling behavior of bulk quantities are found to be quite small @xcite . as for our observables
, there is some evidence to believe that _ ratios _ of correlation lengths of primary operators are universal @xcite , so that we can expect good agreement regardless of the lattice used if cardy s conjecture holds . for
the amplitudes themselves one might find effects of the non - uniformity of the lattice , but would not expect them to become visible until a very high level of precision is reached .
we thus here use the simple cubic approximation to the sphere by six @xmath6 square lattices with suitably chosen boundary conditions ; effects of the discretization will be considered elsewhere @xcite .
having chosen a lattice approximation to the sphere one has to fix a definition of the radius @xmath3 of the sphere the cube with edge length @xmath7 should correspond to .
as is easily checked the three possibilities of assigning a unit area to each lattice site , each pair of bonds , or each lattice square , respectively , lead to total lattice areas of @xmath8 and generate via the relation @xmath9 two different sorts of pseudo radii , which only differ by a constant shift , thus leading to slightly different approaches to the leading fss amplitude .
we consider a classical , ferromagnetic , nearest - neighbor ising model with hamiltonian @xmath10 on lattices compound of the above described cubic discretizations of the sphere times a linear direction of length @xmath11 , modelling the @xmath12-direction . in order to minimize the effect of a finite @xmath11 it was ( self - consistently ) chosen so that @xmath13 and periodic boundary conditions in @xmath14-direction were applied .
simulations were done using the wolff single - cluster update algorithm at an inverse temperature @xmath15 @xcite , checking for the influence of a temperature shift off the critical point by reweighting . to enable a proper fss analysis system sizes ranging from @xmath16 to @xmath17
were used , corresponding to radii from about @xmath18 to @xmath19 and overall lattice volumes of @xmath20 to @xmath21 spins .
since the densities of energy and magnetization are ( the only ) primary operators in the _ two - dimensional _ ising model and the lowest lying states of the two sectors , we here consider the amplitudes of the corresponding ( exponential ) correlation lengths @xmath22 and their ratio .
simulational measurements were done for the ( connected ) correlation functions @xmath23 of these operators using a zero mode projection .
estimators for the correlation lengths that eliminate additive and multiplicative constants in the exponential decay are then given by @xmath24}^{-1 } , \label{diffmethoddelta}\ ] ] with @xmath25 , from which overall estimates @xmath26 are formed by averaging .
a jackknife technique was used to reduce estimator bias and for the error analysis ; for details see @xcite .
( 150,145 ) ( 0 , 0 ) ( 150,145 ) ( 0 , 0 ) using the above described procedures one ends up with scaling plots for the correlation lengths as shown in fig.[fig](a ) for the magnetization density that show an almost perfect linear behavior and invisible differences between the different radii definitions according to eq.([radii ] ) . plotting the amplitudes @xmath27 ( fig.[fig](b ) ) , however , reveals the presence of corrections to scaling that enforce the use of nonlinear fits of the form @xmath28 with a free parameter @xmath29 that introduces a systematic error to the final amplitudes @xmath30 due to higher order corrections .
a combination of the fits for the two sorts of radii gives the amplitudes @xmath31 which agree well with the conjectured amplitudes @xmath32 taking into account the systematic error due to the different radii definitions that is of the same order of magnitude as the statistical error .
the amplitude ratios compare as : @xmath33
the presumably universal ratio of the amplitudes of the spin and energy correlation lengths of the ising model on a cubic model of @xmath34 agrees with the inverse ratio of the scaling dimensions as conjectured by cardy for the continuum case .
moreover , corrections to scaling due to the non - uniform distribution of curvature over the lattice that might influence the amplitudes themselves seem not to be very important at the given level of accuracy ; only the slightly shifted result for the spin amplitude @xmath35 might indicate the onset of such effects . in connection with the results for systems with toroidal cross section @xcite this seems to indicate a deeper analogy between the 2d and 3d situations . in this latter case , however , the system is not conformally related to a flat space which renders impossible an analytical treatment of cardy s kind . in the spherical case ,
on the other hand , conformal flatness is fulfilled , but the algebraic meaning of the primarity of an operator remains unclear .
thus our result shows that spin and energy densities of the 3d ising model are primary operators , taking eq.([conj ] ) as a definition of primarity in three dimensions . | divergent correlation lengths statistical system approaching critical point establish symmetry of scale invariance finite - size scaling theory in continuum critical theory additional symmetries present translational rotational inversion invariance implications symmetries explored in conformal field theory beyond simple scaling theory two - dimensional systems content scaling operators higher correlation functions extracted from symmetry conformal invariance suffices determine critical two - point function in plane conformal transformation wraps plane around infinite - length cylinder arrives expression for correlation function of primary covariant operator on cylinder transformation radial not angular part of coordinates generalization for higher - dimensional geometries large distance exponential decay of correlations determined by correlation length given by inverse of scaling dimension of operator consideration is radius of @xmath4 meaning of _ primary _ operator not defined in three dimensions result considered conjecture numerical analysis more than exercise first attempt establish evidence using hamiltonian formulation ising model platonic solids sphere inconclusive due to restricted number size of regular polyhedrausing quadrangulations check cardy s conjecture for systems large finite - size scaling analysis choices for model lattices with spherical topology natural are variants of rectangular mesh on cube refinements to reduce curvature around corners cube filling in triangles projecting cube on sphere varying link lengths applying site - dependent weight function for cube - like lattices differences in scaling behavior of bulk quantities small evidence ratios of correlation lengths of primary operators universal expect good agreement regardless of lattice if cardy s conjecture holds amplitudes find effects of non - uniformity lattice visible until high level precision reached use simple cubic approximation to sphere by six square lattices with chosen boundary conditions effects discretization considered elsewhere chosen lattice approximation definition of radius of sphere cube with edge length correspond to three possibilities of assigning unit area to each lattice site bonds lattice square lead to total lattice areas generate two different pseudo radii differ by constant shift leading to different approaches to leading fss amplitudeconsider classical ferromagnetic nearest - neighbor ising model hamiltonian @xmath10 lattices cubic discretizations sphere linear direction length @xmath11 modelling @xmath12-direction minimize effect finite @xmath11 chosen @xmath13 periodic boundary conditions in @xmath14-direction applied simulations using wolff single - cluster update algorithm inverse temperature @xmath15 @xcite checking influence temperature shift reweighting system sizes @xmath16 to @xmath17 used radii @xmath18 to @xmath19 lattice volumes @xmath20 to @xmath21 spins densities of energy magnetization primary operators two - dimensional ising model lowest lying states sectors consider amplitudes exponential correlation lengths @xmath22 ratio simulational measurements connected correlation functions @xmath23 zero mode projection estimators correlation lengths additive multiplicative constants given by @xmath24{-1{diffmethoddelta with @xmath25 estimates @xmath26 formed by averaging jackknife technique used reduce estimator bias error analysis details see @xcite 150 procedures scaling plots for correlation lengths fig.[fig](a ) magnetization density linear behavior invisible differences radii definitions eq.([radii ] ) plotting amplitudes @xmath27 fig.(b ) reveals corrections scaling nonlinear fits form @xmath28 free parameter @xmath29 systematic error final amplitudes @xmath30 higher order corrections combination fits radii amplitudes @xmath31 agree conjectured amplitudes @xmath32 systematic error radii definitions same magnitude statistical error amplitude ratios compare @xmath33 universal ratio amplitudes spin energy correlation lengths ising model cubic model @xmath34 agrees inverse ratio scaling dimensions conjectured cardy continuum case corrections scaling non - uniform distribution curvature lattice amplitudes not important accuracy slightly shifted result spin amplitude @xmath35 indicate onset effects results systems toroidal cross section deeper analogy between 2d 3d situations latter case system not conformally related flat space impossible analytical treatment cardy s kind spherical case conformal flatness fulfilled algebraic meaning primarity operator unclear result shows spin energy densities 3d ising model primary operators eq.([conj ] ) definition primarity three dimensions | 0.268583 | 1,646 | 819 | , Saving $0.0 in GPT-4. |
177,411 | we report on an analytical study of the statistics of conductance , @xmath0 , and shot - noise power , @xmath1 , for a chaotic cavity with arbitrary numbers @xmath2 of channels in two leads and symmetry parameter @xmath3 . with the theory of selberg s
integral the first four cumulants of @xmath0 and first two cumulants of @xmath1 are calculated explicitly .
we give analytical expressions for the conductance and shot - noise distributions and determine their exact asymptotics near the edges up to linear order in distances from the edges . for @xmath4
a power law for the conductance distribution is exact .
all results are also consistent with numerical simulations .
+ pacs numbers : 73.23.-b , 73.50.td , 05.45.mt , 73.63.kv @xmath5fachbereich physik , universitt duisburg - essen , 47048 duisburg , germany + @xmath6department of mathematical sciences , brunel university , + uxbridge , ub8 3ph , uk | our goal is to discuss the statistics of conductance and shot - noise power for chaotic cavities using essentially the properties of selberg s integral .
the static conductance @xmath7 relates linearly the time averaged current @xmath8 to the external voltage @xmath9 between two electron reservoirs .
fluctuations of the current around its mean value are conventionally described by the spectral noise power @xmath10 , with @xmath11 .
as temperature goes to zero , the only source of noise that remains non - vanishing is related to the discreteness of the electric charge carriers , the so - called shot - noise . for mesoscopic conductors , an adequate framework for the problematic is based on random - matrix - theory ( rmt ) approach to quantum transport , we refer to @xcite for reviews .
we consider a chaotic cavity with two attached leads supporting @xmath12 and @xmath13 channels , respectively , and coupled perfectly to the interior of the cavity .
according to the landauer - bttiker formalism , the conductance @xmath7 and the shot - noise power @xmath14 can be expressed in terms of the transmission eigenvalues @xmath15 as follows : @xmath16 and @xmath17 where @xmath18 is the conductance quantum and @xmath19 .
the positive numbers @xmath20 are the @xmath21 non - zero eigenvalues of the matrix @xmath22 , with the transmission matrix @xmath23 being the @xmath24 submatrix of the full scattering matrix and consisting of transition amplitudes from @xmath12 left to @xmath13 right channels . for chaotic cavities , universal fluctuations of @xmath15 can be described by rmt @xcite .
considering the moments of @xmath0 and @xmath1 , the exact ( rmt ) results valid at arbitrary channel numbers @xmath2 and repulsion parameter @xmath25 were reported in the literature only for the average and variance of the conductance @xcite : @xmath26 @xmath27 and very recently for the average shot - noise power @xcite : @xmath28 to derive these results in a uniform way , it is convenient to use the known expression for the joint probability density of transmission eigenvalues @xmath15 @xcite @xmath29 to perform the corresponding integrations on eqs .
( [ g])([p ] ) .
the normalization constant @xmath30 above is given by @xmath31 and assures that ( [ jpd ] ) is a probability density .
it is known for discrete positive @xmath32 and continuous @xmath33 and @xmath25 as selberg s integral @xcite . as to the distribution functions
, simple closed expressions can be obtained for the conductance distribution at @xmath34 @xcite and for the shot - noise distribution at @xmath35 only @xcite . to the best of our knowledge ,
no general results valid at arbitrary @xmath2 and @xmath25 have been presented thus far .
to study the cumulants of @xmath0 and @xmath1 , one needs to know what are the moments @xmath36 , with @xmath37 standing for the integration over the joint probability density ( [ jpd ] ) and @xmath38 . moments with all @xmath39 as well as @xmath40 can be found from recursion relations already given in mehta s book @xcite that come from the selberg s integral theory .
the latter can be successfully applied @xcite to derive results ( [ < g>])([<p > ] ) .
this approach was recently extended further to find @xmath41 and @xmath42 exactly @xcite . here
, we have calculated all moments with @xmath43 by using tricks of partial integrations to reduce all moments to forms of selberg s integral .
we will report on that in more detail elsewhere @xcite . by this method
we are able to calculate the so - called skewness ( third cumulant ) of the conductance , which we represent in the following compact form : @xmath44 } { \beta(n-3 + \frac{2}{\beta } ) ( n-1+\frac{2}{\beta})(n-1+\frac{6}{\beta})}\,.\end{aligned}\ ] ] it is worth noting that the skewness vanishes for symmetric cavities ( @xmath45 ) at @xmath46 .
this also holds for any odd cumulant of the conductance , as the conductance distribution becomes symmetric around @xmath47 in this case . ) at @xmath48 by the change of all @xmath49 in @xmath50 . ]
we have also calculated the kurtosis @xmath51 of the conductance ( which is the fourth cumulant of @xmath0 ) and the variance @xmath52 of the shot - noise power ( the second cumulant ) .
these expressions are given explicitly but are too lengthy to be reported here . in the case of the single - mode leads , @xmath35 , one gets @xmath53 and @xmath54 at the values of @xmath55 , 2 , 4 , respectively . in the opposite semiclassical limit of large channel numbers , @xmath56 ,
we write the results as the following @xmath57 expansions : @xmath58 \\ & & + { \cal o } ( 1/n^4)\ , , \label{<g^4 > } \\ \nonumber \frac{\mathrm{var}(p)}{\langle p \rangle } & = & \frac{2}{\beta n^5 } \bigl [ n_1 ^ 4 + n_2 ^ 4 - 4 n_1 n_2 ( n_1-n_2)^2 \\ \nonumber & & + \frac{\beta-2}{\beta n } \bigl ( 9(n_1 ^ 4 + n_2 ^ 4 ) - 42n_1n_2(n_1 ^ 2 + n_2 ^ 2 ) + 70n_1 ^ 2 n_1 ^ 2 \bigr ) \bigr ] \\ & & + { \cal o } ( 1/n^3)\ , .
\label{varp}\end{aligned}\ ] ] one can readily see that higher cumulants contribute at least in the next order of @xmath59 , so that the full distributions get more gaussian - like as @xmath60 grows .
this tendency becomes even stronger for the conductance distribution in symmetric cavities at @xmath46 , as @xmath61 vanishes then in the leading and next - to - leading orders . in this case of symmetric cavities , @xmath62 , we can further find : @xmath63
finally , we discuss shortly the distribution of the conductance , @xmath64 and that of the shot - noise power , @xmath65 , deferring a detailed consideration to a separate publication @xcite . writing the functions @xmath66 and @xmath67 as fourier series over the interval of support , we obtain , e.g. , for the conductance distribution the following representation : @xmath68 where @xmath69 is known at @xmath3 .
for example , @xmath70 is given by the imaginary part of the determinant of the following matrix : @xmath71 in a similar way we may write @xmath72 and @xmath73 as imaginary parts of certain pfaffians .
the sum ( [ prob ] ) can be done numerically where care has to be taken for the pfaffians , which may equivalently be written as quaternion determinants of certain self - dual antisymmetric matrices , which are easier to be calculated .
analogous expressions are also obtained for @xmath74 . the conductance distribution at @xmath55 , @xmath75 and @xmath76 , 2 , 3 , 4 , 5 , 7 , 9 ( from left to right).,width=264 ] it is worth noting that at any finite @xmath32 these distributions are continuous but not analytic functions everywhere .
nonanalyticity in the distribution of conductances in quasi - one - dimensional wires ( the cusp point at @xmath77 ) was recently reported in @xcite , see also @xcite . in the present context of chaotic cavities , it can be understood from the following geometrical consideration . in the case of @xmath66 , calculating the average amounts to an integration over an ( @xmath78)dimensional plane @xmath79 cut by the @xmath32dimensional cube @xmath80 , such that there appear ( sometimes weak ) singularities at all points @xmath81 with an integer @xmath82 , @xmath83 .
a similar situation appears also for @xmath67 . here
, we integrate over the ( @xmath78)dimensional sphere @xmath84 and singularities appear whenever the sphere touches one edge or surface of the cube , that is for @xmath85 with @xmath83 .
figure [ fig1 ] illustrates the conductance distribution at @xmath55 keeping @xmath75 fixed and varying @xmath13 from 1 to 9 . in figure [ fig2 ] , we plot the distribution of the shot - noise power at @xmath86 , keeping @xmath87 fixed and varying @xmath13 from 2 to 14 . in both cases one can see the tendency of the distributions to get peaked around the mean values , so that in the bulk they can be effectively described by a gaussian with the known mean and variance , see eqs .
( [ < g > ] ) ( [ < p > ] ) and ( [ varp ] ) .
distributions of the shot - noise power at @xmath46 , @xmath75 and @xmath88 , 3 , 4 , 6 , 8 ( left plot ) and at @xmath46 , @xmath75 and @xmath89 , 10 , 12 , 14 ( right plot).,width=453 ]
we are able to give some exact asymptotics near the edges of support of the distributions expressed in terms of exact integrals , which are related to some general forms of selberg s integral .
for example , for @xmath4 we find @xmath90 where the proportionality factor is exactly known .
this result follows easily from calculating the powers of @xmath0 in @xmath91 under scaling of all @xmath92 and noticing that the upper integration limit of @xmath93 remains unchanged ( @xmath94 ) as long as @xmath95 .
similarly , @xmath96 behaves near the upper edge as follows : @xmath97 for the distribution of the shot - noise power , one finds @xmath98 whereas near the lower edge ( @xmath99 ) the asymptotics is a bit more complicated , since the contribution of the corners of the cube of integration become disconnected .
these expressions can be extended outside the regions specified above , being valid then in a linear order in distances from the edges .
the expansions can even be improved and all constants can be given explicitly as expressions containing products of gamma functions from selberg s integral .
the distribution of the shot - noise power ( dashed line ) and corresponding asymptotics ( full lines ) at @xmath100 and @xmath75 , @xmath88.,width=264 ] in figure [ fig3 ] , we show as an example the distribution of the shot - noise power at @xmath101 , @xmath75 , @xmath88 ( dashed line ) with calculated asymptotics near the edges ( full lines ) . indicated as dots are the average @xmath102 and @xmath103 known exactly . we have also compared ( and found a perfect agreement ) all the expressions with numerical simulations of the rmt statistics of @xmath15 and resulting distributions for @xmath0 and @xmath1 .
in summary , we have applied essentially the theory of selberg s integral to the problems of quantum transport in chaotic cavities .
the cumulants up to the forth order for the conductance and up to the second order for the shot - noise power have been calculated exactly at arbitrary channel numbers and repulsion parameter @xmath25 .
we have also given the conductance and shot - noise distributions in closed form suitable for subsequent analytic analysis ( e.g. , finding asymptotics ) as well as for numerical implementations .
it would be desirable to compare our findings with relevant experimental results , e.g. , in microwave cavities which became available recently @xcite . this could , however , be not so straightforward , as it requires taking into account effects of dephasing @xcite and absorption @xcite as well .
further work in this direction is needed . c. w. j. beenakker , rev .
phys . * 69 * , 731 ( 1997 ) .
m. blanter and m. bttiker , phys . rep . *
336 * , 1 ( 2000 ) .
h. u. baranger and p. a. mello , phys .
rev . lett . * 73 * , 142 ( 1994 ) .
r. a. jalabert , j .- l .
pichard , and c. w. j. beenakker , europhys . lett .
* 27 * , 255 ( 1994 ) .
d. v. savin and h .- j .
sommers , phys .
b * 73 * , 081307 ( 2006 ) .
m. l. mehta , random matrices , 2nd ed .
( academic press , new york , 1991 ) .
a. garca - martn and j. j. senz , phys .
lett . * 87 * , 116603 ( 2001 ) .
m. h. pedersen , s. a. van langen , and m. bttiker , phys .
b * 57 * , 1838 ( 1998 ) .
m. novaes , phys . rev .
b * 75 * , 073304 ( 2007 ) . d. v. savin , h .- j .
sommers and w. wieczorek , in preparation w. wieczorek , d. v. savin and h .- j .
sommers , in preparation k. a. muttalib , p. wlfle , a. garca - martn and v. a. gopar , europhys .
lett . * 61 * , 95 ( 2003 ) .
s. hemmady , j. hart , x. zheng , t. m. antonsen , e. ott and s. m. anlage , phys . rev .
b * 74 * , 195326 ( 2006 ) . p. w. brouwer and c. w. j. beenakker , phys .
b * 55 * , 4695 ( 1997 ) .
y. v. fyodorov , d. v. savin , h .- j .
sommers , j. phys .
a * 38 * , 10731 ( 2005 ) . | discuss conductance shot - noise power chaotic cavities selberg s integral static conductance @xmath7 relates averaged current @xmath8 external voltage @xmath9 between electron reservoirs fluctuations current described spectral noise power @xmath10 temperature zero noise non - vanishing electric charge carriers shot - noise mesoscopic conductors framework random - matrix - theory quantum transport @xcite chaotic cavity two leads @xmath12 @xmath13 channels coupled interior landauer - bttiker formalism conductance @xmath7 shot - noise power @xmath14 expressed transmission eigenvalues @xmath15 @xmath18 conductance quantum @xmath19 positive numbers @xmath20 @xmath21 non - zero eigenvalues matrix @xmath22 transmission matrix @xmath23 @xmath24 submatrix scattering matrix transition amplitudes @xmath12 to @xmath13 channels chaotic cavities fluctuations @xmath15 described by rmt @xcitemoments @xmath0 @xmath1 exact results at arbitrary channel numbers @xmath2 repulsion parameter @xmath25 reported for average variance conductance @xmath26 @xmath27 average shot - noise power derive results use expression joint probability density of transmission eigenvalues @xmath15 @xcite @xmath29 integrations eqs normalization constant @xmath30 given by @xmath31 probability density known for discrete positive @xmath32 continuous @xmath33 @xmath25 as selberg s integral @xcite distribution functions closed expressions for conductance distribution at @xmath34 @xcite shot - noise distribution @xmath35 no general results at arbitrary @xmath2 @xmath25 presented study cumulants of @xmath0 @xmath1 moments @xmath36 @xmath37 for integration over joint probability density @xmath38 moments with @xmath39 @xmath40 found from recursion relations in mehta s book selberg s integral theory applied @xcite results approach extended to find @xmath41 @xmath42 @xcite calculated all moments with @xmath43 partial integrations reduce moments to forms selberg s integral report detail @xcite method calculate skewness third cumulant conductance compact form : @xmath44 \beta(n-3 + \frac{2}{\beta } ) n-1+\frac{2}{\beta})(n-1+\frac{6}{\beta}).\end{aligned} skewness vanishes cavities @xmath45 ) at @xmath46 holds odd cumulant distribution symmetric around @xmath47 @xmath48 change @xmath49 in @xmath50 calculated kurtosis @xmath51 conductance fourth cumulant @xmath0 ) variance @xmath52 shot - noise power second cumulant expressions lengthy reported single - mode leads @xmath35 gets @xmath53 @xmath54 at values @xmath55 , 2 4 opposite semiclassical limit large channel numbers @xmath56 write results @xmath57 expansions @xmath58 \cal o 1/n^4)\label{<g^4\nonumber\frac{\mathrm{var\frac{2\beta n^5 n_1 ^ 4 + n_2 4 - n_1 n_2-n_2)^2\frac{\beta-2(n_1 ^ 4 + n_2 ^ 4 ) - 42n_1n_2(n_1 ^ 2 + n_2 ^ 2 ) 70n_1 ^ 2 n_1 ^ 2\cal o } ( 1/n^3){varp}{aligned} higher cumulants contribute next order @xmath59 full distributions gaussian - like @xmath60 grows tendency stronger conductance distribution symmetric cavities @xmath46 @xmath61 vanishes leading orders symmetric cavities @xmath62 find @xmath63 discuss distribution conductance @xmath64 shot - noise power @xmath65 separate publication writing functions @xmath66 @xmath67 fourier series interval support obtain conductance distribution representation @xmath68 @xmath69 known @xmath3 @xmath70 given by imaginary part of determinant matrix @xmath71 write @xmath72 @xmath73 as imaginary parts of pfaffians sum done numerically care for pfaffians written as quaternion determinants of self - dual antisymmetric matrices easier analogous expressions for @xmath74 conductance distribution at @xmath55 3 4 5 7 9 to @xmath32 distributions continuous not analytic nonanalyticity in distribution conductances in quasi - one - dimensional wires cusp point at @xmath77 reported in @xcite chaotic cavities understood from geometrical consideration @xmath66 calculating average integration over @xmath78)dimensional plane @xmath79 cut by cube @xmath80 singularities at all points @xmath81 @xmath82 similar for @xmath67 integrate over sphere @xmath84 singularities appear whenever sphere touches edge cube @xmath85 with @xmath83 figure conductance distribution at @xmath55 @xmath75 fixed varying @xmath13 from 1 to 9 distribution shot - noise power at @xmath86 @xmath87 fixed varying @xmath13 from 2 to 14both cases distributions around mean values described by gaussian known mean variance see eqs ( [ < g > ] [ < p > ] ) [ varp ] ) distributions shot - noise power at @xmath46 @xmath88 3 4 6 8 left 10 12 14 right exact asymptotics near edges support distributions exact integrals related to forms selberg s integral example @xmath4 @xmath90 proportionality factor known result calculating powers @xmath0 @xmath91 scaling @xmath92 upper integration limit @xmath93 remains unchanged @xmath94 ) @xmath95 @xmath96 behaves near upper edge @xmath97 distribution shot - noise power finds @xmath98 lower edge @xmath99 ) asymptotics complicated corners cube integration disconnected expressions extended outside regions valid linear order distances from edges expansions improved constants given as expressions products gamma functions from selberg s integral distribution shot - noise power dashed line asymptotics full lines at @xmath100 @xmath75 @xmath88=264 figure distribution shot - noise power at @xmath101 calculated asymptotics near edges lines dots average @xmath102 @xmath103 compared agreement expressions with simulations rmt statistics @xmath15 distributions @xmath0 @xmath1 applied theory selberg s integral to quantum transport in chaotic cavities cumulants forth order conductance second order shot - noise power calculated at arbitrary channel numbers repulsion parameter @xmath25 given conductance shot - noise distributions closed form for analytic analysis asymptotics numerical implementations compare findings with experimental results microwave cavities requires effects dephasing @xcite absorption further work needed w j. beenakker rev . . 69 *, 731 ( 1997 ) m. blanter m. bttiker phys . rep. 336, 1 ( 2000 ) h. u. baranger p. a. mello phys . rev . lett 73 , 142 ( 1994 ) r. a. jalabert l pichard w j beenakker europhys . lett 27 * , 255 ( 1994 ) d. v. savinsommers phys 73 081307 ( 2006 ) mehta random matrices 2nd ed academic press new 1991 ) garca - martn j j senz phys . 87 * 116603 ( 2001 ) pedersen. van langen m bttiker 57 1838 ( 1998 ) novaes phys 75 * 073304 ( 2007 ) d v. savin sommers w. wieczorek muttalib wlfle garca - martn v. gopar europhys . lett 61 95 ( 2003 ) s. hemmady j. hart x. zheng antonsen e ott m. anlage . b 74 195326 ( 2006 ) w. brouwer j beenakker phys 55 * 4695 ( 1997 ) v. fyodorov d v. savin sommers phys . 38 * 10731 ( 2005 ) | 0.501553 | 3,283 | 1,761 | , Saving $0.1 in GPT-4. |
177,419 | we prove that the set of growth rates of permutation classes includes an infinite sequence of intervals whose infimum is @xmath0 , and that it also contains every value at least @xmath1 .
these results improve on a theorem of vatter , who determined that there are permutation classes of every growth rate at least @xmath2 .
thus , we also refute his conjecture that the set of growth rates below @xmath3 is nowhere dense .
our proof is based upon an analysis of expansions of real numbers in non - integer bases , the study of which was initiated by rnyi in the 1950s . in particular , we prove two generalisations of a result of pedicini concerning expansions in which the digits are drawn from sets of allowed values . | we consider a permutation of length @xmath4 simply to be a sequence containing exactly one copy of each element of the set @xmath5 .
a permutation @xmath6 of length @xmath7 is said to be a _ subpermutation _ of another permutation @xmath8 , written @xmath9 , if @xmath8 has a subsequence of length @xmath7 whose terms occur in the same relative order as those of @xmath6 . for example , @xmath10 is a subpermutation of @xmath11 since @xmath12 has the same relative order as @xmath10 .
a _ permutation class _ is a downset of permutations under this subpermutation order ; thus if @xmath8 is a member of a permutation class @xmath13 and @xmath9 , then @xmath14 . if @xmath15 is a set of permutations , we use @xmath16 to denote the members of @xmath15 of length @xmath4 .
the _ growth rate _ of a permutation class @xmath13 is defined by the limit @xmath17{|\ccc_n|},\ ] ] if it exists .
there are no known examples of permutation classes for which this limit does not exist and it is widely believed ( @xcite conjecture 1.1 ) that every permutation class does indeed have a growth rate .
the proof by marcus & tardos @xcite of the stanley wilf conjecture implies that growth rates are finite except in the case of the class of all permutations .
what does the set of possible permutation class growth rates look like ?
the last few years have seen substantial progress on answering this question , with particular focus on significant phase transition values .
kaiser & klazar @xcite characterised all growth rates up to 2 , showing that the golden ratio @xmath18 ( the unique positive root of @xmath19 ) is the least growth rate greater than 1 .
indeed , they prove a stronger statement , known as the _ fibonacci dichotomy _ , that if @xmath20 is ever less than the @xmath4th fibonacci number , then @xmath20 is eventually polynomial ( see also @xcite ) .
kaiser & klazar also determine that 2 is the least limit point in the set of growth rates .
klazar @xcite considered the least growth rate admitting uncountably many permutation classes ( which he denoted @xmath21 ) and proved that @xmath21 is at least 2 and is no greater than approximately @xmath22 ( a root of a quintic ) .
vatter @xcite determined the exact value of @xmath21 to be the unique real root of @xmath23 ( approximately @xmath24 ) and completed the characterisation of all growth rates up to @xmath21 , proving that there are uncountably many permutation classes with growth rate @xmath21 and only countably many with growth rate less than @xmath21 .
the phase transition at @xmath21 also has enumerative ramifications : albert , rukuc & vatter @xcite have shown that every permutation class with growth rate less than @xmath21 has a rational generating function .
balogh , bollobs & morris @xcite extended kaiser & klazar s work to the more general setting of _ ordered graphs_. an ordered graph is a graph with a linear order on its vertices . to each permutation
we associate an ordered graph .
the ( ordered ) _ graph _ of a permutation @xmath8 of length @xmath4 has vertex set @xmath5 with an edge between vertices @xmath25 and @xmath26 if @xmath27 and @xmath28 .
see the figures below for illustrations .
we use @xmath29 to denote the graph of @xmath8 .
an _ induced ordered subgraph _ of an ordered graph @xmath30 is an induced subgraph of @xmath30 that inherits its vertex ordering .
a set of ordered graphs closed under taking induced ordered subgraphs is known as a _
hereditary class_. it is easy to see that @xmath9 if and only if @xmath31 is an induced ordered subgraph of @xmath29 .
thus each permutation class is isomorphic to a hereditary class of ordered graphs , and results can be transferred between the two domains .
balogh , bollobs & morris conjectured that the set of growth rates of hereditary classes of ordered graphs contains no limit points from above , and also that all such growth rates are integers or algebraic irrationals .
these conjectures were disproved by albert & linton @xcite who exhibited an uncountable _ perfect set _ ( a closed set every member of which is a limit point ) of growth rates of permutation classes and in turn conjectured that the set of growth rates includes some interval @xmath32 .
their conjecture was established by vatter @xcite who proved that there are permutation classes of every growth rate at least @xmath2 ( the unique real root of @xmath33 ) .
in addition , he conjectured that this value was optimal , the set of growth rates below @xmath3 being nowhere dense . by generalising vatter
s constructions , we now refute his conjecture by proving the following two results : in order to complete the characterisation of all permutation class growth rates , further investigation is required of the values between @xmath21 and @xmath34 .
this interval contains the first limit point from above ( which we denote @xmath35 ) and the first perfect set ( whose infimum we denote @xmath36 ) , as well as the first interval ( whose infimum we denote @xmath37 ) . in @xcite
, vatter showed that @xmath38 ( the unique real root of @xmath39 ) is the infimum of a perfect set in the set of growth rates and hence a limit point from above , and conjectured that this is the least such limit point ( in which case @xmath40 ) .
if @xmath41 denotes the set of all permutation class growth rates , then the current state of our knowledge can be summarised by the following table : the author is grateful to vince vatter for suggesting that it might be worthwhile investigating whether his conjecture concerning the behaviour below @xmath3 may in fact be false .
he would also like to thank vince , robert brignall and two referees for reading earlier drafts of this paper ; their feedback resulted in significant improvements to its presentation . | permutation length @xmath4 sequence containing one copy each element set @xmath5 permutation @xmath6 length @xmath7 subpermutation permutation @xmath8 @xmath9 if @xmath8 subsequence length @xmath7 terms same order @xmath6 @xmath10 subpermutation @xmath11 @xmath12 same order @xmath10 permutation class downset permutations under subpermutation order if @xmath8 member permutation class @xmath13 @xmath9 @xmath14 if @xmath15 set permutations use @xmath16 denote members @xmath15 length @xmath4 growth rate permutation class @xmath13 defined by limit @xmath17 no examples permutation classes limit every permutation class growth rate proof marcus & tardos @xcite stanley wilf conjecture implies growth rates finite except class all permutations possible permutation class growth rates ? progress answering focus phase transition values kaiser & klazar characterised growth rates up to 2 golden ratio @xmath18 positive root @xmath19 least growth rate greater than 1they prove fibonacci dichotomy if @xmath20 less than @xmath4th fibonacci number polynomial kaiser klazar determine 2 least limit point growth rates klazar considered least growth rate permutation classes denoted @xmath21 proved @xmath21 at least 2 no greater than @xmath22 root of quintic vatter determined @xmath21 root of @xmath23 @xmath24 characterisation growth rates up to @xmath21 uncountably many permutation classes with growth rate @xmath21 many with less than phase transition at @xmath21 enumerative ramifications albert rukuc vatter permutation class with growth rate less than @xmath21 has rational generating function balogh bollobs morris extended s work to ordered graphs linear order on vertices each permutation ordered graph of permutation @xmath8 @xmath4 vertex set @xmath5 edge between vertices @xmath25 and @xmath26 @xmath27 @xmath28 figures use @xmath29 denote graph of @xmath8 induced ordered subgraph of graph @xmath30 inherits vertex orderingset of ordered graphs closed under induced subgraphs hereditary class @xmath9 @xmath31 is induced subgraph of @xmath29 each permutation class isomorphic to hereditary class results transferred between domains balogh bollobs morris conjectured growth rates of hereditary classes no limit points all are integers or algebraic irrationals disproved by albert & linton @xcite exhibited uncountable perfect set limit point of growth rates conjectured includes interval @xmath32 conjecture established by vatter @xcite proved permutation classes of every growth rate at least @xmath2 ( unique root of @xmath33 ) conjectured value optimal growth rates below @xmath3 dense vatter s refute conjecture by proving two results permutation class growth rates further investigation of values between @xmath21 and @xmath34 interval contains first limit point @xmath35 first perfect set @xmath36 first interval @xmath37 ) vatter showed @xmath38 root of @xmath39 ) is infimum of perfect set limit point from above conjectured least limit point @xmath40 ) @xmath41 denotes permutation class growth rates current knowledge summarised by table author grateful to vince vatter for suggesting investigating conjecture behaviour below @xmath3 false vince robert brignall two referees for reading earlier drafts paper feedback resulted improvements presentation | 0.259686 | 1,394 | 730 | , Saving $0.0 in GPT-4. |
177,424 | in this fluid dynamics video we show capillary rise experiments with diethyl ether in short tubes . the height of each short tube is less than the maximum height the liquid can achieve , and therefore the liquid reaches the top of the tube while still rising . over a narrow range of heights ,
the ether bulges out from the top of the tube and spreads onto the external wall . | the rise of diethyl ether ( density @xmath0 , dynamic viscosity @xmath1 , and surface tension @xmath2 ) under normal gravity in 0.8 to 1.0 mm diameter cylindrical capillary tubes is oscillatory due to its low viscosity and therefore delayed development of the viscous boundary layer@xcite . the liquid achieves a maximum height of 13.5 mm in a 0.8 mm diameter tube that is 16 mm high . when the same experiment is performed in a tube that is only 4.7 mm high , the rising liquid reaches the top of the tube , the meniscus inverts , and a bulge of liquid forms and wets the outer surface .
as ether spreads down the outer surface , the ether inside the capillary tube retreats downward and then oscillates vertically .
after inverting several times , the meniscus comes to a rest , pinned at the top of the tube .
external wetting occurs when the kinetic energy of the liquid exceeds a reference surface energy . in this case
, a reasonable choice is the surface energy of a spherical drop that has the same radius as the capillary tube .
if the kinetic energy is too low , the liquid bulges outward , does not spread , and returns into the tube and oscillates .
the kinetic energy is insufficient when the height of the liquid column or the rise speed is too low .
we show that the ether does not spread onto the outer surface when the tube is 2.7 mm high ( too short ) or 11.5 mm high ( too slow ) .
wetting of the external surface also happens with capillary tubes that have a square cross - section ( 0.8 mm inner side width ) .
we highlight the different stages of the capillary rise phenomenon starting with the surface waves that are produced when the tube touches the ether and ending with the oscillations of the meniscus .
9ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) | rise of diethyl ether ( density viscosity surface tension ) under normal gravity in 0.8 to 1.0 mm diameter capillary tubes oscillatory due to low viscosity development viscous boundary layer liquid achieves maximum height 13.5 mm in 0.8 mm tube 16 mm high in tube 4.7 mm high rising liquid reaches top meniscus inverts bulge liquid forms wets outer surface ether spreads retreats downward oscillates vertically after meniscus rest top external wetting occurs when kinetic energy liquid exceeds reference surface energy drop same radius as tube if kinetic energy too low liquid bulges outward returns oscillates kinetic energy insufficient when height liquid column or rise speed too low ether spread outer surface when tube 2.7 mm high short or 11.5 mm high ( too slow wetting external surface happens with capillary tubes square cross - section ( 0.8 mm width ) stages of capillary rise with surface waves tube ether oscillations of meniscus9ifxundefined 1 ] ifx#1 ifnum 1 ] 1firstoftwo secondoftwo secondoftwo 0]secondoftwosanitize 0 ] 12$12 12#1212_12%12[0@innerbibempty * | 0.127478 | 517 | 258 | , Saving $0.0 in GPT-4. |
177,425 | we report the first observation of the doubly - radiative decay mode of pionic hydrogen ( @xmath0 ) and deuterium ( @xmath1 ) using the rmc pair spectrometer at triumf @xcite .
the process is interesting in the context of the @xmath2-compton scattering @xcite and the @xmath3pt predictions of the pion polarizability @xcite .
we present our preliminary values for the b.r.@xmath4}=3.8\times10^{-5}$ ] and b.r.@xmath5}=1.6\times10^{-5}$ ] .
our @xmath6 data indicates the dominance of @xmath7 mechanism .
the @xmath8 data shows no evidence for the @xmath9 dibaryon . | the double radiative decay of pionic atoms @xmath10 was investigated theoretically by ericson and wilkin @xcite a quarter of a century ago .
they predicted that the dominating reaction mechanism was the annihilation of the stopped , real @xmath11 with a soft , virtual @xmath12 , _ i.e. _ @xmath7 .
as they advertised , the annihilation mechanism affords a selective coupling to soft pions , and special sensitivity to renormalization effects in nuclear matter . in order to understand the @xmath13 decay mode of a pionic atom ,
it is necessary to understand the elementary process of @xmath14 .
several authors have made tree
level calculations of the @xmath15 capture mode of pionic hydrogen .
the most recent published calculations are due to beder @xcite , who obtained 5.1@xmath1610@xmath17 for the @xmath15 branching ratio .
beder also pointed out the importance of the pion annihilation diagram , especially at small photon opening angles ( see fig .
[ fig : rmc_theory]a ) . the first attempt to observe the two photon emission in hydrogen pion capture
was made by vasilevsky et.al.@xcite at jinr dubna .
they obtained an upper limit of the @xmath15 branching ratio , b.r .
@xmath18 , i.e. a value ten times larger than the theoretical prediction of ref .
@xcite .
the predicted dominance of the @xmath7 annihilation amplitude in @xmath14 allows to study pion compton scattering @xcite . the feynman diagram in fig.[fig : rmc_theory]b can be viewed as the annihilation of a real pion with a virtual pion @xmath19 or , via crossing symmetry , as the transition of a real pion to a virtual pion via compton scattering @xmath20 ( fig.[fig : rmc_theory]c ) .
[ cols= " < , < , < " , ] [ tab : rmc_pionpol ] the extracted pion polarizabilities have large uncertainties and are ( except mark ii ) substantially larger than the @xmath21 prediction @xcite , @xmath22 , which is based on a robust relationship between radiative pion decay @xmath23 and pion compton scattering @xmath24 @xmath25 @xmath24 .
this discrepancy between theory and experiment obviously calls for more experimental attention .
recently gerasimov@xcite , pointed out , that the double radiative decay of the pionic deuterium ( @xmath26 ) can be used as a means to investigate the existence of the @xmath27 dibaryon ( @xmath28 kev ) that was claimed by the dib-2gamma collaboration@xcite .
as pointed out by gerasimov the @xmath27 does indeed exist , one should be able to detect it via the following chain : @xmath29 based on a toy model , gerasimov estimated that the resonant decay of pionic deuterium via @xmath27 should occur roughly 100 times more frequently than non - resonant decay of @xmath30 .
additionally , a clear signature of the hypothetical @xmath9 would be the photon spectrum with two lines : one at @xmath31 mev ( @xmath32 ) and the other at @xmath33 ( @xmath34 ) .
the experiments presented here were performed at triumf laboratory using the rmc pair spectrometer @xcite on the m9a beam line .
the pionic hydrogen data ( e838 ) were taken in december 1998 and april - may 1999 and the pionic deuterium data ( e864 ) were taken in april 2000 .
an 82 mev / c pion beam was stopped in a liquid hydrogen ( deuterium ) target ( approximate dimensions : length 15 cm , @xmath3616 cm ) , the photons were converted into @xmath37 pairs in a 1 mm thick pb cylinder .
the @xmath38 trajectories were then measured in a set of cylindrical wire and drift chambers , and the magnetic field provided momentum analysis .
trigger scintillators hit patterns were used to identify events .
we collected approximately 1000 @xmath14 events and 500 @xmath26 events .
a cut on the @xmath13 opening angle was used to remove @xmath39 events and a cut on the beam / trigger scintillators was used to remove accidental @xmath13 events .
[ fig : rmc_results ] shows a typical event and the preliminary opening angle distributions for the pionic hydrogen and deuterium .
our preliminary branching ratios are b.r.@xmath4}=3.8\times10^{-5}$ ] and b.r.@xmath5}=1.6\times10^{-5}$ ] . the pionic hydrogen b.r .
is in approximate agreement with beder calculations @xcite and the yield at small opening angles demonstrates the dominance of the pion annihilation graph .
we see no evidence for the the @xmath27 in our deuterium data . in order to compare our results to the existing carbon data
@xcite we quote the relative branching ratios : b.r.@xmath40 for the pionic hydrogen and b.r.@xmath41 for the pionic deuterium .
our preliminary results indicate that the the @xmath42 graph is indeed dominant in the @xmath43 reaction hence this reaction is a potential probe for pion polarizability . we have found no evidence for the @xmath27 in the decay of pionic deuterium .
we plan to extend our studies to nuclear targets . | double radiative decay of pionic atoms investigated by ericson wilkin century ago predicted dominating reaction mechanism annihilation of @xmath11 with virtual @xmath12 annihilation mechanism affords selective coupling to soft pions sensitivity to renormalization effects nuclear matter understand @xmath13 decay understand elementary process of @xmath14 authors made tree level calculations @xmath15 capture mode pionic hydrogen recent calculations beder obtained 5.1@xmath1610@xmath17 for @xmath15 branching ratio importance of pion annihilation diagram small photon opening angles first attempt two photon emission in hydrogen pion capture by vasilevsky et obtained upper limit of @xmath15 branching ratio value ten times larger than theoretical prediction dominance @xmath7 annihilation amplitude in @xmath14 pion compton scattering feynman diagram in viewed as annihilation of real pion with virtual pion @xmath19 or transition of real to virtual pion via compton scatteringcols tab rmc_pionpol extracted pion polarizabilities have large uncertainties mark ii larger than @xmath21 prediction based relationship between radiative pion decay pion compton scattering discrepancy theory experiment calls for experimental attention gerasimov double radiative decay of pionic deuterium @xmath26 investigate @xmath27 dibaryon claimed by dib-2gamma collaboration @xmath27 detect via chain @xmath29 resonant decay of pionic deuterium via @xmath27 100 times more frequently than non - resonant decay of @xmath30 signature of hypothetical @xmath9 photon spectrum with two lines at @xmath31 @xmath33 experiments performed laboratory using rmc pair spectrometer @xcite m9a beam line pionic hydrogen data taken december 1998 april - may 1999 pionic deuterium data e864 april 2000 82 mev / c pion beam stopped in liquid hydrogen ( deuterium ) target length 15 cm @xmath3616 cm photons converted into @xmath37 pairs in 1 mm thick pb cylinder@xmath38 trajectories measured in cylindrical wire drift chambers magnetic field provided momentum analysis trigger scintillators events collected 1000 @xmath14 500 @xmath26 cut @xmath13 opening angle @xmath39 events trigger scintillators accidental @xmath13 events fig rmc_results shows typical event preliminary opening angle distributions pionic hydrogen deuterium branching ratios@xmath4}=3.8@xmath5}=1.6 pionic hydrogen with beder calculations yield at small opening angles dominance pion annihilation graph no evidence @xmath27 in deuterium data results carbon data relative branching ratios@xmath40 pionic hydrogen@xmath41 pionic deuterium results @xmath42 graph dominant in @xmath43 reaction potential probe for pion polarizability no evidence @xmath27 in decay pionic deuterium plan extend studies to nuclear targets | 0.184557 | 1,227 | 614 | , Saving $0.0 in GPT-4. |
177,426 | we present the results of a study of stellar population properties at large galactocentric radii of 14 low - mass early - type galaxies in the fornax and virgo clusters .
we derive radial profiles of age , total metallicity [ z / h ] , and [ @xmath0/fe ] abundance ratios out to @xmath1 effective radii by using nearly all of the lick / ids absorption - line indices in comparison to recent single stellar population models .
we extend our study to higher galaxy mass via a novel literature compilation of 37 early - type galaxies , which provides stellar population properties out to one effective radius .
we find that metallicity gradients correlate with galactic mass , and the relationship shows a sharp change in slope at a dynamical mass of @xmath2 .
the central and mean values of the stellar population parameters ( measured in @xmath3 , and at @xmath4 , respectively ) define positive mass trends .
we suggest that the low metallicities , almost solar [ @xmath0/fe ] ratios and the tight mass - metallicity gradient relation displayed by the low - mass galaxies are indicative of an early star - forming collapse with extended ( i.e. , @xmath5 gyr ) , low efficiency star formation , and mass - dependent galactic outflows of metal - enriched gas .
the flattening of metallicity gradients in high - mass galaxies , and the broad scatter of the relationship are attributed to merger events .
the high metallicities and supersolar abundances shown by these galaxies imply a rapid , high efficiency star formation .
the observed [ z / h]mass and [ @xmath0/fe]mass relationships can be interpreted as a natural outcome of an early star - forming collapse .
however , we find that hierarchical galaxy formation models implementing mass - dependent star formation efficiency , varying imf , energy feedback via agn , and the effects due to merger - induced starbursts can also reproduce both our observed relationships . [ firstpage ] galaxies : abundances galaxies : dwarf galaxies : elliptical and lenticular , cd galaxies : formation galaxies : evolution galaxies : stellar content | the spatial distribution of stellar population properties in early - type galaxies are the chemodynamical fossil imprints of galaxy formation and evolution mechanisms .
the efficiency of different physical processes is expected to vary with radius leaving measurable changes at different galactocentric radii .
thereby , providing unique constraints on the galaxy formation mechanisms predicted by competing scenarios . in the past years an increasing number of observational studies have focused their effort in probing stellar populations beyond the galactic cores ( e.g. , @xcite ; @xcite ; @xcite ) .
however , these previous studies lack to consider low - mass , low - luminosity galaxies because the intrinsic low surface brightness of these objects makes the measurement of their properties at large galactocentric radii observationally challenging .
the aim of this paper is to provide a clear picture of the stellar population properties at large radii , and their relationship to galaxy structural parameters over a comprehensive galaxy mass range . in the classical models of monolithic collapse ( @xcite ;
@xcite ; @xcite ; @xcite ; @xcite ) , early - type galaxies form during dissipational infall of cooling gas .
massive galaxies are predicted to form in a short and intense burst of star formation , whereas dwarf galaxies experience more extended and intermittent star formation ( @xcite ; @xcite ; @xcite ) .
the gas flowing to the centre of the galaxy is chemically enriched by evolving stars , and contributes as metal - rich fuel for star formation .
thus , a negative radial metallicity gradient is established .
metallicity - dependent gas cooling and a time - delay in the occurrence of galactic winds ( i.e. , onsets that vary with the local escape velocity ) can cooperate in steepening any metallicity gradients ( @xcite ; @xcite ) . a strong correlation between metallicity gradient and galaxy mass is expected ( @xcite ; @xcite ) .
[ @xmath0/fe ] abundance gradients are suggested to be a consequence of interplay between local differences in the star formation timescale and gas flows ( @xcite ; @xcite ) . in the hierarchical merging scenario ( e.g. , @xcite ;
@xcite ; @xcite ) , early - type galaxies form by merging events . the properties of merger remnants are related to the progenitors mass ratio , amount of gas content and the geometry of the collision ( @xcite ; @xcite ) .
mergers are expected to flatten any ambient gradients ( e.g. , @xcite ) .
central bursts of star formation induced by the merger can partially regenerate the metallicity gradients and create age gradients ( e.g. , @xcite ) . a weak relation between mass and metallicity gradients
is expected ( @xcite ) .
positive or negative [ @xmath0/fe ] gradients can form depending on the chemical enrichment of the pristine gas and the duration of the burst ( @xcite ; @xcite ) .
recent models include energy feedback due to active galactic nuclei ( agn ) to stop the star formation in massive galaxies , thus producing a mass - dependent star formation history ( @xcite ; @xcite ; @xcite ; @xcite ) . here
we further consolidate the findings of @xcite ( hereafter paper i ) on kinematic and photometric properties at large galactocentric radii of 14 low - mass early - type galaxies in the virgo and fornax clusters .
we study the spatial distribution of luminosity@xmath6weighted age , total metallicity [ z / h ] and @xmath0-element abundance ratio [ @xmath0/fe ] of out to @xmath7 by using nearly all of the lick / ids indices .
the comparison of these line - strength indices with single stellar population models ( ssps ) is found to be successful ( e.g. , @xcite ; @xcite ; @xcite ) in breaking the well known age / metallicity degeneracy ( @xcite ) .
we extended our study to a broad galactic mass range , introducing a novel literature compilation of stellar population properties out to one effective radius for a sample of 37 early - type galaxies of higher mass .
we compare properties measured at different galactocentric radii encompassing different percentages of the total stellar mass .
the implications for mass scaling relations are considered . by comparing our results to model predictions
we constrain the mechanisms for the formation and evolution of early - type galaxies .
this paper is organised as follows . in section 2
, we describe the data sample . in section 3
, we describe our spectroscopic observations and data reduction . in section 4 , we present the stellar population analysis . in section 5 , we introduce the radial profiles for the 14 low - mass galaxies . in section 6 , we describe the stellar population properties of the entire galaxy sample . in section 7
, we discuss our results in the context of predictions from competing galaxy formation scenarios .
section 8 presents a summary of our findings .
our sample consists of 14 low - mass early - type galaxies in the virgo and fornax clusters .
the six galaxies in the fornax cluster are chosen from the catalogue of @xcite , while the eight galaxies in the virgo cluster are from the catalogue of @xcite .
galaxies of low - mass are selected on the basis on their central stellar velocity dispersion @xmath8 and the absolute @xmath9-band total magnitude @xmath10 , both of which are known to be independent proxies of galaxy mass .
the sample uniformly covers the range @xmath11 and @xmath12 .
this translates to a dynamical mass range of about @xmath13 , using the relation @xmath14m@xmath15 where @xmath16 is the effective radius in parsecs ( @xcite ) .
the main properties of the galaxy sample are summarised in table [ gal_prop ] . | spatial distribution of stellar population properties in early - type galaxies are chemodynamical fossil imprints of galaxy formation evolution mechanisms efficiency physical processes with radius changes at different galactocentric radii constraints on galaxy formation mechanisms observational studies stellar populations beyond galactic cores @xcite studies low - mass low - luminosity galaxies low surface brightness measurement at large galactocentric radii challenging aim paper clear picture stellar population properties at large radii relationship to galaxy structural parameters over galaxy mass range classical models of monolithic collapse early - type galaxies form during dissipational infall of cooling gas massive galaxies form in short intense star formation dwarf galaxies experience extended intermittent star formation gas to centre galaxy chemically enriched by evolving stars contributes metal - rich fuel for star formation negative radial metallicity gradient established metallicity - dependent gas cooling time - delay in galactic winds metallicity gradients strong correlation between metallicity gradient and galaxy mass expected@xmath0/fe ] abundance gradients interplay local star formation timescale gas flows hierarchical merging scenario early - type galaxies form by merging events properties of merger remnants related to progenitors mass ratio gas content geometry collision mergers flatten ambient gradients central bursts star formation regenerate metallicity create age gradients weak relation between mass metallicity gradients expected positive or negative [ @xmath0/fe ] gradients form on chemical enrichment pristine gas duration burst models include energy feedback due active galactic nuclei stop star formation in massive galaxies mass - dependent star formation history consolidate findings of @xcite on kinematic photometric properties at large galactocentric radii of 14 low - mass early - type galaxies in virgo fornax clusters study spatial distribution of luminosity@xmath6weighted age total metallicity @xmath0-element abundance ratio/fe of to @xmath7 using lick / ids indices comparison of line - strength indices with single stellar population models successful@xcite age / metallicity degeneracy extended study broad galactic mass range novel literature compilation stellar population properties radius 37 early - type galaxies higher mass properties different galactocentric radii percentages total stellar mass implications mass scaling comparing results to model predictions constrain mechanisms formation evolution early - type galaxies paper organised section 2 data sample section 3 spectroscopic observations data reduction 4 stellar population analysis section 5 radial profiles for 14 low - mass galaxies section 6 stellar population properties sample section 7 results predictions galaxy formation scenarios section 8 summary findings sample 14 low - mass early - type galaxies virgo fornax clusters six galaxies fornax cluster from @xcite eight virgo cluster from @xcite galaxies selected central stellar velocity dispersion @xmath8 absolute @xmath9-band total magnitude @xmath10 independent proxies galaxy mass sample covers range @xmath11 and @xmath12 dynamical mass range about @xmath13 relation @xmath14m@xmath15 @xmath16 effective radius in parsecs @xcite main properties summarised in table [ gal_prop ] | 0.188366 | 1,244 | 594 | , Saving $0.0 in GPT-4. |
177,455 | we have studied the influences of isospin symmetry breaking of nucleon - nucleon interaction on the various properties of asymmetrical nuclear matter and @xmath0-stable matter . for asymmetrical nuclear matter , it is found that by including this isospin symmetry breaking , the changes of bulk properties increase by increasing both density and asymmetry parameter
. however , these effects on the total energy and equation of state of @xmath0-stable matter are ignorable . for asymmetrical nuclear matter
, the validity of the empirical parabolic law in the isospin symmetry breaking case is shown .
it is observed that the isospin symmetry breaking of nucleon - nucleon interaction affects the @xmath0-equilibrium conditions in @xmath0-stable matter .
= 16truecm = 22.5truecm = -1.4truecm = -1.5truecm * key words : * nucleon - nucleon interaction ; isospin symmetry breaking ; nuclear matter ;
@xmath0-stable matter . | in the calculation of the properties of nucleonic matter , the nucleon - nucleon potential has a crucial role .
recent models of two - nucleon potentials which fit the nucleon - nucleon scattering data with high precision , contain terms which break the isospin symmetry ( stoks et al . , 1994 ; machleidt and slaus , 2001 ; li and machleidt , 1998 ) .
isospin symmetry is the invariance under any rotation in the isospin space .
but , due to the mass difference between neutral pion and charged pion ( @xmath1 ) , this symmetry is broken ( stoks et al . , 1994 ; machleidt and slaus , 2001 ; li and machleidt , 1998 ) .
this means that for the new two - nucleon potentials , the neutral pion exchange is distinguished from charged pion exchange .
this distiction implies that the neutron - neutron ( _ nn _ ) , neutron - proton ( _ np _ ) and proton - proton ( _ pp _ ) interactions are different and therefore , these potentials are called isospin symmetry breaking ( * isb * ) potentials ( stoks et al . , 1994 ) . due to the distinction between the neutral pion and charged pion exchange , for the potentials which describe the long range part of the nucleon - nucleon interaction in terms of the one - pion exchange model ,
the one - pion exchange term is one of the origins of * isb * ( stoks et al . , 1994 ; machleidt and slaus , 2001 ; li and machleidt , 1998 ) . for the * isb * potentials , in the isospin @xmath2 states ,
the distinction between the _ nn _ , _ np _ and _ pp _ potentials is necessary ( stoks et al . , 1994 ) . therefore , in the calculations with these potentials , the difference between the _ nn _ , _ np _ and _ pp _ wave functions should be considered . in our calculations
, we use the _ reid-93 _ potential ( stoks et al . , 1994 ) .
this potential describes the longe range part of the nucleon - nucleon interaction in terms of the one - pion exchange model accounting for the mass difference between the neutral pion @xmath3 and charged pions @xmath4 and @xmath5 . in the _ reid-93 _
potential , for the non - one - pion exchange parts , additional * isb * terms have been included in the @xmath6 state potential ( stoks et al . , 1994 ) .
in this work , we intent to calculate the various properties of asymmetrical nuclear matter and @xmath0-stable matter in two different cases with isospin symmetry ( * is * ) and with isospin symmetry breaking ( * isb * ) . by comparing the results of the * is * and * isb * cases , we study the influence of * isb * of nucleon - nucleon interaction on these properties . for the * isb * case
, we use the _ reid-93 _ potential ( stoks et al . , 1994 ) in which the one - pion exchange potentials for the _ np _ and _ pp _ ( _ nn _ ) interactions as well as the _ np _ and _ pp _ ( _ nn _ )
@xmath6 potentials are different . however , for the * is * case , we replace the one - pion exchange and @xmath6 potentials of _ nn _ and _ pp _ interactions with the corresponding _ np _ interaction ( stoks et al . , 1994 ) .
in our calculations , we employ the formalism used in our previous works ( bordbar and modarres , 1997 , 1998 ; modarres and bordbar , 1998 ; bordbar and riazi , 2001 , 2002 ; bordbar , 2002a , b ) in which we consider the distiction between the partial waves with @xmath7 , @xmath8 and @xmath9 .
this method is a variational approach based on the cluster expansion of the energy functional containing the state dependent correlation functions .
the convergence of the results of this approach has been also tested by calculating the three - body cluster term ( bordbar and modarres , 1997 ) .
the asymmetrical nuclear matter properties are important for studying the supernova collapse and heavy - ion reaction ( lattimer and prakash , 2000 ; li et al . , 1997 ) . for asymmetrical nuclear matter
, we define the asymmetry parameter @xmath0 as @xmath10 where @xmath11 and @xmath12 are the number densities of neutrons and protons respectively , and @xmath13 is the total number density .
this implies that for symmetrical nuclear matter ( @xmath14 ) @xmath15 and for the pure neutron matter @xmath16 . in this section
, we present our results for the various properties of asymmetrical nuclear matter and then we investigate the effects of * isb * on these properties .
the total energy of asymmetrical nuclear matter as a function of total number density for various asymmetry parametrs is given in figure [ fig1 ] .
it is seen that for both * is * and * isb * cases as the asymmetry parameter increases , the saturation point shifts to the lower densities .
however , for the large values of @xmath0 , the energy curve does not shows any minimum .
the changes of total energy due to the inclusion of * isb * for nucleon - nucleon interaction are shown in figure [ fig2 ] .
this figure shows that the differences between our results in the * isb * and * is * cases increase by increasing both density and asymmetry parameter . in table
[ tab1 ] , our results for the differences between the * isb * and * is * energies at @xmath17 are compared with those of mpm ( muther et al . , 1999 ) .
it can be seen that our results are in a good agreement with the results of mpm for the _ nijm - ii _ potential .
the @xmath0-dependence of the total energy of asymmetrical nuclear matter in the * isb * case is shown in figure [ fig3 ] for different values of density .
it can be seen that even in the presence of * isb * , the asymmetrical nuclear matter energy is nearly linear in @xmath18 and therefore , the higher orders effects of the asymmetry are very small .
this confirms the validity of the empirical parabolic law . in figure [ fig4 ] , we have shown the effects of * isb * on the potential energy of asymmetrical nuclear matter .
it is seen that the potential energy difference between the * is * and * isb * cases increases by increasing asymmetry parameter @xmath0 .
this is due to the fact that by increasing @xmath0 , the energy contribution of _ nn _ interaction increases . therefore , the difference between the case in which we distinguish between the _ nn _ and _ np _ interactions ( * isb * case ) and the case in which we replace the _ nn _
interaction by the corresponding _ np _ interaction ( * is * case ) increases by increasing @xmath0 .
it is also seen that for each value of density and asymmetry parameter , the potential energy shift due to the * isb * is positive .
this behaviour is reflecting the fact that in the partial wave with isospin @xmath2 , the @xmath19 potential is more attractive than the corresponding @xmath20 and @xmath21 potentials ( bordbar , 2003 ) .
the influence of * isb * on the saturation point properties of asymmetrical nuclear matter is shown in table [ tab2 ] .
we see that the * isb * effects on the saturation density , energy and incompressibility of asymmetrical nuclear matter are generally small , but they are considerable for large @xmath0 .
the nuclear matter asymmetry energy is of particular interest in astrophysics , especially for studying the supernova mechanism ( lattimer and prakash , 2000 ) .
the asymmetry energy can be approximated as the difference between binding energies of pure neutron matter ( @xmath16 ) and symmetrical nuclear matter ( @xmath15 ) .
the validity of using this approximation ( called parabolic approximation ) in calculating the asymmetry energy was shown in figure [ fig3 ] .
our results for the asymmetry energy calculated for both * is * and * isb * cases are presented in figure [ fig5 ] .
it can be seen that the changes of asymmetry energy due to the inclusion of * isb * increase by inceasing density .
the properties of @xmath0-stable matter are of special importance in astrophysics .
the equation of state of @xmath0-stable matter is very important for investigating the neutron stars , particularly in their stability ( akmal et al . , 1998 ; heiselberg and pandharipande , 2000 ) .
the @xmath0-stable matter is a composition of nucleons and leptons which is electrically neutral and in @xmath0-equilibrium .
the @xmath0-equilibrium means that there is a balance between the competing processes of electron capture @xmath22 and @xmath0-decay @xmath23 .
this equilibrium conditions determine the proton fraction ( @xmath24 ) in @xmath0-stable matter .
the proton fraction is important for studing the cooling of a neutron star ( prakash , 1994 ) . in figure
[ fig6 ] , we have plotted our results for the total energy of @xmath0-stable matter versus density in the * is * and * isb * cases .
it is seen that by considering * isb * for the nucleon - nucleon interaction , the change of @xmath0-stable matter energy is small .
for example , at density @xmath25 , the energy correction due to the * isb * effects is about @xmath26 .
therefore , the effects of * isb * on the total energy of @xmath0-stable matter are negligible . the calculated energy contribution of leptons and proton fraction in @xmath0-stable matter are shown in figures [ fig7 ] and [ fig8 ] .
it is found that these properties of @xmath0-stable matter are varied by the inclusion of * isb * of nucleon - nucleon interaction .
this shows that the * isb * of nucleon - nucleon interaction affects the conditions of @xmath0-equilibrium .
the equation of state of @xmath0-stable matter calculated for both * is * and * isb * cases are shown in figure [ fig9 ] .
it can be seen that the influences of * isb * on the equation of state of @xmath0-stable matter are ignorable .
the two - nucleon potential has a crucial role in the nuclear many - body calculations . by comparing the results of two different cases of isospin symmetry ( *
is * ) and isospin symmetry breaking ( * isb * ) for the nucleon - nucleon interaction , we have evaluated the effects of * isb * on the different properties of asymmetrical nuclear matter and @xmath0-stable matter .
it was seen that the changes of total energy , potential energy and asymmetry energy of asymmetrical nuclear matter due to the inclusion of * isb * for the nucleon - nucleon interaction increase by increasing both density and asymmetry parameter .
it was shown that the empirical parabolic law is valid , even by considering * isb * for the nucleon - nucleon interaction .
it was indicated that our results for the total energy and the equation of state of @xmath0-stable matter in the * is * and * isb * cases are nearly identical . for @xmath0-stable matter
, it was shown that the * isb * affects the energy contribution of leptons and proton fraction .
financial support from shiraz university research council and ipm is gratefully acknowledged .
99 akmal , a. , pandharipande , v. r. and ravenhall , d. g. ( 1998 ) .
_ physical review _ * c58 * , 1804 .
+ bordbar , g. h. and modarres , m. ( 1997 ) . _ journal of physics g : nuclear and particle physics _
* 23 * , 1631 .
+ bordbar , g. h. and modarres , m. ( 1998 ) .
_ physical review _ * c 57 * , 714 .
+ bordbar , g. h. and riazi , n. ( 2001 ) .
_ international journal of theoretical physics _ * 40 * , 1671 .
+ bordbar , g. h. and riazi , n. ( 2002 ) .
_ astrophysics and space science _ * 282 * , 563 .
+ bordbar , g. h. ( 2002a ) .
_ international journal of theoretical physics _ * 41 * , 309 .
+ bordbar , g. h. ( 2002b ) .
_ international journal of theoretical physics _ * 41 * , 1135 .
+ bordbar , g. h. ( 2003 ) .
_ international journal of modern physics _ * a18 * , 2629 .
+ heiselberg , h. and pandharipande , v. r. ( 2000 ) .
_ annual review of nuclear and particle science _ , * 50 * , 481 .
+ lattimer , j. m. and prakash , m. ( 2000 ) .
_ physics report _ * 333 * , 121 .
+ li , b. a. , ko , c. m. and ren , z.(1997 ) .
_ physical review letter _ * 78 * , 1644 .
+ li , g. q. and machleidt , r. ( 1998 ) .
_ physical review _ * c58 * , 3135 .
+ machleidt , r. and slaus , i. ( 2001 ) . _ journal of physics g : nuclear and particle physics _
* 27 * , r69 .
+ modarres , m. and bordbar , g. h. ( 1998 ) .
_ physical review _ * c 58 * , 2781 .
+ muther , h. , polls , a. and machleidt , r. ( 1999 ) . _ physics letter _ * b455 * , 259 .
+ prakash , m. ( 1994 ) .
_ physics report _ * 242 * , 191 .
+ stoks , v. g. j. , klomp , r. a. m. , terheggen , c. p. f. and de swart , j. j. ( 1994 ) .
_ physical review _ * c49 * , 2950 . | calculation properties nucleonic matter nucleon - nucleon potential crucial recent models two - nucleon potentials nucleon scattering data contain terms break isospin symmetry ( stoks 1994 isospin symmetry invariance under rotation isospin space due to mass difference between neutral pion and charged pion symmetry broken new two - nucleon potentials neutral pion exchange distinguished from charged pion exchange distiction implies neutron - neutron nn neutron - proton _ np _ proton - proton ( _ pp _ ) interactions different potentials called isospin symmetry breaking ( * isb * ) potentials due to distinction between neutral pion and charged pion exchange for potentials describe long range part nucleon - nucleon interaction one - pion exchange model one - pion exchange term origins of * isb * for * isb * potentials isospin @xmath2 states distinction between _ nn _ , _ np _ _ pp _ potentials necessary calculations potentials difference between _ nn _ _ np _ _ pp _ wave functions should considered calculations use _ reid-93 _ potential stokspotential describes longe range nucleon - nucleon interaction one - pion exchange model for mass difference between neutral pion @xmath3 charged pions @xmath4 @xmath5 _ reid-93 _ potential non - one - pion exchange parts additional * isb * terms included in @xmath6 state potential ( stoks et al 1994 ) calculate properties of asymmetrical nuclear matter @xmath0-stable matter in two cases with isospin symmetry isospin symmetry breaking comparing results study influence of isb * nucleon - nucleon interaction on properties isb * case use _ reid-93 _ potential one - pion exchange potentials for _ np _ _ pp _ ( _ nn _ ) interactions pp @xmath6 potentials different * is * case replace one - pion exchange @xmath6 potentials of _ nn _ pp _ interactions with corresponding _ np _ interaction calculations employ formalism used previous works distiction between partial waves with @xmath7 @xmath8 @xmath9 method variational approach based on cluster expansion energy functional state dependent correlation functions convergence results tested by calculating three - body cluster termasymmetrical nuclear matter properties important studying supernova collapse heavy - ion reaction ( lattimer prakash 2000 li 1997 ) define asymmetry parameter @xmath0 as @xmath10 @xmath11 number densities neutrons protons @xmath13 total number density implies symmetrical nuclear matter @xmath14 ) @xmath15 pure neutron matter @xmath16 present results properties investigate effects of * isb * total energy function total number density asymmetry parametrs given figure fig1 cases asymmetry parameter increases saturation point shifts to lower densities large values @xmath0 energy curve shows minimum changes total energy due to inclusion * isb * nucleon - nucleon interaction shown in figure fig2 differences results isb cases increase by increasing density asymmetry parameter table tab1 results differences @xmath17 compared with mpm ( muther 1999 ) results agreement with mpm for nijm - ii _ potential @xmath0-dependence of total energy asymmetrical nuclear matter in * isb * case shown in figure fig3 ] different values density* isb * asymmetrical nuclear matter energy nearly linear in @xmath18 higher orders effects small confirms empirical parabolic law figure fig4 effects of * isb * on potential energy asymmetrical nuclear matter potential energy difference between * is * isb * cases increases by increasing asymmetry parameter @xmath0 @xmath0 energy contribution _ nn _ interaction increases difference between nn np nn np increases increasing @xmath0 each value density asymmetry parameter potential energy shift due to * isb * positive partial wave with isospin @xmath2 @xmath19 potential more attractive than @xmath20 @xmath21 influence of * isb * on saturation point properties asymmetrical nuclear matter shown in table [ tab2 ] * isb * effects on saturation density energy incompressibility asymmetrical nuclear matter small considerable for large @xmath0 nuclear matter asymmetry energy in astrophysics supernova mechanism asymmetry energy difference between binding energies of pure neutron matter ( @xmath16 ) and symmetrical nuclear matter ( @xmath15 )validity parabolic approximation asymmetry energy shown in figure fig3 results for for * is isb * cases presented in figure fig5 changes energy due to * isb * increase by inceasing density properties of @xmath0-stable matter in astrophysics equation of state important for investigating neutron stars stability-stable matter nucleons leptons electrically neutral in @xmath0-equilibrium balance between electron capture @xmath22-decay equilibrium proton fraction ( @xmath24 ) in matter important for cooling neutron star 1994 in figure fig6 plotted results for total energy of matter versus density in * is * and isb * cases considering * isb * for nucleon - nucleon interaction change of matter energy small at density @xmath25 energy correction due to * isb * effects about @xmath26 effects of * isb * on total energy-stable matter negligible calculated energy contribution of leptons proton fraction in matter shown in figures [ fig7 and fig8 ]properties of @xmath0-stable matter varied by * isb * nucleon - interaction affects conditions @xmath0-equilibrium equation of state matter for * is * isb * cases shown in figure influences of * isb * on ignorable two - nucleon potential crucial in nuclear many - body calculations comparing results isospin symmetry symmetry breaking for nucleon - interaction evaluated effects of * isb * on properties asymmetrical nuclear matter @xmath0-stable matter changes of total energy potential energy asymmetry energy of asymmetrical nuclear matter due to * isb * increase by increasing density asymmetry parameter empirical parabolic law valid * isb * results for total energy equation of state matter in * is * * isb * cases nearly identical @xmath0-stable matter * isb * affects energy contribution of leptons proton fraction financial support from shiraz university research council ipm acknowledged akmal pandharipande ravenhall ( 1998 ) physical review * c58 * 1804 bordbar modarres 1997 )journal physics g nuclear particle physics 23 , 1631 . bordbar modarres 1998 ) physical review 57 , 714 bordbar riazi 2001 international journal theoretical physics 40 , 1671 bordbar riazi 2002 astrophysics space science 282 , 563 . bordbar 2002a international journal theoretical 41 , 309 . bordbar 2002b international journal 41 , 1135 . bordbar 2003 ) international journal modern physics a18 , 2629 . heiselberg pandharipande v 2000 annual review nuclear particle science 50 , 481 . lattimer j prakash m. 2000 ) physics report 333 * , 121 . li b a. ko ren(1997 physical review 78 , 1644 . li g q. machleidt r. 1998 physical review c58 , 3135 . machleidt slaus . 2001 ) journal physics g nuclear particle physics 27 * , r69 . modarres bordbar g h. 1998 ) physical review 58 , 2781 muther polls amachleidt. 1999 ) letter b455, 259 prakash m. 1994 ) physics report 242 191 stoks v g. j. klomp m. terheggen p. f. swart j.. 1994 ) physical review c49 2950 | 0.500825 | 3,264 | 1,623 | , Saving $0.1 in GPT-4. |
177,456 | measurements of differential photon @xmath0 jet and photon @xmath1 jet production cross sections are presented using approximately 1 @xmath2 of data collected by the 0 detector at the tevatron collider at a center - of - mass energy of 1.96 tev .
isolated photons are selected in the rapidity range @xmath3 and jets selected with rapidities @xmath4 .
the measurements are compared to next - to - leading order theoretical predictions . | we present first preliminary measurements of differential @xmath5 c jet and @xmath5 b jet cross sections using @xmath6 of data .
these measurements were performed at the fermilab tevatron collider at @xmath7 = 1.96 tev collected using the 0 detector .
these measurements are differential with respect to the transverse momentum of the leading photon ( @xmath8 ) , the rapidity of the leading jet ( @xmath9 ) and of the leading photon ( @xmath10 ) . the results are presented in five @xmath8 bins and in two regions of photon
jet rapidities : @xmath11 ( region 1 ) and @xmath12 ( region 2 ) .
events are recorded using the 0 detector @xcite and must contain at least one photon candidate and at least one heavy flavor jet candidate .
both the chosen photon and jet candidates are required to be leading with respect to their @xmath13 .
the photon candidate @xmath13 must be greater than 30 gev , be located within the central calorimeter and have @xmath14 .
the leading jet must have @xmath15 gev and @xmath4 .
additionally , events must also contain a primary vertex within 50 cm of the detector s center along the beam pipe .
this vertex is required to have at least three associated tracks , one with @xmath16 gev and a second with @xmath17 gev . to suppress events containing cosmic muons and w bosons , the missing transverse energy ( @xmath18 )
must pass the condition @xmath19 .
events are triggered using electromagnetic ( em ) clusters consistent with photon shower profiles .
the final photon energy is taken as the energy of the cluster within a cone of radius @xmath20 .
an isolation requirement is imposed on the candidate photon such that the total energy in a cone of @xmath21 , excluding @xmath22 , must be less than 7% of @xmath22 .
the cluster must also have a probability less than 0.1% to be spatially matched to any track in the event .
the cluster s transverse profile in the third layer of the em calorimeter must be consistent with that of a photon .
the primary vertex is further constrained to be within 10 cm of the most probable origin of the photon candidate along the beam pipe . to further suppress backgrounds coming from jets , an artificial neural network ( ann ) @xcite , which uses additional single variable discriminants ,
was employed with the requirement of ann @xmath23 .
the energy of the leading jet is determined using 0 s runii jet reconstruction algorithm @xcite with a cone of radius @xmath24 .
heavy flavor jets are identified using their long lifetimes by reconstructing displaced secondary vertices from the jet s tracks .
to best enhance the fraction of jets originating from heavy flavor quarks , an artificial neural network ( bann ) was implemented .
the bann @xcite uses variables characteristic of heavy flavor jets as inputs , trained such that the output of @xmath25 jets tends toward one and that of light ( udsg ) jets tends to zero . for this selection ,
we require the bann output @xmath26 .
to estimate the fraction of photons in the final data sample , the photon ann output in data is fit with a template from signal photon monte carlo ( mc ) pythia @xcite simulation and a background dijet mc template .
the fit is performed in each @xmath8 bin and in each photon - jet rapidity region separately .
the fit results are shown in fig .
[ fig : pur_fit ] . + heavy flavor jet events as a function of @xmath8 for both rapidity regions 1 ( left ) and 2 ( right ) .
the result of the @xmath27 functional fit is shown by the full lines , together with the statistical uncertainty in the default fit ( dashed lines ) and the total uncertainty ( dotted lines).,title="fig : " ] + heavy flavor jet events as a function of @xmath8 for both rapidity regions 1 ( left ) and 2 ( right ) .
the result of the @xmath27 functional fit is shown by the full lines , together with the statistical uncertainty in the default fit ( dashed lines ) and the total uncertainty ( dotted lines).,title="fig : " ] to estimate the fraction of @xmath28 and @xmath25 jets in the final data sample , a different discriminant was used .
it is defined as @xmath29 , where @xmath30 is the probability of a track in the jet cone to originate from the primary vertex , omitting the track with the lowest probability .
@xmath31has large values for @xmath25 jets and values closer to zero for light jets .
a fit to the data @xmath31distribution is performed using mc templates for @xmath25 and @xmath28 jets , and the light jet template is taken from a light jet enriched data sample .
the fit done in each @xmath8 bin , and to verify the quality of the fit , a combined flavor template is compared to the data as shown in fig .
[ fig : cbjet_test ] .
the differential cross sections are presented in five @xmath8 bins and two regions of @xmath32 , corrected for @xmath8 smearing effects due to the finite resolution of the calorimeter using the unfolding method described in @xcite .
the measured cross sections are shown in fig .
[ fig : xsectbc1plot ] for @xmath33and @xmath34processes . statistical uncertainties on the results vary from 0.2% in the lowest @xmath8 bin to 89% in the highest bin , while systematic uncertainties vary between 1525% .
the theoretical predictions from next - to - leading ( nlo ) order calculations are presented in fig .
[ fig : xsectbc1plot ] with the renormalization , factorization and fragmentation scales @xmath35 , and @xmath36 set to @xmath8 .
these predictions are preliminary @xcite and are based on techniques to calculate the analytic cross sections published in ref .
@xcite .
the ratios of the measured and predicted cross sections for both @xmath33and @xmath34cross sections in the two rapidity regions are shown in fig .
[ fig : xsectb1ratio ] . the uncertainty due to the scale choice is estimated by simultaneously varying all three scales by a factor of two , @xmath37 , and @xmath38 .
the theoretical predictions utilize the cteq6.6 m pdf set with uncertainties calculated according to the prescription in @xcite .
( left ) and @xmath34 ( right ) cross sections as a function of @xmath8 in regions 1 and 2 .
the uncertainties on the points in data are the full uncertainties .
the nlo theoretical predictions using the cteq6.6 m pdf set are shown by the dotted lines.,title="fig : " ] ( left ) and @xmath34 ( right ) cross sections as a function of @xmath8 in regions 1 and 2 .
the uncertainties on the points in data are the full uncertainties .
the nlo theoretical predictions using the cteq6.6 m pdf set are shown by the dotted lines.,title="fig : " ] cross section ratio of data to theory as a function of @xmath8 in region 1 ( top left ) and region 2 ( top right ) and the @xmath34cross sections in region 1 ( bottom left ) and region 2 ( bottom right ) . the uncertainties on the data points include both statistical ( inner line ) and full uncertainties ( entire error bar ) also shown are the theoretical scale uncertainties and the cteq6.6 m pdf uncertainties .
the ratio of the central cteq6.6 m prediction to two _ intrinsic charm _
models is shown as well.,title="fig : " ] cross section ratio of data to theory as a function of @xmath8 in region 1 ( top left ) and region 2 ( top right ) and the @xmath34cross sections in region 1 ( bottom left ) and region 2 ( bottom right ) .
the uncertainties on the data points include both statistical ( inner line ) and full uncertainties ( entire error bar ) also shown are the theoretical scale uncertainties and the cteq6.6 m pdf uncertainties .
the ratio of the central cteq6.6 m prediction to two _ intrinsic charm _
models is shown as well.,title="fig : " ] cross section ratio of data to theory as a function of @xmath8 in region 1 ( top left ) and region 2 ( top right ) and the @xmath34cross sections in region 1 ( bottom left ) and region 2 ( bottom right ) . the uncertainties on the data points include both statistical ( inner line ) and full uncertainties ( entire error bar ) also shown are the theoretical scale uncertainties and the cteq6.6 m pdf uncertainties . the ratio of the central cteq6.6 m prediction to two _ intrinsic charm _
models is shown as well.,title="fig : " ] cross section ratio of data to theory as a function of @xmath8 in region 1 ( top left ) and region 2 ( top right ) and the @xmath34cross sections in region 1 ( bottom left ) and region 2 ( bottom right ) .
the uncertainties on the data points include both statistical ( inner line ) and full uncertainties ( entire error bar ) also shown are the theoretical scale uncertainties and the cteq6.6 m pdf uncertainties . the ratio of the central cteq6.6 m prediction to two _ intrinsic charm _
models is shown as well.,title="fig : " ] 9 d0 collaboration , v.m .
et al _ , nucl .
instrum .
methods a * 565 * , 463 ( 2006 ) .
d0 collaboration , v. abazov _ et al .
b 666 , 435 ( 2008 ) . g.c .
blazey _ et al .
_ , arxiv : hep - ex/0005012 ( 2000 ) . t. scanlon , ph.d .
thesis , fermilab - thesis-2006 - 43 .
t. sjostrand , comp .
. comm . * 82 * , 74 ( 1995 ) .
b. abbott _ et al _ ( d0 collaboration ) , phys .
d * 64 * , 032003 ( 2001 ) .
t. stavreva , j.f .
owens , `` photon + heavy flavor cross sections''(_in preparation _ ) , 2008 .
harris , j.f .
owens , `` the two cutoff phase space slicing method '' , arxiv : hep - ph/0102128v2 . j. pumplin _ et al .
_ , j. high energy phys . * 0207 * , 12 ( 2002 ) ; d. stump _ et al .
_ , j. high energy phys . * 0310 * , 046 ( 2003 ) . | preliminary measurements differential @xmath5 c b jet cross sections @xmath6 data measurements performed fermilab tevatron collider @xmath7 = 1.96 tev 0 detector measurements differential transverse momentum leading photon rapidity results presented five @xmath8 bins two regions photon jet rapidities @xmath11 2 events recorded 0 detector @xcite contain one photon heavy flavor jet candidate photon jet candidates leading @xmath13 photon candidate @xmath13 greater than 30 gev within central calorimeter have @xmath14 leading jet have @xmath15 gev @xmath4 events contain primary vertex within 50 cm detector s center beam pipe three tracks one @xmath16 gev @xmath17 gev suppress events cosmic muons bosons missing transverse energy @xmath18 ) pass condition @xmath19 events triggered electromagnetic clusters photon shower profiles final photon energy taken cluster within cone radius @xmath20 isolation requirement candidate photon total energy cone @xmath21 excluding less than 7% of @xmath22 cluster probability less than 0.1% spatially matched trackcluster s transverse profile third layer calorimeter consistent with photon primary vertex constrained within 10 cm of probable origin photon suppress backgrounds from jets artificial neural network @xcite single variable discriminants employed with requirement ann @xmath23 energy leading jet determined using jet reconstruction algorithm with cone radius @xmath24 heavy flavor jets identified reconstructing displaced secondary vertices from jet tracks enhance fraction jets from heavy flavor quarks artificial neural network bann implemented uses variables heavy flavor jets inputs output of @xmath25 jets toward one light jets zero bann output @xmath26 estimate fraction photons final data sample photon ann output with template signal photon monte carlo pythia @xcite simulation background dijet mc template fit performed in each @xmath8 bin photon - jet rapidity region fit results shown in fig . heavy flavor jet events @xmath8 rapidity regions 1 2 result of @xmath27 functional fit shown by full lines statistical uncertainty default fit total uncertainty dotted heavy flavor jet events function @xmath8 rapidity regions 1@xmath27 fit shown by full lines statistical uncertainty default dashed total uncertainty dotted fraction @xmath28 jets final data sample different discriminant used defined as @xmath29 @xmath30 probability track jet cone from primary vertex omitting lowest probability @xmath31has large values for @xmath25 jets closer to zero for light jets fit @xmath31distribution using mc templates for @xmath25 jets light jet template from enriched data sample fit in each @xmath8 bin combined flavor template compared data fig differential cross sections in five @xmath8 bins two regions @xmath32 corrected for @xmath8 smearing effects resolution measured cross sections shown in fig . for @xmath33and @xmath34processes statistical uncertainties vary 0.2% lowest to 89% systematic uncertainties vary 1525% theoretical predictions from next - to - leading order calculations presented in fig . renormalization factorization fragmentation scales @xmath35 @xmath36 set to @xmath8predictions preliminary based on techniques cross sections ratios measured predicted cross sections for @xmath33and @xmath34cross sections in two rapidity regions shown in fig . uncertainty due to scale choice estimated by varying three scales by factor two theoretical predictions utilize cteq6.6 m pdf set uncertainties calculated in @xcite @xmath34 cross sections function of @xmath8 in regions 1 2 uncertainties on points data full uncertainties theoretical predictions using cteq6.6 m pdf set shown by dotted lines cross sections function @xmath8 regions 1 2 uncertainties full uncertainties theoretical predictions using cteq6.6 m pdf set shown by dotted lines cross section ratio data to theory function of @xmath8 in region 1 2 @xmath34cross sections in uncertainties on data points include statistical full uncertainties theoretical scale uncertainties cteq6.6 m pdf uncertainties ratio of central cteq6.6 m prediction to two intrinsic models showncross section ratio data to theory function @xmath8 region 1 2 right @xmath34cross sections 1 left uncertainties include statistical full uncertainties theoretical scale uncertainties cteq6.6 m pdf uncertainties ratio central cteq6.6 m prediction to two intrinsic charm models shown cross section ratio theory function @xmath8 region 1 left 2 top right @xmath34cross sections uncertainties include statistical full uncertainties theoretical scale uncertainties cteq6.6 m pdf uncertainties ratio central cteq6.6 m prediction to two intrinsic charm models shown cross section ratio data to theory function @xmath8 region 1 top left 2 @xmath34cross sections uncertainties include statistical full uncertainties theoretical scale uncertainties cteq6.6 m pdf uncertainties ratio central cteq6.6 m prediction to two intrinsic charm _ models shown d0 collaboration , v.m . et al _ nucl . instrum . methods a * 565 * , 463 ( 2006 ) . d0 collaboration , v. abazov _ et al . b 666 , 435 ( 2008 ) g.c blazey _ et al .arxiv hep - ex/0005012 2000 ) scanlon ph.d thesis fermilab thesis-2006 - 43 sjostrand 82 74 ( 1995 ) abbott al collaboration . 64 032003 ( 2001 ) stavreva j owens photon heavy flavor cross preparation 2008 harris j owens two cutoff phase space slicing method hep - ph/0102128v2 pumplin high energy . 0207 12 ( 2002 ) stump high energy phys. 0310 046 ( 2003 ) | 0.354326 | 2,426 | 1,205 | , Saving $0.1 in GPT-4. |
177,471 | we report recent results on the impact of asymmetric dark matter ( dm ) particles on low - mass stars .
first , we found that the small convective core expected in stars with masses between 1.1 and 1.3 m@xmath0 is suppressed due to dm cooling .
moreover , stars with masses below 1 m@xmath0 have their central temperatures and densities more strongly influenced by dm than in the solar case .
we were able to put limits to the dm mass and spin - dependent dm - proton scattering cross section by comparing the modelling of the nearby star @xmath1 cen b with photometric , spectroscopic and asteroseismic observations . | if the dark matter ( dm ) of the universe is made of weakly interacting massive particles ( wimps ) their accumulation inside stars may lead to observable effects that depend on the unknown characteristics of the dark particles @xcite .
thus , the dm parameter space can be constrained using this approach , in particular in the case of models of dm particles that do not annihilate in stellar interiors .
a good example is asymmetric dm @xcite , a dm model motivated to explain the similar cosmic abundance of dark and baryonic matter and that naturally predicts a dm mass of the order of the few gev ( see for instance the recent reviews @xcite ) .
+ a fraction of the wimps in the galactic halo may scatter with nucleons of the stellar plasma and become gravitationally captured by stars @xcite .
the efficiency of this process increases with the stellar density ( among other factors ) , so compact stars can be in principle strongly influenced @xcite .
the destruction of neutron stars by black hole formation in their interior has been predicted , and limits on the properties of asymmetric dm have been derived using this approach @xcite .
+ in the case of the sun and solar - type stars , the conduction of dm particles in the stellar interior acts as a new energy transport mechanism , cooling the center of the stars @xcite .
the impact on the properties of these stars tends to be less significant than for compact stars , but on the other hand the physics of the interior of solar - type stars is much better known and observations are much more precise , in particular with the complementary diagnostic of the solar interior provided by solar neutrino measurements @xcite and helioseismology @xcite . + in these proceedings we report the study of the impact of asymmetric dm on stars with masses similar to that of the sun @xcite , stars embedded in halos of dm with a density equal to that estimated for the solar neighbourhood , @xmath2gev @xmath3 @xcite .
we focus on stars whose acoustic oscillations have already been precisely identified
. the characteristics of these stars are shown in table [ tab - starchars ] .
our approach is motivated mainly by the following reasons : i ) stars with lower masses are more strongly influenced by dm , ii ) stars with a small convective core ( 1.1 - 1.3 m@xmath0 stars ) can have their dominant energy transport mechanisms altered due to dm , and iii ) asteroseismology ( the study of the stellar oscillations ) is a technique sensitive to properties of the stellar cores , where the dm impact occurs .
+ as we will show , we were able to put the first limits to the nature of dm using an asteroseismic analysis of @xmath1 cen b. the reader is addressed to reference @xcite for a more thorough description of the methods used and the results obtained .
+ l r r r r r r star & @xmath4 ( m@xmath0 ) & @xmath5 ( l@xmath0 ) & @xmath6 ( k ) & @xmath7 & @xmath8 ( @xmath9hz ) & @xmath10 ( @xmath11hz ) + & & & & & + observations @xmath12 & 0.92 - 1.10 & @xmath13 & @xmath14 & @xmath15 & @xmath16 & @xmath17 + stand .
& 0.92 & 0.63 & 5379 & 0.039 & 149.03 & 10.12 + dm model.@xmath18 & 0.92 & 0.63 & 5379 & 0.039 & 149.08 & 9.13 + & & & & & + observations @xmath12 & 1.18 - 1.25 & @xmath19 & @xmath20 & @xmath21 & @xmath22 & @xmath23 + stand . model . & 1.18 & 2.22 & 6170 & 0.028 & 97.92 & 8.16 + dm model.@xmath18 & 1.18 & 2.22 & 6170 & 0.028 & 98.05 & 7.65 + & & & & & + observations @xmath12 & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 + stand . model . &
0.934 & 0.51 & 5245 & 0.031 & 162.56 & 10.23 + dm model.@xmath18 & 0.934 & 0.51 & 5230 & 0.031 & 162.45 & 8.95 + @xmath30 the averages are calculated for the intervals @xmath31(@xmath9hz)@xmath32 for kic 8006161 , @xmath33(@xmath9hz)@xmath34 for hd 52265 , and @xmath35(@xmath9hz)@xmath36 for @xmath1 cen b. + @xmath12data from @xcite and @xcite for kic 8006161 , @xcite for hd 52265 , and @xcite for @xmath1 cen b. + @xmath18 @xmath37gev , @xmath38@xmath39 , @xmath40gev @xmath3 .
we modelled the star kic 8006161 , reproducing all its observed characteristics , including the large and small frequency separations as measured by the _
kepler _ mission .
we found a @xmath41 decline in its central temperature when the influence of the accumulation of asymmetric dm particles with @xmath37gev and @xmath38@xmath39 was taken into account .
+ furthermore , we also studied the star hd 52265 , a 1.2 m@xmath0 star whose oscillation frequencies have been measured by the _ corot _ mission .
we found that , whereas in the standard modelling this star is expected to have a convective core , the inclusion of the dm energy transport removes this core and the star is left with a radiative interior .
specific combinations of frequencies of low - degree modes have been shown to be sensitive to the presence of small convective cores ( see ref . ) .
we leave this interesting diagnostic approach for a further work .
+ we have chosen to study the star @xmath1 cen b because it is close to the sun and belongs to a binary system , which has allowed a very accurate measurement of its mass and other characteristics ( see table [ tab - starchars ] ) .
this fact strongly reduces the uncertainties in the modelling of the star .
we used the stellar oscillations , in particular the mean small separation between the modes of low degree @xmath42 , to identify the modifications introduced by asymmetric dm in the core of the star .
+ we found that the stellar models strongly influenced by the dm cooling mechanism , while reproducing all the global properties of @xmath1 cen b within observational errors , can not account for the observed small frequency separations .
the existence of asymmetric dm particles with @xmath43 and @xmath44 above the blue line in figure [ fig - admconstraints ] can be excluded because they lead to models of @xmath1 cen b with a @xmath42 more than 2 @xmath45 away from the observed value .
the 1 @xmath45 variation on @xmath42 when the classical stellar parameters ( @xmath4 , @xmath5 , @xmath6 and @xmath7 ) vary within the observational errors is plotted as a filled region around the blue line .
further uncertainties not included in this analysis are those in other parameters of stellar modelling , such as nuclear reaction rates and chemical abundances , with an expected minor impact on our diagnostic and , more importantly , the error in the determination of the local dm density , which will introduce an estimated uncertainty of a factor smaller than 2 in @xmath44 @xcite . + our method provides a competitive exclusion plot on @xmath44 for low - mass dm particles .
these constraints are particularly valuable taking into account the present controversy among different direct detection experiments in the same region of the dm parameter space @xcite .
cen b. asymmetric dm particles with properties above the blue line produce a strong impact on the core of the star , leading to a mean small frequency separation more than 2 @xmath45 away from the observations .
the filled region shows the uncertainty in the modelling when the observational errors are taken into account . a density of @xmath46gev @xmath3 was assumed .
figure adapted from ref .
@xcite . ] by 1 cm by -1 cm
10 url # 1#1urlprefix[2][]#2 d n and press w h 1985 _ apj _ * 294 * 663673 | dark matter ( dm ) universe weakly interacting massive particles accumulation inside stars may to effects on unknown characteristics dark particles dm parameter space constrained using approach models of dm particles not annihilate in stellar interiors example is asymmetric dm model similar cosmic abundance of dark baryonic matter predicts dm mass few gev fraction wimps in galactic halo may scatter with stellar plasma gravitationally captured by stars efficiency process increases with stellar density compact stars influenced destruction of neutron stars by black hole formation interior predicted limits on properties of asymmetric dm derived using this approach sun and solar - type stars conduction of dm particles in stellar interior acts new energy transport mechanism cooling center stars impact on properties stars less significant than compact stars physics of interior of solar - type stars better known observations more precise with complementary diagnostic solar interior solar neutrino measurements helioseismology study impact of asymmetric dm on stars with masses similar to sun stars in halos dm with density equal to solar neighbourhoodfocus on stars acoustic oscillations identified characteristics shown in table tab - starchars approach motivated by stars lower masses influenced by dm stars small convective core ( 1.1 - 1.3 m@xmath0 stars energy transport due to dm asteroseismology stellar oscillations sensitive to stellar cores dm impact occurs first limits to nature dm using asteroseismic analysis of @xmath1 cen reader reference @xcite for thorough description of methods results @xmath4 @xmath5 @xmath6 @xmath7 @xmath8 @xmath10 observations @xmath12 0.92 - 1.10 @xmath13 @xmath14 @xmath17 + 0.92 0.63 5379 0.039 149.03 10.12 + dm model.@xmath18 & 0.92 0.63 5379 0.039 149.08 9.13 observations @xmath12 1.18 - 1.25 @xmath19 @xmath20 @xmath21 @xmath22 @xmath23 + model 1.18 2.22 6170 0.028 97.92 & 8.16 + dm model.@xmath18 & 1.18 & 2.22 6170 0.028 98.05 7.65 observations @xmath12 @xmath24 stand model 0.934 0.51 5245 0.031 162.56 10.23 dm model.@xmath18 0.934 0.51 5230 0.031 162.45 8.95 @xmath30 averages calculated intervals @xmath31 kic 8006161 hd 52265 @xmath1 cen b @xmath12data from @xcite kic 8006161 hd 52265 @xmath1 cen b. @xmath18 @xmath37gev @xmath38@xmath39 @xmath40gev @xmath3 modelled star kic 8006161 observed characteristics large small frequency separations measured _ kepler _ mission found @xmath41 decline in central temperature accumulation asymmetric dm particles @xmath37gev studied star hd 52265 1.2 m@xmath0 star oscillation frequencies measured by _ corot _ mission star expected convective core dm energy transport removes core radiative interior specific combinations frequencies low - degree modes sensitive to small convective cores ref . leave diagnostic approach for further workchosen study star @xmath1 cen b close to sun belongs binary system accurate measurement of mass characteristics table starchars reduces uncertainties in modelling star used stellar oscillations small separation between modes low degree @xmath42 identify modifications asymmetric dm in core star stellar models influenced by dm cooling mechanism properties @xmath1 b account for observed small frequency separations asymmetric dm particles with @xmath43 @xmath44 above blue line excluded lead to models @xmath1 with @xmath42 more than 2 @xmath45 away from observed value 1 @xmath45 variation on @xmath42 stellar parameters plotted as filled region around blue line uncertainties not included other parameters stellar modelling nuclear reaction rates chemical abundances minor impact on diagnostic error in determination local dm density estimated uncertainty factor smaller than 2 in @xmath44 @xcite method provides competitive exclusion plot on @xmath44 for low - mass dm particles constraints valuable controversy among direct detection experiments same region dm parameter spaceasymmetric dm particles above blue line impact core star small frequency separation 2 @xmath45 observations filled region shows uncertainty modelling observational errors density @xmath46gev @xmath3 assumed figure adapted from @xcite 1 cm -1 cm 10 url 1#1urlprefix[2]#2 press 1985 294 663673 | 0.311898 | 1,934 | 993 | , Saving $0.1 in GPT-4. |
177,482 | this fluid dynamics video is an entry for the gallery of fluid motion for the 66th annual meeting of the fluid dynamics division of the american physical society .
we show the curious behaviour of a light ball interacting with a liquid jet . for certain conditions
, a ball can be suspended into a slightly inclined liquid jet .
we studied this phenomenon using a high speed camera .
the visualizations show that the object can be ` juggled ' for a variety of flow conditions .
a simple calculation showed that the ball remains at a stable position due to a bernoulli - like effect .
the phenomenon is very stable and easy to reproduce . | it is fascinating : when a liquid jet interacts with a ball , a kind of _ juggling _ can be achieved .
the ball can be trapped in air by its interaction with a fluid jet ; the ball moves up and down around at a stable position .
this experiment is very simple , everyone should try it .
it is a simple demonstration of the beautiful flow behaviour of fluids .
the water jet impacts the bottom part of the ball ; the jet turns into a flowing fluid film that surrounds the ball . due to the change in thickness from the jet to the film , the mean film velocity is higher than that of the jet , resulting in a lower pressure region which in turn make the ball position stable . to successfully produce levitation ,
the liquid in the film needs to be evacuated ( such that its weight does not break the force balance between the momentum flux of the jet and the weigh of the ball ) .
this is the reason why the liquid jet has to be slightly inclined .
the inclination and the flowing film induce rotation in the ball .
the collision of the film with itself ( flowing around the ball ) produces an outgoing jet , which combined with the ball rotation and surface ripples , creates a beautiful series of films , threads and drops that are continuously ejected .
some juggling tricks such as trap , catch , switch , rise up and swirl are shown in the video .
we are eager to study this flow in more detail .
to our knowledge , there has not been a methodical study of the flow shown here .
two sample videos are http://ecommons.library.cornell.edu/bitstream/1813/8237/2/juggling_hi_9oct.mp4[video 1 , hr ] and http://ecommons.library.cornell.edu/bitstream/1813/8237/4/juggling_low_9oct.mpg[video 2 , lr ] . | fascinating liquid jet interacts with ball _ juggling _ achieved ball trapped in air fluid jet ball moves up and down stable position experiment simple try demonstration of flow behaviour of fluids water jet impacts bottom part ball jet turns into flowing fluid film surrounds ball to change in thickness from jet to film mean film velocity higher than jet lower pressure region ball position stable produce levitation liquid in film needs to evacuated weight break force balance between momentum flux jet and weigh ball liquid jet slightly inclined inclination flowing film induce rotation ball collision of film produces outgoing jet with ball rotation surface ripples creates series of films threads drops continuously ejected juggling tricks trap catch switch rise up swirl shown in video eager to study flow more detail not methodical study of flow shown here two sample videos http://ecommons.library.cornell.edu/bitstream/1813/8237/2/juggling_hi_9oct.mp4 and/juggling_low_9oct.mpg 2 | 0.075216 | 387 | 188 | , Saving $0.0 in GPT-4. |
177,483 | the mass differences between spin-1/2 and spin-3/2 baryons are compared to the mass differences between spin-0 and spin-1 mesons .
results of simulations for charmed hadrons in the quenched approximation from a tadpole - improved anisotropic action are discussed in the context of other lattice calculations , quark model predictions , heavy quark symmetry predictions and experimental data . | the mass differences between the lightest vector and pseudoscalar mesons containing one or two heavy quarks are persistently smaller in quenched lattice simulations than the experimental values@xcite
. studies of the unquenched theory have found that the discrepancy persists@xcite .
fewer lattice simulations have been performed for the mass differences between spin-3/2 and spin-1/2 baryons containing one or more heavy quarks@xcite .
it is useful to compare these baryon mass differences to the meson mass differences mentioned above , since both are colour hyperfine ( spin - spin ) effects .
quantitative relationships between the meson and baryon splittings have been claimed using experimental data , quark models or heavy quark effective theory .
if these relations are respected by lattice qcd simulations then the baryon spin splittings will be smaller than experiment ( just like mesons ) , but if lattice results for the baryons agree with experiment then the relations must clearly be violated .
in either case , the information will be helpful when seeking to understand the existing discrepancy between lattice qcd simulations and experiment .
a simple quark model for the hadron masses uses the operators @xmath0 it follows that the spin splittings for heavy - light mesons and for singly and doubly heavy baryons vanish like @xmath1 , @xmath2 where @xmath3 and @xmath4 contain wave function information .
the determination of @xmath5 , using a relation between the wave function at the origin and the derivative of the potential ( plus some other postulates ) , has been discussed by lipkin@xcite and by lipkin and odonnell@xcite .
r@ = l + @xmath6 & @xmath7 + @xmath8 & @xmath9 + + @xmath10 & @xmath11q_c + [ u_a^tc\gamma_5q_b]d_c\right.$ ] + + @xmath12 & @xmath13d_c$ ] + @xmath14 & @xmath13s_c$ ] + @xmath15 & @xmath16s_c+[s_a^tc\gamma_5q_b]u_c\right)$ ] + @xmath17 & @xmath18s_c$ ] + @xmath19 & @xmath13q_c$ ] + @xmath20 & @xmath18q_c$ ] + + @xmath21 & @xmath22q_c$ ] + @xmath23 & @xmath24q_c$ ] + @xmath25 & @xmath26q_c$ ] + @xmath27 & @xmath28u_c$ ] + @xmath29 & @xmath28s_c$ ] + for doubly heavy baryons the pair of heavy quarks forms a colour anti - triplet , so @xmath30 and @xmath31_{qqq } = \frac{3}{4}(m_v - m_p).\end{aligned}\ ] ] a more elegant derivation of this same result comes from incorporating the anti - triplet diquark directly into heavy quark effective theory@xcite .
the resulting lagrangian has a superflavour symmetry among heavy quarks and heavy diquarks : @xmath32\chi \nonumber \\\end{aligned}\ ] ] where @xmath33 and @xmath34 is the block - diagonal spin vector for quarks and diquarks .
a direct calculation from this lagrangian leads to eq .
( [ qqmodel ] ) .
as reported in ref .
@xcite , our simulations used an anisotropic tadpole - improved relativistic action for gauge fields and for quarks .
the gauge action includes @xmath35 rectangular plaquettes which remove @xmath36 classical errors , and the d234 quark action has no @xmath37 nor @xmath36 classical errors .
the hadron creation operators are listed in table [ ops ] .
sink - smeared and local correlators were built from these operators and fit simultaneously to single and double exponential functions .
the `` spin-3/2 '' operators in table [ ops ] actually contain an admixture of spin-1/2 as well .
the spin-3/2 component is isolated by a judicious choice of lorentz components .
.lattice parameters .
[ cols="^,^,^,^",options="header " , ] in our simulations , the bare anisotropy was set to @xmath38 , tadpole improvement was implemented via the mean link in landau gauge , and quarks have dirichlet time boundaries .
configuration details are given in table [ details ] .
as shown in fig . [ fig : ratiosingly ] , the experimental ratio of @xmath39 meson spin splittings to @xmath40 baryon spin splittings is remarkably constant , where @xmath41 is a light quark and @xmath42 is light or strange or charmed .
our lattice results , without any extrapolation in @xmath43 or @xmath44 , are also independent of @xmath45 , but significantly smaller than experiment . fig .
[ fig : ratiodoubly ] shows the ratio of spin splittings for @xmath39 mesons and @xmath46 baryons .
once again , the lattice results are independent of @xmath45 and smaller than the experimental point .
interestingly , they are near the prediction of eq .
( [ qqmodel ] ) .
some representative lattice computations of the heavy - light meson spin splittings are shown in fig .
[ fig : mesons ] , for @xmath42 from light quarks through to bottom .
the experimental results are linear in the average mesons mass to a few percent , but the lattice results show a significant slope .
the lattice bottom splitting is about half of the experimental value .
heavy quark symmetry requires that @xmath47 constant as @xmath48 , but fig .
[ fig : mesons ] does not allow us to determine what the heavy quark limiting value is .
in contrast to the meson situation , the lattice results for the @xmath40 baryon spin splittings , plotted in fig . [
fig : singly ] , are not significantly smaller than experiment ; in fact , some lattice results are larger than experiment .
this work was supported in part by the natural sciences and engineering research council of canada .
some of the computing was done on hardware funded by the canada foundation for innovation , with contributions from compaq canada , avnet enterprise solutions , and the government of saskatchewan .
9 r. lewis , n. mathur and r. m. woloshyn , hep - ph/0107037 , to appear in phys .
d. f. butler et .
al . , nucl .
b430 , 179 ( 1994 ) ; s. aoki et .
al . , phys .
84 , 238 ( 2000 ) ; a. ali khan et .
al . , phys .
d62 , 054505 ( 2000 ) ; k. c. bowler et .
al . , phys .
d62 , 054506 ( 2000 ) .
p. boyle , nucl .
b ( proc . suppl . )
63a - c , 314 ( 1998 ) ; k. i. ishikawa et .
al . , phys .
d61 , 074501 ( 2000 ) ; j. hein et .
al . , phys .
d62 , 074503(2000 ) ; r. lewis and r. m. woloshyn , phys .
d62 , 114507 ( 2000 ) .
s. collins et .
al . , phys . rev .
d60 , 074504 ( 1999 ) ; t. manke et .
al . , phys .
d62 , 114508 ( 2000 ) ; c. stewart and r. koniuk , phys . rev .
d63 , 054503 ( 2001 ) .
t. bhattacharya , r. gupta , g. kilcup and s. sharpe , phys .
d53 , 6486 ( 1996 ) .
n. mathur , r. lewis and r. m. woloshyn , these proceedings . h. j. lipkin , phys .
b171 , 293 ( 1986 ) . h. j. lipkin and p. j. odonnell , phys .
b409 , 412 ( 1997 ) .
m. j. savage and m. b. wise , phys .
b248 , 177 ( 1990 ) . | mass differences between lightest pseudoscalar mesons containing heavy quarks smaller in quenched lattice simulations than experimental values studies theory discrepancy persists fewer lattice simulations performed for between spin-3/2 spin-1/2 baryons containing heavy quarks compare baryon mass differences to meson mass differences colour hyperfine spin effects quantitative relationships between meson baryon splittings claimed using experimental data quark models heavy quark theory if relations respected by lattice simulations baryon spin splittings smaller than experiment if results agree with experiment relations be violated information helpful discrepancy between lattice qcd simulations experiment quark model for hadron masses uses operators spin splittings for heavy - light mesons singly doubly heavy baryons vanish contain wave function information determination of @xmath5 relation between wave function origin derivative potential discussed by lipkin odonnell @xmath6 @xmath7 @xmath8 @xmath9 @xmath10 @xmath11q@xmath12_c$ @xmath14_c$ @xmath15_c @xmath17_c$ @xmath19 @xmath13q_c$ @xmath20_c$ @xmath21_c @xmath23_c @xmath25_c @xmath27 @xmath28u_c$ @xmath29 @xmath28s_c$ doubly heavy baryons heavy quarks forms colour anti - triplet @xmath30 @xmath31_{qqq = \frac{3}{4}(m_v - m_p). derivation anti - triplet diquark into heavy quark effective theory lagrangian superflavour symmetry among heavy quarks diquarks @xmath32\chi @xmath33 @xmath34 block - diagonal spin vector for quarks diquarks direct calculation leads to eq simulations used anisotropic tadpole - improved relativistic action for gauge fields quarks gauge action includes @xmath35 rectangular plaquettes @xmath36 classical errors d234 quark action no @xmath37 classical errors hadron creation operators listed in tablesink - smeared local correlators built from operators fit to single double exponential functions spin-3/2 operators in table contain admixture of spin-1/2 spin-3/2 component isolated by lorentz components.lattice parameters simulations bare anisotropy set to @xmath38 tadpole improvement via mean link landau gauge quarks have dirichlet time boundaries configuration details in table experimental ratio of @xmath39 meson spin to @xmath40 baryon spin splittings constant @xmath41 light quark @xmath42 light lattice results independent of @xmath45 smaller than experiment fig shows ratio spin splittings for @xmath39 mesons @xmath46 baryons lattice results independent of @xmath45 smaller than experimental point near prediction of eq representative lattice computations of heavy - light meson spin splittings shown in fig for @xmath42 experimental results linear lattice results show significant slope lattice bottom splitting half of experimental value heavy quark symmetry requires @xmath47 constant as @xmath48fig mesons heavy quark limiting value lattice results @xmath40 baryon spin splittings fig not smaller experiment some larger work supported by natural sciences engineering research council canada computing hardware funded canada foundation innovation contributions compaq canada avnet enterprise solutions government saskatchewan r. lewis n. mathur r m woloshyn hep - ph/0107037 phys . butler nucl b430 , 179 ( 1994 ) aoki . 84 , 238 ( 2000 ) ali khan phys d62 , 054505 ( 2000 bowler d62 054506 ( 2000 p. boyle nucl 63a, 314 ( 1998 ) ishikawa phys . d61 , 074501 ( 2000 ) j hein phys d62 , 074503(2000 r. lewis woloshyn phys . d62 , 114507 ( 2000 ) collins phys rev d60 , 074504 ( 1999 ) manke phys . d62 , 114508 ( 2000 ) stewart r. koniuk phys . rev d63 054503 2001 ) bhattacharya gupta kilcup sharpe phys d53 6486 1996 ) mathur lewis woloshyn proceedings j lipkin b171 293 1986 ) lipkin p odonnell phys b409, 412 1997 ) m j savage wise b248 177 1990 ) | 0.310014 | 1,881 | 944 | , Saving $0.1 in GPT-4. |
177,496 | we have developed and tested an antireflection ( ar ) coating method for silicon lenses at cryogenic temperatures and millimeter wavelengths .
our particular application is a measurement of the cosmic microwave background .
the coating consists of machined pieces of cirlex glued to the silicon . the measured reflection from an ar
coated flat piece is less than 1.5% at the design wavelength .
the coating has been applied to flats and lenses and has survived multiple thermal cycles from 300 to 4k .
we present the manufacturing method , the material properties , the tests performed , and estimates of the loss that can be achieved in practical lenses . | silicon is a promising lens material for millimeter wavelength observations because it can be machined ; it has a high index of refraction , which is optically advantageous ; and it has a high thermal conductivity , allowing for straightforward cooling of the lenses to cryogenic temperatures .
however , the high index ( @xmath0 ) in the submillimeter region@xcite leads to a reflection at each silicon / vacuum interface of approximately @xmath1 ^ 2\approx30\%$ ] per surface .
this is prohibitively large , especially for multi - lens cameras .
nevertheless , the benefits of silicon have motivated the development of antireflection coatings , @xcite where the referenced ar solutions are in ir and thus easier than mm - wave bands due to the thickness of the ar layer .
this paper details the development and testing of a simple antireflection coating that reduces reflection to @xmath2 per lens at the design wavelength while maintaining @xmath3 transmission at @xmath4 300ghz . at a fixed wavelength
the ideal , normal - incidence antireflection coating for a substrate of index @xmath5 in vacuum has an index of refraction of @xmath6 and is @xmath7 thick . in our application
, we are building lenses for the atacama cosmology telescope ( act ) camera@xcite at 145 , 217 , and 265ghz to measure the fine scale anisotropy of the cosmic microwave background . in the following ,
we focus on the 150ghz band , for which the ideal antireflection coating has @xmath8 m and index @xmath9 .
the coating is a machined piece of cirlex^^ @xcite polyimide glued to silicon with stycast^^ 1266 epoxy @xcite and lord ap-134 adhesion promoter@xcite . for the curved lens surface ,
a piece of cirlex approximately 1 cm thick is machined to the curved shape and then held in a teflon gluing jig shaped to match the lens surface while the epoxy cures .
the low - frequency ( @xmath10khz ) dielectric constant and loss reported in the kapton polyimide data sheet@xcite suggest that polyimide and silicon could be combined in an ar configuration . to ensure accurate modeling and to test sample dependent effects
, we measured the dielectric properties with fourier transform spectrometers ( fts ) , summarized in table [ matprop ] . * * .dielectric properties of the materials for ar coating.[matprop ] [ cols="<,<,>",options="header " , ] of the coated 4mm - thick silicon flat ( flat 7 ) , both modeled ( black ) and measured on the fts ( gray ) .
the measurement is the ratio of a sample to a reference spectrum .
the lower curve shows that the difference ( measurement minus model ) is within 5% of zero through the well - measured range .
the high transmission near 133 and 400ghz is due to the ar coating being @xmath11 and @xmath12 thick .
the slow reduction in @xmath13 with increasing frequency is due to increasing loss in the coating and glue .
this sample was made before precise values of the index of cirlex and stycast 1266 were known .
thus , the center of the passband window , 133ghz , is 15ghz below our target frequency . ] rather than trying to interpret the lens results , we measured the transmission of the two coated flats labeled flat 6 and 7 in table [ befafter ] . figure [ data ] shows the transmission spectra for one of these samples along with a model .
the measurement is the ratio of a sample to a reference spectrum , which are averages over two and six spectra , respectively .
the model is _ not _ a fit to the coated transmission data but is determined instead by the cirlex , stycast , and silicon properties given in table [ matprop ] and by measurements of the component thicknesses .
the one exception is the silicon loss .
the coated flats have somewhat lower resistivity ( between 1300 and @xmath14-cm , as measured by the vendor ) than the uncoated silicon samples ( all specified to exceed @xmath15-cm ) ; both sets have poorly constrained @xmath16 . to handle the uncertain silicon loss ,
we have treated the resistivity of the sample as an unknown and varied it to fit the measured transmission .
a finished design of a silicon antireflection coating for any application requires a complete model of the system .
incident angle range , frequency bandwidth , and polarization all affect the optimal coating thickness . it is helpful to begin the process with a few estimates and approximate guidelines , however , and we offer some here . first , using equation [ dielconst ] and recalling that power loss is one @xmath17 per radian of phase , we find that absorption loss in 10k@xmath18-cm silicon should equal 1% per centimeter , scaling as @xmath19 . for normally incident light , the best ar - coating thickness @xmath20 can be estimated by requiring that the optical path through both materials ( coating and glue ) equal one - quarter wave .
that is , @xmath21 comparing this rule against the calculated reflection , we find that it overestimates the optimal ar thickness @xmath20 by approximately @xmath22 m for a glue layer @xmath23 m thick .
the difference increases quadratically with glue thickness , but the approximate expression is adequate for any reasonable size of the glue layer .
m of glue and enough cirlex so that @xmath24 equals one - quarter of the vacuum wavelength .
solid lines indicate normal incidence ; dashed lines are for @xmath25 incident angle . the lines labeled _ a5 _ and _ a20 _
show the absorption loss .
the _ r20 _
lines give the reflection from the 20 mm lens ; the results are not visibly different for reflection from the 5 mm lens .
absorption increases for oblique angles and at higher frequencies .
we show the expected loss at room temperature in 5000@xmath18-cm silicon .
cryogenically , absorption loss should be reduced . ]
figure [ fig.estimate ] shows the predicted absorption and reflection loss calculated for a pair of realistic coated lenses , both at @xmath26 and @xmath25 incident angles .
the model assumes lenses made of 5000@xmath18-cm silicon , 5 or 20 mm thick , and unpolarized light . over most of the frequency range
it should be possible to achieve better than 2% reflection per lens and less than 10% absorption particularly if the lens is thin or silicon resistivity is greater than the @xmath27-cm assumed here .
it might be possible to use the stycast 1266 alone as an ar coating .
its index of 1.68 is lower than the ideal 1.85 , and the loss is approximately double the loss of cirlex .
however , cutting a single mold to shape the curing epoxy saves three machining steps when compared with the method for cutting cirlex coatings described in this article . at 150 and 300ghz ,
the optimal stycast thickness are 300 and 140@xmath28 m , which offer transmission of 90% and 88% , respectively .
this compares unfavorably with the 95% and 93% transmission offered by a cirlex - coated lens .
the stycast - coated lens would reflect approximately 1.5% in both frequencies , three times the reflection expected from a cirlex - coated lens .
still , a simpler stycast - only coating might suffice in some applications if the thickness and shape could be controlled well enough .
we have developed and tested a technique for antireflection coating silicon lenses at cryogenic temperatures at millimeter wavelengths .
flat samples show @xmath29 1.5% reflection and @xmath30 92% transmission at the design frequency .
the remaining 6.5% is attributable to absorption loss , which will decrease upon cooling .
the authors are very grateful to john ruhl for sharing his fourier transform spectrometer and to norm jarosik for operating a vector network analyzer .
we thank the princeton university physics department machine shop , especially glenn atkinson , for developing techniques to machine and glue the cirlex coatings .
sarah marriage tested different ways of gluing silicon to polyimide ; ted gudmundsen and adrian liu repeatedly cycled coated silicon samples between 77 and 300k .
we are grateful to our colleagues on the act collaboration and to asad aboobaker , andrew bocarsly , mark devlin , simon dicker , phil farese , jeff klein , jeff mcmahon , amber miller , mike niemack , suzanne staggs , and zachary staniszewski for many helpful discussions .
this work was supported by the u.s .
national science foundation through awards ast-0408698 for the act project and phy-0355328 for the princeton gravity group .
n. g. ugras , j. zmuidzinas , and h. g. leduc , `` quasioptical sis mixer with a silicon lens for submillimeter astronomy , '' in _ proceedings of the 5th international symposium space terahertz technology , _ 125 . | silicon promising lens material millimeter wavelength observations machined high index refraction optically advantageous high thermal conductivity cooling lenses cryogenic temperatures high index @xmath0 submillimeter region leads reflection silicon vacuum interface approximately @xmath1 per surface prohibitively large for multi - lens cameras benefits silicon motivated development antireflection coatings paper details development antireflection coating reduces reflection to @xmath2 per lens wavelength transmission at 300ghz wavelength ideal normal - incidence antireflection coating substrate index @xmath5 vacuum index refraction @xmath6 @xmath7 thick building lenses for atacama cosmology telescope camera at 145 217 265ghz measure fine scale anisotropy cosmic microwave background focus on 150ghz band ideal antireflection coating @xmath8 m index @xmath9 coating machined cirlex^^ @xcite polyimide glued to silicon with 1266 epoxy lord ap-134 adhesion promotercurved lens surface cirlex 1 cm thick machined to curved shape held in teflon gluing jig lens while epoxy cures low - frequency @xmath10khz ) dielectric constant loss kapton polyimide data sheet suggest polyimide silicon combined in ar configuration test sample effects measured dielectric properties with fourier transform spectrometers summarized in table properties materials for ar coating coated 4mm - thick silicon flat 7 ) modeled measured fts measurement ratio sample to reference spectrum difference within 5% of zero measured range high transmission near 133 and 400ghz due to ar coating @xmath11 @xmath12 thick slow reduction in @xmath13 increasing frequency due to increasing loss in coating glue sample made before precise values index cirlex stycast 1266 known center passband window 133ghz 15ghz below target frequency measured transmission of two coated flats flat 6 and 7 table figure shows transmission spectra measurement ratio sample to reference spectrum averages over two six spectramodel not to coated transmission data determined by cirlex stycast silicon properties in table matprop measurements component thicknesses exception is silicon loss coated flats lower resistivity between 1300 and @xmath14-cm than uncoated silicon samples exceed @xmath15-cm poorly constrained @xmath16 silicon loss treated resistivity unknown varied to fit transmission finished design silicon antireflection coating requires complete model incident angle range frequency bandwidth polarization affect optimal coating thickness begin process with estimates approximate guidelines power loss per radian phase absorption loss in 10k@xmath18-cm silicon equal 1% per centimeter scaling as @xmath19 for light best ar - coating thickness @xmath20 estimated optical path through materials equal one - quarter wave overestimates optimal ar thickness @xmath20 by @xmath22 m for glue layer @xmath23 m thick difference increases with glue thickness approximate expression adequate for size glue layer glue cirlex @xmath24 equals one - quarter of vacuum wavelengthlines indicate normal incidence dashed lines @xmath25 incident angle lines _ a5 a20 show absorption loss r20 _ lines reflection from 20 mm lens different 5 mm lens absorption increases oblique angles higher frequencies expected loss at room temperature 5000@xmath18-cm silicon cryogenically absorption loss reduced figure shows predicted absorption reflection loss realistic coated lenses @xmath26 @xmath25 incident angles model assumes lenses 5000@xmath18-cm silicon 5 or 20 mm thick unpolarized light frequency range 2% reflection per lens less than 10% absorption lens thin silicon resistivity greater use stycast 1266 coating index 1.68 lower than ideal 1.85 loss double cirlex cutting single mold saves three machining steps at 150 300ghz optimal stycast thickness 300 140@xmath28 m transmission 90% 88% compares with 95% 93% transmission cirlex - coated lens stycast - coated lens 1.5% in both frequencies three times reflection simpler stycast - only coating might suffice if thickness shape controlleddeveloped tested technique antireflection coating silicon lenses cryogenic temperatures millimeter wavelengths samples show 1.5% reflection 92% transmission design frequency remaining 6.5% absorption loss cooling grateful john ruhl norm jarosik vector network analyzer princeton university physics department machine shop glenn atkinson techniques glue coatings sarah marriage tested gluing silicon polyimide ted gudmundsen adrian liu cycled coated silicon samples between 77 300k grateful colleagues asad aboobaker andrew bocarsly mark devlin dicker farese klein mcmahon amber miller niemack staggs staniszewski supported u national science foundation awards ast-0408698 phy-0355328 princeton gravity group ugras zmuidzinas leduc quasioptical sis mixer silicon lens submillimeter astronomy proceedings 5th international symposium space terahertz technology | 0.31929 | 2,006 | 1,026 | , Saving $0.1 in GPT-4. |
177,508 | the spin echo techniques aim at the elimination of the effect of a random magnetic field on the spin evolution .
these techniques conventionally utlize the application of a permanent field which is much stronger than the random one . the strong field
, however , may also modify the magnetic response of the medium containing the spins , thus altering their `` natural '' dynamics .
we suggest an iterative scheme for generating a sequence of pulses which create an echo without an external permanent field .
the approximation to the ideal echo improves with the sequence length . | the use of echoes is a standard technique in spin resonance experiments @xcite , where a large number of echo pulse sequences are used .
echo sequences typically consist of a number of high - frequency pluses that induce controlled rotations on the precessing spin .
the objective is to remove or reduce dephasing due to inhomogeneities in the external magnetic field which give a variation of the precession rate across the sample ( known as inhomogeneous broadening ) .
since any two - level quantum system can be mapped on a spin-@xmath0 in a magnetic field , similar operations can be performed on any such system , including a qubit .
for example , there are recent experiments on superconducting qubits @xcite ( were it is called charge echo , since the physical degree of freedom is two charge states of a superconducting grain ) and electron spins in quantum dots @xcite .
common to the existing echo techniques is the restriction that in order to apply the proper sequence of control rotations one must know the direction of the magnetic field . in other words ,
the sequences are able to remove dephasing due to inhomogeneities in the field amplitude , but not the field direction ( the randomness in the direction of the magnetic field is indeed negligible if a strong external field is applied ) .
starting with the experiments @xcite that showed long spin coherence times of electrons in bulk gaas there has been interest in using electrons confined in quantum dots as qubits .
it appears that the limiting decoherence factor in these systems is the hyperfine interaction between the electron and nuclear spins @xcite .
as long as no special preparation is made , the nuclear spin system will give rise to an effective random magnetic field seen by the electron .
this field will be random both in strength and direction , and in the absence of an external strong permanent magnetic field the usual echo sequences can not be applied . because the response of the nuclear spin system is much slower than the precession of the electron spin we can in a certain approximation assume the nuclear field to be constant . motivated by this we will in this paper investigate the problem of finding an echo sequence that is applicable in the case of a constant effective field with an arbitrary and unknown magnitude and direction .
the goal of this sequence is to return the spin to its starting position at the end of the sequence , independently of the magnetic field .
consider a spin-@xmath0 particle precessing in a constant magnetic field .
let @xmath1 be the unit vector along the precession axis and @xmath2 be the precession angular frequency .
the rotation of the spin state during time @xmath3 is then given by the unitary operator @xmath4 where @xmath5 and @xmath6 are the identity and pauli matrices , and @xmath7 .
the @xmath8 rotations about the coordinate axes are denoted @xmath9 , @xmath10 , and @xmath11 .
the simplest form of a spin echo is for the situation where the spin is precessing in a field with known direction , say along the @xmath12-axis , but with unknown magnitude .
then @xmath13 and @xmath14 , while @xmath15 is arbitrary . the usual echo sequence then consists in waiting for the time @xmath3 , applying an @xmath16 rotation , waiting time @xmath3 and applying a final @xmath16 rotation .
the success of the procedure is expressed by the fact that @xmath17 is the identity ( up to a global sign , which is unimportant ) , independently of @xmath18 . by symmetry ,
the same is true if @xmath16 is replaced by a @xmath8 rotation around any axis in the @xmath19-plane ( perpendicular to the precession axis ) .
in particular we have @xmath20 .
we want to extend this to the case were @xmath1 is not known .
that is , we want to find a sequence of control rotations @xmath21 such that @xmath22 for any @xmath18 .
we make the following assumptions : i ) the control pulses are effectively instantaneous , meaning they can be performed in a time much less than the precession period .
ii ) the external magnetic field is unknown but constant , so that the precession operator @xmath18 does not change in time .
note that we have written eq .
( [ f ] ) as if the time @xmath3 between the pulses is fixed .
different time intervals between the pulses are achieved by choosing some of the @xmath21 to be the identity .
this is sufficient if all intervals are integer multiples of a smallest unit .
intervals with irrational ratios will require a more general form than eq .
( [ f ] ) . in this paper
we only use equal intervals .
as was explained above , if we know the direction @xmath1 of the external field we can create an echo by applying two @xmath8 pulses around any axis perpendicular to @xmath1 .
the sequence @xmath23 can be used if @xmath24 , and similarly we may use @xmath25 if we know that @xmath26 . consider the longer sequence @xmath27 for which , using the general @xmath18 of eq .
( [ u ] ) , we get @xmath28 we see that @xmath29 for either @xmath24 or @xmath26
. this may seem like a small gain , but this sequence is the key to the full solution . let us construct the mapping @xmath30 the idea is that @xmath31 , being composed of the arbitrary @xmath18 and the fixed control rotations , will be `` less arbitrary '' than the original @xmath18 . iterating this mapping
we then construct the set @xmath32 , @xmath33 , ... of pulse sequences that will be better and better approximations to the identity . since @xmath34 contains @xmath35 four times
the time needed for the sequence @xmath35 is @xmath36 .
that is , the length of the sequence grows exponentially in @xmath37 .
this is of course unfortunate as the sequences quickly will become impractically long .
however we will see below that for a large portion of the space of parameters determining @xmath18 , a few iterations @xmath37 are sufficient to reach a good approximation to the identity .
the domain of `` bad '' parameters shrinks exponentially with the increase of @xmath37 . to illustrate this we represent a rotation by the polar angle @xmath38 and azimuth angle @xmath39 of the rotation axis ( so that @xmath40 , @xmath41 and @xmath42 ) and the rotation angle @xmath15 .
+ + figure [ f1 ] shows the evolution of the parameters @xmath43 upon successive transformations given by eq .
( [ t ] ) .
the set of initial rotations fills this space uniformly , see fig .
[ f1 ] ( a ) .
figures [ f1 ] ( b)-(f ) show the successive iterations @xmath44 - @xmath45 of these points , illustrating the convergence of the mapping . from eq .
( [ it ] ) we can write down explicit formulas for the parameters @xmath46 of @xmath47 in terms of @xmath48 , @xmath49 we see that @xmath50 , @xmath14 is a fixed point , and this represents the identity operator .
the stability of the fixed point can be analyzed by expanding close to the fixed point in the independent small quantities @xmath51 , @xmath52 and @xmath53 .
we get @xmath54 were the primed quantities refer to the transformed rotation @xmath55 .
it is clear that if @xmath56 are small quantities , then @xmath57 are even smaller and the fixed point is locally stable .
we studied numerically the convergence of the mapping @xmath31 for points that are not close to the fixed point .
looking at figure [ f1 ] we see that although most points converge to the vicinity of the fixed point in a few iterations , there are some points that do not converge fast . to get a better understanding we do as follows .
let us choose some initial rotation angle @xmath15 and for each point in the @xmath58 plane construct the sequence @xmath44 , @xmath59 , ... stopping when @xmath35 is within a specified distance from the identity ( we used the stopping criterion @xmath60 ) .
+ + + figure [ f2 ] shows the @xmath58 plane for various @xmath15 , the shades of gray representing the number @xmath37 of iterations needed for convergence .
we see that for small @xmath15 convergence is fast and the pattern is simple , but for larger initial @xmath15 the pattern of convergence time is quite complex , and that there exist `` hard '' points , i.e. initial rotations @xmath48 that need a large number @xmath37 if iterations to converge . to estimate the fraction of initial rotations that need a certain number of iterations to converge to the fixed point we do the following .
we start with an ensemble of random initial rotations @xmath18 characterized by a unit vector @xmath1 distributed uniformly on a sphere and by an angle @xmath15 distributed uniformly from in the inerval @xmath61 $ ] .
we proceed with the iterations , eq .
( [ t ] ) , until @xmath60 is reached .
the logarithm of the fraction @xmath62 of initial rotations @xmath18 that needs @xmath37 iterations to converge is shown in figure [ f5 ] as a function of n. of the initial random set of rotations which did not reach the condition @xmath63 till the @xmath37-th iteration .
[ f5],width=302 ] we find that about 74% of the initial rotations will converge after applying the third iterate @xmath64 and about 91% do so after the iterate @xmath65 .
observe that except for the first points , all points fall on a straight line which means that the fraction @xmath62 of `` difficult '' initial rotations decays exponentially with the number of iterations , @xmath66 if @xmath67 is the number of repetitions of @xmath18 in @xmath31 ( for the mapping ( [ t ] ) we have @xmath68 ) , the total time of the sequence @xmath35 is @xmath69 , where @xmath3 is the time between the pulses . the relation ( [ a ] )
can then also be written as the fraction @xmath70 of the initial rotations which did not converge till time @xmath71 , @xmath72 the relations eqs .
( [ a ] ) and ( [ t ] ) are established by running a simulation with the use of a specific convergence criterion , @xmath73 .
changing the criterion affects the proportionality coefficients in these relations , but does not change the values of exponents @xmath74 and @xmath75 . of initial rotations that do not converge in nine iterations , according to the criterion @xmath76 .
[ f3],width=302 ] let @xmath77 denote the set of initial rotations that do not converge in @xmath37 iterations .
[ as an example , the domain @xmath78 of `` difficult '' initial rotations ( according to the criterion @xmath76 ) is presented in fig .
the sequence of sets @xmath77 is such that each set is contained in the previuos one , @xmath79 , and the fraction of points @xmath62 in figure [ f5 ] is proportional to the difference in the volumes of @xmath80 and @xmath81 .
the set @xmath82 of `` infinitely hard '' points appears to be a fractal with fractal dimension @xmath83 according to the box - counting algorithm we used .
we considered the possibility of constructing an echo pulse sequence that does not require applying a high permanent field to the system .
the advantage of such a method is that between the pulses the dynamics of the system is not influenced by external perturbations . in the context of esr of a quantum dot
, this method may help distinguishing between different effects of the hyperfine interaction : this interaction creates some random effective magnetic field acting on the electron spin , but may also lead to electron spin relaxation .
the absence of the external permanent field in this problem is crucial , as its application definitely suppresses the spin relaxation part of the hyperfine interaction effects @xcite .
we have provided a solution to the general echo problem in terms of a set of longer and longer pulse sequences that give successively better approximations to the identity operator .
we have tested several longer sequences _ e.g _ @xmath84 . in all cases we found that the exponent @xmath75 in eq .
( [ t ] ) is independent of the particular mapping chosen .
whether this represents some inherent property of the problem or only is the case for the limited class of mappings we have studied is not known .
for example , we have not studied sequences with control pulses other than @xmath8 rotations about the coordinate axes .
we have also not ruled out the possibility of a solution that will yield an ideal echo with the help of a finite number of control pulses .
the idea of using iterated mappings to generate pulse sequences has been used in nmr applications @xcite but as far as we know this particular problem or the mapping we study was never discussed .
the mapping we have used was also proposed in ref . in the context of dynamical decoupling of a qubit .
this work was supported by the norwegian research council via a storforsk program and by nsf grants dmr 02 - 37296 and dmr 04 - 39026 at the university of minnesota .
the hospitality of mpipks - dresden ( l.i.g . )
is gratefully acknowledged . | echoes standard technique in spin resonance experiments @xcite large echo pulse sequences used echo sequences high - frequency pluses induce controlled rotations on precessing spin objective to remove dephasing due to inhomogeneities external magnetic field variation precession rate inhomogeneous broadening two - level quantum system can be mapped on spin-@xmath0 magnetic field similar operations performed on any system including qubit recent experiments on superconducting qubits @xcite charge echo electron spins in quantum dots @xcite echo techniques restriction know direction magnetic field sequences remove dephasing inhomogeneities field amplitude not field direction randomness field negligible if strong external field experiments @xcite showed long spin coherence electrons in bulk gaas interest in using electrons confined in quantum dots as qubits limiting decoherence factor hyperfine interaction between electron and nuclear spins @xcite no special preparation nuclear spin system random magnetic field seen by electron field random in strength and direction absence external strong permanent magnetic field usual echo sequences not be applied response nuclear spin system slower than electron spin assume nuclear field constantinvestigate finding echo sequence constant field with arbitrary unknown magnitude direction goal return spin to starting position independently magnetic field consider spin-@xmath0 particle precessing in constant magnetic field @xmath1 unit vector precession axis @xmath2 precession angular frequency rotation spin state during time @xmath3 given by unitary operator @xmath4 @xmath5 @xmath6 identity pauli matrices @xmath7 @xmath8 rotations about coordinate axes denoted @xmath9 @xmath10 @xmath11 simplest spin echo spin precessing in field known direction @xmath12-axis unknown magnitude @xmath13 @xmath14 @xmath15 arbitrary echo sequence waiting for time @xmath3 applying @xmath16 rotation @xmath3 final @xmath16 rotation success @xmath17 is identity global sign unimportant independently of @xmath18 same if @xmath16 replaced by @xmath8 rotation around axis @xmath19-plane perpendicular to precession axis @xmath20 extend to @xmath1 not known find sequence of control rotations @xmath21 @xmath22 for @xmath18assumptions control pulses instantaneous performed time less than precession period external magnetic field unknown constant precession operator @xmath18 change time written eq time @xmath3 between pulses fixed different time intervals achieved by choosing @xmath21 identity sufficient if intervals integer multiples of smallest unit intervals irrational ratios require general form use equal intervals if know direction @xmath1 of external field create echo by applying two @xmath8 pulses around axis perpendicular to @xmath1 sequence @xmath23 used if @xmath24 @xmath25 if know @xmath26 consider longer sequence @xmath27 general @xmath18 eq get @xmath28 @xmath29 for @xmath24 or @xmath26 sequence key to full solution construct mapping @xmath30 @xmath31 arbitrary @xmath18 fixed control rotations less arbitrary than original @xmath18 iterating mapping construct @xmath32 @xmath33 of pulse sequences better approximations to identity @xmath34 contains @xmath35 four times time needed for sequence @xmath35 @xmath36 length sequence grows in @xmath37 unfortunate sequences impractically long large portion parameters @xmath18 few iterations @xmath37 sufficient good approximation to identity domain bad parameters shrinks with increase @xmath37 represent rotation by polar angle @xmath38 azimuth angle @xmath39 of rotation axis @xmath40 @xmath41 @xmath42 rotation angle @xmath15 figure f1 shows evolution parameters @xmath43 upon transformations initial rotations fills space uniformly fig f1 figures f1- show successive iterations @xmath44 - @xmath45 points illustrating convergence mapping explicit formulas for parameters @xmath46 @xmath47 @xmath48 @xmath49 @xmath50 @xmath14 fixed point represents identity operator stability fixed point analyzed by expanding close to fixed point in small quantities @xmath51 @xmath52 @xmath53 @xmath54 primed quantities refer to transformed rotation @xmath55 if @xmath56 small quantities @xmath57 smaller fixed point locally stable studied convergence mapping @xmath31 for points not close to fixed point figure f1 most points converge to fixed point in few iterations some not converge fastunderstanding choose initial rotation angle @xmath15 each point @xmath58 plane construct sequence @xmath44 stopping when @xmath35 within distance from identity used stopping criterion @xmath60 figure f2 ] shows @xmath58 plane @xmath15 shades gray @xmath37 iterations needed for convergence small @xmath15 convergence fast pattern simple larger complex hard points initial rotations need large @xmath37 iterations to converge estimate fraction initial rotations certain iterations converge fixed point start with random initial rotations @xmath18 unit vector @xmath1 distributed uniformly sphere angle @xmath15 distributed uniformly inerval @xmath61 proceed iterations until @xmath60 reached logarithm of fraction @xmath62 initial rotations @xmath18 @xmath37 iterations to converge shown in figure [ f5 ] function of n initial random set rotations reach condition @xmath63 till @xmath37-th iteration 74% initial rotations converge after third iterate @xmath64 91% after @xmath65except first points all points fall straight line fraction @xmath62 of difficult initial rotations decays with number iterations @xmath66 @xmath67 is number repetitions of @xmath18 @xmath31 mapping @xmath68 total time sequence @xmath35 is @xmath69 @xmath3 time between pulses relation ( [ a ] ) written as fraction @xmath70 of initial rotations @xmath71 @xmath72 relations t established by simulation specific convergence criterion @xmath73 changing criterion affects proportionality coefficients change values exponents @xmath74 @xmath75 initial rotations converge in nine iterations criterion @xmath76 @xmath77 initial rotations converge in @xmath37 iterations domain @xmath78 of difficult initial rotations presented in fig sequence of sets @xmath77 each set contained in previuos one @xmath79 fraction of points @xmath62 proportional to difference volumes of @xmath80 and @xmath81 set @xmath82 of infinitely hard points fractal with dimension @xmath83 - counting considered possibility constructing echo pulse sequence high permanent fieldadvantage of method is between pulses dynamics system not influenced by external perturbations in context quantum dot method may effects hyperfine interaction interaction creates random magnetic field electron spin may to electron spin relaxation absence of external permanent field crucial suppresses spin relaxation hyperfine interaction effects provided solution to general echo problem longer pulse sequences better approximations to identity operator tested longer sequences @xmath84 found exponent @xmath75 in eq t independent of mapping inherent property problem or limited class mappings not known not studied sequences with control pulses other than @xmath8 rotations about coordinate axes not ruled out possibility of solution ideal echo with finite number of control pulses idea of using iterated mappings to generate pulse sequences used in nmr applications @xcite this problem or mapping never discussed mapping used proposed in context of dynamical decoupling of qubit work supported by norwegian research council storforsk program nsf grants at university of minnesota hospitality of mpipks - dresden acknowledged | 0.440479 | 2,962 | 1,535 | , Saving $0.1 in GPT-4. |
177,518 | the model of community isolation was extended to the case when individuals are randomly placed at nodes of hierarchical modular networks .
it was shown that the average number of blocked nodes ( individuals ) increases in time as a power function , with the exponent depending on network parameters .
the distribution of time when the first isolated cluster appears is unimodal , non - gaussian .
the developed analytical approach is in a good agreement with the simulation data . | recently , hierarchical systems have been attracting attention of scientists working on complex networks @xcite . in fact
many real networks are hierarchically organized , e.g. www network , actor network , or the semantic web @xcite .
dynamics at such networks can be qualitatively and quantitatively different from that at regular lattices ( see @xcite ) . the ising model at a network with a hierarchical topology was studied by komosa and hoyst @xcite .
the analyzed parameters were , among others , magnetization , magnetic susceptibility , critical temperature and correlations of magnetization between different hierarchies .
it was shown that the critical temperature is a power function of the network size and of the ratio @xmath0 , where @xmath1 stands for a node degree .
opinion formation in hierarchical organizations was studied by laguna et al .
agents , belonging to various authority strata , try to influence others opinions .
the probability that an opinion of an agent of a certain authority prevails in the community depends on the size distribution of the authority strata .
phase diagrams can be obtained , where each phase corresponds to a distinct dominant stratum ( or a sequence of the strata , with the decreasing probability of prevailing ) .
fashion phenomena at hierarchical networks were studied by galam and vignes @xcite .
interactions were imposed between social groups at different levels of hierarchy .
a renormalization group approach was used to find the optimal investment level of the producer and to assess the influence of counterfeits on the probability of a new product success .
one of fundamental topics in social dynamics are conflict situations and many different sociophysics approaches @xcite or _
prisoner s dilemma_-type games @xcite have been proposed . recently
a simple model of communities isolation has been introduced by sienkiewicz and hoyst @xcite .
the model can describe such various issues as strategy at battlefields or formation of cultures .
the idea behind this model is similar to the game of go and it takes into account a natural leaning of people to avoid being surrounded by members of another ( potentially hostile ) community @xcite . in this paper
we extend the model of communities isolation studied for chains , hypercubic , random and scale - free networks @xcite to hierarchical networks proposed by @xcite .
the model of hierarchical networks was proposed by ravasz and barabsi @xcite and modified by suchecki and hoyst @xcite .
such networks possess 3 parameters determining their structure : * the degree of hierarchy @xmath2 * the distribution @xmath3 , where @xmath4 , determining number of nodes at each level of hierarchy ( in particular , the size of the cliques at the lowest level of hierarchy is @xmath5 ) * the parameter determining the density of edges @xmath6 $ ] two models ( referred to as p1 and pd models ) were analyzed , which differ in the density of edges .
each network has a central node , referred to as a _ center of hierarchy_. a network of hierarchy @xmath7 is a complete graph of size @xmath5 ( @xmath8 is a random number , chosen with probability @xmath3 ) .
the _ center of hierarchy _ , due to the symmetry , is an arbitrary node . in order to construct a network of hierarchy @xmath9 ,
one has to construct @xmath5 subnetworks of hierarchy @xmath10 and choose one of them its _ center of hierarchy _ becomes a _ center of hierarchy _
@xmath11 of the whole network .
afterwards , new connections ( edges ) are created : for each node @xmath12 of @xmath8 remaining subnetworks a connection ( edge ) @xmath13 is created with probability @xmath14 ( in case of the p1 model ) or @xmath15 ( in case of the pd model ) .
sample networks created this way are presented in fig .
[ fig : sample_networks ] .
let us stress that the subnetworks do not have to be connected , especially if @xmath14 is small .
, with different degrees of hierarchy : ( a ) @xmath7 , ( b ) @xmath16 , ( c ) @xmath17 , ( d ) @xmath18 some basic properties of such networks can be concluded from the construction algorithm : * for @xmath19 , as well as for @xmath20 , models p1 i pd are equivalent . * for @xmath21 the network consists of isolated cliques of size @xmath5 ( @xmath8 random variable ) . * for @xmath22
the number of nodes ( vertices ) of the network equals @xmath23 .
periodic oscillations in degree distribution of such networks can be observed in log - log scale .
the period , the amplitude and the shape of the peaks depend on the parameters of the network @xcite . in this paper
only the case with @xmath22 ( @xmath24 ) was considered , which corresponds to the original ravasz and barabsi model @xcite .
the model of communities isolation was proposed by sienkiewicz and hoyst @xcite .
the rules are similar to those of the game of go .
a number of communities compete with each other , settling nodes of a network . in each step
a random empty node is chosen .
it is then settled by a member of randomly chosen community .
a cluster of nodes occupied by one community becomes blocked when it gets surrounded by another community .
the surrounded nodes are no more active in the game , i.e. they can not take part in surrounding other communities .
the case of communities competing at a chain was analyzed in @xcite .
two functions describing the evolution were studied : the average number of blocked nodes over time and a mean critical time , i.e. the moment , when the first blocked cluster appears . in @xcite the influence of external bias was considered when settling rates of competing communities are different . in this paper
the case of two competing communities at p1 and pd hierarchical networks is considered .
two parameters are analyzed : the average number of blocked nodes @xmath25 and the critical time distribution @xmath26 .
for @xmath21 models p1 are pd equivalent .
the network consists of @xmath28 isolated cliques of size @xmath5 . in such case @xmath29 where @xmath30 probability that in the @xmath31th step @xmath32 nodes will be blocked , @xmath33 after short algebra we obtain @xmath34 the solution of this recursive equation is a @xmath35th degree polynomial , which can be approximated by substituting the sum with the integral : @xmath36 as one can see , @xmath25 is a power function .
the exponent @xmath37 depends only on the @xmath8 parameter , @xmath38 .
in this case models p1 and pd are also equivalent . for networks of hierarchy
@xmath16 : @xmath40 where @xmath41 is a _ reduced density _ : @xmath42 for networks of higher hierarchies , @xmath43 , a recursive equation well approximating @xmath44 can be derived .
the idea behind the formulas is that a clique can only be blocked if all the nodes of higher hierarchies neighboring with it are filled . therefore @xmath45 if the center of hierarchy of the network ( which neighbors with all the other nodes ) is empty . in the opposite case ,
@xmath44 depends on @xmath46 , which describes each of @xmath5 _
subnetworks_. @xmath47 this equation can only be solved numerically .
the solutions are presented in fig .
[ fig : z_p1 ] and [ fig : z_pd ] . it can be noticed , that within a wide range of time @xmath48 , @xmath49 can be with reasonable accuracy approximated with a power function @xmath50 where the exponents @xmath37 are higher than in the case of @xmath27 and they are close to @xmath51 . an analytical approximation of @xmath44 for networks with higher hierarchies ( @xmath52 ) when the parameter @xmath14 is different from zero and one is far more difficult . instead of searching for such a formula , an alternative approach was chosen .
it was assumed that @xmath44 can be estimated from the proportion @xmath53 where @xmath54 $ ] should be an increasing function of @xmath14 which , while not being too complicated , would give a reasonable approximation for the widest possible ranges of @xmath14 and @xmath55 .
it turned out that in the case of the p1 model , choosing @xmath56 results in a good agreement of the @xmath44 function with simulation data . for the pd model
, @xmath57 is a good choice . , for various networks of p1 model .
symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and ) . left side linear scale , right side log - log scale.,title="fig : " ] , for various networks of p1 model .
symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and ) . left side linear scale , right side log - log scale.,title="fig : " ] , for various networks of p1 model .
symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and ) . left side linear scale , right side log - log scale.,title="fig : " ] , for various networks of pd model
. symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and )
. left side linear scale , right side log - log scale.,title="fig : " ] , for various networks of pd model .
symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and ) . left side linear scale , right side log - log scale.,title="fig : " ] , for various networks of pd model .
symbols correspond to data from computer simulations .
lines show analytical approximations ( eq . , and ) .
left side linear scale , right side log - log scale.,title="fig : " ]
as it was previously mentioned , in the case of @xmath27 the network consists of @xmath58 isolated cliques of @xmath5 nodes . in order to find the distribution of critical time ( time , when the first blocked cluster appears )
, one has to consider the probability that at time @xmath48 there are no blocked nodes yet .
it means that at time @xmath48 the only completely filled cliques are those filled with members of one community , which leads to the formula @xmath59 where @xmath60 .
the cumulative critical time distribution can be immediately obtained @xmath61 as well as the critical time distribution in the approximation of continuous time : @xmath62 the mean critical time can be also calculated analytically : @xmath63 where @xmath64 ( euler beta function ) and @xmath65 ( incomplete euler beta function ) . for networks of hierarchy
@xmath7 @xmath66 for networks with hierarchy @xmath16 @xmath67 for networks with any degree of hierarchy , @xmath68 , a recursive formula for the cumulative critical time distribution can be expressed as @xmath69 [ cols="^,^ " , ] the mean critical time can be obtained by numerical integration of @xmath70 : @xmath71
in all cases the function @xmath25 , defined as the average number of blocked nodes at time @xmath48 , can be approximated with a high accuracy by a power function @xmath72 the @xmath37 exponent depends on the parameters of the network . for @xmath73-dimensional hypercubic networks ( including the @xmath74-dimensional ones , i.e. chains ) @xmath75 . for modular hierarchical networks
@xmath37 depends mainly on @xmath8 and @xmath14 parameters , i.e. on the sizes of basic cliques at the lowest hierarchy and on the density of inter - clique connections .
the dependence on the degree of hierarchy @xmath55 ( and on the network size ) is weak , what can be explained by the fact , that increasing the degree of hierarchy @xmath55 is a process similar to system rescaling , therefore @xmath76 for @xmath27 , the parameter @xmath37 can be analytically found : @xmath77 .
the result is in agreement with the simulated data .
increasing the density of connections ( the @xmath14 parameter ) leads to the increase of @xmath37 up to approximately @xmath51 for @xmath39 .
there is an important distinction in the way @xmath25 was approximated for hypercubic and hierarchical networks . for hypercubic networks ,
the number of isolated nodes was calculated using the following approximation : all blocked nodes were blocked alone , i.e. they do not neighbor with other blocked nodes ( of the same community ) .
although this approximation might seem coarse , resulting analytical predictions turned out to be in quite good agreement with simulated data @xcite . for modular hierarchical networks
, such an approximation would not be reasonable . because of the fact that at the lowest level of hierarchy such networks consist of cliques of @xmath5 nodes , the most probable are situations when @xmath78 nodes are simultaneously blocked .
the second analyzed parameter was critical time @xmath79 , i.e. the moment , when the first isolated cluster appears .
it is a random variable .
the critical time distribution @xmath26 was studied , as well as mean critical time @xmath80 .
more precisely , _ a critical density _ ( or _ a critical relative time _ ) @xmath81 was often shown so networks with different parameters could be easily compared . for @xmath27 it was possible to find the analytical formula for both @xmath26 and @xmath80 .
the distribution @xmath26 is a polynomial of degree @xmath85 ( see eq . [ eqn : tcdist_p0 ] ) and the average @xmath80 is a scaled difference of two euler beta functions ( see eq .
[ eqn : avgtc_p0 ] ) . the average @xmath86 decreases with @xmath55 and for a fixed @xmath55 it reaches a minimum for @xmath87 ( see fig .
[ fig : mean_tc ] ) .
for @xmath39 the distribution @xmath82 reaches a constant , non - zero value for @xmath88 $ ] ( for @xmath89 , @xmath90 ) , which means that processes when blocked clusters firstly appear at the very end of the evolution are not unlikely .
the values of @xmath86 can be compared with those obtained for hypercubic networks .
similar trends can be observed in hypercubic and hierarchical networks : @xmath86 decreases with the network size @xmath91 and increases with the average degree .
however , for modular hierarchical networks the dependence of @xmath86 on the average degree ( which equals @xmath8 for @xmath27 and rises with @xmath14 ) is very weak in comparison to hypercubic networks .
typical values of @xmath86 for hierarchical networks correspond to the ones obtained for two- or three - dimensional networks , even for @xmath92 . | hierarchical systems attention scientists complex networks many real networks hierarchically organized www network actor network semantic web dynamics at networks different from regular lattices ising model at network hierarchical topology studied by komosa hoyst analyzed parameters magnetization magnetic susceptibility critical temperature correlations between hierarchies critical temperature function of network size ratio @xmath0 @xmath1 node degree opinion formation in hierarchical organizations studied by laguna et al agents authority strata influence opinions probability depends on size distribution authority strata phase diagrams obtained each phase corresponds to dominant stratum decreasing probability prevailing fashion phenomena at hierarchical networks studied by galam vignes interactions between social groups at levels hierarchy renormalization group approach used to optimal investment level influence of counterfeits on new product success fundamental topics in social dynamics conflict situations sociophysics approaches prisoner s dilemma_-type games proposed model of communities isolation introduced by sienkiewicz hoyst model strategy at battlefields formation of culturesidea model similar to game go natural leaning to avoid potentially hostile community paper extend model of communities isolation for chains hypercubic random scale - free networks to hierarchical networks model hierarchical networks proposed by ravasz and barabsi modified by suchecki and hoyst networks possess 3 parameters determining structure : degree of hierarchy distribution @xmath3 number nodes size cliques at lowest level parameter density of edges @xmath6 two models p1 and pd models analyzed differ in density of edges each network has central node _ center of hierarchy_ network of hierarchy @xmath7 is complete graph of size @xmath8 random number chosen with probability center of hierarchy arbitrary node to construct network of hierarchy construct @xmath5 subnetworks of hierarchy @xmath10 choose one center of hierarchy _ becomes center network new connections ( edges created for each node @xmath12 of @xmath8 connection ( edge ) @xmath13 created with probability @xmath14 or sample networks created presented in fig .fig : sample_networks subnetworks connected especially @xmath14 small different degrees hierarchy : @xmath7 @xmath16 @xmath17 @xmath18 basic properties networks concluded from construction algorithm @xmath19 @xmath20 models p1 i pd equivalent @xmath21 network isolated cliques size @xmath5 @xmath8 random variable @xmath22 number of nodes equals @xmath23 periodic oscillations in degree distribution observed in log - log scale period amplitude shape peaks depend on parameters network @xcite case with @xmath22 ( @xmath24 ) considered corresponds original ravasz and barabsi model model of communities isolation proposed by sienkiewicz and hoyst @xcite rules similar to game of go communities compete settling nodes network random empty node chosen settled by randomly chosen community cluster nodes one community blocked surrounded by another community surrounded nodes active case communities competing at chain analyzed in @xcite two functions evolution studied average number of blocked nodes over time mean critical time first blocked cluster influence external bias considered when settling rates communities differentpaper two competing communities p1 pd hierarchical networks considered parameters analyzed average number blocked nodes @xmath25 critical time distribution @xmath26 @xmath21 models p1 pd equivalent network @xmath28 isolated cliques size @xmath5 @xmath29 @xmath30 probability @xmath31th step @xmath32 nodes blocked @xmath34 solution equation @xmath35th degree polynomial approximated substituting sum with integral @xmath36 @xmath25 power function exponent @xmath37 depends on @xmath8 parameter @xmath38 models p1 pd equivalent networks hierarchy @xmath16 @xmath40 @xmath41 reduced density @xmath42 networks higher hierarchies recursive equation approximating @xmath44 derived clique blocked if nodes higher hierarchies filled @xmath45 center hierarchy network empty opposite case @xmath44 depends on @xmath46 @xmath5 subnetworks @xmath47 equation solved numerically solutions presented in fig . z_p1 z_pd ]noticed wide time @xmath48 , @xmath49 approximated with power function @xmath50 where exponents @xmath37 higher than @xmath27 close to @xmath51 analytical approximation of @xmath44 for networks higher hierarchies ( @xmath52 ) when parameter @xmath14 different from zero one difficult alternative approach chosen assumed @xmath44 estimated from proportion @xmath53 where @xmath54 $ increasing function of @xmath14 reasonable approximation for widest ranges of @xmath14 and @xmath55 p1 model choosing @xmath56 results in good agreement of @xmath44 function with simulation data for pd model @xmath57 good choice various networks of p1 model symbols correspond to data from computer simulations lines show analytical approximations left side linear scale right side log - log scale left linear scale log - log scale networks symbols simulations left side linear scale right side log - log scale networks pd model symbols correspond to data computer simulations analytical approximations left side linear scale right side log - log scale,title="fig : networks pd model symbols correspond data computer simulations lines show analytical approximations eq left side linear scale right side log - log scale networks pd model symbols correspond data computer simulations lines show analytical approximations left side linear scale right side log - log scale @xmath27 network @xmath58 cliques @xmath5 nodes find distribution critical time first blocked cluster consider probability @xmath48 no blocked nodes completely filled cliques members one community leads formula @xmath59 @xmath60 cumulative critical time distribution obtained @xmath61 critical time distribution continuous time @xmath62 mean critical time calculated @xmath63 @xmath64 euler beta function ) @xmath65 incomplete euler beta function ) networks hierarchy @xmath7 @xmath66 @xmath16 @xmath67 hierarchy @xmath68 recursive formula cumulative critical time distribution expressed @xmath69 mean critical time obtained numerical integration @xmath70 : @xmath71 function @xmath25 average number blocked nodes @xmath48 approximated power function @xmath72 @xmath37 exponent depends parameters network@xmath73-dimensional hypercubic networks including @xmath74-dimensional chains ) @xmath75 modular hierarchical networks @xmath37 depends on @xmath8 @xmath14 parameters sizes basic cliques lowest hierarchy density inter - clique connections dependence on degree hierarchy @xmath55 network size weak increasing degree hierarchy @xmath55 similar system rescaling @xmath76 @xmath27 parameter @xmath37 found @xmath77 result agreement with simulated data increasing density connections @xmath14 parameter leads increase @xmath37 @xmath51 @xmath39 distinction @xmath25 approximated hypercubic hierarchical networks hypercubic networks number isolated nodes calculated approximation all blocked nodes blocked alone not analytical predictions agreement with simulated data modular hierarchical networks approximation not reasonable lowest level hierarchy networks cliques @xmath5 nodes probable situations when @xmath78 nodes simultaneously blocked second analyzed parameter critical time @xmath79 moment first isolated cluster appears random variable critical time distribution @xmath26 studied mean critical time @xmath80critical density relative time @xmath81 shown networks compared @xmath27 analytical formula for @xmath26 and @xmath80 distribution @xmath26 polynomial of degree @xmath85 average @xmath80 scaled difference of two euler beta functions average @xmath86 decreases with @xmath55 fixed minimum for @xmath87 @xmath39 distribution @xmath82 constant non - zero value for @xmath88 @xmath89 @xmath90 blocked clusters end of evolution not unlikely values @xmath86 compared with hypercubic networks similar trends in hypercubic hierarchical networks @xmath86 decreases with network size @xmath91 increases with average degree modular hierarchical networks dependence of @xmath86 on average degree equals @xmath8 for @xmath27 rises with @xmath14 weak typical values @xmath86 for hierarchical networks correspond to two- or three - dimensional networks even @xmath92 | 0.505676 | 3,290 | 1,654 | , Saving $0.1 in GPT-4. |
177,525 | limits from neutron beta decay on parameters describing physics beyond the standard model are presented .
new physics is described by the most general lorentz invariant effective hamiltonian involving vector , scalar and tensor operators and standard model fields only .
two - parameter fits to the decay parameters measured in free neutron beta decay have been done , in some cases indicating rather big dependence of the results on @xmath0 ratio of nucleon form factors at zero four - momentum transfer . | for many years nuclear @xmath1-decays have been exploited as laboratories for testing the standard model ( sm ) in the domain of low energies . along with developments of intense sources of cold and ultracold neutrons and improvements of experimental techniques ,
the precision of measurements in the simplest of such systems : the @xmath1-decay of a free neutron , is constantly increasing .
it opens the way to study the limits on physics beyond sm set solely by the parameters of the neutron @xmath1-decay .
we assume that at the quark lepton level @xmath1-decay is described by the general 4-point hamiltonian @xcite @xmath2 where @xmath3 , @xmath4 are quark fields , @xmath5 stands for electron field and @xmath6 , @xmath7 , @xmath8 $ ] , where our metric and gamma matrices are the same as _
e.g. _ in @xcite .
we work in the basis in which mass matrix of charged leptons is diagonal and the left ( @xmath9 ) and right ( @xmath10 ) neutrino fields are given by @xmath11 where @xmath12 is the @xmath13-th neutrino field with a certain mass , @xmath14 and @xmath15 are respectively the pontecorvo
sakata matrix and similar mixing matrix for right - handed neutrinos .
sm is restored when @xmath16 for @xmath17 except @xmath18 , where @xmath19 is the usual fermi constant and @xmath20 is the element of quark cabibbo
maskawa mixing matrix .
when calculating the amplitudes for neutron beta decay at small four - momentum transfer @xmath21 we have used the relations @xcite @xmath22 with @xmath23 , @xmath24 and @xmath25 , @xmath26 being proton and neutron states and bispinors , respectively . from conserved vector current hypothesis one gets @xmath27 . in the quark model with spherically symmetric wave functions of quarks
the following relations have been derived @xcite : @xmath28 and @xmath29 . substituting the sm value for @xmath30 into the above relations leads to : @xmath31 and @xmath32 . however , in our derivations and fits we treat @xmath33 and @xmath34 as free parameters ( independent of @xmath35 ) .
from eq . ( [ hamiltonian ] ) the five - fold differential decay width for polarized neutron without measurement of final electron and proton polarization
is given by ( in analogy to @xcite ) @xmath36 \bigg\}\,\textnormal{,}\end{aligned}\ ] ] where @xmath37 is the neutron polarization vector , @xmath38 , @xmath39 , @xmath40 are , respectively , the mass , momentum and total energy of electron , @xmath41 is the maximum value of @xmath40 , @xmath42 is the antineutrino energy and @xmath43 . ] .
the @xmath44 , @xmath45 denotes the solid angles of electron and antineutrino emission .
we have worked at tree - level ( except : calculation of @xmath46 see below ) and with approximations such that terms proportional to @xmath47 are not present in ( [ d_gamma ] ) .
furthermore , we will consider only cases when : @xmath48 , @xmath49 , @xmath50 , @xmath51 , as well as @xmath52 , @xmath53 , @xmath54 for @xmath17 are real then @xmath55 and time reversal symmetry is preserved , that is well motivated experimentally ( pdg average @xcite : @xmath56 ) .
we express the decay parameters @xmath57 , @xmath58 , @xmath59 , @xmath60 , where @xmath60 has the form of @xmath61 , in terms of the ratio @xmath0 and the following parameters ( see also @xcite ) for @xmath17 @xmath62 where @xmath63 with summation @xmath64 running only over kinematically allowed antineutrino states . in sm and for some cases of physics beyond sm @xmath65 and @xmath66 . as a result of the applied approximations formulas for @xmath57 , @xmath58 , @xmath59 , @xmath67 , @xmath68 depend in general case only on two combinations @xmath69 next ,
following the approach applied in @xcite , we define @xmath70 where @xmath71 , and apply these quantities in the fits to the experimental data .
the @xmath72 , which we will minimize with the fit procedure , is of the form @xmath73 ^ 2 \nonumber \\ & + & \sum_j \left[\frac{a_j - \bar{a}(\langle w^{-1}\rangle_j)}{\delta a_j}\right]^2 \nonumber \\ & + & \sum_k \left[\frac{b_k - \bar{b}(\langle w^{-1}\rangle_k)}{\delta b_k}\right]^2 \,\textnormal{,}\end{aligned}\ ] ] where the selected data are presented in table [ data_table ] : @xmath74 , @xmath75 , @xmath76 and @xmath77 , @xmath78 , @xmath79 denote the central value and the error of the respective decay parameter in a certain experiment .
we calculate the particular value of @xmath80 from @xmath81 where @xmath82 and @xmath83 in general are different for different experiments . at this stage of calculation fermi function @xmath84 ( that is a leading order qed correction )
@xcite has been incorporated and sm was assumed @xmath85 @xmath86 .we have followed the pdg @xcite data selection but took only the most precise data ( the error in measurements of @xmath57 is less than @xmath87 of central value , for @xmath59 and @xmath60 it is less than @xmath88 ) . when experiment report statistic and systematic error separately we add these two errors in quadrature . in the case of asymmetric errors
we have taken the larger of the reported errors .
most of presented values of the @xmath89 have been taken from @xcite .
we have used all 11 `` data points '' in the table above in every fit presented in this paper .
because of unsolved experimental ambiguity of neutron lifetime measurements ( see @xcite ) we have not included this quantity in our analyzes .
[ data_table ] [ cols="^,<,<,<,<,<,<",options="header " , ]
in sm the formulas derived for decay parameters depend on @xmath90 alone and simplify to @xmath91 in this case , the one - parameter fit is performed , which results in @xmath92 with @xmath93 that is in a good agreement with the pdg average @xcite : @xmath94 ( error scaled by pdg by @xmath95 ) . the above results ( [ g_a_fit_68][g_a_fit_95 ] ) apply also when : @xmath96 for @xmath17 except @xmath97 ( and @xmath98 by definition see eqs . ( [ vst ] ) and ( [ kappa_lr ] ) ) .
in this case the formulas ( [ sm_kab ] ) hold for modified @xmath99 @xmath100 in the next step one of the parameters : @xmath101 , @xmath102 , @xmath103 for @xmath104 is nonzero and fitted together with the ratio @xmath0 . among these cases only when the nonzero parameter is @xmath105 or @xmath106 we have @xmath107 and @xmath108 .
the results of such two - parameter fits are presented in the fig .
[ fits_2_param ] . + in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] ) the narrow one and ( [ g_a_fit_95 ] ) the wider one .
note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] ) the narrow one and ( [ g_a_fit_95 ] ) the wider one .
note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] ) the narrow one and ( [ g_a_fit_95 ] ) the wider one .
note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] ) the narrow one and ( [ g_a_fit_95 ] ) the wider one .
note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] ) the narrow one and ( [ g_a_fit_95 ] ) the wider one .
note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in each case ) . the line marked areas correspond to the @xmath109 intervals : ( [ g_a_fit_68 ] )
the narrow one and ( [ g_a_fit_95 ] ) the wider one . note that @xmath110 is the absolute value not the module of a complex number , as all parameters are real .
[ fits_2_param],title="fig:",scaledwidth=48.0% ] in conclusion , standard model describes the neutron beta decay very well .
the fits are minimally better if new physics is included , especially if tensor terms are present . in some cases
there is rather big dependence of the results on @xmath111 ratio . +
this work has been supported by the polish ministry of science and higher education under grant no . . | nuclear @xmath1-decays exploited testing standard model low energies developments sources cold ultracold neutrons improvements experimental techniques precision measurements @xmath1-decay free neutron increasing opens study limits physics beyond sm parameters neutron @xmath1-decay quark lepton level @xmath1-decay described by 4-point hamiltonian @xcite @xmath2 quark fields electron field metric gamma matrices same as @xcite mass matrix of charged leptons diagonal left right neutrino fields by @xmath11 @xmath12 @xmath13-th neutrino field mass @xmath14 pontecorvo sakata matrix mixing matrix for right - handed neutrinos sm restored @xmath16 @xmath17 except @xmath18 fermi constant element quark cabibbo maskawa mixing matrix calculating amplitudes neutron beta decay at small four - momentum transfer @xmath21 used relations @xcite @xmath22 @xmath23 @xmath24 @xmath26 proton neutron states bispinors conserved vector current hypothesis @xmath27 .quark model symmetric wave functions relations derived @xcite @xmath28 @xmath29 substituting sm value @xmath30 relations leads to @xmath31 @xmath32 derivations treat @xmath33 @xmath34 free parameters independent @xmath35 five - differential decay width polarized neutron without measurement electron proton polarization given by @xmath36 @xmath37 neutron polarization vector @xmath38 @xmath39 @xmath40 mass momentum total energy electron @xmath41 maximum value @xmath40 @xmath42 antineutrino energy @xmath43 @xmath44 @xmath45 solid angles electron antineutrino emission worked tree - level calculation @xmath46 approximations terms proportional to @xmath47 not present d_gamma consider cases @xmath48 @xmath49 @xmath50 @xmath51 @xmath52 @xmath53 @xmath54 @xmath17 real @xmath55 time reversal symmetry preserved motivated experimentally @xcite @xmath56express decay parameters @xmath57 @xmath60 @xmath60 form of @xmath61 ratio @xmath0 following parameters see @xcite ) for @xmath17 @xmath62 @xmath63 summation @xmath64 over kinematically allowed antineutrino states sm cases physics beyond sm @xmath65 @xmath66 applied approximations formulas for @xmath57 @xmath58 @xmath59 @xmath67 @xmath68 depend on two combinations @xmath69 approach define @xmath70 @xmath71 apply quantities fits to experimental data @xmath72 minimize with fit procedure form @xmath73 ^ \nonumber \\\sum_j selected data presented in table data_table @xmath74 @xmath75 @xmath76 @xmath77 @xmath78 @xmath79 denote central value error of decay parameter in certain experiment calculate value of @xmath80 from @xmath81 @xmath82 @xmath83 different for experimentsstage calculation fermi function @xmath84 leading order qed correction ) @xcite incorporated sm assumed @xmath85 @xmath86 followed pdg @xcite data selection took precise data error measurements @xmath57 less than @xmath87 @xmath59 @xmath60 less than @xmath88 experiment report statistic systematic error add two errors in quadrature asymmetric errors taken larger reported errors presented values @xmath89 taken from @xcite used 11 data points table every fit unsolved experimental ambiguity neutron lifetime measurements not included quantity analyzes data_table formulas for decay parameters depend on @xmath90 simplify to @xmath91 one - parameter fit performed results in @xmath92 with @xmath93 good agreement with pdg average @xcite : @xmath94 error scaled by pdg @xmath95 results apply @xmath96 @xmath17 except @xmath97 @xmath98 formulas hold for modified @xmath99 @xmath100 next parameters @xmath101 @xmath102 @xmath103 @xmath104 nonzero fitted with ratio @xmath0 .cases nonzero parameter @xmath105 or @xmath106 have @xmath107 and @xmath108 results two - parameter fits presented in fig . fits_2_param ] + each case line marked areas correspond @xmath109 intervals g_a_fit_68 narrow g_fit_95 wider @xmath110 absolute value not module complex number all parameters real fits_2_param]:",scaledwidth=48.0% case line marked areas correspond @xmath109 intervals g_a_fit_68 narrow_fit_95 wider @xmath110 absolute value not module complex number all parameters real fits_2_param]:",scaledwidth=48.0% line marked areas correspond @xmath109 intervals g_a_fit_68 narrow_fit_95 wider @xmath110 absolute value not module complex number all parameters real fits_2_param]:",scaledwidth=48.0% case line marked areas correspond @xmath109 intervals g_a_fit_68 narrow g_a_fit_95 wider one @xmath110 absolute value not module complex number all parameters real fits_2_param]:",scaledwidth=48.0% ] each caseline marked areas correspond @xmath109 intervals ( g_a_fit_68 ] narrow g_a_fit_95 wider @xmath110 absolute value not module complex number all parameters real fits_2_param]:",scaledwidth=48.0% line marked areas correspond @xmath109 intervals g_a_fit_68 narrow_fit_95 wider @xmath110 absolute value not module complex number all parameters real fits_2_param=48.0% standard model describes neutron beta decay well fits minimally better if new physics included especially tensor terms present big dependence results on @xmath111 ratio work supported polish ministry science higher education grant no | 0.350648 | 2,360 | 1,233 | , Saving $0.1 in GPT-4. |
177,526 | we present preliminary results from our spectroscopic search for planets within 1 au of metal - poor field dwarfs using nasa time with hires on keck i. the core accretion model of gas giant planet formation is sensitive to the metallicity of the raw material , while the disk instability model is not . by observing metal - poor stars in the field we eliminate the role of dynamical interactions in dense stellar environments , such as a globular cluster .
the results of our survey should allow us to distinguish the relative roles of the two competing giant planet formation scenarios .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | the two proposed models for the formation of gas giant planets make different , and testable , predictions . for example , core accretion requires several myr to grow a solid core massive enough to accrete a gaseous envelope ( lissauer 1993 ) .
in contrast , disk instability leads to jupiter - mass clumps ( e.g. , boss 1997 ) , which can survive and give rise to actual protoplanets ( mayer et al .
2002 ) , within @xmath0 yr . the time - scale for core accretion to proceed depends strongly on the initial surface density of solids ( pollack et al .
1996 ) , so that formation by this mechanism may be enhanced in metal - rich stars .
instead , disk instability is remarkably insensitive to the primordial metallicity of the protoplanetary disk ( boss 2002 ) .
several observational tests can be performed to probe giant planet formation models ( boss 2003 ) . in order to unambiguously establish whether high primordial metallicity is a requirement , and
thus determine which is the dominant formation mechanism for gas giant planets , we are conducting a spectroscopic search for planets orbiting a sample of metal - poor stars in the field , thus eliminating possible sources of interference with planet formation , or with migration to close - in orbits , or with planet survival , such as dynamical interactions in dense stellar environments .
in this project we are searching for planetary companions within 1 au orbiting a sample of 200 metal - poor dwarfs , chosen from the carney - latham ( e.g. , carney et al .
1994 ) and ryan ( ryan 1989 ) surveys of metal - poor , chromospherically quiet , non rotating ( @xmath1 km / s ) , high - velocity field stars that happen to be passing by the solar neighborhood .
these stars are selected not to have close stellar companions based on long - term radial - velocity ( rv ) monitoring that has been carried out with the cfa digital speedometers .
our survey is orthogonal to other spectroscopic planet searches in that it spans a range of metallicities never investigated before ( @xmath2 [ fe / h ] @xmath3 ) .
we have further refined our sample of metal - poor dwarfs by imposing magnitude ( @xmath4 ) and temperature ( @xmath5 k ) cut - offs , and utilized the metallicity , temperature , and magnitude constraints to compute the relative exposure times needed to achieve 20 m / s rv precision for planet detection .
this precision is sufficient to achieve sensitivity to rv variations due to planetary companions with minimum masses in the average range @xmath6 m@xmath7 , or higher , for orbital periods in the range @xmath8 yr .
our sample of 200 metal - poor stars should eventually provide a robust 3-@xmath9 null result in the case of no detections .
during the first two keck 1/hires observing nights we have obtained spectra ( 1 template + 2 iodine exposures ) for 40 stars covering a large range of metallicities and effective temperatures , with the primary aim of evaluating the rv precision as a function of effective temperature and metallicity , and thus test the validity of our predictions for the dependence of the rv precision on these two parameters .
the resulting rv difference distribution between the two nights exhibits an rms residual velocity @xmath10 m / s , corresponding to an expected single - measurement uncertainty of @xmath11 m / s .
no clear trends in the rv differences as a function of [ fe / h ] and t@xmath12 are present , a further confirmation that the model we developed for the dependence of the rv precision on the above parameters is robust .
our next goal will be to demonstrate the long - term stability of the velocity zero - point and expected single - measurement precision for planet detection .
boss , a. p. 1997 , science , 276 , 1836 boss , a. p. 2002
, , 567 , l149 boss , a. p. 2003
, , in press carney , b. w. , et al .
1994 , , 107 , 2240 lissauer , j. j. 1993 , ara&a , 31 , 129 mayer , l. , et al .
2002 , science , 298 , 1756 pollack , j. b. , et al .
1996 , icarus , 124 , 62 ryan , s. g. 1989 , , 98 , 1693 | two proposed models for formation gas giant planets make different predictions core accretion requires myr to grow solid core envelope lissauer 1993 disk instability leads to jupiter - mass clumps can survive rise to protoplanets 2002 time - scale for core accretion depends on initial surface density of solids pollack 1996 formation may enhanced in metal - rich stars disk instability insensitive to primordial metallicity protoplanetary disk boss 2002 ) observational tests probe giant planet formation models ( 2003 ) to establish high primordial metallicity requirement determine dominant formation mechanism for gas giant planets conducting spectroscopic search for planets orbiting metal - poor stars eliminating sources interference with planet formation migration survival searching for planetary companions within 1 au orbiting 200 metal - poor dwarfs chosen from carney - surveys of metal poor quiet non rotating high - velocity field stars passing by solar neighborhood stars selected not to close stellar companions based on long - term radial - velocity monitoring with digital speedometerssurvey orthogonal to planet searches spans metallicities @xmath2 h @xmath3 ) refined sample metal - poor dwarfs magnitude @xmath4 temperature @xmath5 k cut - offs metallicity temperature magnitude constraints compute exposure times 20 m / s rv precision for planet detection precision sufficient sensitivity to rv variations planetary companions minimum masses range @xmath6 m@xmath7 or higher orbital periods @xmath8 yr sample 200 metal - poor stars provide robust 3-@xmath9 null result no detections first two nights obtained spectra for 40 stars range metallicities effective temperatures rv precision temperature metallicity validity predictions rv precision rv difference distribution exhibits rms residual velocity @xmath10 m / s expected single - measurement uncertainty @xmath11 m / s no clear trends in rv differences [ fe / h ] t@xmath12 model rv precision robust next goal demonstrate long - term stability of velocity zero - point single - measurement precision for planet detection boss 1997 science 276 , 1836 2002 567 , l1492003 press carney 1994 107 , 2240 lissauer 1993 ara&a 31 , 129 mayer 2002 science 298 , 1756 pollack j b. 1996 icarus 124 , 62 ryan 1989 98 , 1693 | 0.181082 | 970 | 483 | , Saving $0.0 in GPT-4. |
177,534 | we analyze the surface morphology of metals after plastic deformation over a range of scales from 10 nm to 2 mm , using a combination of atomic force microscopy and scanning white - light interferometry .
we demonstrate that an initially smooth surface during deformation develops self - affine roughness over almost four orders of magnitude in scale .
the hurst exponent @xmath0 of one - dimensional surface profiles is initially found to decrease with increasing strain and then stabilizes at @xmath1 . by analyzing their statistical properties
we show that the one - dimensional surface profiles can be mathematically modelled as graphs of a fractional brownian motion .
our findings can be understood in terms of a fractal distribution of plastic strain within the deformed samples . according to the traditional paradigm implicit in continuum models of plasticity , plastic deformation under homogeneous loads and in the absence of so - called plastic instabilities
is expected to proceed in a smooth and spatially homogeneous manner .
fluctuations , if any , are supposed to average out above the scale of a `` representative volume element '' which is tacitly assumed to be small in comparison with macroscopic dimensions ( or , indeed , in comparison with the scale of any problem to which continuum modelling is applied ) .
recently for crystalline systems deforming by dislocation glide this paradigm has been challenged both from the theoretical and experimental side @xcite .
weiss et .
al . demonstrated that dislocation glide in single - glide oriented ice single crystals proceeds in a temporally intermittent manner , i.e. , plastic flow is characterized by a sequence of large `` deformation bursts '' @xcite .
acoustic emission due to these bursts was monitored and it was demonstrated that there is no characteristic burst magnitude . instead the energy releases were found to obey a scale - free power - law distribution over more than six orders of magnitude in energy @xcite .
simulations of dislocation dynamics in single - glide single crystals showed qualitatively similar behavior @xcite .
these findings were interpreted as dynamical critical phenomena in a slowly driven non - equilibrium system @xcite , and it has been argued that the dislocation system in a plastically deforming crystal is always close to a critical point ( `` yielding transition '' , @xcite ) .
temporal intermittency of plastic flow goes along with spatial heterogeneity .
this has been long known to metallurgists who observed pronounced surface traces ( `` slip lines '' ) resulting from the collective motion of many dislocations . by monitoring the evolution of surface
features it is therefore possible to study collective behavior in plastic flow .
however , most investigations ( for an overview , see @xcite ) have focused on the evolution of single slip lines or slip line bundles , i.e. , the motion of isolated dislocation groups .
long - range correlations within the slip pattern have only occasionally been studied .
konstantinidis and aifantis @xcite used a wavelet transform to investigate the slip pattern of a fe@xmath2al single crystal on multiple scales between 0.1 and 50 micrometers but did not quantitatively analyze their data in view of scale - invariant behavior .
kleiser and bocek @xcite report a self - similar pattern of slip lines in cu single crystals on scales between 0.06 and 2 micrometers and determine a fractal dimension @xmath3 for the set of intersection points between the slip lines and a line normal to the slip direction .
recently , weiss and marsan @xcite investigated the three - dimensional patterning of slip on macroscopic scales . by recording acoustic emission with multiple transducers and using spatial triangulation , they determined the 3d distribution of slip bursts in the volume of an ice single crystal .
this investigation demonstrated that the slip burst pattern on macroscopic scales ( 0.6 mm to 10 mm ) exhibits features of a fractal set with correlation dimension @xmath4 .
however , due to the limited spatial resolution of the technique no information about the slip distribution on smaller scales could be obtained . in the present study
we pursue a slightly different approach for assessing spatial correlations in plastic deformation from the nanoscopic to the macroscopic scale . instead of directly studying the distribution of plastic strain in the bulk or on the surface of the sample ,
we monitor the evolution of the surface profile .
provided that the initial surface is sufficiently smooth , the evolution of a one - dimensional profile can be directly related to the plastic distortion at the surface : if we define a local coordinate system such that the @xmath5 direction corresponds to the direction of the profile and the @xmath6 direction to the surface normal , then the derivative @xmath7 of the profile @xmath8 equals the component @xmath9 of the plastic distortion tensor .
long - range correlations in the plastic strain pattern can be detected since they give rise to characteristic modifications of the surface morphology - more specifically speaking , power - law correlations in the spatial distribution of plastic strain lead to the emergence of self - affine surface profiles with a hurst exponent @xmath10 as will be discussed below . a self - affine profile @xmath8 is characterized by the property that the statistical properties of the profile remain unchanged under the scaling transformation @xmath11 .
the roughness exponent or hurst exponent @xmath0 can be related to a fractal dimension @xmath12 ( `` box dimension '' ) of the profile through @xmath13 @xcite .
self - affine behavior is abundant in natural surfaces , a most prominent example in materials physics are fracture surfaces whose large - scale roughness can be characterized by a universal ( material - independent ) exponent @xmath14 @xcite . in the present letter we report for the first time an analysis of the surface structure of plastically deformed metal samples in terms of self - affine properties .
we applied a combination of atomic force microscopy ( afm ) and scanning white - light interferometry ( swli ) to quantitatively characterize the surface morphology over a range of scales between 10 nm and 2 mm .
polycrystalline copper samples of 99.9 % nominal purity and an average grain size of 40 @xmath15 m were electropolished to a typical rms roughness of 44 nm as determined by swli over a surface area of @xmath16 m @xmath17 m .
two complementary methods were used for surface imaging : ( i ) afm scans ( 512 @xmath18 512 pixels ) were taken over areas of @xmath19 m @xmath20 m and @xmath21 m @xmath22 m using a picospm ( molecular imaging ) at the constant deflection mode with a medium range scanner and a cantilever ( mikromasch , ultrasharp ) with a nominal spring constant of 1.75 n / m and tip curvature radius 10 nm ; the rms vertical noise amplitude was less than 0.05 nm ; ( ii ) swli ( new view 100 , zygo ) scans were performed over areas of @xmath23 m @xmath24 m with a lateral resolution of 0.6 @xmath15 m and a vertical resolution of 0.5 nm . from the scans , typically 4 - 5 surface profiles were taken in the direction of the specimen axis and 2 - 3 profiles normal to that direction . after characterizing the initial surface ,
the samples were deformed in tension on a standard tensile - testing machine ( instron series 3360 ) .
testing was done at room temperature with an imposed strain rate @xmath25 s@xmath26 . at total ( engineering )
strains @xmath27 and @xmath28 ( see figure [ stressstrain ] ) , the sample was unloaded and afm and swli scans were taken in a similar manner as before deformation .
the sequence of surface morphology investigations was ended at the onset of macroscopic deformation localization ( necking ) .
typical surface profiles are shown in figure [ profiles ] representing a `` long '' afm profile obtained from a @xmath19 m @xmath29 m scan together with a swli profile . to investigate a possible self - affine structure , mean height differences @xmath30 within the profile .
self - affine behavior implies that @xmath31 hence , double - logarithmic plots of @xmath32 vs. @xmath33 should exhibit a linear scaling regime with slope @xmath0 .
such roughness plots are shown in figure [ roughness ] .
typical plots exhibit linear scaling regimes extending between 0.05 and 5 @xmath15 m for the afm and between 0.5 and 100 @xmath15 m for the swli profiles .
hurst exponents deduced from the slope of the scaling regimes are similar for afm and swli profiles , and also the absolute values of @xmath32 are similar across the region of overlap of the roughness plots .
hence , we observe continuous scaling over almost four orders of magnitude .
surface profiles @xmath34 taken in the direction normal to the tensile axis exhibit scaling with similar exponents , although the length of the self - affine scaling regimes is reduced .
roughness plots obtained from the `` short '' afm profiles deduced from @xmath35 m @xmath22 m scans show a crossover to a strain - independent apparent hurst exponent @xmath36 below 50 - 100 nm .
hurst exponents @xmath0 and fractal dimensions @xmath13 determined from the roughness plots are compiled in figure [ dimension ] .
the error bars reflect the scatter of exponents obtained from different profiles taken at the same strain .
it can be seen that the exponents initially decrease with increasing strain and then stabilize above @xmath37 at a value of @xmath1 .
it is important to note that the initial value of @xmath38 does not reflect any self - affine surface roughness but simply stems from large - scale gradients on an almost smooth surface .
the same is true for the steepening of roughness plots obtained from the `` short '' afm profiles on very small scales : below 50 nm , the apparent hurst exponent increases towards @xmath36 as one essentially observes the smooth initial surface in between the deformation - induced surface steps . to further investigate the statistical surface properties in the different scale regimes , probability distributions @xmath39 of surface height differences @xmath40
have been determined for the profiles taken at @xmath41 .
@xmath42 m was taken as a reference and all height differences @xmath43 were normalized by the variance @xmath44 nm of the corresponding height difference distribution . within the scaling regime @xmath45 ( @xmath46 nm , @xmath47 m )
, probability distributions corresponding to different values of @xmath33 can be collapsed by re - scaling @xmath48 with @xmath49 ( see figure [ stat ] ) : @xmath50^h \delta y_l / y_0 ) = p(\delta y_{l_0}/y_0)\;.\ ] ] to assess the shape of the probability distributions , cumulative distributions normalized to unit variance have been represented on a probability scale ( i.e. , the error function appears as a straight line ) .
the insert of figure [ stat ] shows the cumulative distribution for @xmath51 m ( full line ) as well as two distributions for @xmath33 values outwith the scaling regime , namely for @xmath52 nm ( dashed line ) and @xmath53 m ( dotted line ) . whereas distributions throughout the self - affine scaling regime are well approximated by gaussian statistics , the tails of distributions outwith the scaling regime exhibit characteristic deviations from gaussian behavior . on scales below @xmath54 the surface morphology
is characterized by flat regions separated by huge slip steps at locations where slip events have occurred .
this leads to a leptokurtic distribution with heavy tails . on scales above @xmath55 ,
on the other hand , the surface flattens and large height fluctuations are suppressed since the macroscopic slope of the swli profiles is by definition set to zero .
this gives rise to platykurtic distributions with truncated tails .
the fact that the statistics is gaussian throughout the scaling regime implies that the surface profiles @xmath8 can be mathematically interpreted as graphs of fractional brownian motions .
this allows us to draw conclusions on the underlying plastic strain pattern since the increments @xmath56 of a fractional brownian motion exhibit long - range correlations given by @xcite @xmath57 since @xmath56 equals the component @xmath9 of the plastic distortion tensor at the surface , our observations yield strong evidence for power - law correlations in the distribution of plastic strain . from the correlation
integral @xmath58 we deduce a correlation dimension @xmath59 for the one - dimensional strain pattern underlying the surface profile .
the value @xmath60 is consistent with the fractal dimension of slip line patterns reported by kleiser and bocek @xcite .
a dimension of 0.5 for a one - dimensional strain pattern is also consistent with the dimension @xmath61 for the three - dimensional pattern of slip events measured by weiss and marsan @xcite .
these investigations were performed on single crystals ; due to experimental limitations the scaling regimes cover only about one order of magnitude .
our results demonstrate that the strain pattern in polycrystals exhibits an analogous self - similar behavior which by combining different experimental techniques can be traced over almost four orders of magnitude in scale .
more investigations are required to decide whether the hurst exponent @xmath1 which we determine for deformed metal surfaces is universal , i.e. independent of material and deformation conditions .
further investigations are also needed to establish the nature of the boundaries of the self - affine scaling regime .
the lower bound of the scaling regime can be related to the spacing between individual slip events between which the surface remains unchanged by the deformation .
the upper boundary @xmath55 of the scaling regime , which defines the correlation length of the self - affine surface pattern , is of the same order of magnitude as the grain size ( 40 microns ) .
one may reasonably conjecture that the range of correlations in the dislocation dynamics , and therefore in the slip pattern , is limited by the grain size .
however , further experimental work ( preferably on single crystals ) is required to decide whether this is indeed the physical mechanism which determines the correlation length . from the theoretical side , the fact that the hurst exponent of deformed metal surfaces is quite similar to that of fracture surfaces raises the fundamental question whether there is a universal mechanism underlying both observations , or whether the similarity of exponents is incidental .
crystal plasticity is a bulk phenomenon which only indirectly affects the surface , whereas fracture is directly governed by surface and interface properties .
however , this does not preclude the possibility of a common theoretical treatment within the framework of non - equilibrium critical phenomena . in view of plasticity models using coarse - grained strain variables ,
our results indicate that strain fluctuations may average out rather slowly .
the strain fluctuation on scale @xmath33 can be estimated as @xmath62 . at the lower end of the self - affine scaling regime ( @xmath46 nm ) , fluctuations are of the same order of magnitude as the average strain . within the scaling regime ,
strain fluctuations decrease like @xmath63 if averaged over the length @xmath33 ; for @xmath64 we find that the residual strain fluctuations at the upper end of the scaling regime ( @xmath65 m ) are still about 15 % of the average strain .
this observation explains why it is possible to detect significant strain fluctuations even on macroscopic scales @xcite . | financial support by the european commission under rtn / defino hprn - ct 2002 - 00198 and of epsrc under grant nos .
gr / s20406/01 and gr / r43181/01 is gratefully acknowledged . | support european commission rtn 2002 - 00198 epsrc grant nos gr / s20406/01 r43181/01 acknowledged | 0.065106 | 52 | 28 | , Saving $0.0 in GPT-4. |
177,559 | the fregate experiment aboard hete - ii has been successfully integrated into the third interplanetary network ( ipn ) of gamma - ray burst detectors .
we show how hete s timing has been verified in flight , and discuss what hete can do for the ipn and vice - versa . | the fregate experiment aboard hete - ii is an excellent complement to the detectors in the ipn .
it has good sensitivity to bursts , thanks to its large surface area and its steady background in equatorial orbit .
its time resolution in both triggered and untriggered modes is sufficiently high for cross - correlating precisely with other spacecraft . and finally , its energy range is well matched to those of other instruments in the network .
however , before any experiment can be added to the network , its timing must be verified in flight . in this paper
, we will first explain the procedure we have used to demonstrate that the hete timing is accurate .
then we will show how the ipn can be used to improve the location accuracy of hete bursts , and how the fregate data are useful to the ipn .
in principle , we expect the timing of the hete spacecraft to be good , both because it is in low earth orbit , and because it is derived from an onboard gps receiver . while there are several ways to confirm the timing of any spacecraft in flight , the best is to triangulate burst sources with known positions , either soft gamma repeaters , or gamma - ray bursts for which optical or radio counterparts are found . in its first year of operation
, fregate detected numerous bursts from sgr1806 - 20 and 1900 + 14 ; seven of them were observed in conjunction with other spacecraft in the ipn .
the positions of both sgrs are known to arcsecond or sub - arcsecond accuracy , and therefore serve as timing calibration sources . in figures 1 and 2
we show hete / ulysses annuli for bursts from each sgr . from these triangulations
we conclude that hete timing errors are indeed negligible , and moreover , that hete is a sensitive sgr detector when it is pointed towards the galactic plane .
since this occurs for several months around december and june , it is clear that hete will serve as a monitor of the known sgrs , and may also detect new ones when they become active .
at present , the ipn consists of ulysses , in heliocentric orbit , mars odyssey , now in orbit around mars , and several missions close to earth ( cline et al . 2002 ) .
prior to the launch of hete , the two principal grb detectors close to earth were the bepposax gamma- ray burst monitor ( grbm ) and the konus experiment aboard the wind spacecraft .
bepposax has a high duty cycle , but as it is in low - earth orbit , it misses roughly half of all grbs due to earth occultation .
the grbm has a coarse location capability ( 10 - 20 @xmath0 ) .
the mission will end some time in 2002 .
wind is nominally at the l@xmath1 lagrange point , so no bursts are missed due to occultation , but the duty cycle is not as high as that of bepposax , and this causes some events to be missed . because of the placement of the two konus detectors aboard the spacecraft , the experiment can be used to determine the ecliptic latitude of a burst to an accuracy of about 10@xmath0 in many , but not all , cases .
the wind mission is now expected to be supported at least through 2002 , at which time the launch of integral should provide a replacement or near - earth vertex for the ipn .
hete complements these missions in three ways , as far as the ipn is concerned .
first , it fill in the gaps by providing data on bursts missed by the other spacecraft . to date , about one burst per month has been detected by fregate and verified by ulysses or mars odyssey , but has not been detected by konus or by the grbm .
second , it provides burst data in near - real time .
the grbm data are delayed by up to several hours , and the konus data by half a day or more .
thus even when these experiments detect a burst , valuable time can be gained by receiving the hete data and alerting the ulysses and mars odyssey teams at jpl to expedite processing of the data in the small time window where the burst should occur .
finally , fregate data can be used to resolve the ambiguity in a 3 or 4 spacecraft localization . in the case where ulysses , mars odyssey , and fregate observe a burst , fregate s coarse localization capability can serve to indicate which of two alternate triangulation positions is the correct one . in the case where ulysses , mars odyssey , fregate , and konus observe a burst
, the separation between fregate and konus will often be just enough to generate a third , independent annulus of location which again resolves the ambiguity .
it should also be noted that the time resolution of fregate in the untriggered mode is 0.164 s , which is a factor of about 5 to 20 better than the untriggered modes of the grbm and konus , which results in more accurate cross - correlations .
the ipn can assist the hete mission in four ways .
first , it can confirm candidate grbs and sgr bursts .
typically , these are events which are observed only by fregate .
24 such bursts have been confirmed to date , or a rate of about two per month .
second , it can refine wxm error circles .
figure 3 shows one example , grb010613 .
this is an extreme case , in that both the wxm error circle and the ipn annulus are rather large .
however , there are numerous examples of smaller bepposax wfc error circles which were refined by the ipn ( hurley et al .
2001 ) , and we expect most hete - ipn bursts to be similar to them .
third , it can transform `` one dimensional '' wxm error boxes which are too large to search for counterparts into small error boxes which are suitable for optical and radio follow - up observations .
a `` one - dimensional '' wxm error box is one which results when a burst is detected and localized in only one of the two crossed wxm detectors , leading to a long , narrow error box .
figure 4 shows an example , grb010921 .
figure 5 shows an enlargement of the region where the ipn annulus intersects the wxm error box .
this burst and its counterpart are described in detail elsewhere ( ricker et al .
2002 , park et al .
2002 , kulkarni et al .
2002 ) . finally ,
the ipn can , in principle , transform one dimensional sxc error boxes into small error boxes in the same way that it transforms wxm error boxes .
however , this has not yet been done for any burst , since none has been localized by the sxc in this way .
we have successfully integrated the hete spacecraft into the third interplanetary network of gamma - ray burst detectors , and have demonstrated the value of hete to the ipn and vice - versa .
we are now looking forward to several more years of operations . | fregate experiment aboard hete - ii complement to detectors ipn good sensitivity to bursts large surface area steady background equatorial orbit time resolution triggered untriggered modes high for cross - correlating with other spacecraft energy range matched to other instruments network before experiment timing must be verified in flight paper explain procedure hete timing accurate show ipn improve location accuracy of hete bursts fregate data useful to ipn expect timing hete spacecraft good low earth orbit derived from onboard gps receiver confirm timing triangulate burst sources with known positions soft gamma gamma - ray bursts first year fregate detected numerous bursts from sgr1806 - 20 and 1900 + 14 seven observed with other spacecraft in ipn positions of both sgrs known to arcsecond sub - accuracy serve as timing calibration sources figures 1 and 2 show hete / ulysses annuli for bursts from each sgr hete timing errors negligible sensitive sgr detector pointed towards galactic planeoccurs around december june hete monitor known sgrs detect new ones ipn consists of ulysses mars odyssey missions close to earth ( cline et al . 2002 ) prior launch hete two principal grb detectors earth were bepposax gamma- ray burst monitor konus experiment wind spacecraft bepposax high duty cycle low - earth orbit misses half grbs due to earth occultation grbm coarse location capability ( 10 - 20 @xmath0 ) mission in 2002 wind at l@xmath1 lagrange point no bursts missed due to occultation duty cycle not high as bepposax causes some events missed two konus detectors experiment determine ecliptic latitude of burst accuracy of about 10@xmath0 wind mission supported through 2002 launch of integral should provide replacement near - earth vertex for ipn hete complements missions gaps data on bursts missed by other spacecraft one burst per month detected by fregate verified by ulysses mars odyssey not detected by konus or grbm provides burst data in near - real timegrbm data delayed several hours konus data half a day experiments detect burst valuable time gained by receiving hete data alerting ulysses mars odyssey teams jpl expedite processing small window burst fregate data resolve ambiguity in 3 or 4 spacecraft localization ulysses mars odyssey fregate observe burst s coarse localization capability two alternate triangulation positions correct konus observe burst separation between fregate konus enough generate third independent annulus of location resolves ambiguity time resolution of fregate in untriggered mode 0.164 s 5 to 20 better than grbm konus more accurate cross - correlations ipn assist hete mission four confirm candidate grbs sgr bursts fregate 24 bursts confirmed two per month refine wxm error circles figure 3 shows example grb010613 extreme case wxm error circle ipn annulus large numerous examples of smaller bepposax wfc error circles refined by ipn ( hurley et al . 2001 ) expect most hete - ipn bursts be similarone dimensional wxm error boxes into small boxes suitable for optical radio observations one - dimensional wxm error box results when burst detected localized in one two crossed wxm detectors to long narrow error box figure 4 example grb010921 figure 5 region ipn annulus intersects wxm error box burst counterpart described ( ricker park kulkarni 2002 ipn can transform one dimensional sxc error boxes into small error boxes wxm not done for burst none localized by sxc integrated hete spacecraft into third interplanetary network of gamma - ray burst detectors demonstrated value of hete to ipn vice - versa looking forward to more years of operations | 0.264786 | 1,510 | 761 | , Saving $0.0 in GPT-4. |
177,576 | a characteristic feature of thermalized non - equilibrated matter is that , in spite of energy relaxation ( thermalization ) , a phase memory of the way the strongly interacting many - body system was excited remains . in this contribution
we analyze a low energy evaporating proton data in nucleon induced reactions at @xmath062 mev incident energy with @xmath1au , @xmath2pb , @xmath3bi and @xmath4u .
our analysis demonstrates that the thermalized non - equilibrated matter survives a cascade of several evaporating particles .
thus the experiments show that the effect of the anomalously slow phase relaxation , with upper limits of the phase relaxation widths in the range 1 - 10@xmath5 ev , is stable with respect to the multi - step evaporating cascade from the thermalized compound nuclei .
we also briefly mention manifestations and implications of the thermalized non - equilibrated matter for some other fields . | the basic idea of the modern theory of highly excited strongly interacting quantum many - body systems is that the phase relaxation time is much shorter than the energy relaxation time @xcite .
application of this idea to the theory of pre - compound and compound reactions implies the absence of correlations between energy fluctuating around zero s - matrix elements carrying different total spin and parity quantum numbers .
however the assumption of a very fast phase relaxation is in a conflict with many data sets on complex quantum collisions . in particular , the anomalously
long - lived spin off - diagonal s - matrix correlations have been revealed from the data on forward peaking of evaporating protons in nucleon induced @xcite and photonuclear @xcite reactions .
these long - lived correlations reflect an anomalously slow phase relaxation , which is many orders of magnitude longer than the energy relaxation .
this provides a manifestation of a new form of matter : thermalized non - equilibrated matter introduced by one of us @xcite .
the problem is of a primary importance for the nuclear technology applications . the previous analysis @xcite dealt with the data obtained for relatively low , @xmath015 - 20 mev , excitation energies of the compound nuclei .
for such energies a second chance proton evaporation is either forbidden or negligible . in this contribution
we analyze the data sets for the higher energy when there is a high probability for evaporation up to three and even four nucleons from the thermalized compound nucleus .
the data demonstrate that , after evaporation of nucleons , the residual compound nuclei are in ( i ) a coherent superposition of the strongly overlapping resonance states with different total spins and parities , and ( ii ) the phase relaxation between the strongly overlapping resonance states with different total spins and parities is still anomalously long in these highly - excited residual compound nuclei .
in figs . 1 - 6 we present the scaled low energy proton spectra , for forward and backwards angles , produced in nucleon induced reactions with the incident energy @xmath062 mev .
the spectra are in arbitrary units and are scaled with ( i ) the outgoing proton energy @xmath6 , and ( ii ) the cross section , @xmath7 , of the inverse process of the capture of the proton with energy @xmath6 by the residual compound nucleus .
the latter quantities are taken from @xcite for @xmath8=1.3 fm . from the comprehensive analysis @xcite of the @xmath3bi(p , p ) scattering at e@xmath9=62 mev
it follows that the multi - step direct reaction contribution is negligible for e@xmath10 mev . from the structure of the multi - step direct reactions models
it is clear that this should apply to all the data sets analyzed in this contribution .
our analysis , based on the exiton model @xcite , shows that , for e@xmath10 mev , the multi - step compound reaction contribution is also negligible for all the data sets analyzed here ( figs . 1 - 6 ) .
thus , for e@xmath10 mev , we deal exclusively with evaporation cascade from the chain of thermalized cooling down compound nuclei . indeed the scaled spectra show locally exponential @xmath6-dependence , in agreement with the statistical evaporation model @xcite .
the slopes of the scaled spectra the inverse nuclear `` temperatures '' decrease with decrease of the energy of the evaporated protons .
this reflects a process of cooling down of the compound nuclei by means of particle evaporation : the smallest energies of the evaporating protons correspond to the mainly last chance of evaporation .
notice that the slightly different slopes for forward and backward angles do not mean different `` temperatures '' but reflect different @xmath6-dependence of the transmission coefficients for different exit channel orbital momenta of evaporating protons .
the scaled outgoing proton spectra , for different angles , from the @xmath1au(p , p ) reaction .
the lines connecting the data full points are for eye guide .
the data errors reported are about of the size of the full dots .
the data are from @xcite .
, width=377,height=377 ] the same as in fig . 1 but for the @xmath2pb(p , p ) reaction .
the data are from @xcite .
, width=377,height=377 ] the same as in fig . 1 but for the @xmath3bi(p , p ) reaction
. the data are from @xcite .
, width=377,height=377 ] the same as in fig . 1 but for the @xmath3bi(n , p ) reaction
. the data are from @xcite .
, width=377,height=377 ] the same as in fig . 1 but for the @xmath4u(n , p ) reaction
. the data are from @xcite .
, width=377,height=377 ] the same as in fig . 1 but for the @xmath4u(p , p ) reaction
. the data are from @xcite .
, width=377,height=377 ] yet , in spite of a complete energy relaxation ( thermalization ) one observes that angular distributions of the evaporating protons are not symmetric around 90@xmath11 up to the last chance thermal proton emission .
the forward thermal emission intensities exceed the backward ones by factors of 2 to 6 .
the data @xcite are in good agreement with each other and show that forward peaking is basically present for many angular distributions of @xmath12 , @xmath13 , @xmath14he and @xmath15-particles in the low energy , typically evaporation , domain of the outgoing spectra . in ref .
@xcite , a forward peaking of evaporating neutrons is also reported .
thus , the data in figs . 1 - 6 demonstrate that thermalized non - equilibrated matter is stable with respect to a multi - step evaporation cascade .
this also means that the phase relaxation widths @xmath16 , introduced by one of us @xcite , are about or less than compound nucleus total decay widths ( defined as inverse average compound nuclei life - times ) , on each step of the cascade .
the smallest compound - nucleus total - decay width corresponds to the last chance evaporation , i.e. to the smallest proton outgoing energy . from the slopes of the scaled spectra at the smallest energies in figs . 1 - 6
we evaluate the `` temperatures '' of the residual compound nuclei .
this allows us to evaluate excitation energies of the compound nuclei from which protons , with the lowest measured energies in figs . 1 - 6 ,
were evaporated . finally , using the statistical model expression for the compound nucleus total decay width , which is in a good agreement with the data systematics @xcite
, we find the upper limits for the phase relaxation widths for the data in figs . 1 - 6 .
these upper limits are in the range from the maximal value of @xmath01 ev for the data @xcite ( fig .
3 ) ( because of the lowest energy reported @xcite is maximal among all the other data sets analyzed here ) down to the value of @xmath010@xmath5 ev for the data @xcite(fig .
6 ) . for a comparison ,
typical nuclear energy relaxation widths are evaluated as @xmath171 mev @xcite .
we note that in all the data sets in figs . 1 - 6 , the excitation energy of the compound nuclei for the last chance proton evaporation is @xmath18 10 mev .
this means that the average level - spacing @xmath19 mev yielding effective dimensions of hilbert spaces @xmath20 .
thus , the effect of anomalously slow phase relaxation for the data sets analyzed in this contribution is again revealed for exponentially large effective dimensions of hilbert space @xcite . yet , for the upper limit of lowest @xmath21 ev obtained above , we have @xmath22 .
we briefly mention that the new form of matter thermalized non - equilibrated matter is of primary importance for many - qubit quantum computation since anomalously long - lived phase memory can extend the time for quantum computation far beyond the quantum chaos border @xcite . an effect of a very slow phase relaxation has also been strongly supported by numerical calculations for h+d@xmath23 @xcite , f+hd @xcite and he+h@xmath24 @xcite chemical reactions . in these calculations ,
a slow phase relaxation ( @xmath25 ) manifests itself in stable rotating wave - packets of the excited intermediate complexes @xcite .
interestingly , this same effect of stable coherent rotation @xcite was originally revealed and described for heavy - ion elastic and inelastic scattering , e.g. for @xmath26c+@xmath27 mg @xcite , @xmath27mg+@xmath27 mg and @xmath28si+@xmath28si @xcite , @xmath29ni+@xmath30ti and @xmath29ni+@xmath31ni @xcite scattering ( see also @xcite ) as well as for strongly dissipative heavy - ion collisions , e.g. ,
@xmath32f+@xmath33y @xcite , @xmath28si+@xmath34ni @xcite and @xmath28si+@xmath35ti @xcite collisions .
slow phase relaxation has been found responsible for quantum - classical transition and schrdinger cat states in heavy - ion collisions and chemical reactions @xcite .
another manifestation of the thermalized non - equilibrated matter is a strong channel - channel correlation in dissipative heavy - ion collisions @xcite .
the correlation has nothing to do with s - matrix unitary condition for fixed total spin and parity quantum numbers but originates from the correlation and slow phase relaxation between the strongly overlapping resonance states carrying different total spin and parity values .
this correlation results in non - self - averaging of excitation function oscillations in these processes .
this signifies that @xmath16 values are much smaller than @xmath36 and yet requires finite @xmath16 values , comparable with the total decay width of the intermediate system @xcite .
recently , a fine energy resolution measurement of excitation functions for @xmath15-particle yields in @xmath32f(110 - 118 mev)+@xmath37al with the energy step 250 kev in laboratory system has been performed @xcite . for each incident energy ,
the @xmath15-particle energy spectra have typical evaporation shape .
yet , in spite of a summation over the whole evaporation spectra for each energy step , the corresponding excitation functions show strong oscillations .
this provides yet another demonstration that even though the intermediate system is thermalized it is not equilibrated @xcite .
thermalization does not guarantee equilibrium !
there exists an indication that the anomalously slow phase relaxation and the thermalized non - equilibrated matter may lead to anomalous sensitivity of the cross sections in complex collisions @xcite .
the phenomenon is intimately related to the channel - channel correlation and non - self - averaging of the excitation function oscillations indicating deterministic randomness in complex collisions .
so far the experiments @xcite have been supporting the theoretical conjecture @xcite .
the data analyzed in this contribution demonstrate that , after the evaporation cascade , the residual compound nuclei are in ( i ) a coherent superposition of the strongly overlapping resonance states with different total spins and parities , and ( ii ) the phase relaxation between the strongly overlapping resonance states with different total spins and parities in the highly - excited residual compound nuclei is still anomalously long corresponding to the upper limit of the phase relaxation widths to be in the range 1 - 10@xmath5 ev .
this is from 6 to 10 orders of magnitude less than typical nuclear energy relaxation ( spreading ) widths @xcite .
a possibly interesting aspect of the present discussion is the following . after emitting a first proton the highly excited residual nucleus
is entangled with this proton .
the thermally emitted proton is expected to readily interact with , say , atoms in the target , i.e. with `` environment '' .
therefore , this emitted proton very quickly ( possibly quicker than the average time to complete the whole evaporation cascade ) experiences decoherence .
accordingly , the reduced density matrix of the residual nucleus may also be expected to correspond to a fully mixed state unless it belongs to a decoherence free subspace of the highly excited strongly interacting many - body system . on the other hand , for the all data sets analyzed in this paper
, it is not excluded that there is a considerable contribution of events when the proton is emitted on the last stage of the cascade only , while all the previous cascade steps proceeded through evaporation of neutrons .
then , since effective interaction of neutrons with the `` environment '' is much weaker than that of protons , the decoherence effects should be expected to be suppressed / postponed .
it could be of interest to design experiments to try to find out which of the two possibilities may actually be realized , leading to the residual nucleus being in coherent superposition of strongly overlapping resonance states with different total spin and parity values ( not to mention anomalously long phase relaxation of such a coherent superposition ) .
one does not a priory exclude a possible relationship between the robust , against decoherence , quantum superpositions and their anomalously slow phase relaxation in a view that the latter was introduced by means of ensemble averaging over couplings of a given ( fixed ) intermediate system with continuum ( `` environment '' ) .
we are grateful to erwin raeymackers for kindly providing us with the data tables from refs @xcite .
we are thankful to thomas seligman for useful discussions .
we acknowledge financial support from the projects in-111607 ( dgapa - unam ) and 79988 ( conacyt ) .
saul hernndez - quiroz acknowledges a conacyt phd scholarship . s.yu .
kun , in non - equilibrium and nonlinear dynamics in nuclear and other finite systems , edited by zhuxia li , ke wu , xizhen wu , enguang zhao , fumihiko sakata , aip conf .
597 ( aip , melville , ny , 2001 ) , p. 319 , ( 2001 ) . | modern theory quantum systems phase relaxation time shorter than energy relaxation @xcite application to - compound reactions implies correlations between energy fluctuating around zero s - matrix elements spin parity numbers assumption fast phase relaxation with data complex quantum collisions long - lived spin off - diagonal s - matrix correlations revealed from data on evaporating protons in nucleon induced @xcite photonuclear @xcite reactions reflect slow phase relaxation longer than energy relaxation new form matter thermalized non - equilibrated matter introduced by @xcite problem importance for nuclear technology applications previous analysis dealt data for low 20 mev excitation energies compound nuclei second chance proton evaporation forbidden or negligible this contribution data sets for higher energy high probability for evaporation up to three four nucleons from thermalized compound nucleus data demonstrate after evaporation residual compound nuclei in superposition overlapping resonance states with spins parities phase relaxation between long in highly - excited residual compound nuclei1 - 6 scaled low energy proton spectra forward backwards angles nucleon induced reactions incident energy @xmath062 mev spectra arbitrary units scaled with outgoing proton energy @xmath6 cross section @xmath7 inverse capture proton energy @xmath6 residual compound nucleus quantities from @xcite @xmath8=1.3 fm analysis @xmath3bi scattering@xmath9=62 mev multi - step direct reaction contribution negligible e@xmath10 mev all data sets analysis exiton model @xcite shows e@xmath10 mev multi - step compound reaction contribution negligible all data sets figs . 1 - 6 ) e@xmath10 mev evaporation cascade thermalized cooling compound nuclei scaled spectra show exponential @xmath6-dependence statistical evaporation model @xcite inverse nuclear temperatures decrease with energy evaporated protons reflects cooling down compound nuclei particle evaporation smallest energies evaporating protons last chance evaporation different slopes forward backward angles different temperatures reflect different @xmath6-dependence transmission coefficients different exit channel orbital momenta evaporating protonsscaled outgoing proton spectra different angles from @xmath1au(p , p ) reaction lines connecting data full points for eye guide data errors size full dots data from @xcite . width=377,height=377 same fig . 1 @xmath2pb(p , p ) reaction data from @xcite @xmath3bi(p , p ) reaction data from @xcite. @xmath3bi(n , p ) reaction data @xmath4u(n , p ) reaction data from @xmath4u(p , p ) reaction data complete energy relaxation thermalization angular distributions evaporating protons not symmetric around 90@xmath11 last chance thermal proton emission forward thermal emission intensities exceed backward ones by factors 2 to 6 data @xcite show forward peaking present for many angular distributions of @xmath12 @xmath13 @xmath14he @xmath15-particles in low energy evaporation domain of outgoing spectra @xcite forward peaking of evaporating neutrons reporteddata in figs 1 - 6 demonstrate thermalized non - equilibrated matter stable multi - step evaporation cascade phase relaxation widths @xmath16 introduced by @xcite or less than compound nucleus total decay widths on each step cascade smallest total decay width corresponds to last chance evaporation smallest proton outgoing energy smallest energies in figs 1 - 6 evaluate temperatures of residual compound nuclei excitation energies of nuclei from protons evaporated statistical model expression for compound nucleus total decay width with data systematics @xcite upper limits for phase relaxation widths for data figs 1 - 6 from maximal @xmath01 ev @xcite down to @xmath010@xmath5 ev for @xcite typical nuclear energy relaxation widths evaluated as @xmath171 mev @xcite data - excitation energy of compound nuclei for last chance proton evaporation is @xmath18 10 mev average level - spacing @xmath19 mev effective dimensions of hilbert spaces @xmath20 effect slow phase relaxation for data sets revealed for large dimensions hilbert space @xcite upper limit of lowest @xmath21 have @xmath22 new form thermalized non - equilibrated matter for many - qubit quantum computation long - lived phase memory time quantum computation beyond quantum chaos border @xcite slow phase relaxation supported by numerical calculations for h+d@xmath23 f+hd he+h@xmath24 @xcite chemical reactions slow phase relaxation ( @xmath25 ) manifests in stable rotating wave - packets of intermediate complexes @xcite effect rotation revealed for heavy - ion elastic inelastic scattering @xmath26c+ mg strongly dissipative heavy - ion collisions slow phase relaxation responsible for quantum - classical transition schrdinger cat states in heavy - ion collisions chemical reactions @xcite thermalized non - equilibrated matter strong channel - channel correlation in dissipative heavy - ion collisions @xcitecorrelation s - matrix condition for spin parity numbers originates from correlation slow phase relaxation between overlapping resonance states different spin parity values results in non - self - averaging excitation function oscillations @xmath16 values smaller than @xmath36 requires finite values decay width intermediate system fine energy resolution measurement of excitation functions for @xmath15-particle yields @xmath32f(110 - 118 mev)+@xmath37al 250 kev performed @xmath15-particle energy spectra have typical evaporation shape excitation functions show strong oscillations intermediate system thermalized not equilibrated equilibrium indication slow phase relaxation non - equilibrated matter may lead to anomalous sensitivity cross sections in complex collisions related to channel - channel correlation non - self - averaging excitation function oscillations deterministic randomness in complex collisions experiments theoretical conjecturedata analyzed in demonstrate after evaporation cascade residual compound nuclei in coherent superposition of overlapping resonance states with different spins parities phase relaxation between states in anomalously long corresponding to upper limit phase relaxation widths range 1 - 10@xmath5 ev 6 to 10 orders of magnitude less than typical nuclear energy relaxation widths @xcite after emitting first proton excited residual nucleus entangled with proton thermally emitted proton expected to interact with atoms in target environment proton quickly experiences decoherence reduced density matrix of residual nucleus may expected correspond to fully mixed state unless to decoherence free subspace of excited many - body system for considerable contribution of events when proton emitted on last stage of cascade while previous cascade steps through evaporation of neutrons effective interaction of neutrons with environment weaker than protons decoherence effects should be expected to be suppressed / postponedinterest design experiments two possibilities realized leading residual nucleus coherent superposition overlapping resonance states different total spin parity values long phase relaxation relationship robust quantum superpositions slow phase relaxation latter introduced ensemble averaging couplings intermediate system continuum grateful to erwin raeymackers data tables refs @xcite thankful thomas seligman discussions acknowledge financial support projects in-111607 dgapa - unam 79988 conacyt ) saul hernndez - quiroz acknowledges conacyt phd scholarship s.yu . kun non - equilibrium nonlinear dynamics in nuclear finite systems edited zhuxia li ke wu xizhen wu enguang zhao fumihiko sakata aip conf . 597 melville ny 2001 p. 319 | 0.504883 | 3,103 | 1,489 | , Saving $0.1 in GPT-4. |
177,594 | we discuss recent results on spectral properties of discrete alloy - type random schrdinger operators .
they concern wegner estimates and bounds on the fractional moments of the green s function . | a discrete alloy - type model is a family of operators @xmath0 on @xmath1 where @xmath2 is the negative discrete laplacian given by @xmath3 , and @xmath4 the multiplication operator by the function @xmath5 here @xmath6 , is an i.i.d .
sequence of random variables and @xmath7 a so - called single - site potential .
the minimal conditions which are assumed throughout this note are that the distribution of @xmath8 has a density @xmath9 and that @xmath10 belongs to @xmath11 .
there are several well - studied relatives of the discrete alloy - type model : the continuum alloy - type model @xmath12 on @xmath13 where @xmath14 has the same form as in . the anderson model on @xmath1 : this is the special case of the discrete alloy - type model where @xmath15 .
the correlated anderson model , which corresponds to @xmath15 and the case where the random coupling constants @xmath16 are no longer assumed independent . obviously , the discrete alloy - type model is a special case of case ( iii ) .
however , previous results on ( iii)-type models require certain regularity conditions which are typically not satisfied for alloy - type potentials , cf . section [ sec : example ] .
the main challenge of models of the type is that @xmath10 may change its sign which leads to negative correlations of the potential .
this results in a non - monotone dependence on the random variables of certain spectral quantities . for this reason many established tools for the spectral analysis of random schrdinger operators are not directly applicable in this model . for literature concerning the discussion in this section we refer to the references of @xcite .
our results concern averages of resolvents and projectors . the presented results are closely related to dynamical and spectral localisation .
we use the symbol @xmath17 to denote the average over the collection of random variables @xmath16 , @xmath18 . for @xmath19
we denote by @xmath20 the natural restriction of @xmath21 to the set @xmath22 .
we present here a selection of the results proven in @xcite . for @xmath23
we denote the set @xmath24^d \cap \mathbb{z}^d$ ] by @xmath25 .
the number of points in the support of @xmath10 is denoted by @xmath26 . [ thm : degenerate ] assume that the single - site potential @xmath10 has support in @xmath27 and @xmath28 is of bounded variation .
then there exists a constant @xmath29 depending only on @xmath10 such that for any @xmath23 , @xmath30 and @xmath31 we have @xmath32}(h_{\omega,\lambda_l})\big]\right\ } \le c_u \ , \|\rho\|_{\mathrm{var } } \ , \mathrm{rank}\ , u \ \epsilon\ , ( l+n)^{d\cdot ( n+1 ) } .\ ] ] here @xmath33 denotes the total variation of @xmath28 . by the assumption on the support of the single - site potential , @xmath34 .
our bound is linear in the energy - interval length and polynomial in the volume of the cube @xmath25 .
this implies that the wegner bound can be used for a localisation proof via multiscale analysis , as soon as an appropriate initial length scale estimate is at disposal .
[ thm : non - degenerate ] assume @xmath35 and that @xmath36 has compact support .
let @xmath37 be such that @xmath38 .
then we have for any @xmath23 , @xmath30 and @xmath31 @xmath32}(h_{\omega,\lambda_l})\big]\right\ } \le \frac{8}{\bar u } \ , \|\rho\|_{\mathrm{var } } \ , \min\big(l^d , \mathrm{rank}\ , u\big ) \
\epsilon\ , ( l+m)^{d } .\ ] ] in the case that the support of @xmath10 is compact we obtain a bound which is linear in the volume of the box and thus yields the lipschitz continuity of the integrated density of states .
in this subsection we assume that the function @xmath39 , defined by @xmath40 does not vanish , @xmath41 is compact , and @xmath42 .
let @xmath19 be finite .
for @xmath43 and @xmath44 we set @xmath45 .
[ thm : apriori - trans ] let @xmath46 and @xmath19 be finite .
then there exists a constant @xmath47 depending only on @xmath10 , such that for all @xmath48 and all @xmath43 @xmath49 theorem [ thm : apriori - trans ] is proven in section [ sec : proof ] .
the proof is a combination of ideas from @xcite and @xcite .
the proof also gives rise to a quantitative estimate on the constant @xmath47 . to our knowledge , theorem [ thm : apriori - trans ] does not immediately imply exponential decay of fractional moments of the green s function by standard methods .
in this subsection we assume @xmath50 and @xmath41 compact .
denote the diameter of the support by @xmath51 .
we define the following conditions which may or may not hold : * @xmath52 , * @xmath53 compact , * @xmath42 . for @xmath54 and @xmath48
we denote the green s function by @xmath55 and set @xmath56 let @xmath46 and assume that ( a ) holds .
then there exists a constant @xmath57 , depending on @xmath58 , @xmath59 and @xmath10 , such that for all @xmath60 with @xmath61 and all @xmath48 we have @xmath62 where @xmath63 and @xmath64 is defined by @xmath65 .
if @xmath66 , or equivalently @xmath67 , ineq . describes exponential decay .
[ thm : expd1_gen ] let @xmath68 and assume that either ( b ) holds with @xmath69 sufficiently small , * or * ( c ) holds with @xmath70 sufficiently small . then there exist constants
@xmath71 , such that eq . holds for all @xmath72 with @xmath73 and all @xmath48 .
theorem [ thm : expd1 ] and [ thm : expd1_gen ] are proven in a joint paper @xcite with a. elgart .
the estimates of theorem [ thm : expd1 ] and [ thm : expd1_gen ] concern only off - diagonal elements .
if we assume ( b ) and @xmath74 , then @xmath75 is uniformly bounded for @xmath72 and @xmath48 , see ( * ? ? ?
* appendix ) . in spite these estimates on the green s function , neither dynamical nor spectral localisation follow immediately using the existent methods . currently , the analogues of theorem [ thm : expd1 ] and [ thm : expd1_gen ] in higher dimension are an open question . in a forthcoming paper with a.
elgart we intend to extend theorem [ thm : expd1_gen ] to operators on a one - dimensional strip , i.e. @xmath76 with @xmath77 finite .
we show that discrete alloy - type models do not satisfy in general the regularity conditions required in @xcite for correlated anderson models .
we consider @xmath50 .
let the density function @xmath28 be of bounded support and let @xmath78 , @xmath79 and @xmath80 for @xmath81 . for simplicity
we require that the infimum of the support of @xmath28 is zero .
let @xmath82 and @xmath83 be the events @xmath84\}$ ] and @xmath85\}$ ] with @xmath86 . then ,
if @xmath87 one calculates @xmath88 for all @xmath86 and consequently we have @xmath89 this shows that the regularity assumptions required in @xcite are violated . in the case
@xmath90 one can proceed analogously , but with a different choice of the sets @xmath83 and @xmath82 .
let @xmath91 be such that the cube @xmath92^d \cap \mathbb{z}^d$ ] contains @xmath93 , which is the set of all lattice sites whose coupling constant influence the potential in @xmath22 .
let @xmath94 be the linear operator whose coefficients in the canonical orthonormal basis are @xmath95 for @xmath96 . since @xmath10 has compact support , the operator @xmath97 is bounded . since @xmath98 does not vanish , @xmath99 , see @xcite for details .
moreover , there exists an invertible matrix @xmath100 satisfying @xmath101 and @xmath102 , see ( * ? ? ?
* proposition 5 ) .
we set @xmath103 . from @xcite and @xcite , respectively , we infer that for @xmath43 with @xmath104 @xmath105 here @xmath106 are functions of @xmath107 , @xmath108 , and @xmath109 is a function of @xmath110 , @xmath111 .
set @xmath112 , @xmath113 , @xmath114 and @xmath115 . using the substitution @xmath116 and eq .
we obtain @xmath117 where @xmath118 .
since by construction @xmath119 for all @xmath120 , @xmath121 is now a function of @xmath122 , @xmath108 . for non - negative functions
@xmath123 with @xmath124 and @xmath125 one has for all @xmath126 the estimate @xmath127 where the first inequality is due to @xcite and the second is the fundamental theorem of calculus . set @xmath128 . using fubini s theorem and ineq .
we obtain @xmath129 % & = \lambda^{-s } + \frac{\lambda^{1-s}}{1-s } \int_{\omega_{\lambda^+ } } \bigl\lvert \frac{\partial}{\partial \zeta_x } k(b_{\lambda^+ } \zeta_{\lambda^+ } ) \bigr\rvert \lvert\det b_{\lambda^+}\rvert \ensuremath{\mathrm{d } } \zeta_{\lambda^+ } .\end{aligned}\ ] ] we calculate the partial derivative by the product rule , substitute back into original coordinates and obtain @xmath130 % & \leq \lambda^{-s } + \frac{\lambda^{1-s}}{1-s } \sum_{j \in \lambda^+ } \bigl\lvert b_{\lambda^+}(j , x ) \bigr\rvert \int_{\omega_{\lambda^+ } } \bigl\lvert \rho ' ( \omega_j ) \bigr\rvert \prod_{\substack{k \in \lambda^+ \\ k \not = j } } \rho ( \omega_k ) \ensuremath{\mathrm{d } } \omega_{\lambda^+ } \\%[1ex ] % & = \lambda^{-s } + \frac{\lambda^{1-s}}{1-s } \lvert b_{\lambda^+ } \rvert_1 \lvert \rho ' \rvert_{l^1 } \leq \lambda^{-s } + \frac{\lambda^{1-s}}{1-s } c_u \lvert \rho ' \rvert_{l^1}.\end{aligned}\ ] ] if we choose @xmath131 we obtain @xmath132 .
thus we have shown @xmath133 .
analogously we obtain for @xmath134 the estimate @xmath135 .
the additional factor @xmath136 arises in the case @xmath104 , since in eq .
there are two summands of the type @xmath137 , and each summand has an additional factor @xmath138 in its numerator .
a. elgart , m. tautenhahn and i. veseli , exponential decay of green s function for anderson models on @xmath139 with single - site potentials of finite support , preprint , arxiv:0903.0492 [ math - ph ] , ( 2009 ) . | discrete alloy - type model family operators @xmath0 on @xmath1 @xmath2 negative discrete laplacian @xmath3 @xmath4 multiplication operator function @xmath5 @xmath6 sequence random variables @xmath7 single - site potential minimal conditions assumed distribution of @xmath8 density @xmath9 @xmath10 belongs to @xmath11 relatives continuum alloy - type model @xmath12 on @xmath13 @xmath14 same form anderson model on @xmath1 special case @xmath15 correlated anderson model @xmath15 random coupling constants @xmath16 no longer independent discrete alloy - model special case of case ( iii ) previous results-type models require regularity conditions not satisfied for alloy - type potentials challenge @xmath10 change sign negative correlations potential non - monotone dependence on random variables spectral quantities tools spectral analysis of random schrdinger operators not applicable literature refer to references of @xcite results concern averages of resolvents projectors related to dynamical spectral localisation symbol @xmath17 average over random variables @xmath16 @xmath18@xmath19 denote @xmath20 restriction @xmath21 to set @xmath22 present selection results proven in @xcite @xmath23 denote set @xmath24^d\mathbb{z^d by @xmath25 number of points support @xmath10 denoted by @xmath26 assume single - site potential @xmath10 support in @xmath27 @xmath28 bounded variation constant @xmath29 depending on @xmath10 @xmath23 @xmath30 @xmath31 @xmath32}\omega\lambda_l @xmath33 denotes total variation of @xmath28 assumption support single - site potential @xmath34 bound linear in energy - interval length polynomial in volume of cube @xmath25 wegner bound used for localisation proof via multiscale analysis appropriate initial length scale estimate disposal assume @xmath35 @xmath36 has compact support @xmath37 @xmath38 .@xmath23}(h_{\omega,\lambda_l})\big]\right \frac{8}{\bar u }|\rho\{\mathrm{var }\min\big(l^d \mathrm{rank} u\big )\epsilon l+m)^{d } support @xmath10 compact bound linear in volume box yields lipschitz continuity integrated density states assume function @xmath39 defined @xmath40 vanish @xmath41 compact @xmath42 . let @xmath19 finite @xmath43 @xmath44 set @xmath45 @xmath46 @xmath19 finite constant @xmath47 on @xmath10 all @xmath48 @xmath43 @xmath49 theorem - proven in section proof ] proof combination ideas @xcite @xcite quantitative estimate on constant @xmath47 theorem apriori - trans ] imply exponential decay of fractional moments green s function assume @xmath50 @xmath41 compact denote diameter support by @xmath51 define conditions may not hold : @xmath52 @xmath53 compact @xmath42 .@xmath54 @xmath48 denote green s function @xmath55 set @xmath56 @xmath46 assume ( a ) holds constant @xmath57 depending @xmath58 @xmath10 @xmath60 @xmath61 @xmath48 @xmath62 @xmath63 @xmath64 defined @xmath65 @xmath66 @xmath67 ineq . describes exponential decay thm : expd1_gen ] @xmath68 assume ( b ) holds @xmath69 ( c ) holds @xmath70 constants @xmath71 eq . holds @xmath72 @xmath73 @xmath48 theorem [ thm : expd1 ] expd1_gen ] proven joint paper @xcite a. elgart estimates concern off - diagonal elements assume ( b ) @xmath74 @xmath75 uniformly bounded for @xmath72 @xmath48 appendix ) estimates green s function dynamical spectral localisation follow methods analogues higher dimension open question forthcoming paper extend theorem expd1_gen ] operators one - dimensional strip @xmath76 @xmath77 finitediscrete alloy models satisfy regularity conditions @xcite for anderson models consider @xmath50 density function @xmath28 bounded support @xmath78 @xmath79 @xmath80 for @xmath81 require infimum support @xmath28 zero @xmath82 @xmath83 events @xmath84 with @xmath86 @xmath87 calculates @xmath88 for @xmath86 @xmath89 regularity assumptions @xcite violated @xmath90 proceed analogously different choice sets @xmath83 @xmath82 @xmath91 cube @xmath92 contains @xmath93 set lattice sites coupling influence potential @xmath22 @xmath94 linear operator coefficients are @xmath95 for @xmath96 @xmath10 has compact support operator @xmath97 bounded @xmath98 vanish @xmath99 see @xcite details invertible matrix @xmath100 satisfying @xmath101 @xmath102 set @xmath103 infer @xmath43 with @xmath104 @xmath105 @xmath106 functions of @xmath107 @xmath109 function @xmath110 @xmath111 set @xmath112 @xmath113 @xmath114 @xmath115substitution @xmath116 eq obtain @xmath117 @xmath118 construction @xmath119 @xmath120 @xmath121 function @xmath122 @xmath108 non - negative functions @xmath123 @xmath124 @xmath125 @xmath126 estimate @xmath127 first inequality due @xcite second fundamental theorem calculus set @xmath128 fubini s theorem ineq obtain @xmath129 \lambda^{-s\frac\lambda^\omega_{\lambda^+\lvert\frac{\partial\zeta_x\lambda^+\lambda^+\lambda^+{\mathrm{d\zeta{\lambda^+{aligned calculate partial derivative product rule substitute original coordinates obtain @xmath130 % \lambda^{-s\frac{\lambda^{1-s}}{1-s_ \lambda^+\lvert b{\lambda^+}(j x\rvert\omega_{\lambda^+\omega_j_{\substack{k \lambda^+ = j \omega_k )\ensuremath{\mathrm{d }\omega_{\lambda^+ }[1ex % \lambda^{-s } \frac{\lambda^{1-s}}{1-s\lvert b{\lambda^+ _{l^1 } \lambda^{-s } +\frac{\lambda^{1-s}}{1-s }_{l^1}{aligned choose @xmath131 obtain @xmath132 shown @xmath133 obtain @xmath134 estimate @xmath135 additional factor @xmath136 arises @xmath104 two summands @xmath137 each additional factor @xmath138 numerator elgart tautenhahnexponential decay green anderson models single site potentials finite support preprint arxiv:0903.0492 math ph 2009 | 0.489477 | 2,795 | 1,520 | , Saving $0.1 in GPT-4. |
177,606 | using the renormalization group based smoothing technique we have studied the gauge invariant field strength correlator at @xmath0 and @xmath1 in pure @xmath2 gauge theory . in conjunction with a cluster analysis , the field strength correlator is used to study correlations between the clusters in space and color orientation . | with the rise of perfect lattice gauge actions , inverse blocking @xcite has become a viable tool in numerical lattice field theory . a natural requirement for the action
is that a classical solution , if interpolated in this way , remains classical on all finer lattices .
equilibrium configurations are not classical , nevertheless , a topological charge can be unambiguously assigned to the interpolating field @xcite for generic lattice configurations .
this has led to the hope to describe the topological ( instanton ) structure of monte carlo ( mc ) configurations with the help of the new method .
our variant of rg smoothing @xcite consists of ( 1 ) coarse graining a mc configuration ( blocking in order to strip off uv fluctuations of scale @xmath3 ) and ( 2 ) inverse blocking to interpolate it by a smooth configuration ( sm ) on the original lattice .
rg - smoothing by our method does not lead to classical configurations .
clustering of topological charge and action in dominantly ( anti)selfdual clusters @xcite is a property of sm lattice fields inherited from the mc vacuum . smoothed fields preserve the string tension , its abelian dominance and contain a topological susceptibility of the right magnitude . in earlier work we were concentrating on the monopole content @xcite of sm configurations in certain gauges , relating it to the topological charge distribution and to other gauge - invariant signatures . in our present study on sm samples at @xmath1
we found the dimensionless ratio @xmath4 , remarkably independent of @xmath5 , in the maximal abelian and laplacian gauges ( @xmath6 monopole density , @xmath7 topological susceptibility ) .
coming back to the topological structure of the @xmath2 vacuum , we have explored new capabilities of the rg smoothing technique : ( 1 ) to study the gauge invariant field strength correlators at finite temperature @xcite and , more recently , at @xmath1 , ( 2 ) to investigate further the clustering of topological charge ( started in @xcite with a study of correlations in euclidean and in color space ) . for the perfect action used in our studies
see @xcite .
the gauge invariant field strength correlator @xcite , _ , ( x_1-x_2 ) = 0| \ { g_(x_1 ) s(x_1,x_2 ) + g_(x_2 ) s^(x_1,x_2 ) } |0 describes the non - perturbative vacuum structure in a model - independent way .
decomposing it into two basic structure functions , _ , ( x ) = ( _ _ - _ _ ) ( d(x^2 ) + d_1(x^2 ) ) + + ( x_x _ _ + x_x _ _ - ... ) , exposes a confining part @xmath8 , while @xmath9 is related to the dominance of ( anti)selfdual local excitations @xcite contributing to it . using smoothing ( instead of cooling @xcite ) we want to study eventual effects of smoothing as a noise reduction method on the topological content .
the `` schwinger line '' @xmath10 , a path dependent transporter , requires ( for distances not on - axis on the lattice ) to perform a random sampling of paths of shortest length .
we have measured at finite @xmath11 electric and magnetic structure functions @xmath12 , @xmath13 , @xmath14 and @xmath15 at space - like distances for samples of @xmath16 smoothed @xmath17 configurations at @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 and @xmath23 ( a temperature interval @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 ) .
the lattice spacing @xmath30 has been calibrated using the zero temperature @xmath2 string tension by comparison with the monopole creutz ratios which reach a plateau early on a @xmath31 lattice ( except for @xmath32 ) . for @xmath1 the same functions have been studied on @xmath31 sm configurations at @xmath33 ( @xmath34 fm ) corresponding to the deconfinement @xmath35 found on @xmath17 @xcite for this action . in the confinement phase ,
the finite temperature correlators have an almost perfect electric - magnetic symmetry , even on our unsymmetric lattice .
the signal of the deconfining transition is the breakdown of @xmath36 ( fig .
1 ) . + ( a ) ( b ) +
a fit with exponential structure functions @xmath37 and @xmath38 does nt give a satisfactory description for all @xmath11 .
in @xcite we solve the equations defining @xmath39 and @xmath40 .
this leads to condensates and _ integrated _ correlation lengths given in table 1 .
[ tab : expfit ] .fit of condensates and integrated correlation lengths . [ cols="<,^,^,^,^ " , ] for @xmath1 we find @xmath41 and @xmath42 compatible with the lowest temperature there .
the values of the correlation lengths are in the expected ballpark . in terms of @xmath8 and @xmath43 the transition appears as the result of continuous changes . the degree of ( anti)selfduality begins to decrease already below the deconfining transition @xcite . for a better understanding
a detailed comparison with the cooling method ( presently not available for @xmath2 ) is under way .
the uncorrelated instanton liquid provides a good description of the qcd ground state .
there are no indications from the field strength correlator at @xmath1 @xcite for strong correlations between ( anti)instantons ( @xmath44 and @xmath45 ) . at higher temperatures interactions
are expected to be important in the instanton ensemble .
contrary to fermionic interactions , anstze for gluonic interactions rely on specific @xmath46 and @xmath47 superpositions . analyzing the clustered sm configurations we hoped to learn about instantons in the real yang mills vacuum . instead of supporting the instanton liquid picture
our results are more suggestive for a different interpretation of the clusters exposed by smoothing . the charge density has been measured by means of lscher s charge which is inexpensive for sm configurations .
we have defined topological clusters using a @xmath48 site percolation algorithm for assigning neighboring ( same sign ) lattice sites , marked to have @xmath49 ( threshold ) , to the same cluster .
clusters are characterized by a center ( identified by @xmath50 ) , a cluster volume @xmath51 and cluster charge @xmath52 .
eventual color correlations between different clusters can be studied by the normalized _ cluster overlap _ for a pair of clusters , = with the field strength correlator between the centers involved .
correlators of specified components would give , for instanton pairs , access to the relative color orientation matrix @xmath53 .
the overlap @xmath54 ( 2 ) can be identified with the adjoint trace @xmath55 . for random orientations
the average of @xmath54 over all pairs would vanish .
plotting the average for different pairs as a function of the distance shows that the clusters are _ not uncorrelated _
this average is sensitive to deviations of the histogram of @xmath56 from the haar measure .
maximizing the overlap by a gauge rotation one can find @xmath53 for each pair .
this method can be applied in all instanton finding studies .
our analysis has shown for @xmath46 pairs strong _ aligning _ correlations in the confined phase ( decreasing with distance ) but none for @xmath47 pairs and _ aligning _ correlations _ similar _ for both types of pairs in the deconfined phase .
a recent @xmath1 study confirms strong @xmath46 correlations and rather weak ones for @xmath47 .
these results are difficult to reconcile with any instanton picture .
therefore we have carefully studied the influence of the cutoff @xmath57 ( over an interval of space filling fraction between 1 and 10 % ) on the cluster composition .
the cluster multiplicity goes through a maximum before a cluster percolation transition towards huge and multiply charged clusters is observed . with a cutoff keeping the cluster multiplicity below the maximum we get clustering properties being rather cutoff independent ( apart from distance correlations between centers ) with an almost poissonian multiplicity distribution .
for instance , on the @xmath31 lattice at @xmath33 ( @xmath58 ) we find for @xmath59 an average number of clusters of @xmath60 .
a strong correlation exists relating charge and volume , @xmath61 ( with almost all @xmath62 ) .
if these clusters identified on sm configurations by lscher s charge could be interpreted as subclusters of instantons ( `` instanton quarks''= half - instantons for @xmath2 ) this would explain the unexpected pattern of color correlations .
equal sign clusters would be bound into instantons ( dipoles ) in the confinement phase .
they could dissociate in the deconfined phase with opposite charge correlations becoming of equal importance .
further investigations are worthwhile , relating this to the behavior of the field strength correlators above .
if this picture can be consolidated this would be a step beyond the present search for instanton structure by various cooling techniques @xcite .
99 t. a. degrand et al . , _ nucl .
_ * b475 * ( 1996 ) 321 ; _ ibid . _
* b478 * ( 1996 ) 349 . m. feurstein et al . , _ nucl . phys . _
* b511 * ( 1998 ) 421 .
ilgenfritz et al . , _ phys .
rev . _ * d58 * ( 1998 ) 094502 .
ilgenfritz et al . , hep - lat/9904010 .
ilgenfritz and s. thurner , hep - lat/9905012 .
ilgenfritz and s. thurner , hep - lat/9810010 .
a. simonov , _ phys .
* 39 * ( 1996 ) 313 .
ilgenfritz et al . , _ phys .
rev . _ * d58 * ( 1998 ) 114508 .
a. di giacomo et al . , _ nucl .
* b * ( proc . suppl . ) * 54a * ( 1997 ) 343 ( 1997 ) ; _ nucl .
_ * b483 * ( 1997 ) 371 ; m. delia et al .
, _ phys .
* b408 * ( 1997 ) 315 .
m. teper , plenary talk at this conference . | rise perfect lattice gauge actions inverse blocking @xcite viable numerical lattice field theory classical solution interpolated remains classical on finer lattices equilibrium configurations not classical topological charge assigned to interpolating field @xcite for generic lattice configurations hope describe topological structure monte carlo mc configurations new method variant rg smoothing @xcite coarse graining mc configuration uv fluctuations inverse blocking smooth configuration original lattice rg - smoothing lead to classical configurations clustering topological charge action in anti)selfdual clusters @xcite property sm lattice fields mc vacuum smoothed fields preserve string tension abelian dominance contain topological susceptibility right magnitude earlier work monopole content @xcite of sm configurations in gauges topological charge distribution gauge - invariant signatures study sm samples found dimensionless ratio @xmath4 independent of @xmath5 in maximal abelian laplacian gauges densitytopological structure @xmath2 vacuum explored rg smoothing technique study gauge invariant field strength correlators at finite temperature @xcite @xmath1 investigate clustering topological charge @xcite correlations euclidean color space see @xcite gauge invariant field strength correlator @xcite ( x_1-x_2 ) = 0 g_(x_1 ) s(x_1,x_2 ) +_2 ) s^(x_1 describes non - perturbative vacuum structure model - decomposing into structure functions x ) = - ) ( d(x^2 ) + d_1(x^2 ) ) + ( x_x x_x - ) exposes confining part @xmath8 @xmath9 related to anti)selfdual local excitations @xcite smoothing study effects noise reduction on topological content schwinger line '' @xmath10 path dependent transporter requires random sampling paths shortest lengthmeasured finite @xmath11 electric magnetic structure functions @xmath12 space - like distances samples @xmath16 smoothed @xmath17 configurations temperature interval @xmath24 lattice spacing @xmath30 calibrated using zero temperature @xmath2 string tension monopole creutz ratios plateau early @xmath31 lattice except @xmath32 @xmath1 same functions studied on @xmath31 sm configurations @xmath33 @xmath34 fm ) deconfinement @xmath35 found on @xmath17 confinement phase finite temperature correlators have almost perfect electric - magnetic symmetry even unsymmetric lattice signal deconfining transition breakdown of @xmath36 ( fig . 1 ) fit exponential structure functions @xmath37 @xmath38 satisfactory description for all @xmath11 solve equations defining @xmath39 @xmath40 leads to condensates integrated correlation lengths table 1 @xmath1 @xmath41 @xmath42 compatible with lowest temperature values correlation lengths in expected ballpark @xmath8 @xmath43 transition result continuous changesanti)selfduality below deconfining transition @xcite comparison with cooling method not @xmath2 under uncorrelated instanton liquid qcd ground state no indications field strength correlator at @xmath1 @xcite for strong correlations between anti)instantons @xmath44 @xmath45 higher temperatures interactions important in instanton ensemble gluonic interactions rely on @xmath46 @xmath47 superpositions analyzing clustered sm configurations instantons in yang mills vacuum instanton liquid results different interpretation clusters exposed by smoothing charge density measured lscher s charge inexpensive for defined topological clusters using @xmath48 site percolation algorithm neighboring lattice sites @xmath49 cluster clusters characterized by center @xmath50 volume @xmath51 charge @xmath52 color correlations between clusters studied by normalized cluster overlap field strength correlator between centers correlators of components access to relative color orientation matrix @xmath53 overlap @xmath54 identified with adjoint trace @xmath55 random orientations average of @xmath54 over pairsplotting average for pairs distance shows clusters not uncorrelated average sensitive to deviations histogram @xmath56 from haar measure maximizing overlap by gauge rotation @xmath53 for each pair method in instanton finding studies analysis shown @xmath46 pairs strong aligning correlations in confined phase decreasing distance none for @xmath47 pairs similar for in deconfined phase recent @xmath1 study confirms strong @xmath46 correlations weak for @xmath47 results difficult to reconcile with instanton picture studied influence cutoff @xmath57 1 10 % on cluster composition cluster multiplicity maximum before transition towards huge charged clusters cutoff multiplicity below maximum clustering properties cutoff independent almost poissonian multiplicity distribution @xmath31 lattice at @xmath33 @xmath59 average number of clusters of @xmath60 strong correlation charge volume @xmath61 almost all @xmath62 ) clusters interpreted as subclusters of instantons explain unexpected color correlations equal sign clusters bound into instantons in confinement phase could dissociate in deconfined phase opposite charge correlations equal importance further investigations worthwhile behavior field strength correlatorspicture consolidated step beyond search structure cooling techniques degrand et al . nucl b475 * 1996 ) 321 ibid b478 * 1996 ) 349 feurstein et al . nucl . b511 * 1998 ) 421 ilgenfritz al phys . rev. d58 * 1998 ) 094502 ilgenfritz hep - lat/9904010 hep lat/9905012/9810010 simonov phys . 39 1996 ) 313 ilgenfritz phys . rev . d58 * 1998 ) 114508 di giacomo et al . nucl . 54a 1997 ) 343 1997 nucl b483 * 1997 ) 371 delia et al . phys . b408 * 1997 ) 315 teper plenary talk conference | 0.356642 | 2,328 | 1,183 | , Saving $0.1 in GPT-4. |
177,609 | a theorem is proved concerning approximation of analytic functions by multivariate polynomials in the @xmath0-dimensional hypercube .
the geometric convergence rate is determined not by the usual notion of degree of a multivariate polynomial , but by the _
euclidean degree , _ defined in terms of the 2-norm rather than the 1-norm of the exponent vector @xmath1 of a monomial @xmath2 . | the aim of this paper is to prove a theorem concerning an effect identified in section 6 of @xcite .
if an analytic function @xmath3 is approximated by multivariate polynomials in the @xmath0-dimensional hypercube @xmath4^s$ ] , the usual notion of polynomial degree , namely the total degree , is not the right predictor of approximability . in the hypercube ,
the set of polynomials of a given total degree has @xmath5 times finer resolution along a direction aligned with an axis than along a diagonal .
conversely , the set of polynomials of a given maximal degree has @xmath5 times finer resolution along a diagonal than along an axis . to achieve balanced resolution in all directions one should work with polynomials of a given _ euclidean degree . _
our definitions are as follows . with @xmath6 ,
we consider functions @xmath7 in @xmath4^s$ ] , with @xmath8^s}^{}$ ] representing the maximum norm over this set . for a monomial @xmath9 we define @xmath10 \mbox{\em euclidean degree } : & ~~ & \de = \|\k\|_2^ { } , \\ [ 2pt ] \mbox{\em max degree } : & ~~ & \dmax = \|\k\|_\infty^{},\end{aligned}\ ] ] where @xmath11 , @xmath12 , and @xmath13 are the 1- , 2- , and @xmath14- norms of the @xmath0-vector @xmath15 , and the degree of a multivariate polynomial is the maximum of the degrees of its nonzero monomial constituents .
the total and maximal degree definitions are standard and appear in publications like @xcite and @xcite where multivariate polynomial approximation in the hypercube is discussed , but the euclidean degree seems to be new in @xcite .
note that @xmath16 is not in general an integer .
our interest is in leading order exponential effects , not algebraic fine points , and accordingly , we will make use of the notation @xmath17 defined as follows : @xmath18 if for all @xmath19 , @xmath20 as @xmath21 .
by @xmath22 we mean @xmath23 , or equivalently , for all @xmath19 , @xmath24 .
the case @xmath25 suffices for a numerical illustration .
let @xmath26 be the 2d runge function @xmath27 which is analytic for all real values of @xmath28 and @xmath29 and isotropic in the sense that it is invariant with respect to rotation in the @xmath28-@xmath29 plane .
figure [ fig ] gives an indication of the minimal error in approximation of @xmath26 on @xmath4 ^ 2 $ ] by bivariate polynomials of various total , euclidean , and maximal degrees .
( bivariate chebyshev coefficients of @xmath26 are plotted in figure 6.4 of @xcite . )
the figure is actually based on @xmath30 rather than @xmath31 approximations , since these are much easier to compute , but this is enough to give an indication of the separation between the convergence rates when the degree is defined by @xmath32 and when it is defined by @xmath16 or @xmath33 .
.1 in as a function of degree @xmath34 in the unit square , for three different definitions of degree .
the approximations come from least - squares minimization over a fine grid in @xmath4 ^ 2 $ ] .
straight lines mark the convergence rates of theorem @xmath35.,title="fig : " ] the function ( [ runge ] ) satisfies assumption a of our theorem , theorem [ thm1 ] , with @xmath36 , and the data in the figure show convincing agreement with the predictions of the theorem .
this function is analytic when @xmath28 and @xmath29 are real but not when they are complex . on the other hand the similar function
@xmath37 has real singularities just outside the unit square .
theorem [ thm1 ] applies with @xmath36 for this function too , and a plot of convergence rates ( not shown ) looks almost exactly like figure [ fig ] .
in the standard theory for a single variable @xmath28 , for any @xmath38 , let @xmath39 denote the open set bounded by the bernstein @xmath40-ellipse in the complex @xmath28-plane , i.e. , the image of the circle @xmath41 under the map @xmath42 .
a lipschitz continuous function @xmath26 defined on @xmath4 $ ] has an absolutely and uniformly convergent chebyshev series @xmath43 where @xmath44 is the chebyshev polynomial of degree @xmath45 .
truncating the series at degree @xmath34 gives the polynomial approximation @xmath46 the following result goes back to bernstein s prize - winning memoir of 1912 @xcite . here and elsewhere , when we say that @xmath26 is analytic in a region , we mean that if necessary @xmath26 can be analytically continued to that region .
[ lemma1 ] if @xmath26 is analytic in @xmath39 , its chebyshev coefficients and truncated chebyshev expansions satisfy @xmath47}^ { } = \oe(\kern .7pt\rho^{-n } ) .
\label{rhoests}\ ] ] the second estimate follows from the first , whose proof can be based on contour integrals over @xmath48-ellipses with @xmath49 for arbitrarily small @xmath50 , or equivalently on contour integrals over circles in the @xmath51-plane after a change of variables from @xmath52 $ ] to @xmath51 on the unit circle .
see theorems 8.1 and 8.2 of @xcite . for our purposes
it will be important to consider @xmath53 as well as @xmath28 .
when @xmath28 ranges over @xmath39 , with foci @xmath54 and @xmath55 and topmost point @xmath56 , @xmath53 ranges over another ellipse , with foci @xmath57 and @xmath55 and leftmost point @xmath58 , where @xmath59 and @xmath60 are related by @xmath61 arnold calls @xmath39 a _ hooke ellipse _ and @xmath62 a _ newton ellipse _ @xcite .
we wish to parametrize the latter by @xmath63 rather than @xmath60 , so we make the following definition . for any @xmath64 , @xmath65 is the open region in the complex plane bounded by the ellipse with foci @xmath57 and @xmath0 and leftmost point @xmath66 .
equivalently , it is the region consisting of points @xmath28 satisfying @xmath67 .
thus @xmath68 , and lemma [ lemma1 ] can be equivalently restated as follows .
[ cor1 ] suppose that for some @xmath69 , @xmath70 is analytic for all @xmath71 such that @xmath72 .
then @xmath73 holds with @xmath74 .
now let @xmath26 be a function of @xmath75^s$ ] for some @xmath6 .
if @xmath26 is smooth , it has a uniformly and absolutely convergent multivariate chebyshev series @xmath76 ( see e.g. theorem 4.1 of @xcite ) . here is our analyticity assumption , generalizing that of lemma [ cor1 ] .
_ * assumption a. * for some @xmath69 , @xmath77 is analytic for all @xmath78 in the @xmath0-dimensional region defined by the condition @xmath79 . _
note that a sufficient condition for assumption a to hold is that @xmath77 is analytic for all @xmath80 with @xmath81 .
the following lemma will be proved in the next section .
[ lemma2 ] if @xmath26 satisfies assumption a , its multivariate chebyshev coefficients satisfy @xmath82 where @xmath74 . based on this result
, our theorem bounds the convergence rates of polynomial approximations defined by total , euclidean , and maximal degree .
[ thm1 ] if @xmath26 satisfies assumption a , then @xmath83^s}^ { } = \begin{cases } \oe(\kern .7pt\rho^{-n/\sqrt s } ) & \hbox{if\/ } d=\dt , \\ \oe(\kern .7pt\rho^{-n } ) & \hbox{if\/ } d=\de , \\
\oe(\kern .7pt\rho^{-n } ) & \hbox{if\/ } d=\dmax , \end{cases}\ ] ] where @xmath74 .
_ proof of theorem @xmath35 , assuming lemma @xmath84 . _
the second ( middle ) assertion of the theorem follows from lemma [ lemma2 ] by truncating the multivariate chebyshev series ( [ chebs ] ) , since @xmath85 for all @xmath1 for all @xmath75^s$ ] .
the third assertion is a consequence of the second , since @xmath86 for any multivariate polynomial @xmath87 .
the first assertion is also a consequence of the second since @xmath88 .
to complete the proof of theorem [ thm1 ] we must prove lemma [ lemma2 ] . for this we will make use of a result in the book by bochner and martin @xcite .
let @xmath89 be an @xmath0-vector with @xmath90 for each @xmath91 , and let @xmath92 be the _
elliptic polycylinder _ defined as the set of all points @xmath78 such that @xmath93 for each @xmath91 .
the result in question is an @xmath0-dimensional generalization of lemma [ lemma1 ] .
[ lemma3 ] let @xmath26 be analytic in @xmath94 .
then its multivariate chebyshev coefficients satisfy @xmath95 as @xmath96 .
equation ( [ eq12 ] ) means that for any @xmath19 , @xmath97 .
this is essentially theorem 11 on p. 95 of @xcite , which is derived by contour integrals . for further discussion
see @xcite . for
any @xmath0-vector @xmath1 of nonnegative indices , define @xmath98 and @xmath99 then @xmath100 , so by assumption a , @xmath77 is analytic in the subset of @xmath101 defined by the condition @xmath79 . from lemma [ minklem ] below , we have @xmath102 where @xmath103 denotes the standard minkowski sum of sets .
it follows that @xmath77 is analytic whenever @xmath104 for each @xmath91 . in other words
, @xmath77 is analytic in the elliptic polycylinder @xmath105 with @xmath106 defined by @xmath107 it can be shown ( lemma [ lemma4 ] , below ) this this final quantity is greater than or equal to the number @xmath108 which we define by @xmath109 therefore if @xmath110 is the @xmath0-vector with components given by this formula , then the associated polycylinder satisfies @xmath111 , and @xmath77 is analytic in @xmath94 .
we now calculate @xmath112 and inserting this identity in ( [ eq12 ] ) gives ( [ multicoeffs ] ) , as required
. here are the two lemmas just used .
[ minklem ] for any @xmath113 and @xmath114 , @xmath115 if @xmath116 and @xmath117 , then we have @xmath118 therefore by the triangle inequality , @xmath119 that is , @xmath120 which implies @xmath121 .
[ lemma4 ] for any @xmath122 and @xmath123 $ ] , @xmath124 .
given @xmath59 , define @xmath125 .
we must show @xmath126 , or equivalently @xmath127 where @xmath128 since @xmath129 a sufficient condition for ( [ convex ] ) to hold is that @xmath130 is convex in the sense that @xmath131 for @xmath123 $ ] .
this follows from the identity @xmath132 .
this work is motivated by computational applications , since for computation in higher dimensions , a hypercube is usually the domain of choice .
theorem [ thm1 ] suggests that any method for computation in a hypercube that is based on one of the familiar definitions of the degree of a multivariate polynomial , namely total degree or maximal degree , is likely to be suboptimal .
for example , a standard idea of multidimensional quadrature ( cubature ) is the exact integration of multivariate polynomial approximations of a given total degree , an idea going back to maxwell @xcite . for functions @xmath26 with approximately angle - independent complexity in the hypercube , the suboptimality factors may be exponentially large as @xmath133 , since the number of degrees of freedom increases by @xmath134 when the degree increases by @xmath5 .
for example , in @xmath135 dimensions , our results suggest that isotropic analytic functions can be resolved to a given accuracy by polynomials of fixed euclidean degree with @xmath136 times fewer degrees of freedom than polynomials of fixed total degree , and @xmath137 times fewer degrees of freedom than polynomials of fixed maximal degree
. we shall not give details here , since these are discussed in section 6 of @xcite with formulas and a table to quantify the exponential effects .
theorem [ thm1 ] is only an upper bound , so in principle , the difference it suggests between @xmath32 and the other degrees @xmath16 and @xmath33 might be illusory
. however , numerical experiments such as that of figure [ fig ] and those reported in @xcite make it clear that the difference is genuine .
this could be made rigorous by the development of a converse theorem , as has been long established in the 1d case , again thanks to bernstein ( see theorem 8.3 of @xcite ) . in closing
i would like to highlight a conceptual link between this note and my earlier paper @xcite with nick hale .
the central observation of @xcite is that the resolving power of ( univariate ) polynomials on an interval @xmath4 $ ] is nonuniform , making polynomials fall short of optimality by a factor of @xmath138 in representing functions whose complexity on @xmath4 $ ] is uniform . in the present work ,
the issue is again nonuniformity of polynomials , but now they are multivariate and the uniformity issue pertains to rotation rather than translation . as pointed out in section 7 of @xcite , the translational issue is present in multiple dimensions too .
i have benefitted from extensive discussions of approximation and cubature in the hypercube with hadrien montanelli and klaus wang .
i thank also jared aurentz , stefan gttel , nick hale , allan pinkus , and alex townsend for helpful suggestions .
it was aurentz who proposed the term `` euclidean degree . '' | aim paper prove theorem effect in section 6 @xcite if analytic function @xmath3 approximated by multivariate polynomials in @xmath0-dimensional hypercube @xmath4^s$ ] usual polynomial degree not right predictor approximability in hypercube polynomials total degree has @xmath5 times finer resolution along direction aligned with axis than diagonal conversely polynomials maximal degree @xmath5 times finer resolution along diagonal axis balanced resolution work with polynomials of euclidean degree definitions with @xmath6 consider functions @xmath7 in @xmath4^s$ ] @xmath8^s} maximum norm set for monomial @xmath9 define @xmath10 euclidean degree max degree @xmath11 @xmath12 @xmath13 are 1- 2- @xmath14- norms of @xmath0-vector @xmath15 degree of multivariate polynomial is maximum of degrees nonzero monomial constituents total and maximal degree definitions standard in publications @xcite polynomial approximation euclidean degree new in @xcite@xmath16 not integer interest in order exponential effects not algebraic points notation @xmath17 defined @xmath18 @xmath19 @xmath21 @xmath22 @xmath23 @xmath19 @xmath24 @xmath25 suffices for numerical illustration @xmath26 be 2d runge function @xmath27 analytic for values @xmath28 and @xmath29 invariant rotation in @xmath28-@xmath29 plane figure minimal error in approximation of @xmath26 on @xmath4 ^ 2 $ ] by bivariate polynomials of total euclidean maximal degrees bivariate chebyshev coefficients of @xmath26 plotted in figure 6.4 based on @xmath30 @xmath31 approximations separation between convergence rates defined by @xmath32 @xmath16 or @xmath33 function of degree @xmath34 unit square three definitions approximations from least - squares minimization over fine grid in @xmath4 ^ 2 $ straight lines mark convergence rates of theorem @xmath35 function satisfies assumption a theorem with @xmath36 data show agreement with predictionsfunction analytic when @xmath28 @xmath29 real not complex similar function @xmath37 has singularities outside unit square theorem applies with @xmath36 plot convergence rates like figure standard theory single variable @xmath28 @xmath39 open set by bernstein @xmath40-ellipse in complex @xmath28-plane image circle @xmath41 under map @xmath42 lipschitz continuous function @xmath26 @xmath4 has convergent chebyshev series @xmath43 @xmath44 polynomial of degree @xmath45 truncating series at @xmath34 gives polynomial approximation @xmath46 result bernstein s memoir 1912 @xmath26 analytic in region continued region @xmath26 analytic in @xmath39 chebyshev coefficients truncated chebyshev expansions satisfy @xmath47 second estimate follows first proof based on contour integrals over @xmath48-ellipses with @xmath49 @xmath50 contour integrals over circles in @xmath51-plane after change variables from @xmath52 to @xmath51 unit circletheorems 8.1 8.2 @xcite consider @xmath53 @xmath28 @xmath28 ranges over @xmath39 foci @xmath54 @xmath55 topmost point @xmath56 @xmath53 ranges over ellipse foci @xmath57 @xmath55 leftmost point @xmath58 @xmath59 @xmath60 related by @xmath61 @xmath39 hooke ellipse @xmath62 newton ellipse @xcite parametrize latter by @xmath63 @xmath60 definition @xmath64 @xmath65 open region complex plane ellipse foci @xmath57 @xmath0 leftmost point @xmath66 region points @xmath28 satisfying @xmath67 @xmath68 lemma restated @xmath69 @xmath70 analytic for @xmath71 @xmath73 holds with @xmath74 @xmath26 function @xmath75 @xmath6 @xmath26 smooth convergent multivariate chebyshev series @xmath76 theorem 4.1 @xcite analyticity assumption @xmath69 @xmath77 analytic for all @xmath78 @xmath0-dimensional region defined condition @xmath79sufficient condition for assumption a @xmath77 analytic for all @xmath80 with @xmath81 lemma proved next section if @xmath26 satisfies assumption a multivariate chebyshev coefficients satisfy @xmath82 @xmath74 theorem bounds convergence rates of polynomial approximations defined by total euclidean maximal degree if @xmath26 satisfies assumption a @xmath83^s\rho d=\dt=\dmax @xmath74 proof of theorem @xmath35 assuming lemma @xmath84 second assertion follows from lemma by truncating multivariate chebyshev series since @xmath85 for all @xmath1 for @xmath75^s$ third assertion consequence of second since @xmath86 for any multivariate polynomial @xmath87 first assertion consequence of second since @xmath88 complete proof of theorem prove lemma [ lemma2 ] result in book by bochner and martin @xcite@xmath89 @xmath0-vector with @xmath90 @xmath91 @xmath92 elliptic polycylinder set points @xmath78 @xmath93 each @xmath91 @xmath0-dimensional generalization of lemma @xmath26 analytic in @xmath94 multivariate chebyshev coefficients satisfy @xmath95 @xmath96 equation eq12 ] @xmath19 @xmath97 theorem 11 p. 95 @xcite derived by contour integrals @xcite @xmath0-vector @xmath1 nonnegative indices define @xmath98 @xmath99 @xmath100 @xmath77 analytic subset @xmath101 @xmath79 @xmath102 @xmath103 standard minkowski sum of sets @xmath77 analytic @xmath104 for each @xmath91 @xmath77 analytic elliptic polycylinder @xmath105 with @xmath106 defined @xmath107 final quantity greater than or equal to number @xmath108 @xmath110 @xmath0-vector components formula polycylinder satisfies @xmath111 @xmath77 analytic in @xmath94 calculate @xmath112 identity eq12 ] gives multicoeffs ] ) two lemmas usedminklem @xmath113 @xmath114 @xmath115 @xmath116 @xmath117 @xmath118 triangle inequality @xmath119 @xmath120 implies @xmath121 lemma4 @xmath122 @xmath123 @xmath124 @xmath59 define @xmath125 show @xmath126 @xmath127 @xmath128 @xmath129 condition convex @xmath130 convex @xmath131 @xmath123 $ follows identity @xmath132 work motivated computational applications computation higher dimensions hypercube domain choice theorem thm1 suggests method computation hypercube based familiar definitions degree multivariate polynomial total degree maximal degree suboptimal standard idea multidimensional quadrature integration multivariate polynomial approximations total degree maxwell @xcite functions @xmath26 angle - independent complexity hypercube suboptimality factors exponentially large @xmath133 degrees freedom increases by @xmath134 degree increases @xmath5 @xmath135 dimensions results suggest isotropic analytic functions resolved polynomials fixed euclidean degree @xmath136 times fewer degrees freedom total degree @xmath137 fewer degrees freedom fixed maximal degreedetails discussed in section 6 @xcite formulas table exponential effects theorem [ thm1 ] upper bound difference between @xmath32 degrees @xmath16 @xmath33 might illusory numerical experiments figure ] @xcite difference genuine rigorous by converse theorem established in 1d case thanks to bernstein theorem 8.3 @xcite highlight conceptual link between note earlier paper @xcite with nick hale observation resolving power of univariate polynomials on interval @xmath4 $ ] nonuniform fall short of optimality factor @xmath138 functions complexity uniform present work issue nonuniformity polynomials multivariate uniformity pertains to rotation translation section 7 @xcite translational issue present in multiple dimensions benefitted from discussions of approximation cubature hypercube with hadrien montanelli klaus wang thank jared aurentz stefan gttel nick hale allan pinkus alex townsend for suggestions aurentz proposed term `` euclidean degree | 0.501952 | 3,412 | 1,829 | , Saving $0.1 in GPT-4. |
177,627 | with an effective lagrangian approach , we give a full analysis on the @xmath0 and @xmath1 reactions by exploring the roles of various resonances with mass up to 1.72 gev .
we find large contributions from @xmath2 , @xmath3 , @xmath4 and @xmath5 resonances .
our calculations also indicate sizeable contributions from nucleon poles for the energies close to the threshold . a good description to the existing data of different isospin channels of @xmath6 and @xmath7 for beam energies up to 2.2 gev
is reached .
our results provide important implications to the abc effect and guildlines to the future experimental projects at cosy , hades and hirfl - csr .
we point out that the panda at fair could be an essential place for studying the properties of baryon resonances and the data with baryon and anti - baryon in final states are worth analyzing . | our understanding on the low energy strong interaction physics is still unsatisfactory though quantum chromodynamics ( qcd ) has been established as the standard theory of strong interaction for many years .
because of its color confinement non - perturbative properties , we have difficulties to calculate the properties of mesons and baryons directly from qcd .
constituent quark models are developed to explain the mesonic and baryonic spectrum and they are successful in some aspects .
but one of severe problems is that they predict more excited states around 2 gev than what have been observed experimentally . as a result
, a lot of efforts have been devoted to the meson production in pion- , photo- , and electro - induced reactions in order to reduce the uncertainties of extracted parameters of resonances , deepen our knowledge on the structure of the resonances and search for missing resonances .
in the past few years , the double pion production in nucleon - nucleon collisions tends to be a fascinating field for studying resonances properties and has been accurately measured at the facilities of celsius@xcite , cosy@xcite , kek@xcite , and pnpi - gatchina@xcite . in the table [ data ] , we list all the data measured after the year of 2000 . the celsius@xcite and cosy@xcite collaboration measured the differential cross sections of @xmath8 and @xmath9 channels for the beam energies from the threshold to 1.3 gev .
one of the promising findings was that the @xmath3 contribution dominates at the close - to - threshold region , as expected by the valencia model@xcite .
this provided us a good place to explore the properties of @xmath3 whose theoretical interpretation is still under controversial . however the valencia model , though compatible with the old bubble chamber and magnetic spectrometer data@xcite , overestimated the new data at the close - to - threshold energies by several times .
another finding was that the celsius data of @xmath10 invariant mass spectra for the beam energies above 1.0 gev demonstrated a single peak at the low invariant mass while the model gave double hump structure which was inconsistent to the data .
the model also predicted a preferential parallel emission of the two pion - meson , which was contrary to the celsius data .
the third interesting finding was that there was a level - off behavior around the beam energies of 1.0 gev in the energy dependence of the @xmath9 , which was thought to be the result of interference between different contributions .
kek collaboration made another progress and they measured the total cross section of @xmath11 and @xmath12 channels which were important for our understanding on the excitation of @xmath3 . on the other side , though the role of @xmath3 in nucleon - nucleon collisions was firmly established , its contribution in @xmath7 reactions has never been considered .
jetset@xcite measured the @xmath13 channel in order to search for narrow resonances decaying to @xmath14 but they only included the double-@xmath2 diagram in their monte - carlo simulation .
moreover , it has never been explored whether it was possible to extract the properties of other resonances from antinucleon - nucleon collisions .
the one - pion exchange ( ope ) models@xcite of more than 45 years ago only considered the double-@xmath2 diagram and the valencia model of more than 10 years ago made a further step by including also the @xmath3 resonance and the @xmath15- and @xmath16-meson exchange . with the new accurate data ,
it is very necessary to perform a more comprehensive analysis including more resonances and matching all the data of ( anti)nucleon - nucleon collisions .
our full model is demonstrated in ref .
, so herein we only give a brief description to the main features .
the effective lagrangians for the resonances in our model are based on a lorentz covariant orbital - spin ( l - s ) scheme @xcite . in view of the overall system invariant mass about 2.8 gev for tp = 2.2 gev , we have checked contributions from all the well - established @xmath17 and @xmath18 resonances below 1.72 gev .
the coupling constants appearing in relevant resonances are determined by the empirical partial decay width of the resonances taken from particle data group ( pdg ) book@xcite .
our calculated results show that the @xmath3 , @xmath19 , @xmath20 , and @xmath21 resonances play the relatively significant role in the considered energies and other resonances give negligible contributions @xcite . in our model ,
the adjustable parameters are the cut - off values in form factors at vertices and resonances .
we include five types of form factors : those at the meson-(anti)nucleon-(anti)nucleon vertices , those at the meson-@xmath17(@xmath18)-(anti)nucleon vertices , the blatt - weisskopf barrier factors at the @xmath3-@xmath2-@xmath22 vertex @xcite , the form factor at the @xmath15-@xmath22-@xmath22 vertex , and the form factor of resonances and nucleon poles .
the @xmath23 channel is very useful to pin down these cut - off values in form factors of the relevant @xmath2 and @xmath18 resonances because it has negligible @xmath17 contribution , and then we could easily determine the cut - off values of @xmath17 resonances in other channels .
if the double-@xmath2 mechanism is dominant at high energies as in the ope model and valencia model , the total cross section of @xmath24 should be a factor of about four larger than that of @xmath23 according to isospin coefficients .
however , the new exclusive measurements indicate an approximately equal value of these two channels , which is consistent to the old bubble - chamber data@xcite . after calculating all the possible contributions , our model got two important conclusions :
our model reduces the relative branching ratio of @xmath25 and assumes a smaller cutoff parameter for the @xmath26 coupling so the relative contribution from the @xmath27 term increases significantly .
our model introduces significant contributions from @xmath28 at energies near threshold and from @xmath20 and @xmath21 at energies above 1.5 gev .
this is more obvious in the @xmath23 channel because the isospin coefficients of these terms are much bigger than those in other channels . as a result ,
our description of the total cross sections of all isospin channels is considerably improved and our results agree well with the new data of @xmath29 collisions in the close - to - threshold region .
however , we find that it is difficult to interpret the @xmath10 invariant mass spectra for the beam energies above 1.0 gev and the level - off behavior around 1.0 gev in the total cross section of the @xmath9 channel . in fig .
[ pipidis ] , we give the invariant mass spectra and angular distribution of the opening angle for @xmath30 system in the overall center - of - mass system in the @xmath9 channel at the beam energies of 795 mev and 1300 mev . at 795 mev , which is close to the threshold , the @xmath3 is dominant and the given spectra is consistent to the data . at 1300
mev the model fails to describe the data because the @xmath25 , which is significant at high energies , gives a double hump structure in the @xmath30 invariant mass spectrum .
but if the @xmath3 contribution is negligible which is caused by some kind of destructively interference in this energy region , then the double-@xmath2 diagram , which is dominant by @xmath22-meson exchange in our model , describe the data well@xcite .
the situation is similar in the @xmath31 channel .
our model provide essential hints to the abc effect in @xmath32 reaction@xcite . in fig [ pppipi00 ]
we show the total cross section of @xmath33 channel .
it can be seen that the @xmath34 and @xmath25 are significant in all the considered energies . at the close - to - threshold region ,
the nucleon poles are very important .
so it is suggested that the @xmath3 resonance and nucleon poles should be carefully considered before we reach the right conclusion@xcite .
the agreement between our model and the data of @xmath35 collisions is good@xcite .
it should be addressed that the @xmath3 and other resonances should be included in the monte - carlo simulation of the experimental analysis in order to get the correct total cross sections from limited measured phase space , which is overlooked by the previous measurements .
we find that the @xmath36 reaction is useful to determine the model parameters because like the @xmath23 reaction , the @xmath17 contribution is small .
we also point out that other channels in antinucleon - nucleon collisions could serve as a good place for studying baryon resonances .
a lot of precise measurements in nucleon - nucleon collisions are being carried out by celsius and cosy collaborations in order to study @xmath17 and @xmath18 resonances . besides ,
the hades collaboration has measured some channels of @xmath29 collisions at the beam energies of 1.25 gev@xcite and their results are expected to come soon , so this will constitute a good test of our model . recently , a cooler storage ring hirfl - csr , which can produce the proton beam with similar beam energy range of cosy , has already been successfully installed at lanzhou . with the scheduled 4@xmath22 hadronic detector for complete measurement of diverse differential cross sections
, it will have a special advantage for studying excited nucleon states through nucleon - nucleon collisions .
we also suggest that the double pion production in antinucleon - nucleon collisions should be measured at panda ( anti - proton annihilation at darmstadt ) at the gsi facility of antiproton and ion research ( fair ) , which could provide the antiproton beam of kinetic energy ranging from 1 to 15 gev with the luminosity of about @xmath37 .
we address that panda could play an important role in the hadronic physics with a 4@xmath22 solid angle detector with good particle identification for charged particles and photons .
the planned new experiments are anticipated to offer more tools and information to help us understanding the low energy physics better .
useful discussions with c. wilkin , members of celsius and hades collaborations are gratefully acknowledged .
this work was supported by the national natural science foundation of china ( grant nos .
10635080 , 10875133 , 10821063 , and 10925526 ) .
0 j. johanson et al . , _ nucl .
* a712 * , 75 ( 2002 ) ; w. brodowski et al .
lett . _ * 88 * , 192301 ( 2002 ) ; j. ptzold et al . ,
c _ * 67 * , 052202(r ) ( 2003 ) ; e. doroshkevich1 et al .
j. a _ * 18 * , 171 ( 2003 ) ; t. skorodko et al . , _ eur .
j. a _ * 35 * , 317 ( 2008 ) ; t. skorodko et al .
lett . _ * b679 * , 30 ( 2009 ) ; h. clement et al . , _ int .
* 20 * , 1747 ( 2005 ) ; t. skorodko , _ arxiv:1007.0405 _ , ( 2010 ) .
s. abd el - bary et al .
j. a _ * 37 * , 267 ( 2008 ) ; s. abd el - samad et al .
j. a _ * 42 * , 159 ( 2009 ) .
t. tsuboyama , f. sai , n. katayama , t. kishida and s. s. yamamoto , phys .
c * 62 * , 034001 ( 2000 ) . v. v. sarantsev et al . , _ phys .
nucl . _ * 70 * , 1885 ( 2007 ) .
l. alvarez - ruso , e. oset and e. hernndez , _ nucl .
phys . _ * a633 * , 519 ( 1998 ) . jetset collab .
( a. buzzo _
et al_. ) , _ z. phys .
c _ * 76 * , 475 ( 1997 ) .
e. ferrari , _ nuovo cimento _ * 30 * , 240 ( 1963 ) ; e. ferrari , s. gennarini and p. lariccia , _
nuovo cimento _ * 39 * , 169 ( 1965 ) ; g. wolf , _ phys .
rev . _ * 182 * , 1538 ( 1966 ) .
x. cao , b .- s .
zou , and h .- s .
xu , _ phys .
c _ * 81 * , 065201 ( 2010 ) . b .- s .
zou and f. hussain , _ phys .
c _ * 67 * , 015204 ( 2003 ) .
z. ouyang , j .- j .
xie , b .- s .
zou , and h .- s .
xu , _ nucl .
* a821 * , 220 ( 2009 ) ; z. ouyang , j .- j .
xie , b .- s .
zou , and h .- s .
phys . _ * e18 * , 281 ( 2009 ) . the review of particle physics , c. amsler et al .
lett . _ * b667 * , 1 ( 2008 ) . b .- s .
zou , and d. v. bugg , _ eur .
j. a _ * 16 * , 537 ( 2003 ) .
a. baldini , v. flamino , w.g .
moorhead , and d. r. o. morrison ( eds . ) , _ landolt - brnstein , numerical data and functional relationships in science and technology _ , vol . 12 , ( springer - verlag , berlin , 1988 ) .
m. bashkanov et al . , _ phys .
_ * b637 * , 223 ( 2006 ) ; m. bashkanov et al .
lett . _ * 102 * , 052301 ( 2009 ) ; s. dymov et al .
lett . _ * 102 * , 192301 ( 2009 ) .
l. alvarez - ruso , _ phys .
_ * b452 * , 207 ( 1999 ) .
x. cao , b .- s .
zou , and h .- s .
xu , _ in preparation_. hades collab . ,
_ this proceeding and private conmmunication_. | understanding low energy strong interaction physics unsatisfactory quantum chromodynamics ( qcd ) standard theory interaction color confinement non - perturbative properties difficulties calculate properties mesons baryons from qcd constituent quark models developed explain mesonic baryonic spectrum successful predict more excited states around 2 gev than observed experimentally efforts devoted to meson production in pion- photo- electro - induced reactions reduce uncertainties deepen knowledge structure search for missing resonances double pion production in nucleon - nucleon collisions fascinating field measured at celsius@xcite cosy@xcite kek@xcite pnpi - gatchina@xcite table data data measured after 2000 celsius@xcite cosy@xcite collaboration measured differential cross sections of @xmath8 @xmath9 channels for beam energies from threshold to 1.3 gev @xmath3 contribution dominates at close - to - threshold region explore properties @xmath3 theoretical interpretation controversial valencia model compatible overestimated new data at close - to - threshold energies several timesfinding celsius data @xmath10 invariant mass above 1.0 gev single peak at low invariant mass model double hump structure inconsistent model predicted parallel emission two pion - meson contrary to celsius data level - off behavior around 1.0 gev in energy dependence @xmath9 interference contributions collaboration measured cross section @xmath11 @xmath12 channels important for understanding excitation @xmath3 role @xmath3 in nucleon - nucleon collisions established contribution in @xmath7 reactions never considered jetset@xcite measured @xmath13 channel resonances decaying to @xmath14 included double-@xmath2 diagram in monte - carlo simulation never explored extract properties other resonances from antinucleon - nucleon collisions one - pion exchange models 45 ago considered double-@xmath2 diagram valencia model 10 years ago @xmath3 resonance @xmath15- @xmath16-meson exchange new data necessary comprehensive analysis more resonances data anti - nucleon collisions full model demonstrated in ref . effective lagrangians for resonances based on lorentz covariant orbital - spin scheme @xcitesystem invariant mass 2.8 gev for tp = 2.2 gev checked contributions @xmath17 @xmath18 resonances below 1.72 gev coupling constants in resonances determined by partial decay width from particle data group book@xcite results show @xmath3 resonances significant role in energies other negligible contributions @xcite adjustable parameters are cut - off values in form factors at vertices resonances five types form factors meson- blatt - weisskopf barrier factors @xmath3-@xmath2-@xmath22 vertex @xmath15-@xmath22-@xmath22 vertex of resonances nucleon poles @xmath23 channel useful down cut - off values @xmath2 resonances negligible @xmath17 contribution determine cut - off values resonances other channels if double-@xmath2 mechanism dominant at high energies cross section of @xmath24 should be four larger than @xmath23 new measurements indicate equal value of two channels consistent to old bubble - chamber data@xcitecalculating contributions model conclusions reduces branching ratio @xmath25 assumes smaller cutoff parameter @xmath26 coupling contribution @xmath27 increases introduces contributions from @xmath28 near threshold @xmath20 @xmath21 above 1.5 gev obvious in @xmath23 channel isospin coefficients bigger description cross sections isospin channels improved results agree with new data @xmath29 collisions close - to - threshold region difficult interpret @xmath10 invariant mass spectra energies above 1.0 gev level - off behavior around 1.0 gev @xmath9 channel fig . invariant mass spectra angular distribution opening angle for @xmath30 system center - of - mass system @xmath9 channel at beam energies 795 1300 mev 795 mev @xmath3 dominant spectra consistent data at 1300 mev model fails describe data @xmath25 significant high energies double hump structure in @xmath30 invariant mass spectrum if @xmath3 contribution negligible double-@xmath2 diagram dominant by @xmath22-meson exchange data well similar in @xmath31 channel hints to abc effect in @xmath32 reactionfig [ pppipi00 ] total cross section @xmath33 channel @xmath34 @xmath25 significant energies close - to - threshold region nucleon poles important @xmath3 resonance nucleon poles before conclusion agreement between model data @xmath35 collisions good @xmath3 other resonances included in monte - carlo simulation correct total cross sections limited phase space overlooked @xmath36 reaction useful model parameters @xmath17 contribution small other channels in antinucleon - nucleon collisions studying baryon resonances precise measurements nucleon - nucleon collisions by celsius cosy collaborations study @xmath17 @xmath18 resonances hades collaboration measured channels @xmath29 collisions beam energies 1.25 gev results expected soon good test model cooler storage ring hirfl - csr proton beam similar beam energy range cosy installed at lanzhou scheduled 4@xmath22 hadronic detector measurement diverse differential cross sections special advantage studying excited nucleon states nucleon - nucleon collisionssuggest double pion production antinucleon - nucleon collisions measured panda anti - proton annihilation darmstadt gsi facility antiproton ion research provide antiproton beam kinetic energy 1 to 15 gev luminosity @xmath37 panda role hadronic physics 4@xmath22 solid angle detector particle identification charged particles photons planned new experiments offer tools information low energy physics discussions with c. wilkin celsius hades collaborations acknowledged work supported national natural science foundation china grant nos 10635080 10875133 10821063 10925526 ) j johanson al nucl a712 *, 75 ( 2002 ) brodowski 88 * 192301 ( 2002 ) ptzold 67 * 052202(r ) 2003 ) doroshkevich1 18 * 171 ( 2003 ). skorodko 35 *, 317 ( 2008 ) b679 * 30 ( 2009 ) clement 20 * 1747 ( 2005 ) skorodko arxiv:1007.0405 2010 ) abd el - bary 37 *, 267 ( 2008 ) abd el - samad 42 * 159 ( 2009 ) tsuboyama saikatayama kishida yamamoto phys . 62 034001 ( 2000 ) sarantsev phys . 70 1885 ( 2007 ) alvarez - ruso. oset hernndez nucl a633 519 ( 1998 ) jetset collab buzzo . 76 475 ( 1997 ) ferrari nuovo cimento 30 240 ( 1963 ) ferrari gennarini p. lariccia nuovo cimento 39 169 ( 1965 ) wolf phys . rev 182 1538 ( 1966 ) x. cao b .- s zou h . s xu . 81 065201 ( 2010 ) b .- s zou f. hussain . 67 015204 ( 2003 ) ouyang j xie b . s zou h s xu nucl a821 220 ( 2009 ) ouyang xie s zou h s phys . e18 281 ( 2009 ) review particle physics amsler b667 1 ( 2008 ) b .- s zou d v. bugg 16 537 ( 2003 ). baldini flamino moorhead morrison eds landolt - brnstein numerical data functional relationships science technology vol . 12 springer - verlag berlin 1988 ). bashkanov al . b637 223 ( 2006 ) bashkanov 102 052301 ( 2009 ) s. dymov lett 102 192301 ( 2009 ) alvarez - ruso . b452 * 207 ( 1999 ) x. cao s zou s xu preparation hades collab. proceeding private conmmunication | 0.505451 | 3,330 | 1,742 | , Saving $0.1 in GPT-4. |
177,641 | if dark energy can be described as a perfect fluid , then , apart from its equation of state relating energy density and pressure , we should also especify the corresponding rest frame .
since dark energy is typically decoupled from the rest of components of the universe , in principle such a frame could be different from that of matter and radiation . in this work we consider the potential observable effects of the motion of dark energy and the possibility to measure the dark energy velocity relative to matter .
in particular we consider the modification of the usual interpretation of the cmb dipole and its implications for the determination of matter bulk flows on very large scales .
we also comment on the possible origin of a dark energy flow and its evolution in different models .
address = dept .
fsica terica , universidad complutense de madrid , 28040 madrid , spain | the content of the universe can be appropriately described as a four - fluid system : baryons , radiation , dark matter and dark energy .
such fluids can be considered in good approximation to be decoupled from each other after recombination time .
each fluid carries its own rest frame .
thus for instance , the cmb radiation , being highly homogeneous allows us to define its rest frame by means of the cmb dipole anisotropy .
indeed , in the usual interpretation @xcite , the cmb dipole is due to the doppler effect caused by the motion of the observer with respect to the last scattering surface
. therefore the radiation rest frame would be that in which an observer measures a vanishing dipole .
this fact in turn has been used to calculate the velocity of our local group with respect to the cmb radiation , just by substracting the velocity of the sun with respect to the milky way and the milky way velocity with respect to the local group .
the result is @xmath0 km s@xmath1 in the direction @xmath2 in galactic coordinates .
the rest frame of ( dark ) matter is not so easily defined since we know that matter distribution is only homogeneous on very large scales .
the presence of density inhomogeneities makes the velocity of a given matter volume to deviate from the pure hubble flow .
such deviations @xmath3 are known as peculiar velocities and they are related to the density perturbations @xmath4 by this simple expression to first order in perturbation theory : @xmath5 according to the cosmological principle , the universe is homogeneous on very large scales and the amplitude of density perturbations on scales of size @xmath6 will decline as we take larger and larger values of @xmath6 .
this in turn means that the peculiar velocity of a matter volume of radius @xmath6 will also decline for larger @xmath6 ( see fig .
1 ) , the matter rest frame eventually converging to the radiation frame on very large scales .
in other words , matter and radiation would share a common rest frame according to the cosmological principle .
however as shown in fig .
1 , this theoretical framework is not conclusively confirmed by observations . indeed , in recent years several peculiar velocity surveys @xcite have tried to determine the volume size at which the streaming motion of matter with respect to the cmb vanishes . in the figure the results of different observations are compared with the _ rms _ expected bulk velocity @xmath7 for standard @xmath8cdm model in a sphere of radius @xmath6 .
the results seem to agree with the theoretical expectations only at scales @xmath9 mpc . at larger scales , @xmath10 mpc ,
different data sets lead to different bulk velocities both in amplitude and direction . moreover , there are indeed measurements in which large matter volumes are moving at speeds @xmath11 km s@xmath1 with respect to the cmb frame , several standard deviations away from the theoretical predictions .
these results have been argued to be affected by systematic errors in distance indicators , but if confirmed by future surveys , a revision of some of the underlying ideas in standard cosmology would be required in order to understand the origin of such large flows . for different peculiar velocity surveys @xcite centered around the local group in the cmb frame .
the dot labelled by cobe denotes the velocity of the local group measured by cobe .
the solid line corresponds to the expected _ rms _ velocity in the standard @xmath8cdm cosmology , together with the 90% deviation in dashed lines .
figure from s. zaroubi @xcite ] concerning the dark energy rest frame , it is usually assummed that dark energy does not cluster on small scales and therefore can be considered as a highly homogeneous fluid which is almost decoupled from the rest of components , its only interaction being of gravitational nature .
in such a case there is no reason to expect that its rest frame should necessarily agree with the matter / radiation frame and therefore an observational determination of the dark energy bulk velocity would be needed . in the following
we will explore this problem , considering cosmological models with four decoupled homogeneous fluids and different rest frames @xcite .
let us therefore consider a cosmological scenario with four perfect fluids : baryons , radiation , dark matter and dark energy , whose equations of state read @xmath12 with @xmath13 . for the sake of generality , we will allow the dark energy equation of state to have a smooth dependence on redshift @xmath14 .
the energy - momentum tensor of each fluid will take the form : @xmath15 since we are only interested in the effects of fluids motion on the cmb dipole , it is sufficient to take into account the evolution of the largest - scale velocity perturbations , i.e. we will just consider the zero - mode equations .
the presence of inhomogeneities will contribute to higher multipoles .
therefore , for this particular problem we can write : @xmath16 we will assume that @xmath17 and we will work at first order in perturbation theory . in the particular case we are considering
, the form of the space - time metric will be given by the following vector - perturbed spatially - flat friedmann - robertson - walker metric : @xmath18 accordingly , the total energy - momentum tensor reads : @xmath19 notice that we are considering only the epoch after matter - radiation decoupling , assuming that dark energy is also decoupled and for that reason we will ignore possible energy and momentum transfer effects .
we now calculate the linearized einstein equations using ( [ metric ] ) and ( [ t ] ) .
they yield just the condition : @xmath20 in general relativity the combination @xmath21 appearing in ( [ s ] ) plays the role of inertial mass density of the corresponding fluid , and accordingly @xmath22 can be understood as the _ cosmic center of mass velocity_. notice that a pure cosmological constant has no inertial mass density . on the other hand ,
the energy conservation equations are trivially satisfied , whereas from momentum conservation we see that the velocity of each fluid relative to the center of mass frame scales as : @xmath23 notice that for dark energy the scaling properties will depend on the particular model under consideration @xcite .
once we know the form of the perturbed metric , we can calculate the effect of fluids motion on photons propagating from the last scattering surface using standard tools @xcite .
the energy of a photon coming from direction @xmath24 with @xmath25 as seen by an observer moving with velocity @xmath26 is given by @xmath27 , i.e. to first order in the perturbation : @xmath28 where @xmath29 parametrizes the photon energy and the perturbed trajectory of the photon reads @xmath30 , with @xmath31 . in order to obtain @xmath32
, we solve the geodesics equations to first order in the perturbations , and for the 0-component we get @xmath33 . by defining @xmath34 ,
the temperature fluctuation reads : @xmath35 where the indices @xmath36 , @xmath37 denote the present and decoupling times respectively .
today the only relevant contributions to the center of mass motion are those of matter and dark energy , radiation being negligible , so that : @xmath38 where @xmath39 is the present equation of state of dark energy and we have assumed that today the relative velocity of baryons and dark matter is negligible .
for that reason we have used @xmath40 to denote them both simultaneously .
notice that for a pure cosmological constant @xmath41 and there would be no contribution from dark energy in such a case . at decoupling ,
the universe is matter dominated and we can neglect the contribution to @xmath22 from dark energy .
therefore : @xmath42 where the emitter velocity is nothing but the baryonic velocity @xmath43 .
as we will see below , when dark energy is absent , it is in the form of a pure cosmological constant or it is at rest with respect to radiation , we will have @xmath44 and this term vanishes .
according to this result , the cmb dipole has two different types of contributions : one is the usual doppler effect due to the change of velocity between emitter and observer ; the second contribution comes from the fact that the photon is propagating in an anisotropic medium which is changing in time .
this second contribution is precisely given by the change in the velocity of the cosmic center of mass between emission and reception .
the possibility that the dipole is not enterely due to a doppler effect has been considered previously in the litereature in different contexts ( see @xcite ) .
when all the components share a common rest frame then the previous result reduces to the usual expression for the dipole : @xmath45 .
however in general it is possible that an observer at rest with radiation @xmath46 can measure an nonvanishing dipole according to ( [ dipolo ] ) . in the absence of dark energy or in the case
in which it is in the form of a pure cosmological constant ( @xmath47 ) , dark energy would not contribute to the center of mass motion .
moreover , today the radiation contribution is negligible and accordingly the center of mass rest frame would coincide with the matter rest frame .
this implies that the relative motion of matter and radiation today could not explain the existence of bulk flows on the largest scales , since the frame in which the dipole vanishes would coincide with the matter rest frame .
conversely , the existence of non - vanishing bulk flows would require the presence of moving dark energy with @xmath48 .
indeed , if moving dark energy is responsible for the existence of cosmic bulk flows on very large scales , then the amplitude and direction of such flows would provide a way to measure the relative velocity of matter and dark energy .
as commented above , the bulk flow @xmath49 can be understood as the average velocity of a given matter volume with respect to an observer who measures a vanishing cmb dipole , i.e. @xmath50 . such an observer has a velocity which can be obtained from ( [ dipolo ] ) , and accordingly : @xmath51 notice that , according to these results , even if matter is at rest with respect to the cmb radiation , @xmath52 , it would be possible to have a non - vanishing flow @xmath53 , provided dark energy is moving with respect to matter .
so far we have considered only the effects of the different components of the universe having different rest frames .
however we have not discussed what is the origin of such a velocity offset . in standard cosmology , baryonic matter and radiation were coupled before recombination , and this was also so for dark matter , in the case in which it is in the form of weakly interacting massive particles .
however the nature of dark energy is still a mistery and we ignore what is the type of interaction , if any , between dark energy and radiation / matter .
for that reason , the primordial value of the dark energy rest frame velocity with respect to radiation / matter should be considered as a completely free cosmological parameter , which could be determined only by observations .
although the primordial value is unknown , its subsequent evolution is given by ( [ velocity ] ) . in fig .
2 we show the evolution of the velocities of the different fluids with respect to the center of mass frame . we compare two models for dark energy , one in which the equation of state is a constant close to @xmath54 and a scaling @xcite dark energy model in which the equation of state mimics that of the dominant component . we see that the velocity of radiation is almost constant throughout the cosmological evolution , whereas although matter is initially dragged by radiation , its velocity starts decaying as @xmath55 after decoupling .
dark energy velocity is damped very fast for constant equation of state , whereas it could have appreciable amplitude today in scaling models .
apart from the effects on the dipole , the metric anisotropies generated by the motion of dark energy can give rise in some cases to non - negligible quadrupole contributions and generate a net polarization of the cmb radiation .
such effects offer additional possibilities for the determination of the dark energy rest frame .
these results will be presented elsewhere @xcite
. 0 p.j.e .
peebles and d.t .
wilkinson , _ phys .
rev . _ * 174 * , 2168 ( 1968 ) ; p.j.e .
peebles , _ principles of physical cosmology _ , princeton university press ( 1993 ) ; a. kogut et al . , _ astrophys.j . _ * 419 * , 1 ( 1993 ) a.g .
riess et al . , _ astrophys .
j. _ * 488 * , l1 ( 1997 ) ; d.a .
dale et al .
_ astrophys.j .
_ * 510 * , l11 ( 1999 ) ; m. colles et al .
, _ mnras _ , * 321 * , 277 ( 2001 ) ; t.r . lauer and m.
postman , _
_ * 425 * , 418 ( 1994 ) ; m.j .
hudson et al .
_ astrophys.j .
_ * 512 * , l79 ( 1999 ) ; j.a .
willick , _ astrophys.j . _ * 522 * , 647 ( 1999 ) ; m.j .
hudson et al . , _ mon . not .
* 352 * , 61 ( 2004 ) ; m.j .
hudson , astro - ph/0311072 s. zaroubi , astro - ph/0206052 .
contribution to xiii recontres de blois `` frontiers of the universe '' , 17 - 23 june 2001 a.l .
maroto , _ jcap _ * 0605*:015 , ( 2006 ) ; a.l .
maroto , astro - ph/0605381 m. giovannini , _ int .
phys . _ * d14 * , 363 , ( 2005 ) r.a .
matzner , _ astrophys .
j. _ * 241 * , 851 ( 1980 ) ; b. paczynski and t. piran , _ astrophys .
j. _ * 364 * , 341 ( 1990 ) ; m.s .
turner , _ phys .
rev . _ * d44 * , 3737 ( 1991 ) ; j.f .
pascual - sanchez .
grav . _ * 17*:4913 - 4918 , ( 2000 ) | universe described as a four - fluid system : baryons radiation dark matter dark energy fluids decoupled after recombination time each fluid carries own rest frame cmb radiation homogeneous rest frame by cmb dipole anisotropy cmb dipole due to doppler effect by motion observer last scattering surface radiation rest frame observer measures vanishing dipole used to calculate velocity of local group cmb radiation by substracting velocity sun milky way and milky way velocity local group result is @xmath0 km s@xmath1 in direction @xmath2 in galactic coordinates rest frame of dark matter not easily defined matter distribution only homogeneous on very large scales density inhomogeneities velocity of matter volume deviate from pure hubble flow deviations as peculiar velocities related to density perturbations @xmath4 universe homogeneous on large scales amplitude of density perturbations decline as larger values peculiar velocity of matter volume of radius @xmath6 decline for larger @xmath6 matter rest frame converging to radiation frame on large scales matter and radiation share common rest frame cosmological principlein fig . 1 theoretical framework not confirmed by observations recent velocity surveys @xcite tried determine volume size streaming motion matter cmb vanishes figure results compared with expected bulk velocity @xmath7 for standard @xmath8cdm model radius @xmath6 results agree with expectations at scales @xmath9 mpc at larger scales @xmath10 mpc data sets lead to different bulk velocities measurements large matter volumes moving at speeds @xmath11 km s@xmath1 cmb frame standard deviations from theoretical predictions results argued affected by errors in distance indicators if confirmed revision of ideas standard cosmology required origin large flows velocity surveys centered around local group in cmb frame dot by cobe denotes velocity local group measured solid line corresponds to expected rms velocity in standard @xmath8cdm cosmology 90% deviation in dashed lines figure from dark energy rest frame on small scales homogeneous fluid decoupled from components interaction gravitational no reason to expect rest frame agree with matter / radiation frame observational determination of dark energy bulk velocity neededexplore problem cosmological models with four decoupled fluids rest frames scenario four perfect fluids baryons radiation dark matter dark energy equations state read @xmath12 @xmath13 dark energy equation dependence on redshift @xmath14 energy - momentum tensor fluid @xmath15 effects fluids motion on cmb dipole evolution largest - scale velocity perturbations zero - mode equations inhomogeneities contribute to higher multipoles write @xmath16 assume @xmath17 work at first order in perturbation theory form space - time metric by vector - perturbed spatially - flat friedmann - robertson - walker metric : @xmath18 total energy - momentum tensor reads @xmath19 considering epoch after matter - radiation decoupling dark energy decoupled ignore energy momentum transfer effects calculate linearized einstein equations using ( [ metric ] ) ( [ t ] ) yield condition : @xmath20 combination @xmath21 s ] inertial mass density fluid @xmath22 cosmic center of mass velocity pure cosmological constant no inertial mass densityenergy conservation equations satisfied momentum conservation velocity fluid to center mass scales @xmath23 dark energy scaling properties depend on model @xcite form perturbed metric calculate effect fluids motion on photons from scattering surface using standard tools energy of photon from direction @xmath24 @xmath25 given by @xmath27 first order perturbation @xmath28 @xmath29 parametrizes photon energy perturbed trajectory @xmath30 @xmath31 obtain @xmath32 solve geodesics equations first order 0-component @xmath33 defining @xmath34 temperature fluctuation reads @xmath35 indices @xmath36 @xmath37 denote present decoupling times relevant contributions to center mass motion matter dark energy radiation negligible @xmath38 @xmath39 present equation of state dark energy relative velocity of baryons dark matter negligible used @xmath40 denote both simultaneously pure cosmological constant @xmath41 no contribution from dark energy decoupling universe matter dominated neglect contribution @xmath22 from dark energy @xmath42 emitter velocity is baryonic velocity @xmath43dark energy absent pure cosmological constant or at rest radiation @xmath44 term vanishes cmb dipole has two contributions one usual doppler effect change velocity between emitter observer second contribution photon propagating in anisotropic medium changing in time second contribution given by change velocity cosmic center of mass between emission and reception possibility dipole not due to doppler effect considered previously @xcite components share common rest frame result reduces to usual expression dipole @xmath45 possible observer at rest with radiation @xmath46 can measure nonvanishing dipole dipolo absence of dark energy or pure cosmological constant @xmath47 ) dark energy contribute to center of mass motion radiation contribution negligible center of mass rest frame coincide with matter rest frame relative motion of matter and radiation explain bulk flows on largest scales dipole vanishes with matter rest frame non - vanishing bulk flows moving dark energy with @xmath48 if moving dark energy responsible for cosmic bulk flows large scales amplitude direction of flows measure relative velocity of matter and dark energybulk flow @xmath49 average velocity of matter volume observer vanishing cmb dipole @xmath50 observer has velocity obtained from ( [ dipolo ] ) @xmath51 if matter at rest cmb radiation @xmath52 possible non - vanishing flow @xmath53 dark energy moving matter considered effects of different components universe different rest frames not discussed origin of velocity offset baryonic matter and radiation coupled before recombination for dark matter weakly interacting massive particles nature of dark energy mistery ignore type interaction between dark energy radiation / matter primordial value of dark energy rest frame velocity radiation matter free cosmological parameter determined by observations primordial value unknown subsequent evolution given by ( [ velocity ] ) fig . 2 evolution of velocities of different fluids center of mass frame two models for dark energy equation of state constant to @xmath54 and scaling @xcite dark energy model mimics dominant component velocity of radiation almost constant throughout evolution matter initially dragged by radiation velocity starts decaying as @xmath55 after decouplingdark energy velocity damped for constant equation state amplitude in scaling models effects dipole metric anisotropies dark energy quadrupole contributions net polarization cmb radiation effects offer possibilities determination dark energy rest frame results presented elsewhere peebles wilkinson phys . rev . 174 *, 2168 ( 1968 ) principles physical cosmology princeton university press 1993 ) kogut astrophys.j . 419 *, 1 ( 1993 ) riess astrophys . 488 *, l1 ( 1997 ) dale astrophys.j 510 *, l11 ( 1999 ) colles 321 * , 277 ( 2001 ) lauer postman * 425 * , 418 ( 1994 ) hudson astrophys.j 512 * l79 ( 1999 ) willick astrophys.j 522 * , 647 ( 1999 ) hudson 352 *, 61 ( 2004 ) hudson astro - ph/0311072 zaroubi astro - ph/0206052 contribution xiii recontres de blois frontiers universe 17 - 23 june 2001maroto jcap 0605*:015 2006 a maroto astro - ph/0605381 giovannini int .. d14 363 2005 matzner astrophys . 241 851 1980 ) b. paczynski piran astrophys. 364 341 1990 ) m.s turner . rev. d44 3737 1991 ) j pascual - sanchez grav 17*:4913 - 4918 2000 | 0.506007 | 3,160 | 1,614 | , Saving $0.1 in GPT-4. |
177,648 | in this paper , the dynamical attractor and heteroclinic orbit have been employed to make the late - time behaviors of the model insensitive to the initial condition and thus alleviate the fine - tuning problem in the torsion cosmology .
the late - time de sitter attractor indicates that torsion cosmology is an elegant scheme and the scalar torsion mode is an interesting geometric quantity for physics .
the numerical solutions obtained by nester et al .
are not periodic solutions , but are quasi - periodic solutions near the focus for the coupled nonlinear equations . | the current observations , such as sneia ( supernovae type ia ) , cmb ( cosmic microwave background ) and large scale structure , converge on the fact that a spatially homogeneous and gravitationally repulsive energy component , referred as dark energy , accounts for about @xmath0 % of the energy density of universe .
some heuristic models that roughly describe the observable consequences of dark energy were proposed in recent years , a number of them stemming from a certain physics @xcite and the others being purely phenomenological @xcite .
about thirty years ago , the bouncing cosmological model with torsion was suggested in ref.@xcite , but the torsion was imagined as playing role only at high densities in the early universe .
goenner et al . made a general survey of the torsion cosmology @xcite , in which the equations for all the pgt ( poincar gauge theory of gravity ) cases were discussed although they only solved in detail a few particular cases .
recently some authors have begun to study torsion as a possible reason of the accelerating universe @xcite .
nester and collaborators @xcite consider an accounting for the accelerated universe in term of pgt : dynamic scalar torsion . with the usual assumptions of homogeneity and isotropy in cosmology
, they find that torsion field could play a role of dark energy .
this elegant model has only a few adjustable parameters , so scalar torsion may be easily falsified as `` dark energy '' .
the fine - tuning problem should be one of the most important issues for the cosmological models , and a good model should limit the fine - tuning as much as possible .
the dynamical attractor of the cosmological system has been employed to make the late - time behaviors of the model insensitive to the initial condition of the field and thus alleviates the fine - tuning problem @xcite . in this paper
, we study attractor and heteroclinic orbit in the torsion cosmology .
we show that the late - time de sitter behaviors cover a wide range of the parameters .
this attractor indicates that torsion cosmology is an elegant scheme and the scalar torsion mode is an interesting geometric quantity for physics .
furthermore , there are only exact periodic solutions for the linearized system , which just correspond to the critical line ( line of centers ) .
the numerical solutions in ref.@xcite are not periodic , but are quasi - periodic solutions near the focus for the coupled nonlinear equations .
pgt @xcite based on a riemann - cartan geometry , allows for dynamic torsion in addition to curvature .
the affine connection of the riemann - cartan geometry is @xmath1 where @xmath2 is the levi - civita connection and @xmath3 is the torsion tensor .
meanwhile , the ricci curvature and scalar curvature can be written as @xmath4 where @xmath5 and @xmath6 are the riemannian ricci curvature and scalar curvature , respectively , and @xmath7 is the covariant derivative with the levi - civita connection and @xmath8 . according as ref.@xcite we take the restricted form of torsion in this paper @xmath9\rho } \label{restrictedt}\end{aligned}\ ] ] therefore , the gravitational lagrangian density for the scalar mode is ( for a detailed discussion see ref.@xcite ) @xmath10 since current observations favor a flat universe , we will work in the spatially flat robertson - walker metric .
according to the homogeneity and isotropy , the torsion @xmath11 should be only time dependent , so one can let @xmath12 and the spatial parts vanish since we have taken the restricted form ( [ restrictedt ] ) of torsion . for the general form ,
the torsion tensor have two independent components @xcite-@xcite . from the field equations
one can finally give the necessary equations for the matter - dominated era to integrate ( for a detailed discussion see ref.@xcite ) @xmath13 where @xmath14 and the energy density of matter component @xmath15 one can scale the variables and the parameters as @xmath16 where @xmath17 is the present value of hubble parameter and @xmath18 is the planck length .
under the transform ( [ scale ] ) , eqs .
( [ dth])-([dtr ] ) remain unchanged .
after transform , new variables @xmath19 , @xmath20 , @xmath21 and @xmath22 , and new parameters @xmath23 , @xmath24 , @xmath25 and @xmath26 are all dimensionless . furthermore , the newtonian limit requires @xmath27 .
obviously , eqs .
( [ dth])-([dtr ] ) is an autonomous system , so we can use the qualitative method of ordinary differential equations .
it is worth noting that in the analysis of critical points , copeland et al .
@xcite introduced the elegant compact variables which are defined from the friedmann equation constraint , but in our case , the friedmann equation can not be written as the ordinary form , so the compact variables are not convenient here .
therefore , we will analyze the system of eqs.([dth])-([dtr ] ) using the variables @xmath20 , @xmath21 and @xmath22 under the transform ( [ scale ] ) .
in the case of scalar torsion mode , the effective energy - momentum tensor can be represented as @xmath28\,,\label{torpre}\end{aligned}\ ] ] and the off - diagonal terms vanish .
the effective energy density @xmath29 which is deduced from general relativity .
@xmath30 is an effective pressure , and the effective equation of state is @xmath31 which is induced by the dynamic torsion . according to equations ( [ dth])-([dtr ] ) ,
we can obtain the critical points and study the stability of these points .
there are five critical points @xmath32 of the system as follows @xmath33 where @xmath34 , @xmath35 and @xmath36 .
consider that the parameter @xmath26 is associated with the quadratic scalar curvature term @xmath37 , so that @xmath26 should be positive @xcite .
evidently , the critical points @xmath32 are not real values except ( 0 , 0 , 0 ) in the cases of parameters @xmath38 and @xmath39 or @xmath40 .
if we consider the linearized equations , then eqs .
( [ dth])-([dtr ] ) are reduced to @xmath41 in the case @xmath39 , there is only a critical point @xmath42 for the nonlinear system .
( [ dthphirlinear ] ) has an exact periodic solution @xmath43 where @xmath44 , @xmath45 , @xmath46 and @xmath47 , @xmath48 and @xmath49 are initial values .
obviously , ( @xmath20 , @xmath50 , @xmath50 ) is a critical line for the linearized system .
however , there is only a critical point @xmath42 for the nonlinear system which is an asymptotically stable focus . in other words
, there is no periodic solution for the nonlinear system since the corresponding eigenvalue is ( @xmath50 , @xmath51 , @xmath52 ) . in fig.[focusa1g0 ] , we plot the orbits near the point @xmath42 for the nonlinear systems . for nonlinear system ( [ dth])-([dtr ] ) with @xmath53 .
we take @xmath54 and the initial values @xmath55 .
@xmath56 is an asymptotically stable focus point.,width=172,height=172 ] next , the parameters are restricted within @xmath38 and @xmath57 or @xmath58 in this paper . to study the stability of the critical points @xmath59
, we write the variables near the critical points in the form @xmath60 , @xmath61 , and @xmath62 with @xmath63 the perturbations of the variables near the critical points . substituting the expression into the system of equations ( [ dth])-([dtr ] )
, one can obtain the corresponding eigenvalues of critical points ( i)-(v ) @xmath64 the properties of the critical points are shown in tables [ cripointsa1l-1 ] and [ cripointsa1l0 ] .
we find that critical point ( ii ) is a late time de sitter attractor in the case of @xmath65 .
in previous sections , we have studied the phase space of a torsion cosmology .
the de sitter attractor indicates that torsion cosmology @xcite-@xcite is an elegant scheme and the scalar torsion mode is an interesting geometric quantity for physics . in this section
, we study their dynamical evolution numerically .
.the physical properties of critical points for @xmath66 [ cols="^,^,^,^",options="header " , ] for different initial values@xmath67 with @xmath68 .
we take @xmath69.,width=201,height=134 ]
in this paper , we investigate the dynamics of a torsion cosmology , in which we consider only the `` scalar torsion '' mode .
this mode has certain distinctive and interesting qualities .
we show that the late - time asymptotic behavior does not always correspond to an oscillating aspect .
in fact , only in the focus case can we declare that `` scalar torsion '' mode can contribute a quasinormal oscillating aspect to the expansion rate of the universe .
there are only exact periodic solutions for the linearized system , which just correspond to the critical line ( line of centers ) . via numerical calculation of the coupled nonlinear equations nester et al .
@xcite plot that quasi - periodic solution near the focus .
the late - time de sitter attractor indicates that torsion cosmology is an elegant scheme and the scalar torsion mode is an interesting geometric quantity for physics .
we show that the late - time de sitter behaviors cover a wide range of the parameters and thus alleviate the fine - tuning problem .
furthermore , the torsion cosmology has considered the possibility that the dynamics scalar torsion ( geometric field ) could fully account for the accelerated universe , which is naturally expected from spacetime gauge theory .
t. padmanabhan , phys . rept . * 380 * , 235 ( 2003 ) ; j. g. hao and x. z. li , phys . rev . * d67 * , 107303 ( 2003 ) ; d. j. liu and x. z. li , phys
* d68 * , 067301 ( 2003 ) ; x. z. li and j. g. hao , phys . rev . * d69 * , 107303 ( 2004 ) .
e. j. copeland , m. sami and s. tsujikawa , int .
d15 * , 1753 ( 2006 ) ; j. g. hao and x. z. li , phys . lett . *
b606 * , 7 ( 2005 ) ; d. j. liu and x. z. li , phys . lett . *
b611 * , 8 ( 2005 ) ; d. j. liu , c. b. sun and x. z. li , phys .
b634 * , 442 ( 2006 ) .
s. capozziello , s. carloni and a. troisi , recent res .
. astrophys . * 1 * , 625 ( 2003 ) ; c. g. bhmer , j. burnett , phys . rev . *
d78 * , 104001 ( 2008 ) ; c. g. bhmer , acta phys . polon . *
b36 * , 2841 ( 2005 ) ; e. w. mielke and e. sanchezromero , phys . rev . *
d73 * , 043521 ( 2006 ) ; a. v. minkevich , a. s. garkin and v. i. kudin , class .
* 24 * , 5835 ( 2007 ) f. w. hehl , j. d. mccrea , e. w. mielke and y. neeman , phys . rept .
* 258 * , 1 ( 1995 ) ; m. blagojevi , _ gravitation and gauge symmetries _
( institute of physics , bristol , 2002 ) ; t. ortn , _ gravity and strings _ , ( cambridge university press , cambridge , 2004 ) . | current observations sneia cmb large scale structure converge on spatially homogeneous gravitationally repulsive energy component dark energy accounts for @xmath0 % of energy density universe heuristic models consequences of dark energy proposed from physics @xcite others phenomenological thirty years ago bouncing cosmological model with torsion suggested in torsion at high densities in early universe goenner et al . made survey of torsion cosmology @xcite equations for pgt cases discussed solved few cases authors torsion as possible reason accelerating universe @xcite nester collaborators consider accounting for accelerated universe in pgt : dynamic scalar torsion torsion field could play role of dark energy model has few adjustable parameters scalar torsion may be falsified as dark energy fine - tuning problem important for cosmological models good model should limit fine - tuning dynamical attractor employed to make late - time behaviors model insensitive to initial condition alleviates fine - tuning problem in paper study attractor heteroclinic orbit in torsion cosmologylate - time de sitter behaviors cover wide range parameters torsion cosmology elegant scheme scalar torsion mode interesting geometric quantity for physics exact periodic solutions for linearized system correspond to critical line numerical solutions in ref.@xcite not quasi - periodic near focus for coupled nonlinear equations @xcite riemann - cartan geometry allows dynamic torsion curvature affine connection is @xmath1 @xmath2 levi - civita connection @xmath3 torsion tensor ricci curvature scalar curvature written as @xmath4 @xmath5 @xmath6 riemannian ricci curvature scalar curvature @xmath7 covariant derivative with levi - civita connection @xmath8 ref.@xcite restricted form of torsion @xmath9 gravitational lagrangian density for scalar mode is @xmath10 observations favor flat universe work in spatially flat robertson - walker metric torsion @xmath11 time dependent @xmath12 spatial parts vanish restricted form torsion torsion tensor two independent components @xcite-@xcite field equations necessary equations for matter - dominated era@xcite @xmath13 @xmath14 energy density matter @xmath15 scale variables parameters @xmath16 @xmath17 present value hubble parameter @xmath18 planck length transform scale eqs . ( dth])-([dtr ] unchanged transform new variables @xmath19 parameters @xmath23 newtonian limit requires @xmath27 eqs . ( [ dth])-([dtr ] autonomous system use qualitative method ordinary differential equations analysis critical points copeland et al @xcite introduced compact variables defined friedmann equation constraint friedmann equation written ordinary form compact variables not convenient analyze system eqs.([dth])-([dtr ] using variables @xmath20 @xmath21 @xmath22 under transform scale scalar torsion mode effective energy - momentum tensor represented @xmath28 off - diagonal terms vanish effective energy density @xmath29 deduced from general relativity @xmath30 effective pressure effective equation state @xmath31 induced dynamic torsion equations ( [ dth])-([dtr ] obtain critical points study stabilityfive critical points @xmath32 @xmath33 parameter @xmath26 associated with quadratic scalar curvature @xmath37 @xmath26 positive @xcite critical points @xmath32 not real values except 0 parameters @xmath38 @xmath39 or @xmath40 linearized equations dth-([dtr reduced to @xmath41 @xmath39 critical point @xmath42 nonlinear system exact periodic solution @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 @xmath49 initial values @xmath20 @xmath50 critical line for linearized system critical point @xmath42 nonlinear system asymptotically stable focus no periodic solution nonlinear system corresponding eigenvalue is ( @xmath50 @xmath51 @xmath52 ) fig.[focusa1g0 orbits near point @xmath42 nonlinear systems nonlinear system dth])-([dtr @xmath53 @xmath54 initial values @xmath55 @xmath56 asymptotically stable focus point parameters restricted within @xmath38 @xmath57 or @xmath58study stability critical points @xmath59 write variables near @xmath60 @xmath61 @xmath62 @xmath63 perturbations substituting expression equations dth])-([dtr ] obtain eigenvalues critical points ( i)-(v ) @xmath64 properties critical points shown tables [ cripointsa1l-1 cripointsa1l0 ] critical point ( ii ) late time de sitter attractor @xmath65 studied phase space torsion cosmology de sitter attractor indicates torsion cosmology @xcite elegant scheme scalar torsion mode interesting geometric quantity study dynamical evolution numerically physical properties critical points @xmath66 different initial values@xmath67 @xmath68 @xmath69=134 investigate dynamics torsion cosmology consider `` scalar torsion '' mode distinctive qualities late - time asymptotic behavior always correspond oscillating aspect focus case scalar torsion '' mode contribute quasinormal oscillating aspect expansion rate universe exact periodic solutions linearized system correspond critical line centers numerical calculation coupled nonlinear equations @xcite plot quasi - periodic solution near focuslate - time de sitter attractor torsion cosmology scheme scalar torsion mode interesting geometric quantity physics late behaviors cover parameters alleviate fine - tuning problem torsion cosmology scalar torsion accelerated universe expected spacetime gauge theory padmanabhan . rept. 380 235 ( 2003 ) j g. hao x. z. li d67 107303 2003 j liu x. z. li d68 067301 ( 2003 x. z. li j g. hao phys d69 107303 ( 2004 ) j copeland m sami tsujikawa d15 1753 ( 2006 ) j g. hao x. z. li phys b606 7 2005 d j liu x. z. li b611 8 2005 j liu sun x. z. li b634 442 ( 2006 ) capozziello s carloni a. troisi astrophys . 1 625 ( 2003 ) c g. bhmer j. burnett phys d78 104001 ( 2008 ) bhmer phys b36 2841 ( 2005 ) mielkesanchezromero phys . rev. d73 043521 2006 ). minkevich garkin kudin class 24 5835 ( 2007 ). hehl d. mccrea mielke neeman . rept . 258 1 1995 ) blagojevi gravitation gauge symmetries institute bristol 2002 t ortn gravity strings cambridge university press cambridge 2004 | 0.436688 | 2,735 | 1,423 | , Saving $0.1 in GPT-4. |
177,651 | nonlinearity introduced in slow - roll inflation will produce weakly non - gaussian cmb temperature fluctuations .
we have simulated non - gaussian large scale cmb maps ( including _ cobe_-dmr constraints ) introducing an additional quadratic term in the gravitational potential.the amount of nonlinearity being controlled by the so called nonlinear coupling parameter @xmath0 .
an analysis based on the spherical mexican hat wavelet was applied to these and to the _
cobe_-dmr maps .
skewness values obtained at several scales were combined into a fisher discriminant .
comparison of the fisher discriminant distributions obtained for different nonlinear coupling parameters with the _ cobe_-dmr values , sets a constraint of @xmath1 at the @xmath2 confidence level .
this new constraint being tighter than the one previously obtained by using the bispectrum by komatsu et al .
( 2002 ) .
= = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] | the _ cobe_-dmr data have been found to be compatible with gaussianity by several methods based on real ( torres et al .
1995 ; kogut et al 1996 ; schmalzing & grski 1997 ; novikov et al .
2000 ) , spherical harmonic ( banday et al . 2000 ; sandvik & magueijo 2000 ; komatsu et al . 2002 ) and wavelet space ( mukherjee , hobson & lasenby 2000 ; aghanim , forni & bouchet 2000 ; barreiro et al . 2000 ; cayn et al . 2001 ) .
cmb observations by boomerang , dasi and maxima - i ( netterfield et al . 2001 ; pryke et al .
2001 ; stompor et al .
2001 ) seem to indicate that any possible non - gaussianity present in the data will more likely be produced by non - standard inflationary models .
it is important at the moment to study the characteristics of the non - gaussianity introduced by the alternative theories to standard inflation .
one of the simplest alternative scenarios that will generate weakly non - gaussian cmb fluctuations is one based on slow - roll inflation
. investigated by salopek & bond ( 1990 ) and gangui et al .
( 1994 ) , this model generates non - gaussianity in matter and radiation fluctuation fields through features appearing in the inflaton potential . in the cmb context , the model has been studied by komatsu & spergel ( 2001 ) , verde et al .
( 2001 ) and komatsu et al .
the gravitational potential includes now a quadratic term whith an amplitude regulated by the so called non - linear coupling parameter ( see section 2 ) .
verde et al .
( 2001 ) conclude in their paper that cmb data will be more sensitive to the non - gaussianity generated by these models than observations of high redshift objects .
a semianalytical expression of the bispectrum generated by this model is presented in komatsu & spergel ( 2001 ) .
the minimum non - linear coupling parameter values that will produce detectable non - gaussianity ( based on the bispectrum and skewness statitics ) in different cmb data sets are estimated .
these values are however larger than the ones expected from slow - roll inflation theories .
komatsu et al . ( 2002 ) revised the constraint impossed by _
cobe_-dmr data on the non - linear coupling parameter based on the normalized bispectrum .
this constraint ( slightly larger than the one estimated by komatsu & spergel 2001 ) , though very weak to say much about slow - roll inflation is in any case interesting in itself .
deviations from slow - roll inflation might include larger values of the non - linear coupling parameter generating non - gaussianities that could be detected by cmb experiments .
we want to asses in this paper which is the minimum value of the non - linear coupling parameter that would be constraint by the _ cobe_-dmr data based on a wavelet space statistic .
comparison of this constraint with the one obtained using the normalized bispectrum is an important comparison between two methods that will potentially be used in the analysis of future cmb data . at the moment , simulations of cmb temperature fluctuations generated by the slow - roll inflation model here considered are possible at large angular scales .
the outline of the paper is the following .
the simulations used in this work are described in section 2 .
the method used to constrain the non - linear coupling parameter is based on the fisher discriminant built on skewness values corresponding to wavelet coefficient maps at several scales .
these method is presented in section 3 .
results are summarized in section 4 .
section 5 is dedicated to discussion and conclusions .
at large scales , the cmb temperature fluctuations are dominated by the sachs - wolfe effect and are therefore proportional to the fluctuations of the gravitational potential .
weakly non - gaussian cmb temperature fluctuations will be generated by the introduction of a nonlinear term in the gravitational potential @xmath3 @xmath4 where @xmath5 refers to the linear part of the gravitational potential , @xmath0 is the nonlinear coupling parameter controlling the amount of non - gaussianity introduced and the brackets indicate a volume average .
the linear part of the gravitational potential gives rise to cmb temperature fluctuations characterised by a power spectrum that in our case is assumed to be a cdm power spectrum ( computed with the cmbfast software ) with @xmath6 km / s / mpc and @xmath7 and normalized to @xmath8 .
we generate gaussian as well as non - gaussian ( as explained above ) healpix ( grski , hivon & wandelt 1999 ) projected cmb maps that include the observational constraints affecting the _ cobe_-dmr maps .
we apply the method to the 4-year 53 and 90 ghz _
cobe_-dmr maps after adding them using inverse - noise - variance weights .
these maps have a resolution of @xmath9 corresponding to a pixel size of @xmath10 .
only the pixels outside the extended galactic cut described in banday et al ( 1997 ) but explicitly recomputed for the healpix sky maps are considered in the computation .
best - fit monopole and dipole are removed from the pixels under analysis . to follow these _
cobe_-dmr observational constraints , the simulated gaussian and non - gaussian maps are convolved with a window function characterising the _
cobe_-dmr one as defined in wright et al .
noise maps , generated with the same observation pattern and rms noise levels as the _
cobe_-dmr data , are added to them .
the extended galactic cut is also impossed to these simulated maps and a best - fit monopole and dipole is as well removed from the pixels outside the cut .
5000 simulated maps are used in the statistical comparison described in the next section .
the aim of this work is to obtain the constraint impossed by the _ cobe_-dmr data on the nonlinear coupling parameter @xmath0 .
we use an approach based on the information provided by wavelets at several scales .
there are at the moment only two wavelets implemented on the sphere that have been used in cmb analyses ; the spherical mexican hat wavelet ( cayn et al . 2001 ; martnez - gonzlez et al . 2001 ) and the spherical haar wavelet ( tenorio et al . 1999 ; barreiro et al .
the former has proved to be more efficient than the spherical haar wavelet in detecting certain kind of non - gaussianity as shown in martnez - gonzlez et al .
moreover , the non - gaussianity introduced in the cmb maps by the nonlinear term added to the gravitational potential , is not expected to be dominated by any prefered direction .
therefore a wavelet with spherical symmetry , as the spherical mexican hat , appears to be more appropriate to analyse these data .
simulated gaussian and non - gaussian maps as well as the original _
cobe_-dmr maps are convolved with spherical mexican hat wavelets with sizes ranging from 2 ( 220 arcmins ) to 8 pixels . for these maps
we compute skewness and kurtosis .
the non - gaussianity introduced by the nonlinear term added to the gravitational potential will be visible when we compare distributions for these statistics for the gaussian and non - gaussian cases .
moreover , since the nonlinear term is small in amplitude and quadratic , we will expect the third order cumulant to be larger than any of the higher order cumulants . as an example , the skewness and kurtosis distributions of gaussian and non - gaussian maps at 220 arcmins are compared in figures 1 and 2 .
the weakly non - gaussianity introduced in the cmb simulations reflects in skewed distributions .
as expected , the kurtosis seems not to be affected by the nonlinear term .
we will therefore only use the skewness in our final analysis . in order to combine all the information we have at the seven scales used in the analyses we use the fisher discriminant . being @xmath11 the 7-dimensional data vector including the skewness values at the seven scales , the fisher discriminant @xmath12 is defined as @xmath13 where the @xmath14 subscript refers to the gaussian case and the @xmath15 subscript refers to the non - gaussian cases .
@xmath16 gives the sum of the covariance matrixes and @xmath17 refer to the mean values obtained for the gaussian and non - gaussian case respectively .
this statistic has been previously used in other analysis related to detecting non - gaussianity in the cmb ( barreiro & hobson 2001 ; martnez - gonzlez et al .
we establish the _
cobe_-dmr constraint on @xmath0 by comparing the _ cobe_-dmr fisher discriminant value with the fisher discriminant distributions obtained for non - gaussian simulations .
fisher discriminant distributions are presented in figure 3 for several non - gaussian cases characterised by the value of the nonlinear coupling coefficient .
comparison of figures 1 and 3 shows the expected fisher discriminant result of increasing the power to distinguish two distributions .
vertical bars in figure 3 are drawn at the _ cobe_-dmr fisher discriminant values for each non - gaussian case considered .
in order to obtain a constraint on @xmath0 we calculate the probability of observing a fisher discriminant value larger than the _
cobe_-dmr one for each of the non - gaussian cases . these probabilities are presented in table 1 . as one can see the _
cobe_-dmr data set a constraint on @xmath18 at the @xmath2 confidence level .
c|c @xmath0 & prob(@xmath19 ) 1000 & 63.0 1100 & 68.7 -1100&98.9 1200 & 70.2 1300 & 76.6 -1300&99.4 1400 & 78.6 1500 & 81.2 2000 & 93.4 2500 & 97.9 3000 & 99.3 in all previous calculations we have only considered positive values of the nonlinear coupling coefficient . to study the effect of having a negative @xmath0 on the fisher discriminant values , we have simulated 5000 non - gaussian maps with @xmath20 .
the distribution of temperature fluctuations in the map would be expected in this case to be shifted towards negative values in the same way as the distribution of temperature fluctuations is shifted towards positive values for @xmath21 ( as can be seen from the skewness distributions presented in figure 1 ) .
therefore , we would expect the distribution of skewness values in the case of @xmath22 to be towards the left of the distribution for @xmath23 .
same behaviour is observed for the fisher discriminant distributions as can be seen in figure 4 .
the cases for @xmath24 are also included in that figure for comparison . the power to distinguish the gaussian and non - gaussian distributions is approximately the same for the positive and negative values of @xmath0 . at the @xmath25 confidence level , the power of the fisher discriminant test is @xmath26 for @xmath27 respectively .
the same `` symmetric '' behaviour would therefore be expected for the any negative @xmath0 value in relation to the positive one .
as one can see from figure 4 the locus of the _ cobe_-dmr fisher discriminant value does not change much ( the probability of obtaining gaussian values larger than the _
cobe_-dmr ones is approximately the same ) .
the probability of having non - gaussian values smaller than _ cobe_-dmr ones is @xmath28 for the case of @xmath29 and @xmath30 for the case of @xmath31 .
the negative values of @xmath0 are therefore better constrained than the positive ones .
we can conclude that the _ cobe_-dmr data set a constraint on the absolute value of the nonlinear coupling parameter of @xmath1 at the @xmath2 confidence level . in order to account for the possible influence of foregrounds we have also analysed the _
cobe_-dmr data after foreground removal ( using _ cobe_-dirbe data ) .
the results obtained are consistent with the ones presented above .
weakly non - gaussian cmb temperature fluctuations can be generated by the presence of a nonlinear term in the gravitational potential .
the amount of nonlinearity can be controlled by the so called nonlinear coupling parameter @xmath0 . at large scales ,
the cmb temperature fluctuations are proportional to the gravitational potential .
simulations of weakly non - gaussian cmb temperature fluctuations maps at these large scales are therefore reasonably easy to make .
the aim of this paper is to set a constraint on the value of the nonlinear coupling parameter by using the _ cobe_-dmr data .
we perform 5000 simulations of gaussian and several non - gaussian models characterised by the value of @xmath0 , impossing the _ cobe_-dmr observational constraints . with the aid of the spherical mexican hat wavelet , we are able to obtain the skewness of these maps at several scales . all this information is combined into the fisher discriminant .
distributions of this statistic are used to set the constraint on @xmath0 by comparing the ones obtained from the non - gaussian simulations to the _ cobe_-dmr values . with this method
we are able to set @xmath1 at the @xmath2 confidence level .
this is a tighter contraint than the one previously found by komatsu et al .
they obtain the best - fit value of @xmath0 , by fitting the observed _ cobe_-dmr bispectrum to a theoretical non - gaussian model .
they estimate statistical uncertainties of this parameter and place a @xmath2 confidence limit on @xmath0 as @xmath32 for the extended cut . a method based on wavelets in which an efficient combination of information from different scales is used ,
has been shown to be very powerful in order to test the non - gaussianity introduced in cmb maps by a nonlinear term in the gravitational potential .
this work has been carried out at large scales at which the sachs - wolfe effect dominates .
however tighter constraints are expected to be obtained from the analysis of higher resolution maps .
grski , k.m . ,
hivon , e. & wandelt , b.d .
( astro - ph/9812350 ) 1999 , proceedings of the mpa / eso conference on evolution of large - scale structure : from recombination to garching , 2 - 7 august 1998 ; eds .
banday , r.k .
sheth and l. da costa , printpartners ipskamp nl ( 1999 ) | cobe_-dmr data compatible with gaussianity by methods real ( torres 1995 kogut 1996 schmalzing 1997 novikov 2000 spherical harmonic ( banday komatsu 2002 wavelet space cayn 2001 cmb observations by boomerang dasi maxima - i netterfield pryke stompor indicate possible non - gaussianity likely produced by non - standard inflationary models important to study characteristics of non - gaussianity by alternative theories to standard inflation alternative non - gaussian fluctuations slow - roll inflation investigated by salopek bond gangui 1994 model generates non - gaussianity in matter radiation fluctuation fields through features in inflaton potential cmb context model studied by komatsu & spergel 2001 verde et al . 2001 komatsu gravitational potential includes quadratic term amplitude regulated by non - linear coupling parameter 2 verde et al . 2001 ) cmb data more sensitive to non - gaussianity by models than high redshift objects semianalytical expression of bispectrum model presented in komatsu & spergel ( 2001 )minimum non - linear coupling parameter values non - gaussianity in cmb data sets estimated values larger than slow - roll inflation theories komatsu et al 2002 revised constraint by cobe_-dmr data on non - linear coupling parameter normalized bispectrum constraint larger than spergel 2001 weak about slow - roll inflation interesting deviations might include larger values non - linear coupling parameter non - gaussianities asses minimum value of non - linear coupling parameter by cobe_-dmr data wavelet space statistic comparison constraint with normalized bispectrum important future cmb data simulations of cmb temperature fluctuations by slow - roll inflation model possible at large angular scales outline paper simulations described in section 2 method non - linear coupling parameter based on fisher discriminant skewness values wavelet coefficient maps presented in section 3 results summarized in section 4 section 5 discussion conclusions at large scales cmb temperature fluctuations dominated by sachs - wolfe effect proportional to fluctuations gravitational potentialnon - gaussian cmb temperature fluctuations generated by nonlinear term in gravitational potential @xmath3 @xmath4 @xmath5 refers to linear part potential @xmath0 nonlinear coupling parameter non - gaussianity brackets indicate volume average linear part gravitational potential cmb temperature fluctuations power spectrum assumed cdm power spectrum computed with cmbfast software @xmath6 km / s / mpc @xmath7 normalized to @xmath8 generate gaussian non - gaussian healpix projected cmb maps observational constraints affecting _ cobe_-dmr maps apply method to 4-year 53 90 ghz _ cobe_-dmr maps inverse - noise - variance weights maps have resolution @xmath9 pixel size @xmath10 pixels outside extended galactic cut considered best - fit monopole dipole removed from pixels simulated gaussian non - gaussian maps convolved with window function _ cobe_-dmr one noise maps pattern-dmr data added extended galactic cut impossed to maps best - fit monopole dipole removed from pixels outside cut 5000 simulated maps used in statistical comparisonaim work obtain constraint _ cobe_-dmr data on nonlinear coupling parameter @xmath0 approach on information wavelets at scales two wavelets sphere used in cmb analyses spherical mexican hat wavelet spherical haar wavelet former more efficient detecting non - gaussianity non - gaussianity introduced in cmb maps by nonlinear term added gravitational potential not dominated by prefered direction wavelet with spherical symmetry spherical mexican hat appropriate analyse data simulated gaussian non - gaussian maps original _ cobe_-dmr maps convolved with spherical mexican hat wavelets sizes 2 to 8 pixels compute skewness kurtosis non - gaussianity introduced by nonlinear term added to gravitational potential visible when distributions gaussian non - cases nonlinear term small in amplitude quadratic third order cumulant larger than higher order cumulants skewness kurtosis distributions of gaussian non - gaussian maps at 220 arcmins compared in figures 1 and 2 non - gaussianity introduced in cmb simulations reflects in skewed distributions kurtosis not affected by nonlinear termuse skewness final analysis combine information seven scales use fisher discriminant @xmath11 7-dimensional data vector skewness values seven fisher discriminant @xmath12 @xmath13 @xmath14 gaussian @xmath15 non - gaussian cases @xmath16 sum covariance matrixes @xmath17 mean values gaussian non - gaussian statistic used non - gaussianity establish cobe_-dmr constraint on @xmath0 comparing fisher discriminant value with distributions non - gaussian simulations fisher discriminant distributions figure 3 non - gaussian cases nonlinear comparison figures shows expected fisher discriminant result increasing power distinguish distributions vertical bars at cobe_-dmr fisher discriminant values non - gaussian case constraint on @xmath0 calculate probability observing fisher discriminant value larger than cobe_-dmr non - gaussian cases probabilities table 1 cobe_-dmr data constraint on @xmath18 @xmath2 confidence level @xmath0@xmath19 ) 1000 63.0 1100 68.7 -1100&98.9 1200 70.2 1300 76.6 -1300&99.4 1400 78.6 1500 81.2 2000 93.4 2500 97.9 3000 99.previous calculations considered positive values nonlinear coupling coefficient study effect negative @xmath0 on fisher discriminant values simulated 5000 non - gaussian maps with @xmath20 distribution temperature fluctuations expected shifted towards negative values positive values @xmath21 figure distribution skewness values @xmath22 left of distribution @xmath23 same behaviour for fisher discriminant distributions in figure 4 cases @xmath24 included comparison power to distinguish gaussian and non - gaussian distributions same for positive and negative values @xmath0 @xmath25 confidence level power fisher discriminant test is @xmath26 for @xmath27 same symmetric behaviour expected for negative @xmath0 value positive locus of _ cobe_-dmr fisher discriminant value change probability obtaining gaussian values larger same probability non - gaussian values smaller than is @xmath28 for @xmath29 and @xmath30 @xmath31 negative values of @xmath0 better constrained than positive ones _ cobe_-dmr data on absolute value nonlinear coupling parameter of @xmath1 at @xmath2 confidence levelinfluence foregrounds analysed cobe_-dmr data after foreground removal_-dirbe data results consistent with non - gaussian temperature fluctuations generated by nonlinear term in gravitational potential controlled by nonlinear coupling parameter @xmath0 at large scales cmb temperature fluctuations proportional to gravitational potential simulations non - gaussian easy aim constraint on value nonlinear coupling parameter cobe_-dmr data 5000 simulations of gaussian non - gaussian models @xmath0 cobe_-dmr observational constraints spherical mexican hat wavelet skewness of maps at several scales information combined into fisher discriminant distributions set constraint on @xmath0 non - gaussian simulations to cobe_-dmr values set @xmath1 at @xmath2 confidence level tighter contraint than by komatsu et al obtain best - fit value of @xmath0 fitting cobe_-dmr bispectrum to non - gaussian model estimate statistical uncertainties place @xmath2 confidence limit on @xmath0 as @xmath32 for extended cutmethod wavelets combination information scales non - gaussianity cmb maps nonlinear gravitational potential work carried large scales sachs - wolfe effect dominates tighter constraints higher resolution maps grski hivon. wandelt b.d astro - ph/9812350 ) 1999 proceedings mpa / eso conference evolution large - scale structure recombination to garching 2 - 7 august 1998 eds banday r sheth l. da costa printpartners ipskamp 1999 | 0.507572 | 3,187 | 1,616 | , Saving $0.1 in GPT-4. |
177,655 | in this brief note , we show that chaotic boltzmann machines truly yield samples from the probabilistic distribution of the corresponding boltzmann machines if they are composed of only two elements .
this note is an english translation ( with slight modifications ) of the article originally written in japanese [ h. suzuki , seisan kenkyu 66 ( 2014 ) , 315316 ] . | recently , chaotic boltzmann machines @xcite were proposed as a deterministic implementation of boltzmann machines .
the apparently stochastic behavior of chaotic boltzmann machines is achieved , without any use of random numbers , by chaotic dynamics that emerges from pseudo - billiard dynamics .
it was shown numerically that the chaotic billiard dynamics of chaotic boltzmann machines can be used for generating sample sequences from the probabilistic distribution of boltzmann machines , and it was successfully applied to other spin models such as the ising model and the potts model @xcite . despite these numerical evidences ,
there have been no theoretical proof that chaotic boltzmann machines yield samples from the probabilistic distribution of the corresponding boltzmann machines . in this brief note , as a first step of theoretical approach , we investigate the simplest system .
namely , we show that chaotic boltzmann machines truly yield samples for the corresponding boltzmann machines if they are composed of only two elements .
although our approach can not be applied to larger chaotic boltzmann machines with more than two elements , we expect that it gives some insights into the dynamics of larger chaotic boltzmann machines . since the proof is not entirely trivial , it is considered worth making available on arxiv .
this note is an english translation ( with slight modifications ) of the article @xcite originally written in japanese .
let @xmath0 and @xmath1 be random variables that take values on @xmath2 .
we assume that the probabilistic model @xmath3 $ ] is given in the form of conditional probabilities @xmath4 $ ] and @xmath5 $ ] , which is common in boltzmann machines and many other spin models in statistical physics .
let us consider a chaotic boltzmann machine for the probabilistic model .
in addition to the states @xmath6 and @xmath7 of the elements , we introduce internal states @xmath8 and @xmath9 $ ] of the elements . therefore
, the state space of the chaotic boltzmann machine is @xmath10 ^ 2 $ ] .
the internal states @xmath8 and @xmath11 evolve according to the following differential equations : @xmath12^{-1 } , \label{eq : dx1}\\ \frac{dx_2}{dt } & = ( 1 - 2s_2 ) c_2(s_1 ) p[s_2\mid s_1]^{-1 } , \label{eq : dx2}\end{aligned}\ ] ] where @xmath13 represents arbitrary positive constants defined for each state of the other element . when the internal state @xmath14 reaches @xmath15 or @xmath16 , the state @xmath17 of the element changes as follows : @xmath18 the internal state @xmath19 moves straight inside the unit square @xmath20 ^ 2 $ ] , and changes its direction when it hits a side of the square .
therefore , this system can be understood as a pseudo - billiard in the billiard table @xmath20 ^ 2 $ ] .
what we expect here is that the state variables @xmath6 and @xmath21 observed from this system follow the distribution @xmath22 $ ] .
namely , our goal in this note is to show the following proposition . for any initial values
@xmath23 ^ 2 $ ] , the chaotic boltzmann machine defined by eqs .
( [ eq : dx1])([eq : ds0 ] ) satisfies @xmath24 \label{eq : pss}\ ] ] for @xmath25 , provided that @xmath26 is irrational . here ,
@xmath27 is kronecker s delta function , and @xmath28 $ ] .
to show this proposition , we rewrite eqs . ( [ eq : dx1 ] ) and ( [ eq : dx2 ] ) of the chaotic boltzmann machine . specifically , instead of the state @xmath29 of each element , we introduce a new state variable @xmath30 using the map @xmath31 then the state space @xmath10 ^ 2 $ ] of the chaotic boltzmann machine is mapped to the rectangular @xmath32\times[-r_2(1),r_2(0)]$ ] ( see fig .
[ fig : sspace ] ) . on this state space ,
the dynamics of the chaotic boltzmann machine ( eqs .
( [ eq : dx1])([eq : ds0 ] ) ) is rewritten as follows : @xmath33^{-1 } = \frac{r_1(s_1)r_2(s_2)}{p[s_1,s_2 ] } , \label{eq : dy1 } \\
\frac{dy_2}{dt } & = r_2(s_2 ) c_2(s_1 ) p[s_2|s_1]^{-1 } = \frac{r_1(s_1)r_2(s_2)}{p[s_1,s_2 ] } , \label{eq : dy2}\end{aligned}\ ] ] and @xmath34 where the state @xmath17 can be determined from @xmath35 as @xmath36 when @xmath37 , and @xmath38 when @xmath39 . the state space for @xmath40 . ]
the differential equations ( [ eq : dy1 ] ) and ( [ eq : dy2 ] ) always satisfy @xmath41 .
hence , the state @xmath40 moves in the direction of @xmath42 in the rectangular state space . here
we introduce @xmath43 which moves along the orbit of @xmath40 at a constant velocity @xmath44 .
then the orbit of @xmath43 is equidistributed in the state space in the following sense .
let us consider the poincar section on @xmath45 . while the value of @xmath46 changes by @xmath47 , the value of @xmath48 changes by the same amount
. therefore , the poincar map on @xmath45 is a rigid rotation with rotation number @xmath49 . if the rotation number is irrational , the poincar map is ergodic with respect to the uniform distribution . then the orbit is also equidistributed in the state space , and we have @xmath50 since @xmath40 moves along the same orbit as @xmath43 with the velocity described in eqs .
( [ eq : dy1 ] ) and ( [ eq : dy2 ] ) , we have @xmath51,\ ] ] which completes the proof .
in this brief note , we have shown that chaotic boltzmann machines with two elements have quasi - periodic dynamics and yield samples from the probabilistic distribution of the corresponding boltzmann machines , provided that the rotation number is irrational .
the set of parameter values @xmath52 that make rotation numbers rational has lebesgue measure zero .
this is analogous to probabilistic monte carlo sampling , which does not work with probability ( measure ) zero .
however , when we design chaotic boltzmann machines artificially , it may be possible that the rotation number easily becomes rational .
in such a case , we have to adjust @xmath52 by multiplying some constants to make the rotation number irrational .
the performance as an sampling algorithm depends on the characteristics of the rotation number . at least , a chaotic boltzmann machine with a good rotation number that yields a low - discrepancy sequence is expected to exhibit better performance than the corresponding boltzmann machine .
our approach in this note can not be applied to larger chaotic boltzmann machines with more than two elements , which exhibit chaotic behavior . however
, some aspects are expected to be shared with larger systems ; for example , our preliminary numerical study shows that irrationality is important , for chaotic boltzmann machines composed of not many but more than two elements , to generate faithful sample sequences .
therefore , we expect that our approach gives some insights into the dynamics of larger chaotic boltzmann machines . | chaotic boltzmann machines @xcite proposed as deterministic implementation boltzmann stochastic behavior achieved without random numbers by chaotic dynamics from pseudo - billiard dynamics shown chaotic billiard dynamics used for generating sample sequences from probabilistic distribution applied to other spin models ising model potts model @xcite no theoretical proof that chaotic boltzmann machines yield samples from probabilistic distribution first investigate simplest system show chaotic boltzmann machines yield samples for if only two elements approach not to larger machines more than two elements expect gives insights into dynamics larger proof not trivial worth making available on arxiv english translation of article @xcite in japanese @xmath0 and @xmath1 be random variables take values on @xmath2 probabilistic model @xmath3 in of conditional probabilities @xmath4 and @xmath5 $ ] common in boltzmann machines other spin models in statistical physics consider chaotic boltzmann machine for probabilistic model states @xmath6 @xmath7 introduce internal states @xmath8 and @xmath9state space chaotic boltzmann machine is @xmath10 ^ 2 $ internal states @xmath8 @xmath11 evolve differential equations @xmath12^{-1 }{eq : dx1}\\{dx_2}{dt } ( 1 - 2s_2 ) c_2(s_1 ) p[s_2 s_1]^{-1 }{eq : dx2}{aligned} @xmath13 represents positive constants each state internal state @xmath14 reaches @xmath15 or state @xmath17 changes internal state @xmath19 moves inside unit square @xmath20 ^ 2 $ changes direction hits side square system pseudo - billiard billiard table state variables @xmath6 @xmath21 follow distribution @xmath22 $ goal show proposition @xmath23 2 chaotic boltzmann machine defined dx1])( ds0 satisfies @xmath24 : pss} for @xmath25 @xmath26 irrational @xmath27 is kronecker s delta function @xmath28 $ ] rewrite eqs dx1 dx2 ] chaotic boltzmann machineinstead state @xmath29 introduce new state variable @xmath30 map @xmath31 state space @xmath10 chaotic boltzmann machine mapped rectangular @xmath32[-r_2(1),r_2(0)] fig . dynamics chaotic boltzmann machine rewritten @xmath33^{-1 = \frac{r_1(s_1)r_2(s_2)}{p[s_1,s_2{eq : dy1\frac{dy_2} r_2(s_2 ) c_2(s_1 ) p[s_2|s_1]{-1 }\frac{r_1(s_1)r_2(s_2)}{p[s_1,s_2 ]\label{eq : dy2}{aligned @xmath34 state @xmath17 determined from @xmath35 @xmath36 @xmath37 @xmath38 @xmath39 state space @xmath40 differential equations eq : dy1 dy2 satisfy @xmath41 state @xmath40 moves direction @xmath42 space introduce @xmath43 moves orbit @xmath40 constant velocity orbit @xmath43 equidistributed state space poincar section @xmath45 value @xmath46 changes by @xmath47 @xmath48 samepoincar map on @xmath45 rigid rotation with rotation number @xmath49 if rotation number irrational poincar map ergodic uniform distribution orbit equidistributed in state space have @xmath50 @xmath40 same orbit as @xmath43 velocity in @xmath51 completes proof chaotic boltzmann machines with two elements have quasi - periodic dynamics yield samples from probabilistic distribution rotation number irrational parameter values @xmath52 make rotation numbers rational has lebesgue measure zero analogous to probabilistic monte carlo sampling work with probability measure zero chaotic boltzmann machines artificially rotation number becomes rational adjust @xmath52 by multiplying constants to make rotation number irrational performance depends on characteristics rotation number chaotic boltzmann machine with good rotation number low - discrepancy sequence better performance approach not applied to larger chaotic boltzmann machines with more than two elements some aspects shared with larger systems study shows irrationality important for chaotic boltzmann machines two generate faithful sample sequences approach gives insights into dynamics of larger chaotic boltzmann machines | 0.26804 | 1,697 | 915 | , Saving $0.0 in GPT-4. |
177,679 | properties of the disrupted nlc beam at the interaction point ( ip ) and particle loss in the extraction line are analyzed as a function of beam - to - beam position and angular offset at ip .
the simulations show that disruption and beam loss maximize when the vertical beam separation at ip is about 20 times the rms vertical beam size .
the horizontal offset does not increase the disruption and the beam loss .
the angular offsets cause particle loss in the extraction line mainly because of the beam orbit oscillations .
pub8872 + june 2001 + * nlc beam properties and extraction line performance + with beam offset at ip [ multiblock footnote omitted ] * * abstract * + | in the nlc @xcite , the strong beam - beam interaction significantly distorts beam distribution at ip . this effect ,
called disruption @xcite , increases the beam emittance and angular divergence after collision and generates a huge energy spread in the outgoing beam .
the nlc extraction line was designed to minimize particle loss caused by these effects @xcite .
so far , the nlc beam disruption and extraction line performance were studied for the ideal beam conditions at the ip . however , various incoming beam errors may affect the beam - beam interaction and the resultant disrupted distribution . in this paper
, we discuss the effects of beam - to - beam transverse position and angular offsets at ip on the beam distribution and particle loss in the extraction line .
beam - beam effects such as energy loss due to beamstrahlung occur as particles in each bunch go through the strong coherent field of the other bunch .
when the beams are vertically flat , as they are in nlc , the beamstrahlung can be enhanced significantly for some vertical position and angle offsets , because the field seen by the bulk of the particles in one beam increases as the offset from the other beam increases .
of course , the field seen by each beam eventually falls off when position offsets become large enough .
the interaction is further complicated by the fact that the two beams distort each other s shape during the collision .
thus we resort to numerical simulations , where care must be taken in choosing the grids to perform beam - beam calculations as accurately as possible .
the beam parameters for the nlc design can be found in ref .
@xcite . in this study
, we used the stage 1 parameters where the center - of - mass energy is 500 gev ; the ip parameters listed in tables 1 and 2 , where @xmath0 is the beam emittance , @xmath1 the beam size , and @xmath2 angular divergence in the horizontal and vertical plane , respectively .
the listed parameters are for ideal ip conditions and zero initial energy spread .
the realistic incoming energy spread is about @xmath3 , but it is negligible compared to the disrupted energy spread and does not affect the results of this study .
.ip parameters before and after collision .
[ cols= " < , < " , ] at ip vs. @xmath4.,width=188 ] vertical divergence at ip vs. @xmath4.,width=188 ] , @xmath5 and @xmath6.,width=264 ] offset.,width=188 ] as a result of the increased energy spread and vertical beam size , one can expect higher beam loss in the extraction line for @xmath4 offsets near 15 to 20@xmath7 .
the results of particle loss versus @xmath8-offset are shown in fig .
[ partloss - offy - h490 ] .
the maximum loss occurs at @xmath9 and amounts to 0.7 kw of power loss with 15 lost particles out of @xmath10 in tracking . some irregularities in fig .
[ partloss - offy - h490 ] are due to low statistics of the lost particles .
as expected , tracking with the horizontal ip offsets showed no particle loss .
more detailed analysis of the beam loss with @xmath11 offset revealed that all of the particles except one were lost in the vertical plane .
this indicates that the blow up of vertical emittance may be the source of the particle loss .
our expectation is that the power loss caused by ip offset , even at the maximum value of 0.7 kw , can be safely disposed . in practice , the losses should be much lower since the beam offset will be controlled at the level of @xmath12 for a maximum luminosity . we expect that with such control of the vertical offsets the power loss will be on the order of @xmath130.05 kw for the stage 1 parameters .
disrupted beam distributions with angular offset at ip were generated in the guinea pig code by changing the nominal initial angle at ip by @xmath14 or @xmath15 in the two beams .
the resultant distributions of @xmath16 particles were then tracked in the extraction line . at ip vs. @xmath17.,width=188 ] vertical divergence at ip vs. @xmath17.,width=188 ] offset.,width=188 ] offset.,width=188 ] as in the case of ip position offset , a vertical angular offset results in somewhat increased beam energy spread and vertical divergence at ip as shown in fig .
[ deave - offangy - h490 ] and [ sigpy - offangy - h490 ] , while the effect of a horizontal angle is small .
note that angular offset in fig .
[ deave - offangy - h490 ] and [ sigpy - offangy - h490 ] is normalized to @xmath18 , the undisrupted initial divergence at ip .
particle tracking in the extraction line showed , however , that large offset angles have much stronger effect on the beam loss than the position offset .
the particle loss versus @xmath19 and @xmath17 is shown in fig .
[ partloss - offangx - h490 ] and [ partloss - offangy - h490 ] .
the large particle losses are caused by the increased beam orbit oscillations in the extraction line proportional to the ip angular offsets . for a beam power loss below 1 kw at a center - of - mass energy of 500 gev
, the particle loss needs to be lower than 0.05% . according to fig .
[ partloss - offangx - h490 ] and [ partloss - offangy - h490 ] , the corresponding maximum full angular offsets are about 40@xmath20 ( 1.2 mrad ) and 10@xmath21 ( 0.25 mrad ) for @xmath22 and @xmath8 planes , respectively .
note that the beam loss is strongly reduced at smaller angles .
however , to avoid unnecessary large beam loss , the angular offsets should be kept well below the above tolerances .
the beam position and angular offsets at ip may increase the beam disruption . for the stage 1 ( 500 gev _ cms _ ) nlc parameters , the beam vertical offset may result in up to 0.7 kw beam loss in the extraction line . in practice
, the losses will be much lower for reasonably controlled ip beam positions .
the angular offsets should be kept lower than 40@xmath20 ( 1.2 mrad ) and 10@xmath21 ( 0.25 mrad ) in @xmath22 and @xmath8 planes for a power loss below 1 kw . http://www-project.slac.stanford.edu/lc/ and _ 2001 report on the next linear collider _ , slac
report571 , fermi
conf01/075e , lbnl
pub47935 , ucrl id144077 ( 2001 ) .
r. hollebeek , _ nim _ , * 184 * , p. 333
y. nosochkov , _ et al .
_ , slac pub8096 ( 1999 ) . y.m .
nosochkov and t.o .
raubenheimer , slac pub8313 and lcc0034 ( 1999 ) .
d. schulte , tesla9708 ( 1996 ) .
p. tenenbaum , _ et al .
_ , slac pub8136 ( 1999 ) . | nlc @xcite strong beam - beam interaction distorts beam distribution at ip disruption increases beam emittance angular divergence after collision generates energy spread in outgoing beam nlc extraction line designed minimize particle loss nlc beam disruption extraction line performance studied for ideal beam conditions at ip incoming beam errors may affect beam interaction disrupted distribution effects of beam - to - beam transverse position angular offsets at ip on beam distribution particle loss beam beam effects energy loss beamstrahlung occur as particles go through field other bunch beams vertically flat beamstrahlung enhanced for vertical position angle offsets field increases as offset increases field falls off when position offsets large interaction complicated two beams distort shape during collision resort to numerical simulations care choosing grids beam calculations beam parameters for nlc design in ref . @xcite used stage 1 parameters center - of - mass energy 500 gev ip parameters in tables 1 2 @xmath0 beam emittance @xmath1 beam size @xmath2 angular divergenceparameters for ideal ip conditions zero initial energy spread realistic incoming energy spread about @xmath3 negligible compared to disrupted energy spread affect results study .ip parameters before after collision vs @xmath4 vertical divergence @xmath5 @xmath6 increased energy spread vertical beam size expect higher beam loss in extraction line for @xmath4 offsets near 15 to 20@xmath7 results particle loss versus @xmath8-offset shown in fig . maximum loss at @xmath9 0.7 kw power loss 15 lost particles out of @xmath10 irregularities in due to low statistics lost particles horizontal ip offsets no particle loss beam loss with @xmath11 offset particles except one lost in vertical plane blow up of vertical emittance may source particle loss expectation power loss ip offset maximum 0.7 kw safely disposed losses lower beam offset controlled at level @xmath12 for maximum luminosity control vertical offsets power loss @xmath130.05 kw for stage 1 parametersbeam distributions with angular offset at ip generated by changing initial angle at ip by @xmath14 or @xmath15 in two beams resultant distributions of @xmath16 particles tracked in extraction line vs @xmath17 vertical divergence vertical angular offset results in increased beam energy spread and vertical divergence at ip shown in effect horizontal angle small angular offset in normalized to @xmath18 undisrupted initial divergence at ip large offset angles stronger effect on beam loss than position offset particle loss versus @xmath19 and @xmath17 shown in fig - - large particle losses caused by increased beam orbit oscillations in line proportional to ip angular offsets for beam power loss below 1 kw at center - of - mass energy 500 gev particle loss needs lower than 0.05% maximum full angular offsets are about 40@xmath20 ( 1.2 mrad ) and 10@xmath21 ( 0.25 mrad ) for @xmath22 and @xmath8 planesbeam loss reduced at smaller angles large loss angular offsets below tolerances offsets ip increase beam disruption stage 1 500 gev nlc parameters beam vertical offset 0.7 kw loss extraction line losses lower for controlled ip beam positions angular offsets lower than 40@xmath20 1.2 mrad 10@xmath21 ( 0.25 mrad ) in @xmath22 @xmath8 planes for power loss below 1 kw http://www-project.slac.stanford.edu 2001 report next linear collider slac report571 conf01/075e lbnl pub47935 ucrl id144077 2001 hollebeek 333 nosochkov slac pub8096 ( 1999 ) raubenheimer slac pub8313 lcc0034 ( 1999 schulte tesla9708 1996 tenenbaum slac pub8136 ( 1999 | 0.265785 | 1,614 | 788 | , Saving $0.0 in GPT-4. |
177,687 | fast kicker magnets are used to inject beam into and eject beam out of the cern accelerator rings .
these kickers are generally transmission line type magnets with a rectangular shaped aperture through which the beam passes .
unless special precautions are taken the impedance of the yoke can provoke significant beam induced heating , especially for high intensities .
in addition the impedance may contribute to beam instabilities .
the results of longitudinal and transverse impedance measurements , for various kicker magnets , are presented and compared with analytical calculations : in addition predictions from a numerical analysis are discussed . | ferrite kicker magnets are used extensively within the cern accelerator complex to inject and extract beams from the various machines for transfer to experimental areas , other particle accelerators or to dump beams .
they have been known to be a significant source of beam coupling impedance in the accelerator complex at cern for some time @xcite .
this is due to the proximity of very lossy materials to the beam . to counteract this a number of impedance reduction techniques
have been applied to these magnets , both retroactively to existing pieces of equipment @xcite and during the design stages @xcite . due to the difficulty of simulating ferrite kickers with the impedance reduction measures in place , it is often more convenient to measure their beam impedance using a coaxial wire setup as opposed to using simulations .
further work can subsequently be done to understand the measurements and give guidance to simulation studies .
the coaxial wire method has been used for a number of decades as a bench - top method of measuring the beam coupling impedance of accelerator structures .
it is based on the concept that the electromagnetic field profile around an ultrarelativistic charged particle is similar in nature to that of a short electrical pulse propagating along a coaxial line . for the measurements of the lhc - mki ( large hadron collider injection kicker magnets )
the resonant coaxial wire method has been used . for the other measurements presented here
the classical transmission method has been used .
both of these are described in more detail in @xcite .
the lhc - mkis are a set of transmission line ferrite kicker magnets . unlike most existing kicker magnets
, they were designed with impedance reduction mechanisms in place , in the form of a ceramic tube within which a series of longitudinal ( from the beam point of view ) conducting strips are inserted to provide a low resistivity path for the image currents .
this was done to reduce both beam induced heating @xcite and the likelihood of impedance based instabilities in the circulating beam . the longitudinal impedance of the lhc - mki was measured using the resonator method . for reference measurements ,
the beampipe of the counter - rotating beam in the kicker assembly was used .
this is a copper tube of equal length to the main magnet aperture .
values without the beam screen are given by the tsutsui formalism for two parallel plates @xcite .
as can be seen in fig .
[ fig : mki - long ] , the inclusion of the beam screen is highly effective in reducing the real impedance of the magnet . the imaginary impedance is also greatly reduced across the majority of the measured frequency spectrum , however two large peaks occur at @xmath0800 mhz and @xmath01200 mhz . a series of measurements were also done on an lhc - mki magnet at different stages of preparation to see how operation may condition the impedance of the device .
[ fig : mki - long - comp ] demonstrates that although the beam screen remains effective in suppressing the real impedance , the preparations for insertion in the lhc and operation with the lhc can cause significant changes in the impedance of devices built to the same specification , particularly at frequencies greater than 1 ghz .
simulations to understand the source of these peaks are currently under investigation . in common with the longitudinal impedance , the transverse ( both dipolar and quadrupolar ) impedances are greatly reduced when compared to that expected for unscreened ferrite .
the large peak observed in the horizontal dipolar impedance at 1800 mhz ( see fig .
[ fig : mki - dipolar ] ) is thought to be a higher order mode ( hom ) present in the measuring setup and thus does not represent a real impedance .
the quadrupolar impedance is drastically reduced at all frequencies by the beam screen .
the behaviour of the transverse impedances with respect to the state of the beam screen is similarly under investigation using simulation models .
the sps - mkes ( extraction kicker magnet ) are transmission line ferrite kicker magnets to allow the extraction of beam from the sps to the lhc .
two families of mkes exists ; the l - type and the s - type , differing in the dimensions of the kicker aperture .
originally designed without impedance reduction techniques in place , recent desires for increased beam performances were the cause for a concerted effort to devise effective retroactive impedance reductions techniques to be applied to these magnets @xcite .
comparisons between the measured impedances of two magnets with and without the reduction , as well as an analytical calculation with the tsutsui model can be seen in fig .
[ fig : mke - long ] .
the impedance of the sps - mke was measured using the classical transmission method .
for the reference measurement an analytical calculation was made .
the longitudinal impedance of the sps - mke is greatly suppressed by the inclusion of serigraphy on the surface of the magnet ( fig .
[ fig : mke - long ] ) .
furthermore a comparison of measurements of the magnets without serigraphy to the tsutsui model indicate a very good agreement up to 600 mhz , with divergence beyond this frequency due to differences between the internal structure of the magnet and the tsutsui model . again
the large peak at @xmath01700 mhz is postulated to be an artifact of the measurement .
the sps - mkp ( injection kicker magnet ) are a set of transmission line magnets used to inject beam into the sps .
one of their features is that they are composed of a many segments of ferrite separated by high voltage plates as can be seen in fig .
[ fig : mkp - geo ] . measurements and simulations have been carried out to determine how the inclusion of the segmentation causes the measurements to differ from the analytic models.the impedance of the sps - mkp was measured using the classical transmission method .
simulations were carried using cst particle studio @xcite . as can be seen from the simulations ( see fig .
[ fig : mkp - long ] ) , the addition of segmentation and the c - core shape cause an increase in the peak value of the real impedance at @xmath0600 mhz and also the generation of a lower frequency peak at 30 - 40 mhz .
we have demonstrated the effectiveness of the implemented impedance reduction techniques in lowering the beam coupling impedance of a number of ferrite kicker magnets at the cern accelerator complex .
in particular we have a significant reduction in the real impedance compared to their unshielded cases thereby reducing the problem of beam - induced heating .
furthermore we can see the theoretical models currently in use are not sufficient to fully explain the measured impedance profiles due to not truly representing the internal magnet structure .
further work is underway to develop models that more accurately represent the magnets internal structure .
thanks to yves sillanoli and salim bouleghlimat for help in the mechanical setup of the measurement system . | ferrite kicker magnets used within cern accelerator complex to inject extract beams from machines for transfer experimental or dump beams significant source of beam coupling impedance due to proximity of lossy materials to beam impedance reduction techniques applied to magnets design stages difficulty simulating ferrite kickers with impedance reduction measures convenient to measure beam impedance using coaxial wire setup simulations further work can to understand measurements guidance to simulation studies coaxial wire method used for decades measuring beam coupling impedance of accelerator structures based on concept electromagnetic field profile around ultrarelativistic charged particle similar to short electrical pulse along coaxial line for measurements of lhc - mki kicker magnets resonant coaxial wire method used for other measurements classical transmission method used described in @xcite lhc - mkis are transmission line ferrite kicker magnets designed with impedance reduction mechanisms form ceramic tube longitudinal conducting strips inserted low resistivity path for image currents reduce beam induced heating impedance based instabilities in circulating beamlongitudinal impedance lhc - mki measured using resonator method for reference beampipe of counter - rotating beam in kicker assembly used copper tube equal length to main magnet aperture values without beam screen given by tsutsui formalism for two parallel plates @xcite inclusion beam screen effective real impedance imaginary impedance reduced across measured frequency spectrum two large peaks occur at @xmath0800 mhz and @xmath01200 mhz measurements done on lhc - mki magnet at different stages preparation operation impedance beam screen effective real impedance preparations for insertion operation can cause significant changes in impedance of devices particularly at frequencies greater than 1 ghz simulations source of peaks under investigation transverse dipolar and quadrupolar impedances reduced unscreened ferrite large peak in horizontal dipolar impedance at 1800 mhz thought higher order mode ( ) in setup not represent real impedance quadrupolar impedance reduced at all frequencies by beam screen behaviour of transverse impedances state beam screen under investigationsps - mkes extraction kicker ) are transmission line ferrite magnets extraction beam from sps to lhc two families mkes l - and s - type differing in dimensions kicker aperture designed without impedance reduction techniques for performances retroactive impedance reductions techniques comparisons between impedances magnets with and without reduction analytical calculation with tsutsui model in fig . impedance sps - mke measured classical transmission method for reference analytical calculation made longitudinal impedance of sps - mke suppressed by serigraphy on surface magnet comparison measurements without serigraphy to tsutsui model good agreement up to 600 mhz divergence beyond due to differences internal structure magnet tsutsui model large peak at @xmath01700 mhz artifact of measurement sps - mkp ( injection kicker magnet ) are transmission line magnets inject beam into sps segments of ferrite separated by high voltage plates in fig measurements simulations to segmentation impedance of sps - mkp measured using classical transmission method simulations using cst particle studio @xcitesimulations fig . segmentation c - core shape cause increase peak real impedance at @xmath0600 mhz lower frequency peak at 30 - 40 mhz demonstrated effectiveness impedance reduction techniques lowering beam coupling impedance ferrite kicker magnets at cern accelerator complex significant reduction real impedance unshielded cases reducing beam - induced heating theoretical models not sufficient explain impedance profiles not representing internal magnet structure work underway to develop models structure thanks to yves sillanoli salim bouleghlimat for mechanical setup measurement system | 0.264852 | 1,427 | 679 | , Saving $0.0 in GPT-4. |
177,689 | we summarize the current top - quark mass measurements from the cdf and d experiments at fermilab .
we combine published ( 19921996 ) measurements with the most precise published and preliminary ( 20012011 ) measurements based on data sets corresponding to up to @xmath0 of @xmath1 collisions . taking correlations of uncertainties into account , and combining the statistical and systematic uncertainties , the resulting preliminary tevatron average mass of the top quark is @xmath2 , corresponding to a relative precision of @xmath3 .
fermi national accelerator laboratory fermilab - conf-13 - 164-ppd - td + tevewwg / top 2013/01 + cdf note 10976 + d note 6381 + august 2013 + * combination of cdf and d results on the mass of the top quark using up to at the tevatron + * | this note reports the tevatron average top - quark mass obtained by combining the most precise published and preliminary measurements of the top - quark mass .
it is an update of the combination presented in ref .
@xcite , where further details can be found .
the atlas and cms collaborations have also performed a combination of their most recent top quark mass measurements @xcite .
the cdf and d collaborations have performed several direct experimental measurements of the top - quark mass ( @xmath4 ) using data collected at the tevatron proton - antiproton collider located at the fermi national accelerator laboratory .
these pioneering measurements were first based on approximately @xmath5 of data @xcite-@xcite collected from 1992 to 1996 , and included results from the decay channels @xmath6 ( alljets ) , @xmath7 ( @xmath8+jets ) , and @xmath9 ( @xmath10 ) , where @xmath11 or @xmath12 .
decays with @xmath13 are included in the direct @xmath14 and @xmath15 channels . in ( 20012011 ) , many top mass measurements have been performed , and those considered here are the most recent results in these channels , using up to @xmath0 of data for cdf ( corresponding to the full cdf run ii dataset ) @xcite , and up to @xmath16 of data for d @xcite . the cdf analysis based upon charged particle tracking for exploiting the transverse decay length of @xmath17-tagged jets ( @xmath18 ) and the transverse momentum of electrons and muons from @xmath19 boson decays ( @xmath20 ) uses a data set corresponding to a luminosity of 1.9 @xmath21 @xcite , and there are no plans to update this analysis .
the d measurements presented in this note include the most recent run ii measurement in the @xmath10 @xcite channel using 5.4 fb@xmath22 of data and in the @xmath8+jets channel @xcite with 3.6 fb@xmath22 of data .
both results are now published .
since the combination performed in 2011 @xcite , a new final state signature was introduced by cdf that requires events to possess missing transverse energy ( @xmath23 ) and jets , but no identified lepton ( `` met '' ) @xcite .
this sample is statistically independent from the previous three cdf channels . with respect to the july 2011 combination@xcite and the published version of the combination @xcite
, the cdf measurement in the @xmath8+jets channel has been updated using 8.7 fb@xmath22 of data , an improved analysis technique , and improved jet energy resolution @xcite .
the cdf measurement in the met channel was updated to use the full data set for cdf of 8.7 fb@xmath22 of data as well @xcite .
the now published cdf measurements in the @xmath10 channel@xcite and alljets channel@xcite are unchanged .
the measurement based on charged particle tracking @xcite was incorporated as described in the past combinations @xcite . from the corresponding analysis
only the measurement of the top quark mass using the mean decay length @xmath24 of @xmath25 hadrons in @xmath17-tagged lepton+jets events has been used .
it is independent of energy information in the calorimeter , and its main source of systematic uncertainty is uncorrelated with the dominant ones from the jet energy scale calibration in other measurements .
this measurement of @xmath26 is essentially uncorrelated with the higher precision cdf result from the lepton+jets channel .
the overlap between the data samples used for the decay - length method and the lepton+jets sample has therefore no effect .
the tevatron average top - quark mass is obtained by combining five published measurements @xcite with four published cdf results @xcite , one preliminary cdf result @xcite , and two published d results @xcite .
this combination supersedes previous combinations .
the definition and evaluation of the systematic uncertainties and the understanding of the correlations among channels , experiments , and tevatron runs is the outcome of many years of joint work between the cdf and d collaborations and is described in detail elsewhere @xcite .
the input measurements and uncertainty categories used in the combination are detailed in sections [ sec : inputs ] and [ sec : uncertainty ] , respectively .
the correlations assumed in the combination are discussed in section [ sec : corltns ] and the resulting tevatron average top - quark mass is given in section [ sec : results ] .
a summary is presented in section [ sec : summary ] .
twelve measurements of @xmath4 used in this combination are shown in table [ tab : inputs ] .
the measurements all have relatively large statistical uncertainties and their systematic uncertainties are dominated by the total jet energy scale ( jes ) uncertainty . in both cdf and d
take advantage of the larger @xmath27 samples available and employ new analysis techniques to reduce both of these uncertainties .
in particular , the d analysis in the @xmath8+jets channel and the cdf analyses in the @xmath8+jets , alljets , and met channels constrain the response of light - quark jets using the kinematic information from @xmath28 decays ( so - called _ in situ _ calibration ) @xcite .
residual jes uncertainties associated with @xmath29 and @xmath30 dependencies as well as uncertainties specific to the response of @xmath17 jets are treated separately .
the d @xmath10 measurement uses the jes determined in the @xmath8+jets channel by _ in situ _ calibration @xcite .
.summary of the measurements used to determine the tevatron average @xmath4 .
integrated luminosity ( @xmath31 ) has units of @xmath32 , and all other numbers are in @xmath33 .
the uncertainty categories and their correlations are described in section [ sec : uncertainty ] .
the total systematic uncertainty and the total uncertainty are obtained by adding the relevant contributions in quadrature .
`` n / a '' stands for `` not applicable '' , `` n / e '' for `` not evaluated '' . [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^ " , ] [ tab : three_observables ] to test the influence of the choices in modeling the correlations , we performed a cross - check by changing all non - diagonal correlation coefficients of the correlation matrix defined in section [ sec : corltns ] from 100% to 50% and re - evaluated the combination .
the result of this large variation of degree of correlation is a @xmath34 shift of the top - quark mass and reduces the total uncertainty negligibly .
the chosen approach is therefore conservative .
we also performed separate combinations of all the cdf and d measurements .
the results of these combinations are @xmath35 for cdf and @xmath36 for d . taking all correlations into account
, we calculate the chi - square value @xmath37 corresponding to a probability of 13% .
an update of the combination of measurements of the mass of the top quark from the tevatron experiments cdf and d has been presented .
this preliminary combination includes five published measurements , six published measurements , and one preliminary measurement , but the majority of these measurements are not yet performed on the full datasets available . taking into account the statistical and systematic uncertainties and their correlations , the preliminary result for the tevatron average is @xmath38 , where the total uncertainty is obtained assuming gaussian systematic uncertainties .
the central value is 0.02gev/@xmath39 higher than our july 2012 average @xcite of @xmath40gev/@xmath39 . adding in quadrature the statistical and systematic uncertainties yields a total uncertainty of @xmath41 which represents an improvement of @xmath42 .
the mass of the top quark is now known with a relative precision of 0.50% , limited by the systematic uncertainties , which are dominated by the jet energy scale uncertainty .
this result will be further improved when all analysis channels from cdf and d using the full run ii data set are finalized .
we thank the fermilab staff and the technical staffs of the participating institutions for their vital contributions .
this work was supported by doe and nsf ( usa ) , conicet and ubacyt ( argentina ) , cnpq , faperj , fapesp and fundunesp ( brazil ) , crc program , cfi , nserc and westgrid project ( canada ) , cas and cnsf ( china ) , colciencias ( colombia ) , msmt and gacr ( czech republic ) , academy of finland ( finland ) , cea and cnrs / in2p3 ( france ) , bmbf and dfg ( germany ) , ministry of education , culture , sports , science and technology ( japan ) , world class university program , national research foundation ( korea ) , krf and kosef ( korea ) , dae and dst ( india ) , sfi ( ireland ) , infn ( italy ) , conacyt ( mexico ) , nsc(republic of china ) , fasi , rosatom and rfbr ( russia ) , slovak r&d agency ( slovakia ) , ministerio de ciencia e innovacin , and programa consolider - ingenio 2010 ( spain ) , the swedish research council ( sweden ) , swiss national science foundation ( switzerland ) , fom ( the netherlands ) , stfc and the royal society ( uk ) , and the a.p .
sloan foundation ( usa ) .
b. abbott _
et al . _
[ d collaboration ] , phys . rev . *
d60 * , 052001 ( 1999 ) .
f. abe _ et al . _
[ cdf collaboration ] , phys .
80 * , 2767 ( 1998 ) .
t. affolder _ et al . _
[ cdf collaboration ] , phys . rev . *
d63*. s. abachi _
et al . _
[ d collaboration ] , phys . rev
. lett . * 79 * , 1197 ( 1997 ) .
b. abbott _
et al . _
[ d collaboration ] , phys . rev . *
d58 * , 052001 ( 1998 ) .
v. m. abazov _ et al . _
[ d collaboration ] , nature * 429 * , 638 ( 2004 ) .
f. abe _ et al . _
[ cdf collaboration ] , phys .
lett . * 79 * , 1992 ( 1997 ) .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
lett . * 109 * , 152003 ( 2012 ) .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
d * 81 * ( 2010 ) 032002 , phys .
d * 83 * , 111101 ( 2011 ) .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
b * 714 * , 24 ( 2012 ) .
v. m. abazov _ et al . _
[ d collaboration ] , phys .
* 101 * , 182001 ( 2008 ) . v.
m. abazov _ et al . _
[ d collaboration ] , phys .
d * 84 * , 032004 ( 2011 ) .
v. m. abazov _ et al .
_ [ d collaboration ] , phys .
d * 86 * , 051103 ( 2012 ) .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
d * 81 * , 032002 ( 2010 ) .
, arxiv:1107.5255 .
s. frixione and b. webber , jhep * 029 * , 0206 ( 2002 ) .
m. l. mangano , m. moretti , f. piccinini , r. pittau , and a. d. polosa , jhep * 07 * , 001 ( 2003 ) g. marchesini _ et al .
_ , comput .
phys . commun . *
67 * , 465 ( 1992 ) . | note reports tevatron average top - quark mass combining precise preliminary measurements update combination ref @xcite further details atlas cms collaborations performed recent top quark mass measurements @xcite cdf d collaborations performed direct measurements top - quark mass using data tevatron proton - antiproton collider fermi national accelerator laboratory measurements based on @xmath5 data @xcite 1992 to 1996 included results decay channels @xmath6 @xmath7 decays with @xmath13 included in @xmath14 @xmath15 channels 20012011 top mass measurements performed recent results channels using @xmath0 data for cdf @xcite @xmath16 for d @xcite cdf analysis charged particle tracking transverse decay length of @xmath17-tagged jets transverse momentum electrons muons from @xmath19 boson decays uses data set luminosity 1.9 @xmath21 @xcite no plans to update d measurements include recent run ii measurement @xmath10 @xcite channel using 5.4 fb@xmath22 data @xmath8+jets channel @xcite with 3.6 fb@xmath22 dataresults published combination 2011 new final state signature introduced cdf requires events missing transverse energy jets no lepton sample statistically independent from previous three cdf channels july 2011 combination cdf measurement @xmath8+jets channel updated using 8.7 fb@xmath22 data improved analysis technique improved jet energy resolution cdf measurement met channel updated full data set cdf 8.7 fb@xmath22 cdf measurements @xmath10 alljets channel unchanged measurement charged particle tracking incorporated past combinations measurement top quark mass using mean decay length @xmath24 @xmath25 hadrons in @xmath17-tagged lepton+jets events used independent of energy information calorimeter uncorrelated with jet energy scale calibration measurement @xmath26 uncorrelated with higher precision cdf result from lepton+jets channel overlap between data samples decay - length method lepton+jets sample no effect tevatron average top - quark mass obtained combining five measurements four cdf results one preliminary cdf result two d results combination supersedes previous combinationsdefinition evaluation of systematic uncertainties understanding correlations among channels experiments tevatron runs outcome of joint work cdf d collaborations described input measurements uncertainty categories detailed in sections inputs uncertainty correlations discussed in tevatron average top - quark mass in results ] summary in section [ : summary ] twelve measurements of @xmath4 shown in table inputs measurements have large statistical uncertainties dominated by total jet energy scale jes ) uncertainty cdf d larger @xmath27 samples employ new analysis techniques reduce uncertainties d analysis @xmath8+jets cdf analyses constrain response light - quark jets using information from @xmath28 decays in situ calibration residual jes uncertainties @xmath29 @xmath30 dependencies response @xmath17 jets treated separately d @xmath10 measurement uses jes determined @xmath8+jets channel.summary of measurements tevatron average @xmath4 integrated luminosity ( @xmath31 ) has units of @xmath32 other numbers in @xmath33 uncertainty categories correlations described in section [ sec : uncertainty ]total systematic uncertainty obtained adding contributions in quadrature n / a for not applicable n / e not evaluated three_observables test influence choices modeling correlations cross - check non - diagonal correlation coefficients from 100% to 50% re - evaluated combination large variation is @xmath34 shift top - quark mass reduces total uncertainty negligibly chosen approach conservative performed separate combinations of cdf d measurements results are @xmath35 for cdf @xmath36 for d chi - value @xmath37 probability 13% update of combination measurements top quark from tevatron experiments d presented preliminary combination includes five published six published one preliminary measurement majority not performed on full datasets uncertainties preliminary result tevatron average is @xmath38 total uncertainty obtained assuming gaussian systematic uncertainties central value 0.02gev/@xmath39 higher than july 2012 average @xmath40gev/@xmath39 adding quadrature statistical systematic uncertainties yields total uncertainty @xmath41 improvement of @xmath42mass top quark known precision 0.50% limited systematic uncertainties dominated jet energy scale uncertainty result improved analysis channels cdf d full run ii data set finalized thank fermilab staff technical staffs participating institutions contributions work supported doe nsf usa conicet ubacyt argentina cnpq faperj fapesp fundunesp brazil crc program cfi nserc westgrid project canada cas cnsf china colciencias colombia msmt gacr academy finland cea cnrs in2p3 france bmbf dfg germany ministry education culture sports science technology japan world class university program national research foundation korea krf kosef dae dst india sfi ireland infn italy conacyt mexico nsc china fasi rosatom rfbr russia slovak r&d agency ministerio ciencia innovacin programa consolider - ingenio 2010 spain swedish research council swiss national science foundation fom netherlands stfc royal society a.p sloan foundation usa abbott et al collaboration . rev. d60 052001 1999 abe alcdf collaboration phys . 80 * 2767 ( 1998 ) affolder phys . rev d63* abachi collaboration phys . rev 79 * 1197 ( 1997 ). abbott et al collaboration phys . rev d58 052001 ( 1998 ) abazov collaboration nature * 429 * 638 ( 2004 ). abe cdf collaboration phys . 79 * 1992 ( 1997 ) t. aaltonen al cdf collaboration phys . 109 * 152003 ( 2012 ) phys . d 81 * ( 2010 ) 032002 . d 83 * 111101 ( 2011 ) collaboration . 714 * 24 ( 2012 ) abazov collaboration . 101 * 182001 ( 2008 ) abazov collaboration phys . d * 84 * 032004 ( 2011 ) abazov collaboration phys . d * 86 * 051103 ( 2012 ) aaltonen collaboration phys . d * 81 * 032002 ( 2010 ) arxiv:1107.5255 frixione b. webber jhep * 029 * 0206 ( 2002 )mangano moretti piccinini pittau polosa * 07 * 001 ( 2003 ) marchesini comput. . commun 67 * 465 ( 1992 ) | 0.431883 | 2,830 | 1,396 | , Saving $0.1 in GPT-4. |
177,695 | motivated by recent experiments , we calculate both the average current and the current fluctuations for a metallic island which oscillates between two symmetric electrodes .
electrons can only tunnel on or off the island when it is close to one of the electrodes . using a master equation
we investigate the accuracy of such an electron shuttle both analytically and numerically .
it is shown that optimum operation is reached when the contact time is much larger than the @xmath0-time . | one of the most exciting prospects of coulomb blockade is the realization of a current standard based on periodically transferring single electrons through the device @xcite . in order to minimize higher order tunnelling processes which limit the accuracy ,
a chain of tunnelling junctions was used @xcite .
quite recently , a single electron shuttle mechanism which is based on mechanical oscillations between two electrodes was suggested by gorelik _
the major advantage of a mechanically driven single electron transfer is that higher order processes are negligible . in the original suggestion of a mechanical single electron shuttle a metallic island
is connected via organic molecules to two electrodes .
one of the electrodes is positive and the other is negatively biased . for mechanically soft molecules self - excitation
is expected to internally drive oscillations of the island which transfers charge at frequency @xmath1 .
because of coulomb blockade an integer number of electrons is loaded onto the grain close to one turning point , and the same number of electrons is unloaded close to the other . the number of electrons transferred per cycle is thus expected to be @xmath2\ , , \label{eq : n_average}\ ] ] where @xmath3 is the applied voltage and @xmath4 the island capacitance relative to the electrodes .
the average current is given by @xmath5 .
if the island is situated between the two electrodes there should be either @xmath6 or @xmath7 electrons on it depending on which electrode it is coming from .
a first step towards an experimental realization of a mechanical single electron shuttle was undertaken by erbe _
the authors designed a semiconductor device where a mechanical clapper oscillates between two electrodes - _ i.e. _ the oscillation is now externally driven with the average number of electrons transferred in each cycle of order @xmath8 @xcite .
this experiment which works similar to a bell is still in the classical regime , _
i.e. _ coulomb blockade is not relevant here .
the aim in a future experimental setup with a metallic island on the tip of the clapper is to obtain an electron shuttle mechanism which is limited to single electrons by coulomb repulsion .
since the tunnelling resistance @xmath9 increases exponentially with the distance @xmath10 between the island and the electrodes ( @xmath11 ) , tunnelling is suppressed exponentially when the island is in the middle between the two electrodes ( both the oscillation amplitude and the distance between the two electrodes is much larger than @xmath12 ) .
for the same reason both cotunnelling and simultaneous tunnelling between the island and the electrodes can be neglected . in
order for electrons to tunnel the amplitude of the oscillation has to be large compared with the typical tunnelling distance @xmath12 . in the experiments by erbe _
the oscillation amplitude is of order 100 nm ( compared to a few nanometers as suggested by gorelik _
et al . _ ) and thus much larger than @xmath13 . in our present work
we investigate the accuracy of a mechanically driven single electron shuttle using a master equation approach @xcite .
the master equation is solved numerically and also analytically in a simplified form . in particular , we calculate both the average number of electrons transferred per period which in general is different from the simple result ( [ eq : n_average ] ) and the root mean square fluctuations @xmath14 . as one of our main results , it turns out that optimum operation with @xmath15 is obtained for long contact times ( @xmath16 ) and sufficiently low temperatures ( @xmath17 ) .
in order to calculate the probability @xmath18 to find @xmath19 additional electrons on the island at time @xmath20 we use a master equation . this technique has been successfully used to describe both quantum dots and metallic islands exhibiting coulomb blockade @xcite-@xcite . in our case the island oscillates mechanically which leads to time dependent transition rates @xmath21 at the leads . collecting gain and loss terms
the master equation reads @xmath22p(m , t ) \nonumber\nopagebreak\\ & + & \left[\gamma^{(+)}_{\rm l}(m-1,t)+\gamma^{(+)}_{\rm r}(m-1,t)\right]p(m-1,t)\nonumber\nopagebreak\\ & + & \left[\gamma^{(-)}_{\rm l}(m+1,t)+\gamma^{(-)}_{\rm r}(m+1,t)\right]p(m+1,t)\ , .
\label{eq : mastereq}\end{aligned}\ ] ] in a golden rule approach the tunnelling rates are of the form @xcite @xmath23 since both @xmath9 and the capacitance @xmath4 in @xmath24 are time dependent now , with the dominant time dependence associated with the exponentially varying @xmath25 , the tunnelling rates factorize into : @xmath26 here @xmath27 is a strongly varying function of time , dominated by the exponential decrease of tunnelling when the island moves away from the electrode ( @xmath28 is the time where the island is at its closest point to the right electrode ) .
the motion is taken as purely harmonic , _
i.e. _ @xmath29 . for the rates @xmath30
whose time dependence is determined by that of the capacitance only , we take the standard result for tunnelling rates when a voltage of @xmath31 is applied at the left electrode and a voltage of @xmath32 is applied at the right electrode : @xmath33}\ ] ] where @xmath34 and @xmath35 . for the left electrode
the indices @xmath9 and @xmath36 have to be interchanged , @xmath3 has to be replaced by @xmath37 and @xmath38 by @xmath39 . for the numerical calculations the full master equation was used . for simplicity
, we assumed that the capacitance depends linearly on separation like @xmath40 .
the function @xmath41 is of the form @xmath42\ ] ] which is sharply peaked around @xmath28 . since tunnelling through the left and right electrodes
does not take place simultaneously ( @xmath43 ) , the tunnelling at the left or the right electrodes can be treated separately . for the analytic calculations we consider only one electrode : @xmath44 which is approached by an island with @xmath45 electrons from the left , _
i.e. _ we are calculating the conditional probability @xmath46 to find @xmath19 electrons on the island given that initially there where @xmath45 electrons on the island .
it has to be noted that when the island is close to the right electrode , @xmath47 changes much slower than @xmath48 - independent of the precise model for the capacitance .
as long as the applied voltage @xmath3 is not too large ( _ i.e. _ as long as @xmath49 is of order @xmath50 , @xmath51 , ) @xmath47 can be replaced by @xmath52 . in order to be able to solve the master equations analytically
, we replace the function @xmath53 by a step function @xmath54 having the same height and area : @xmath55 the width @xmath56 corresponds to an effective contact time .
as expected , the contact time decreases for increasing oscillation frequencies ( @xmath57 , @xmath58 ) .
the solutions @xmath46 of the master equation can be used to calculate the probability @xmath59 that @xmath19 electrons are transferred : @xmath60 where @xmath61 is the initial probability to find @xmath62 additional electrons on the island ( @xmath63 ) . to calculate this probability
, we consider a symmetric situation where the probability for the island to approach the right electrode with @xmath45 electrons is the same as the probability for the island to approach the left electrode with @xmath64 electrons : @xmath65 where @xmath66 this leads to a set of linear equations for @xmath67 which in principle can be solved for arbitrary values of @xmath6 and @xmath68 .
however , for practical purposes , small values for @xmath6 should be used . from @xmath69
we then calculate both @xmath70 and @xmath71 ( @xmath72 ) .
for very low temperatures @xmath73 coulomb blockade fixes the number of additional electrons @xmath45 on the island between @xmath7 and @xmath6 strictly .
we will focus on the behaviour of the single electron shuttle for small integer values of @xmath6 ( @xmath74 ; @xmath75 ) . with the approximation described in eq .
( [ eq : gres ] ) the master equation at @xmath76 simplifies to : @xmath77 with @xmath78 we solve eq .
( [ eq : master_tilde ] ) for @xmath79 : @xmath80^k \label{eq : p_t0}\ ] ] where @xmath81 ( @xmath82 ) . eq .
( [ eq : p_t0 ] ) can easily be proved by induction using the fact that @xmath83 . for @xmath84
, the normalization condition leads to the following result : @xmath85 . together with eqs .
( [ eq : p_d_def ] ) and ( [ eq : p_f_eq ] ) this can be used to calculate the current and the current fluctuations for @xmath86 .
the results obtained from the analytical expression ( [ eq : p_t0 ] ) and also from the numerical solution of the full master equation are shown in fig .
[ fig : coulombblockade ] .
coulomb blockade is clearly visible : for low voltages no electrons are transferred . above a critical voltage of @xmath87
the shuttle starts to work . for @xmath88 , @xmath89 the analytic expressions for the average number of electrons transferred ( @xmath90 ) and the mean square fluctuations ( @xmath91 ) are comparatively simple : @xmath92\left[1+\frac 12a+a^2\right ] } \label{eq : nmeant0 } \\ ( \delta n)^2 & = & 2a(1-a)\frac{6 + 9a+22a^2 + 13a^3 + 10a^4}{\left[2a^2+a+2\right]^2\left[a+1\right]^2}\end{aligned}\ ] ] where @xmath93 as expected , in the limit of very short contact times @xmath94 ( @xmath95 ) ,
no electrons are transferred whereas very long contact times @xmath96 ( @xmath97 ) lead to the simple result ( [ eq : n_average ] ) .
the fluctuations disappear for both short and long contact times .
their maximum ( @xmath98 ) is reached for @xmath99 as shown in fig .
[ fig : nt0 ] .
for finite temperatures the number of additional electrons on the island is no longer as restricted as for @xmath76 . here
, the master equation has the more complicated form : @xmath100 \nonumber\\ + \frac { g(t)}{\tau } \frac{k+\frac { 1}{2}+\delta}{\left(1-\varepsilon^{2k+1}\xi\right ) } \left[p(-n_{\rm max}\!+\!k\!+\!1|n , t ) -{\varepsilon^{2k+1}}\xi p(-n_{\rm max}\!+\!k|n , t)\right]\,,\end{aligned}\ ] ] where @xmath101 and @xmath102 . for @xmath103 and constant @xmath104 ( _ i.e. _ constant @xmath105 and @xmath106 ) the stationary solution ( @xmath107 , which is independent of @xmath45 ( @xmath108 ) is given by : @xmath109 with the normalization constant @xmath110 together with eq .
( [ eq : p_d_def ] ) this leads to : @xmath111 in reality however , @xmath112 is time - dependent and thus the results ( [ eq : p_ft ] ) and ( [ eq : p_deltat ] ) are not exact .
it turns out however , that if one replaces @xmath112 by @xmath113 where @xmath114 is defined by @xmath115 one obtains good agreement between numerical results obtained by the solution of the time dependent problem and the analytic result ( [ eq : p_ft ] ) .
( changes which happen on time - scales much smaller than @xmath116 , _
i.e. _ effective time - scales much smaller than @xmath117 , are too fast to be followed , cf . fig .
[ fig : nt0 ] . ) fig .
[ fig : coulombtg0 ] shows two steps of the symmetric resulting coulomb staircase for several temperatures . in the middle of one of the steps of the coulomb staircase ( @xmath118 )
the average number of electrons transferred per period agrees with the simple result @xmath119 even for @xmath120 .
the root mean square fluctuations are approximately given by @xmath121 as expected , the fluctuations disappear exponentially for decreasing temperatures .
using a simple master equation we have derived analytic expressions for both the average current and its root mean square fluctuations in a mechanically driven single electron shuttle . we find a good agreement between our analytical results and a numerical solution of the fully time - dependent master equation . to minimize fluctuations
the shuttle should be operated under conditions where the interaction time @xmath68 is much greater than the @xmath0-time @xmath117 ( @xmath122 ) and the temperature @xmath123 is much lower than the coulomb energy @xmath124 . | exciting prospects of coulomb blockade realization current standard transferring electrons through minimize tunnelling processes chain of tunnelling junctions used single electron shuttle mechanism on mechanical oscillations between two electrodes suggested by gorelik advantage mechanically driven transfer higher order processes negligible original metallic island connected via organic molecules to two electrodes one positive other negatively biased soft molecules self - excitation drive oscillations island charge frequency coulomb blockade integer number electrons loaded onto grain close one turning point same unloaded other number of electrons transferred per cycle expected @xmath2_average @xmath3 applied voltage @xmath4 island capacitance to electrodes average current given by @xmath5 if island between two electrodes should @xmath6 or @xmath7 electrons on depending electrode first step towards mechanical single electron shuttle by erbe designed semiconductor device mechanical clapper oscillates between two electrodes externally driven with average number of electrons transferred in each cycle experiment in classical regime coulomb blockade not relevantaim future setup with metallic island on tip clapper obtain electron shuttle mechanism limited to single electrons by coulomb repulsion tunnelling resistance @xmath9 increases with distance between island electrodes tunnelling suppressed when island middle between two electrodes oscillation amplitude distance electrodes larger than @xmath12 cotunnelling simultaneous tunnelling between island electrodes can be neglected for electrons to tunnel amplitude oscillation large compared typical tunnelling distance experiments erbe oscillation amplitude 100 nm gorelik larger than @xmath13 present work investigate accuracy of mechanically driven single electron shuttle using master equation approach master equation solved numerically analytically calculate average number of electrons transferred per period root mean square fluctuations @xmath14 optimum operation with @xmath15 for long contact times low temperatures to calculate probability to find additional electrons on island at time use master equation technique used to describe quantum dots metallic islands exhibiting coulomb blockade island oscillates mechanically leads to time dependent transition rates @xmath21 at leadsgain loss terms master equation reads @xmath22p(m t )\gamma(+)\rm l}(m-1,t)+ r}(m-1,t]p(m-1,t)\gamma{-\rm l}(m+1,t)+ r},t]p(m+1,t){eq mastereq}{aligned golden rule approach tunnelling rates form @xcite @xmath23 @xmath9 capacitance @xmath4 @xmath24 time dependent dominant time dependence varying @xmath25 tunnelling rates factorize into @xmath26 @xmath27 varying function time decrease tunnelling when island moves away from electrode @xmath28 island closest right electrode motion harmonic @xmath29 rates @xmath30 time dependence determined capacitance standard result tunnelling rates voltage @xmath31 left electrode @xmath32 right electrode @xmath33 @xmath34 @xmath35 left electrode indices @xmath9 @xmath36 interchanged @xmath3 replaced by @xmath37 @xmath38 @xmath39 numerical calculations full master equation usedassumed capacitance depends on separation like @xmath40 function @xmath41 form @xmath42 peaked around @xmath28 tunnelling through left and right electrodes simultaneously treated separately analytic calculations consider one electrode : @xmath44 approached by island with @xmath45 electrons left calculating conditional probability @xmath46 to find @xmath19 electrons on island island close to right electrode @xmath47 changes slower than @xmath48 independent of model capacitance applied voltage @xmath3 not large @xmath49 of order @xmath47 replaced by @xmath52 replace function @xmath53 by step function @xmath54 same height and area width @xmath56 corresponds to effective contact time contact time decreases for increasing oscillation frequencies solutions @xmath46 of master equation used to calculate probability @xmath59 that @xmath19 electrons transferred @xmath60 @xmath61 initial probability to find @xmath62 additional electrons on islandcalculate probability consider symmetric situation island approach right electrode with @xmath45 electrons same as left @xmath64 electrons @xmath65 @xmath66 leads linear equations for @xmath67 solved for arbitrary values @xmath6 @xmath68 small values for @xmath6 used from @xmath69 calculate @xmath70 @xmath71 low temperatures @xmath73 coulomb blockade fixes additional electrons @xmath45 island between @xmath7 @xmath6 focus on behaviour single electron shuttle for small integer values of @xmath6 @xmath74 ; @xmath75 ) approximation eq master equation at @xmath76 simplifies to @xmath77 @xmath78 solve eq master_tilde @xmath79 @xmath80^k : p_t0 @xmath81 @xmath82 eq p_t0 proved by induction @xmath83 @xmath84 normalization condition leads to result : @xmath85 p_d_def_f_eq calculate current fluctuations for @xmath86 results from analytical expression eq : p_t0 numerical solution full master equation shown in fig . coulombblockadecoulomb blockade visible low voltages no electrons transferred above critical voltage @xmath87 shuttle starts @xmath88 @xmath89 analytic expressions average electrons transferred mean square fluctuations simple @xmath92\left[1+ 12a+a^2\right{eq : nmeant0\delta n^2 2a(1-a)\frac{6 + 9a+22a^2 + 13a^3 + 10a^4}[2a^2+a+2\right][a+1\right]^2{aligned @xmath93 short contact times @xmath94 no electrons transferred long contact times @xmath96 @xmath97 ) simple result eq n_average fluctuations disappear short long times maximum reached @xmath99 fig . finite temperatures additional electrons restricted @xmath76master equation complicated form : @xmath100 \nonumber\\ + \frac { g(t)}{\tau } \frac{k+\frac { 1}{2}+\delta}{\left(1-\varepsilon^{2k+1}\xi\right ) } \left[p(-n_{\rm max}+\!k!1|n , t ) -{\varepsilon^{2k+1}}\xi p(-n max}+!k|n t)\right]\end{aligned} @xmath101 and @xmath102 for @xmath103 @xmath104 stationary solution @xmath107 independent of @xmath45 @xmath108 given by @xmath109 with normalization constant @xmath110 p_d_def leads to @xmath111 @xmath112 time - dependent results_ft not exact @xmath112 by @xmath113 @xmath114 defined by @xmath115 agreement between numerical results analytic result eq : p_ft ] changes on time - scales smaller than @xmath116 too fast fig . shows two steps symmetric coulomb staircase for several temperaturescoulomb staircase @xmath118 average electrons transferred period agrees with result @xmath119 @xmath120 root mean square fluctuations given by @xmath121 disappear decreasing temperatures derived analytic expressions average current root square fluctuations mechanically driven single electron shuttle agreement analytical results numerical solution time - dependent equation minimize fluctuations shuttle interaction time @xmath68 greater than @xmath0-time @xmath117 temperature @xmath123 lower than coulomb energy @xmath124 | 0.494827 | 2,980 | 1,541 | , Saving $0.1 in GPT-4. |
177,696 | a trend in high performance computers that is becoming increasingly popular is the use of symmetric multiprocessing ( smp ) rather than the older paradigm of mpp .
mpi codes that ran and scaled well on mpp machines can often be run on an smp machine using the vendor s version of mpi .
however , this approach may not make optimal use of the ( expensive ) smp hardware .
more significantly , there are machines like blue horizon , an ibm sp with 8-way smp nodes at the san diego supercomputer center that can only support 4 mpi processes per node ( with the current switch ) . on such a machine
it is imperative to be able to use openmp parallelism on the node , and mpi between nodes .
we describe the challenges of converting milc mpi code to using a second level of openmp parallelism , and benchmarks on ibm and sun computers . | [ cols="<,<,<",options="header " , ] open multiprocessing ( openmp or omp @xcite ) and message passing interface ( mpi ) are two strategies for using multiple processors for a single problem .
the key difference between them is that in mpi , different nodes have their own memory and they communicate with each other when needed ; but with openmp , the memory is shared between threads .
here is an example .
suppose we have a two dimensional lattice with 4 sites in each direction , and we are using four nodes or threads , as shown below .
+ 1in@cc@cc 1&2&3&4 + 5&6&7&8 + + 9&10&11&12 + 13&14&15&16
+ suppose node / thread 1 corresponds to sites 1 , 2 , 5 and 6 .
in mpi , node 1 has information only about sites on that node , namely 1 , 2 , 5 and 6 .
if it needs information about other sites , for example about sites 3 or 7 which are nearest neighbors of sites 2 and 6 respectively , it has to use communication routines .
contrast this with openmp , where all threads have access to data for all sites , but thread 1 does computations only for sites 1 , 2 , 5 and 6 .
table 1 summarizes the differences between the two strategies . a trend towards shared memory parallel machines or clusters of symmetric multiprocessing ( smp ) nodes rather than the older paradigm of massively parallel processing ( mpp ) machines makes a study of openmp parallelism timely .
openmp was designed to exploit certain characteristics of shared - memory architectures .
the ability to directly access memory throughout the system ( with minimum latency and no explicit address mapping ) combined with very fast shared memory locks , makes shared - memory architectures best suited for supporting openmp .
the advantage of openmp is that it is easier to program . unlike mpi
, one does not have to worry about passing messages between nodes . in this paper , we study how openmp performs relative to mpi , and whether combining the two strategies gives better performance .
in this section , we give some details about how a c code that works for a single processor is changed to work on multiple threads .
the number of threads is determined by an environment variable , ` omp_num_threads ` .
the code is executed serially , on a single thread , until a parallel construct is encountered , which is executed on multiple threads and then serial execution is resumed . to define a parallel construct ,
lines beginning with ` # pragma omp ` are added to the code .
such pragmas are ignored by the usual c compiler , so the code may also be run as an ordinary serial code .
they are , however , interpreted by an openmp compiler to identify parallel regions .
there are several constructs that can be made to execute in parallel ; here is an example for a `` ` for ` '' construct .
+ ` # pragma omp parallel for for(i=0;i <
n;i++ ) { my_job(i ) ; } ` this will run the function ` my_job ` in parallel on different threads .
note that though the memory is shared , each thread must have a private copy of some variables , like ` i ` in the above example .
loop variables are made private by default but other such variables have to be declared `` ` private ` '' .
some variables may need to be summed over all the sites .
this is accomplished with a ` reduction ` statement .
the syntax is as follows : + + ` j=0 ; ` + ` # pragma omp parallel for reduction(+:j ) ` + ` for(i=0;i <
n;i++ ) { j+=my_function(i ) ; } ` this is equivalent to + @xmath0 .
+ + note that the sum is performed over all threads though each thread works only on part of the total number of iterations .
identifying ` private ` and ` reduction ` variables is necessary for getting correct results .
the milc @xcite code is a set of publicly available codes developed by the mimd lattice computation ( milc ) collaboration for doing qcd simulations .
this code has been run on a variety of parallel computers , using mpi , for many physics projects .
the files are organized in different directories the ` libraries ` directory contains low level routines like matrix multiplication , the ` generic ` directory contains oft - needed but somewhat higher level routines , including the updating and inversion routines .
then there are various application directories .
for this project , we only concentrated on the conjugate gradient inverter , file ` d_congrad5.c ` in ` generic_ks ` directory in version 6 of milc code .
the code uses a macro ` forallsites ` defined as + ` # define forallsites(i , s)\backslash for(i=0,s = lattice;i
< sites_on_node;\backslash i++,s++ ) ` + where ` lattice ` is an array of sites , ` site ` is a structure containing variables defined at each lattice point and ` sites_on_node ` is the number of lattice points on a given node .
we needed to redefine the macro ` forallsites ` because the openmp compiler we used could not deal with two variables ( ` i ` and ` s ` ) in a parallel ` for ` statement .
here is the macro redefinition .
+ ` # define forallsites(i , s)\backslash for(i=0;i < sites_on_node;i++){\backslash s=&(lattice[i ] ) ; ` we used another macro ` end_loop ` , which is just defined to be a closing brace ` } ` to match the opening brace in the above macro .
we used the kap / pro toolset @xcite for this project .
it includes the following : * ` guidec ` : openmp compiler for c. * ` guideview ` : openmp parallel performance visualization tool .
it gives details of program execution , in particular , time spent in serial and parallel execution , imbalance in different regions of the code , etc .
* ` assurec ` : compiler to be used with debugger which works by comparing single thread and multiple thread executions . * ` assureview ` : openmp programming correctness tool for viewing details of errors or conflicts which occur if different threads try to read / write the same variables at the same time .
to add openmp parallelism to the milc code , the following steps were required .
first , we had to redefine the macro as explained above .
then we added the parallel for pragmas , specifying ` private ` and ` reduction ` variables .
for example , in the ` forallsites ` loop , ` s ` was made ` private ` .
we changed ` cc ` to ` guidec ` in our makefiles , and we had to modify those compiler options that ` guidec ` did not recognize . adding ` backend ` before a compiler option forces ` guidec ` to use ` cc ` compiler options .
then , we ran ` assurec ` and ` assureview ` to locate and remove conflicts .
finally , we ran the executable on different number of threads and verified that the output agreed with the mpi output . even after one has a working openmp code , there are some issues to consider when comparing its performance with that of mpi .
some performance problems are openmp issues , while others are not .
if single thread openmp performance does not match that with a single node under mpi , that may indicate a culprit other than openmp .
for example , thread safe compilation requires the ` -mt ` switch on sun .
if using this switch on the original serial code decreases performance substantially , then the performance issue lies with the sun compiler and its runtime libraries , not with openmp . if the code uses many ` malloc / free ` pairs then thread - safe memory allocation is likely the culprit .
again , this is not an openmp issue , but an issue with the quality of the vendor s thread - safe compiler / runtime implementation .
we verified that the ` -mt ` option on the serial version on sun did not affect the performance significantly . except for the case where we combine openmp and mpi , no ` malloc / free ` statements are used in the region of the code where the performance is evaluated .
thus , to the best of our knowledge , this is a fair comparison between openmp and mpi performance .
we first ran the modified codes on a sun e10000 at indiana university .
the details of architecture for this computer can be found online @xcite .
benchmarks were done for various lattice sizes and numbers of threads .
as the number of threads increased , the lattice dimensions were increased to keep the volume per thread constant at @xmath1 .
the number of threads @xmath2 was increased from 1 to 16 by factors of 2 .
for example , for a given @xmath3 , the lattice size for 2 threads is @xmath4 and for 16 threads it is @xmath5 .
@xmath3 was increased from 4 to 14 in steps of 2 .
reported in fig .
1 is the performance on the kogut - susskind quark conjugate gradient routine in megaflop / s per cpu for both omp and mpi . for smaller number of threads / nodes , the omp rates are quite comparable to mpi .
they lag behind for larger number of threads .
there are two factors involved here : the overhead for setting up threads and the use of the cache . for small lattice sizes , since there is only a small number of computations to be performed , the former degrades the performance , but if significant portion of the problem can fit in the cache , the execution is speeded up . on the other hand , for larger lattices
the thread initialization overhead is a much smaller fraction of the total computation time , but the problem size is too big to fit into the cache .
we see that omp has a `` sweet spot '' at size 6 , much as the mpi performance peaks at size 8 .
since we keep the load per thread constant , for the same lattice size the performance monotonically decreases in most cases as we increase the number of threads .
= 7.0 cm = 6.0 cm = 7.0 cm = 6.0 cm next we benchmarked the code on blue horizon @xcite .
this ibm sp machine at the san diego supercomputer center has 8-way smp nodes but with the current switch can support only 4 mpi processes per node .
figure 2 contains the preliminary results from blue horizon .
these results are qualitatively similar to the e10000 results .
a hybrid approach combining openmp parallelism within mpi processes may offer better performance than either individual approach .
we tried different combinations of threads and mpi processes on blue horizon .
the hybrid approach fared better at times .
figure 3 shows the results for a total of eight processors .
it can be seen again that the mpi performance peaks at size 8 ( the left - most bars in fig .
3 ) and the omp at size 6 ( the right - most bars ) . the combination of 2 threads and 4 nodes works best for smaller sizes .
the processors on blue horizon were upgraded after these runs .
we should repeat these calculations and extend the study to a larger number of cpus .
= 7.0 cm = 6.0 cm
on both computers studied , openmp performance was very similar to mpi performance for a small number of threads , but it deteriorated much faster as the number of threads increased , for smaller lattice sizes .
thus , openmp may be a viable option for someone writing a code to be used with a modest number of processors on smp machines .
the milc collaboration , however , already has a working mpi code that scales well on many machines . for almost all the combinations of problem sizes and number of cpus studied in this paper , mpi is at least as good as openmp , if not better .
the only case where we get a considerable improvement over mpi is when we combine openmp and mpi on blue horizon for @xmath6 and 6 .
not only does the hybrid approach give the best performance on a single smp node , it should allow us to run multi - node jobs using all eight processors on each node rather than the limit of four with the current switch .
we have added openmp parallelism to the milc code only for the conjugate gradient inverter for this test project .
it will require considerably more effort to modify the whole code to run on multiple openmp threads .
it is our pleasure to thank bill magro and henry gabbs at kai for many useful discussions .
we gratefully acknowledge the help provided by the staff at research and technical services at iu , especially david hart , mary papakhian and stephanie burks .
this work was supported by the doe under grant de - f002 - 91er 40661 .
we thank the san diego supercomputer center and npaci for use of blue horizon . | cols="<",options="header " open multiprocessing ( openmp ) and message passing interface ( mpi ) are two strategies for multiple processors for single problem difference in mpi nodes have own memory communicate when needed openmp memory shared between threads example two dimensional lattice with 4 sites each direction using four nodes or threads 5 9 13 node / thread 1 corresponds to sites 1 2 5 6 in mpi node 1 has information only about sites needs information about other sites 3 7 communication routines contrast with openmp all threads have access to data for all sites thread 1 computations only for sites 1 2 5 6 table 1 summarizes differences strategies trend towards shared memory parallel machines multiprocessing nodes massively parallel processing machines makes study openmp parallelism timely openmp designed to characteristics shared - memory architectures directly access memory minimum latency no address mapping fast shared memory locks for supporting openmp advantage openmp easier to program unlike mpi worry about passing messages between nodes paper study openmp relative to mpi combining two strategies better performancesection details c code single processor changed multiple threads number threads determined by environment variable omp_num_threads ` code executed serially single thread until parallel construct encountered executed on multiple threads serial execution resumed define parallel construct lines with # pragma omp ` added code pragmas ignored by c compiler run as ordinary serial code interpreted by openmp compiler identify parallel regions several constructs execute parallel example for ` ` '' construct # pragma omp parallel for for(i=0;i < n;i++ ) { my_job(i ) ; function_job parallel on different threads memory shared each thread private copy of variables loop variables private default other variables declared private ` some variables summed over all sites accomplished with ` reduction ` statement syntax j=0 # pragma omp parallel for reduction(:j ) for(i=0;i < n;i++ )_function(i ) equivalent to @xmath0 sum performed over all threads each thread works part total iterations identifying ` private ` ` reduction ` variables necessary for correct results milc @xcite code publicly available codes mimd lattice computation collaboration for qcd simulationscode run on parallel computers using mpi for physics projects files organized in different directories libraries directory contains low level routines like matrix multiplication generic ` directory contains higher level routines updating inversion routines various application directories concentrated on conjugate gradient inverter file d_congrad5.c ` in generic_ks ` directory in version 6 milc code uses macro ` forallsites ` define forallsites(i s lattice sites_on_node lattice ` is array of sites site ` is structure variables at each lattice point sites_on_node ` number of lattice points on node needed redefine macro openmp compiler deal with two variables i s in parallel statement macro redefinition define forallsites(i s sites_on_node(lattice[i used macro ` end_loop ` closing brace match opening brace used kap / pro toolset @xcite includes guidec ` openmp compiler guideview ` : openmp parallel performance visualization tool gives details of program execution time spent in serial parallel execution imbalance in regions codeassurec ` : compiler with debugger single multiple thread executions assureview ` : openmp programming correctness tool viewing errors conflicts if different threads read write same variables add openmp parallelism to milc code steps required macro added parallel for pragmas specifying private ` reduction ` variables forallsites ` loop ` s ` made ` private ` changed ` cc ` to guidec ` makefiles compiler options adding ` backend ` before compiler option forces guidec use ` cc ` options ran assurec ` assureview ` to locate remove conflicts ran executable on threads verified output agreed with mpi output working openmp code issues comparing performance with mpi some performance problems openmp issues others not single thread openmp performance match node under mpi culprit other than openmp thread safe compilation requires ` -mt ` switch on sun performance issue with sun compiler runtime libraries not openmp if code uses many ` malloc / free ` pairs thread - safe memory allocation likely culprit not openmp issue quality vendor s thread - safe compiler / runtime implementation ` -mt ` option on sun affect performanceopenmp mpi no malloc / free statements used in code performance evaluated fair comparison between openmp mpi performance ran modified codes on sun e10000 at indiana university details architecture online @xcite benchmarks for lattice sizes numbers threads number threads increased lattice dimensions increased volume per thread constant at @xmath1 number threads @xmath2 increased 1 to 16 by factors 2 @xmath3 lattice size for 2 threads @xmath4 16 threads @xmath5 @xmath3 increased from 4 to 14 in steps 2 fig . 1 performance on kogut - susskind quark conjugate gradient routine in megaflop / s per cpu for omp mpi smaller threads omp rates comparable to mpi lag behind for larger number threads two factors overhead for setting up threads use cache small lattice sizes computations degrades performance cache execution speeded up larger lattices thread initialization overhead smaller computation time problem size too big to fit cache omp sweet spot at size 6 mpi performance peaks at size 8 load per thread constant same lattice size performance decreases increase number threads7.0 cm = 6.0 cm benchmarked code on blue horizon @xcite ibm sp machine san diego supercomputer center has 8-way smp nodes current switch 4 mpi processes per node figure 2 preliminary results blue horizon similar to e10000 results hybrid approach openmp parallelism mpi processes better performance tried different combinations of threads mpi processes hybrid approach fared better figure 3 results for eight processors mpi performance peaks at size 8 omp at size 6 combination 2 threads 4 nodes works best for smaller sizes processors upgraded after runs repeat calculations extend study to larger number cpus cm openmp performance similar to mpi performance for small threads deteriorated faster as threads increased for smaller lattice sizes openmp viable option for code modest number processors on smp machines milc collaboration has working mpi code scales well on many machines combinations problem sizes cpus mpi as good as openmp not better improvement over mpi openmp and mpi on blue horizon for @xmath6 and 6 hybrid approach best performance on single smp node multi - node jobs using all eight processors each node fouradded openmp parallelism milc code conjugate gradient inverter test more effort modify code multiple openmp threads bill magro henry gabbs discussions acknowledge help staff research technical services iu david hart mary papakhian stephanie burks supported doe grant de - f002 - 91er 40661 thank san diego supercomputer center npaci blue horizon | 0.441742 | 2,751 | 1,331 | , Saving $0.1 in GPT-4. |
177,698 | we analyze the eigenvalues and eigenvectors of the staggered dirac operator in quenched lattice qcd in the vicinity of the deconfinement phase transition using the lscher - weisz gauge action .
the spectral and localization properties of the low - lying eigenmodes show characteristic differences between the @xmath0 sectors above the critical temperature @xmath1 .
these findings can be interpreted in terms of calorons . | the chiral phase transition is a main feature of qcd . in the instanton picture @xcite
the chirally broken phase is represented by an ensemble of weakly interacting instantons and anti - instantons . due to the fermionic quasi zero modes of the ( anti- ) instantons , a finite density of eigenvalues , @xmath2 ,
is generated near zero in a ( large ) volume @xmath3 , and according to the banks - casher formula @xcite @xmath4 chiral symmetry is broken . in the chirally symmetric phase instantons and anti - instantons form
strongly bound molecules , the eigenvalues of the dirac operator move towards the bulk of the spectrum and the eigenvalue density vanishes near the origin . of course one would like to verify the essential parts of this picture in qcd .
therefore we study the localization properties of the eigenmodes of the lattice dirac operator in the neighborhood of the chiral phase transition . in a previous paper @xcite
we have investigated this question with the wilson gauge action on smaller lattices . here
we present results obtained on larger lattices with the lscher - weisz gauge action .
the temporal extension @xmath5 of the lattice is related to the temperature @xmath6 by @xmath7 , where @xmath8 is the lattice spacing . due to the periodic boundary condition in the time direction ,
an instanton becomes a chain of instantons which one calls caloron @xcite .
the properties of a caloron are similar to that of an instanton : a caloron has topological charge one and a localized left - handed fermionic zero mode .
however , the localization pattern of the zero mode is different .
whereas in the case of an instanton the zero mode is localized in space and time , the zero mode of a caloron is localized in space but delocalized in time .
let us now compare the localization of the zero mode in the different sectors of the polyakov loop @xmath9 .
we know that in the deconfined phase @xmath9 clusters around the values @xmath10 , with @xmath11 ( real sector ) and @xmath12 ( complex sector ) @xcite .
because the fermionic action does not share the @xmath0 symmetry of the gauge action , the localization of the eigenmodes of the dirac operator can depend on the polyakov loop sectors .
indeed one finds that the zero mode @xmath13 of the caloron is more localized in the real sector than in the complex sector , as can be seen from @xcite @xmath14/r^2.\ ] ] next we will search for caloron - like configurations concentrating on the fermionic properties of the caloron described above .
we work in the quenched approximation with the lscher - weisz gauge action @xmath15 & = & \textstyle \beta_1 \sum_{pl } \frac{1}{3 } { \mathrm{re}}\ ; { \mathrm{tr}}\ ; [ 1 - u_{pl } ] + \nonumber \\ & & \textstyle \beta_2 \sum_{rt } \frac{1}{3 } { \mathrm{re}}\ ; { \mathrm{tr}}\ ; [ 1 - u_{rt } ] + \nonumber \\ & & \textstyle \beta_3 \sum_{pg } \frac{1}{3 } { \mathrm{re}}\ ; { \mathrm{tr}}\ ; [ 1 - u_{pg } ] , \end{aligned}\ ] ] where @xmath16 , @xmath17 and @xmath18 means a summation over all plaquettes , @xmath19 rectangles and parallelograms , respectively .
the coefficients @xmath20 and @xmath21 are computed from @xmath22 via tadpole improved pertubation theory .
further we use the staggered dirac operator @xmath23 \:,\ ] ] where @xmath24 .
the eigenvalues of @xmath25 come in pairs of @xmath26 with @xmath27 real , so we can restrict ourselves to positive @xmath27 in the following . in the continuum limit
this action corresponds to four quark flavors . from now on we set @xmath8 to 1 .
we have calculated the lowest eigenvalues and eigenvectors for lattice sizes @xmath28 , @xmath29 and @xmath30 below the deconfinement phase transition at @xmath31 ( @xmath32 @xcite ) and above at @xmath33 ( @xmath34 @xcite ) with the arnoldi method @xcite .
we will present data for the largest lattice at @xmath35 .
when searching for calorons on the lattice with the help of their fermionic properties , the zero mode of the dirac operator plays the central role . unfortunately the staggered dirac operator has no exact zero modes , but , as we can see clearly in the left plot of fig .
[ fig1 ] , there are for both polyakov loop sectors low - lying eigenmodes with a high value of @xmath36 and eigenmodes in the bulk of the spectrum with values of @xmath37 around zero .
we find that the low - lying eigenvalues are quasi 4-fold degenerate in the sense that the number of low - lying eigenmodes is almost always 4 , 8 , 12 , ( in fig .
[ fig1 ] this is illustrated by the encircled eigenmodes , which belong to the same gauge field configuration .
note that the total number of low - lying eigenmodes of that configuration is 8 . )
so one would expect that those low - lying eigenmodes correspond to exact zero modes with definite handedness in the continuum .
let us now look at the localization properties of these approximate zero modes .
we define a gauge - invariant measure of the localization of a quark eigenmode @xmath38 ( @xmath27 is the dirac eigenvalue , @xmath39 a color index ) , the inverse participation ratio @xmath40 ^ 2}\ : , \label{idef}\ ] ] where @xmath3 is the number of lattice sites and @xmath41 is the gauge - invariant probability density @xmath42 . for a completely delocalized state ( all @xmath43 the same ) one finds @xmath44 , whereas a state localized on a single lattice site ( only one non - zero @xmath43 ) would have @xmath45 . in the right part of fig .
[ fig1 ] we plot @xmath46 vs. @xmath47 .
we see that the modes in the real sector are , in general , more localized than those in the complex sector . in both polyakov loop sectors
the approximate zero modes have a greater value of @xmath46 than the modes in the bulk spectrum , which means that they are more localized .
these are exactly the properties we expect for caloron configurations .
finally we consider the localization pattern . in fig .
[ fig2 ] isosurfaces of @xmath48 and the local chirality @xmath49 are plotted at fixed @xmath50 and at fixed @xmath51 for one of the encircled eigenmodes .
this ( typical ) mode is clearly localized in space but not in time and the isosurfaces for @xmath48 and @xmath52 are very similar .
this is again as is expected for caloron - like configurations .
studying the localization and chirality properties of low - lying quark eigenstates in quenched lattice qcd , we could characterize semiclassical properties of the gauge field configurations without any cooling . for temperatures above @xmath1 we found isolated modes with definite handedness .
they are localized in space but not in time with characteristic differences between the polyakov loop sectors .
thus they show essential properties of quark states associated with calorons .
+ + * acknowledgements : * we thank the lrz in munich for computer time on the hitachi sr8000 . support from bmbf and dfg is acknowledged .
ilgenfritz and e.v .
shuryak , phys . lett . *
b325 * , 263 ( 1994 ) ; t. schfer and e.v .
shuryak , rev .
phys . * 70 * , 323 ( 1998 ) ; r. rapp , t. schfer , e.v .
shuryak , and m. velkovsky , ann . phys . *
280 * , 35 ( 2000 ) .
t. banks and a. casher , nucl .
* b 169 * , 103 ( 1980 ) . m. gckeler , p.e.l .
rakow , a. schfer , w. sldner and t. wettig , phys .
lett . * 87 * , 042001 ( 2001 ) .
harrington and h.k .
shepard , phys .
d * 17 * , 2122 ( 1978 ) ; phys .
d * 18 * , 2990 ( 1978 ) .
a.m. polyakov , phys . lett . *
72b * , 477 ( 1978 ) ; b. svetitsky and l.g .
yaffe , nucl .
* b 210 [ fs6 ] * , 423 ( 1982 ) . c. gattringer , m. gckeler , p.e.l .
rakow , s. schaefer and a. schfer , hep - lat/0105023 . c. gattringer , m. gckeler , p.e.l .
rakow , s. schaefer and a. schfer , hep - lat/0107016 . | chiral phase transition main feature qcd instanton picture chirally broken phase represented by weakly interacting instantons anti - instantons to fermionic quasi zero modes anti- instantons finite density of eigenvalues generated near zero in large volume banks - casher formula chiral symmetry broken in chirally symmetric phase instantons anti - instantons form strongly bound molecules eigenvalues move towards bulk spectrum eigenvalue density vanishes near origin essential parts picture in qcd study localization properties eigenmodes lattice dirac operator in neighborhood chiral phase transition investigated question with wilson gauge action on smaller lattices present results on larger lattices with lscher - weisz gauge action temporal extension of lattice related to temperature @xmath8 lattice spacing periodic boundary condition time instanton becomes chain of instantons caloron properties caloron similar to instanton topological charge one localized left - handed fermionic zero mode localization pattern zero mode different space caloron space delocalized in time compare localization zero mode in sectors polyakov loopdeconfined phase @xmath9 clusters around values @xmath10 @xmath11 real @xmath12 complex sector ) fermionic action share @xmath0 symmetry gauge action localization eigenmodes dirac operator on polyakov loop sectors zero mode @xmath13 of caloron more localized in real sector than complex sector from @xcite @xmath14/r^2. search for caloron - like configurations fermionic properties caloron work in quenched approximation with lscher - weisz gauge action @xmath15 \textstyle \beta_1 \sum_{pl\frac{1}{3 1 - u\nonumber \beta_2 \sum_{rt }{1{3 1 - u\nonumber \beta_3 \sum_{pg }{1}{3 - @xmath16 @xmath17 @xmath18 summation over plaquettes @xmath19 rectangles parallelograms coefficients @xmath20 @xmath21 computed from @xmath22 via tadpole improved pertubation theory use staggered dirac operator @xmath23 @xmath24 eigenvalues @xmath25 in pairs @xmath26 @xmath27 restrict to positive @xmath27 continuum limit action corresponds to four quark flavors set @xmath8 to 1 calculated lowest eigenvalues eigenvectors for lattice sizes @xmath28 @xmath30 below deconfinement phase transition @xmath31 above @xmath33 with arnoldi method present data largest lattice at @xmath35 searching for calorons lattice fermionic properties zero mode dirac operator central role staggered dirac operator no exact zero modes sectors low - lying eigenmodes with high value @xmath36 values @xmath37 around zero low - lying eigenvalues quasi 4-fold degenerate number of low - lying eigenmodes almost always 4 , 8 , 12 , same gauge field configuration total number low - lying eigenmodes 8 low - lying eigenmodes correspond to exact zero modes with definite handedness in continuum at localization properties of approximate zero modesdefine gauge - invariant measure localization quark eigenmode @xmath38 @xmath27 dirac eigenvalue @xmath39 color index inverse participation ratio @xmath40 @xmath3 number lattice sites @xmath41 gauge - invariant probability density @xmath42 delocalized state @xmath43 finds @xmath44 state localized single lattice site non - zero @xmath43 @xmath45 plot @xmath46 vs. @xmath47 modes real sector more localized complex sector polyakov loop sectors approximate zero modes greater value @xmath46 bulk spectrum more localized properties expect caloron configurations consider localization pattern fig isosurfaces @xmath48 local @xmath49 plotted fixed @xmath50 @xmath51 mode localized space not time isosurfaces @xmath48 @xmath52 similar expected caloron - like configurations studying localization chirality properties low - lying quark eigenstates quenched lattice qcd characterize semiclassical properties gauge field configurations without cooling temperatures above @xmath1 isolated modes definite handedness localized space not time differences polyakov loop sectors show essential properties quark states caloronsacknowledgements thank lrz munich computer time hitachi sr8000 support bmbf dfg acknowledged ilgenfritz shuryak phys . lett b325 263 ( 1994 ) t schfer shuryak 70 323 ( 1998 ) r rapp t schfer shuryak velkovsky ann 280 35 ( 2000 ) banks casher nucl 169 103 ( 1980 ) m. gckeler rakow schfer w sldner t wettig phys . 87 042001 2001 ) harrington shepard phys 17 2122 ( 1978 ) 18 2990 ( 1978 ) polyakov . lett 72b 477 ( 1978 ) svetitsky l yaffe nucl 210 423 ( 1982 ) gattringer m gckeler rakow schaefer a. schfer hep - lat/0105023 gattringer gckeler rakow schaefer schfer - lat/0107016 | 0.318051 | 2,113 | 1,079 | , Saving $0.1 in GPT-4. |
177,701 | we have correlated a sample of 99 seyfert 1 & 98 seyfert 2 ( rafanelli et al . , 1995 ) interacting galaxies with the survey and pointed rosat data and have detected 91 seyferts 1 & 47 seyfert 2s in the x - ray band .
we have performed spectral and timing analysis of the x - ray detected sources , and we have examined the relation of the x - ray luminosity @xmath0 with the interaction strength @xmath1 .
galaxies : active , seyfert 1 , seyfert 2 , merger ; x - rays : galaxies | interacting galaxies are considered to have a wide importance on triggering of starburst and agn activity .
therefore , we have investigated two subsamples of spectroscopically - selected , x - ray detected interacting seyfert 1s , including nls1 , and seyfert 2s galaxies ( rafanelli et al.,1995 ) with @xmath2 .
+ the pairs have been selected on poss plates using the following criteria given by rafanelli et al .
( 1995 ) : separation between the components @xmath3 , where @xmath4 is the apparent major axis of theseyfert galaxy , and a magnitude - difference between seyfert and the companion@xmath5 .
the identification is based on x - ray contour lines overlaid on optical images taken from dss . in fig .
1 we show as an example the overlay from the nls1 galaxy mkn 896 .
2 shows the spectrum of this object .
the deviation from the fit at energy 0.7 kev in fig .
2 indicates the presence of a warm absorber .
the pointing and survey lightcurves in fig .
3 show the variability of this source on short time scales .
to derive the interaction strength @xmath1 , we concentrate on the tidal force per unit mass produced by a companion on a primary galaxy , which is proportional to @xmath6 .
@xmath7 is the mass of the companion and @xmath8 is its distance from the center of the primary galaxy . in most cases ,
@xmath7 and the absolute value of @xmath8 are unknown . instead
, these parameters are related to the dimensions of the pair .
rubin et al . (
1982 ) describe the dependence of the mass @xmath9 of a galaxy on the size of its major axis @xmath10 as @xmath11 and we use @xmath12 ( dahari , 1984 ) .
if we utilize the apparent diameter of the primary galaxy @xmath4 as a scaling factor , we obtain : @xmath13 using these relations , we get as the dimensionless gravitational interaction strength @xmath1 : @xmath14 this parameter is obviously large for close and relatively large companions .
fig.4 shows interaction strength @xmath1 vs. the x - ray luminosity for seyfert 1 galaxies .
noticeable is a faint tendency for the luminosity increase for increasing values of interaction strength .
a possible explanation for the low luminosity sources are the presence of a high hydrogen column density in the nucleus and the emission from the circumnuclear starbursts .
+ the galaxy interactions or mergers are able to trigger a burst of star formation , which can be a reason for the rise in luminosity .
several studies about physical mechanisms which could trigger a burst of strong star formation have been investigated by barnes & hernquist ( 1991 ) , jog & das ( 1992 ) and jog & solomon ( 1992 ) .
also many observations suggest that galactic encounters enhance star formation rates , as demonstrated by studies of the optical ( e.g. , kennicut et al . 1987 ; bushouse 1987 ) , infrared ( e.g. , bernlhr 1993 ; telesco et al .
1993 ) and radio ( e.g. , hummel 1981 ; smith & kassim 1993 ) emission from interacting systems .
+ for each source detected in the all - sky survey with more than 100 counts in an exposure time of 400 sec and for all sources within pointed observations , we performed a spectral analysis and created a lightcurve .
studies of variability were possible , if more than one pointed observation was available or if both a detection in the survey and in a pointed observation existed .
we are extending our studies on starburst galaxies to examine the influence of interaction with companions on the x - ray luminosity .
a paper with data concerning x - ray properties , luminosity , flux , count rate variation and lightcurves for seyfert 1s & 2s is in preparation . | interacting galaxies starburst agn activity investigated two spectroscopically - selected x - ray detected interacting seyfert 1s seyfert 2s galaxies rafanelli et al.,1995 ) with @xmath2 pairs selected plates criteria separation between components @xmath3 @xmath4 major axis magnitude - difference between seyfert companion@xmath5 identification based on x - ray contour lines optical images dss fig . 1 overlay from nls1 galaxy mkn 896 deviation energy 0.7 kev indicates warm absorber lightcurves show variability source short time scales interaction strength @xmath1 on tidal force per unit mass companion primary galaxy proportional to @xmath6 @xmath7 mass companion @xmath8 distance from center primary galaxy @xmath7 absolute value @xmath8 unknown parameters related to dimensions pair rubin et al . 1982 dependence mass @xmath9 galaxy on size major axis @xmath10 @xmath11 @xmath12 ( dahari , 1984 )utilize diameter primary galaxy @xmath4 scaling factor obtain : @xmath13 dimensionless gravitational interaction strength @xmath1 : @xmath14 parameter large for close large companions fig.4 shows interaction strength @xmath1 vs x - ray luminosity seyfert 1 galaxies faint tendency luminosity increase increasing values interaction strength possible explanation low luminosity sources high hydrogen column density nucleus emission circumnuclear starbursts galaxy interactions trigger star formation rise luminosity studies mechanisms star formation investigated by barnes & hernquist 1991 jog & das 1992 solomon 1992 ) galactic encounters enhance star formation rates demonstrated studies optical infrared radio emission from interacting systems each source detected all - sky survey more than 100 counts exposure time 400 sec all sources pointed observations performed spectral analysis created lightcurve studies variability possible if more than one pointed observation available or extending studies on starburst galaxies examine influence interaction companions x - ray luminosity paper with data x - ray properties luminosity flux count rate variation lightcurves for seyfert 1s & 2s in preparation | 0.137483 | 874 | 442 | , Saving $0.0 in GPT-4. |
177,706 | we calculate the expected spectrum and light curves of the early afterglow . for short grbs
the peak of the afterglow will be delayed , typically , by a few dozens of seconds after the burst .
the x - ray and @xmath0-ray characteristics of this delayed emission provide a way to discriminate between late internal shocks emission ( part of the grb ) and the early afterglow signal .
detection of this delayed emission will prove the internal shock scenario as producing the grb , and will pinpoint the initial lorentz factor @xmath1 . in the optical band ,
the dominant emission arises from the reverse shock .
this shock , carries an energy comparable to that of the forward shock .
it radiates this energy at much lower frequencies , producing a short optical flash of 15th magnitude or brighter . | in the internal - external scenario , the grb is produced by internal shocks while the afterglow is produced by the interaction of the flow with the ism .
the original fireball model was invoked to explain the gamma - ray bursts ( grbs ) phenomena .
it requires extreme relativistic motion , with a lorentz factor @xmath2 . the afterglow observations , which fit the theory rather well , are considered as a confirmation of the fireball model .
however , the current afterglow observations , which detect radiation from several hours after the burst onwards , do not probe the initial extreme relativistic conditions . by the time of the present observations , several hours after the burst , the lorentz factor is less than @xmath3 , and it is independent of the initial lorentz factor .
afterglow observations , a few seconds after the burst , can provide the missing information concerning the initial phase and prove the internal shock scenario .
such rapid observations are possible , in principle , with future missions ( kulkarni and harrison , private communication ) .
the synchrotron spectrum from relativistic electrons that are continuously accelerated into a power law energy distribution is always given by four power law segments , separated by three critical frequencies : @xmath4 the self absorption frequency , @xmath5 the cooling frequency and @xmath6 the characteristic synchrotron frequency .
using the relativistic shock jump conditions and assuming that the electrons and the magnetic field acquire fractions @xmath7 and @xmath8 of the equipartition energy , we obtain : @xmath9 @xmath10 @xmath11 @xmath12 these scalings generalize the adiabatic scalings obtained by sari , piran & narayan ( 1998 ) to an arbitrary hydrodynamic evolution of @xmath13 . for typical parameters , @xmath14 ,
so fast cooling occurs .
the spectrum of fast cooling electrons is described by four power laws : ( i ) for @xmath15 self absorption is important and @xmath16 .
( ii ) for @xmath17 we have the synchrotron low energy tail @xmath18 . (
iii ) for @xmath19 we have the electron cooling slope @xmath20 .
( iv ) for @xmath21 @xmath22 , where @xmath23 is the index of the electron power law distribution . in the early afterglow ,
the lorentz factor is initially constant . after this phase ,
the evolution can be of two types ( sari 1997 ) .
thick shells , which correspond to long bursts , begin to decelerate with @xmath24 .
only later there is a transition to deceleration with @xmath25 .
the light curves for such bursts can be obtained by substituting these scalings in equations [ num]-[nusa ] . however , for these long bursts , the complex internal shocks grb signal would overlap , the smooth external shock afterglow signal .
the separation of the observations to grb and early afterglow would be rather difficult . for thin shells , that correspond to short bursts ,
there is no intermediate stage of @xmath24 .
there is a single transition , at the time @xmath26 , from a constant velocity to self - similar deceleration with @xmath25 .
the possible light curve are illustrated in figure 1 . as the intial afterglow peaks several dozen seconds after the grb there should be no difficulty to detect it .
the detection of delayed emission which fits the light curves of figure 1 , would enable us to determine @xmath27 .
using @xmath27 we could proceed to estimate the initial lorentz factor : @xmath28 if the second peak of grb 970228 , delayed by 35s , is indeed the afterglow rise , then @xmath29 for this burst .
there are many attempts to detect early optical emission and there is a good chance that this emission will be observed in the near future .
a strong 5th magnitude optical flash would have been produced if the fluence of a moderately strong grbs , @xmath30erg / s/@xmath31 would have been released on a time scale of 10s in the optical band .
even a small fraction of this will be easily observed .
it is important , therefore , to explore the expected optical emission from the grb and the early afterglow . during the grb and the initial emission from the forward shock the emission peaks in @xmath0-rays , and only an extremely small fraction
is emitted in the optical band .
for example , the prompt optical flash from the grb would be of 21st magnitude if the flux drops according to the synchrotron low energy tail of @xmath32 .
a considerably stronger flux is obtained from the reverse shock .
the reverse shock contains , at the time it crosses the shell , a comparable amount of energy to the forward shock . however , its effective temperature is significantly lower ( typically by a factor of @xmath33 ) than that of the forward shock .
the resulting peak frequency is therefore lower by @xmath34 .
a more detailed calculation shows that the reverse shock frequency is @xmath35 the cooling frequency is similar to that of the forward shock , since both have the same magnetic field and the same lorentz factor . using the parameters obtained by granot , piran and sari ( 1998 ) , from the afterglow of grb 970508 , and using @xmath36 we get for the reverse shock @xmath37hz and @xmath38hz leading to an 8th magnitude flash . with higher initial lorentz factor of @xmath39
the flash drops to 13th magnitude .
inverse compton cooling , if exists , can reduce the flux by @xmath40 magnitudes , while self absoption can influence only very short bursts with small surface area .
therefor , quite concervatively , the optical flash should be stronger than 15th magnitude and should be soon seen with modern experiments .
the reverse shock signal is very short living . after the shock crosses the shell ,
no new electrons are injected and there is no emission above @xmath41 .
moreover , @xmath41 drops fast with time as the shell s material cools adiabatically .
the early afterglow multi - wavelength radiation could provide interesting and invaluable information on the extreme relativistic conditions that occur at this stage . the initial emission from the forward shock , which continues later as the observed afterglow signals is in the @xmath0-rays or x - rays .
the reverse shock emission could have a strong , short living , optical component which we expect to be brighter than 15th magnitude .
some of the current basic ideas concerning the fireball models should be revised if such signals will not be seen by new detectors that should become operational in the near future . | internal - external scenario grb produced by internal shocks afterglow interaction flow with ism original fireball model gamma - ray bursts requires extreme relativistic motion lorentz factor @xmath2 afterglow observations fit theory fireball model current afterglow observations detect radiation hours after burst probe initial extreme relativistic conditions observations after burst lorentz factor less than @xmath3 independent of initial lorentz factor afterglow observations after burst provide missing information prove internal shock scenario rapid observations possible with future missions synchrotron spectrum from relativistic electrons accelerated power law energy distribution four power law segments separated by three frequencies @xmath4 self absorption @xmath5 cooling characteristic synchrotron relativistic shock jump conditions electrons magnetic field acquire @xmath7 @xmath8 energy obtain @xmath9 @xmath10 @xmath12 scalings generalize scalings sari piran narayan to arbitrary hydrodynamic evolution of @xmath13 parameters fast cooling occurs spectrum fast cooling electrons described by four power laws @xmath15 self absorption important @xmath16@xmath17 synchrotron low energy tail @xmath18 @xmath19 electron cooling slope @xmath20 @xmath21 @xmath22 @xmath23 index electron power law distribution early afterglow lorentz factor constant evolution two types 1997 ) thick shells long bursts decelerate with @xmath24 transition to deceleration with @xmath25 light curves substituting scalings equations num]-[nusa ] long bursts complex internal shocks grb signal overlap smooth external shock afterglow signal separation observations grb afterglow difficult thin shells short bursts no intermediate stage @xmath24 single transition @xmath26 constant velocity to self - similar deceleration with @xmath25 possible light curve figure 1 afterglow peaks seconds after grb no difficulty detect detection delayed emission light curves determine @xmath27 @xmath27 estimate initial lorentz factor @xmath28 second peak of grb 970228 delayed 35s afterglow rise @xmath29 burst many attempts detect early optical emission good chance emission observed near futurestrong 5th magnitude optical flash produced if moderately strong grbs @xmath30erg@xmath31 released 10s in optical band small fraction observed to explore expected optical emission from grb early afterglow during grb initial forward shock emission peaks in @xmath0-rays small fraction emitted in optical band prompt optical flash from grb 21st magnitude if flux drops synchrotron low energy tail @xmath32 stronger flux from reverse shock contains comparable energy to forward shock effective temperature lower by @xmath33 than peak frequency lower by @xmath34 reverse shock frequency is @xmath35 cooling frequency similar to forward shock same magnetic field lorentz factor parameters granot piran sari afterglow grb 970508 @xmath36 reverse shock @xmath37hz @xmath38hz to 8th magnitude flash higher initial lorentz factor of @xmath39 flash drops to 13th magnitude inverse compton cooling reduce flux by @xmath40 magnitudes self absoption short bursts small surface area optical flash should be stronger than 15th magnitude soon seen with modern experiments reverse shock signal short livingshock crosses shell no new electrons injected no emission above @xmath41 @xmath41 drops fast shell s material cools early afterglow multi - wavelength radiation provide information on extreme relativistic conditions initial emission from forward shock in @xmath0-rays x - reverse shock emission strong short living optical component brighter than 15th magnitude current ideas fireball models should revised if signals not seen by new detectors operational | 0.26192 | 1,459 | 758 | , Saving $0.0 in GPT-4. |
177,710 | we review our recent analysis of @xmath0 scattering data in terms of roy equations and forward dispersion relations , and present some preliminary results in terms of a new set of once - subtracted coupled equations for partial waves .
the first analysis consists of independent fits to the different @xmath0 channels that satisfies rather well the dispersive representation . in the second analysis
we constrain the fit with the dispersion relations .
the latter provides a very precise and model independent description of data using just analyticity , causality and crossing .
address = departamento de fsica terica ii , facultad de ciencias fsicas , universidad complutense de madrid , e-28040 , madrid , spain .
address = departamento de fsica terica ii , facultad de ciencias fsicas , universidad complutense de madrid , e-28040 , madrid , spain .
address = department of theoretical physics henryk niewodniczaski institute of nuclear physics , polish academy of sciences , 31 - 342 , krakw , poland .
address = departamento de fsica terica , c - xi , universidad autnoma de madrid , canto blanco , 28049 , madrid , spain . | a precise knowledge of pion - pion scattering is of interest since it provides a test of chiral perturbation theory ( chpt ) as well as useful information about quark masses and the chiral condensate @xcite .
the reaction , at least in the elastic regime , is also remarkably symmetric in terms of isospin and crossing symmetries .
unfortunately , the existing experimental information from @xmath0 scattering has many conflicting data sets at intermediate energies and no data at all close to the interesting threshold region . for many years
this fact has made it very hard to obtain conclusive results on @xmath0 scattering at low energies or in the sigma region .
however , recent @xcite and precise experiments on kaon decays , related to @xmath0 scattering at very low energies , have renewed the interest on this process .
the dispersive integral formalism is model independent , just based on analyticity and crossing , and relates the @xmath0 amplitude at a given energy with an integral over the whole energy range , increasing the precision and providing information on the amplitude at energies where data are poor , or in the complex plane .
in addition , it makes the parametrization of the data irrelevant once it is included in the integral and relates different scattering channels among themselves .
for all these reasons it is well suited to study the threshold region or the poles in the complex plane associated to resonances ( see h. leutwyler and r. garca martn @xcite talks on this conference and references therein ) .
our recent works make use of two complementary dispersive approaches , in brief : @xmath1 _ forward dispersion relations ( fdrs ) : _ they are calculated at @xmath2 so that the unknown large - t behavior of the amplitude is not needed .
we consider two symmetric and one asymmetric isospin combinations , to cover the isospin basis .
the symmetric ones , @xmath3 and @xmath4 , have two subtractions and can be written as @xmath5 where @xmath6 stands for the @xmath7 or @xmath8 amplitudes .
all contributions to their integrands are positive , which makes them very precise .
the antisymmetric isospin combination @xmath9 does not require subtractions : @xmath10 we have implemented all of them up to @xmath11 mev @xmath1 _ roy equations ( re ) @xcite _ : they are an infinite set of coupled equations fully equivalent to nonforward dispersion relations , plus some @xmath12 crossing symmetry , written in terms of partial waves of definite isospin i and angular momentum @xmath13 .
the complicated left cut contribution is rewritten as series of integrals over partial waves in the physical region : @xmath14 where the @xmath15 , @xmath16 constants and @xmath17 kernels are known . in practice , the calculation is truncated at @xmath18 and at some cutoff energy @xmath19 . the @xmath20 waves and the high energy are treated as input .
re are well suited to study poles of resonances but are limited to @xmath21mev . at present , we have implemented them up to @xmath22 .
the use of re has gained interest with three aims : to improve the precision of scattering data , to test chpt , or to use chpt to obtain the subtraction constants at low energies , which can be recast in terms of the scattering lengths @xmath23 and @xmath24 , and obtain precise predictions on @xmath0 scattering .
in particular a series of re analysis in @xcite using @xmath0 data parametrizations for the @xmath25 waves and above 800 mev for the rest , as well as some regge input , was performed with and without chpt constraints .
the latter provided , @xmath26 and @xmath27 , an extremely precise claim , together with predictions for other scattering lengths and the s and p wave phase shifts up to 800 mev .
some of the input , particularly the regge theory and the d waves , was questionable @xcite , but it certainly seems to have a very small influence in the threshold region of the scalar waves @xcite . in recent years the krakow - paris @xcite and paris @xcite groups have performed other re analysis .
the former resolved the long - standing ambiguity , discarding the so - called `` up'' solution , including in their analysis an study using polarized target data .
the latter checked the calculation in @xcite and claimed an small discrepancy in the olsson sum rule .
the approach followed in the series of works by our group @xcite can be summarized as follows : ( 1 ) we first obtain simple fits to each @xmath0 channel independently , that we call `` unconstrained data fits '' ( udf ) . in this way
all waves are uncorrelated and can be easily changed if new , more precise data becomes available .
this actually happened , for example , in one of our most recent works @xcite where we have included the newest @xmath28 data @xcite , which is very precise . at different stages of our approach
we have also fitted regge theory to @xmath0 high energy data , and as our precision was improving , we have improved some of the udf fits with more flexible parametrizations .
( 2 ) next , we check how well these udf satisfy the dispersion relations .
surprisingly some of the most widely used parametrizations fail to satisfy the fdrs .
we then select the data parametrizations in better agreement with fdrs .
( 3 ) finally , we impose in the fits the dispersion relations .
this provides very precise data fits where all waves are correlated thus we call them `` constrained data fits '' ( cdf) which are consistent with analyticity , unitarity , crossing , etc ... initially we only considered fdrs but in our most recent work , whose results we review next , we have included re . in fig.1
we show our unconstrained fits to data ( ufd ) for several waves .
the details of their simple parametrizations can be found in @xcite .
we only plot explicitly the constrained fits ( cdf ) for the isospin-2 waves since in all other cases they are indistinguishable from the udf , showing their remarkable stability . actually , all cdf waves differ from the udf waves by less than one @xmath29 , except for @xmath30 , @xmath31 that deviates by 1.5 standard deviations from the unconstrained case . in the bottom row of fig.1
we show in greater detail both the ufd and cfd results for the s0 wave , which is probably very controversial and the relevant one for this conference .
partial waves .
almost all constrained fits are indistinguishable except those for s2 , d2 and s0 waves that are also shown here . ] in order to quantify how well the dispersion relations are satisfied , we define six @xmath32 as the difference between the left and right sides of each dispersion relation in eqs . , and , whose uncertainties we call @xmath33 .
next , we define the average discrepancies @xmath34 where the values of @xmath35 are taken at intervals of 25 mev .
note the similarity with an averaged @xmath36 and thus @xmath37 implies fulfillment of the corresponding dispersion relation . in table 1
we show the average discrepancies of the udf for each fdr , up to two different energies , and each re up to @xmath38 .
although the total average discrepancy of the udf set is practically one , they can be clearly improved in the high energy region of the antisymmetric fdr and in the scalar isospin-2 re .
this is actually done in the cdf set , which is obtained by minimizing : @xmath39 where @xmath40 are all the parameters of the different udf parametrization for each wave or regge trajectory , thus ensuring the data description , and @xmath41 and @xmath42 are the discrepancies for a couple of crossing sum rules .
the weight @xmath43 was estimated from the typical number of degrees of freedom needed to describe the shape of the dispersion relations .
.average discrepancies @xmath44 of the udf and cdf for each fdr and re . on average ,
the udf are consistent with dispersion relations .
note the remarkable cdf consistency .
[ cols="<,^,^,^,^ " , ] from the table it is clear that the cdf set satisfies remarkably well all dispersion relations within uncertainties , and hence can be used directly and inside the olsson sum rule to obtain the following precise determination _ from data _ : @xmath45 and @xmath46 .
this is in remarkable agreement with the predictions of re and chpt of @xcite .
nevertheless , the agreement is fairly good only up to roughly 450 mev , but from that energy up to 800 mev those predictions deviate from our data analysis .
we should stress that we are nevertheless talking about a disagreement of a few degrees at most and would affect the determination of the sigma mass by tens of mev at most , which is a remarkable improvement compared with the situation just a few years ago and the huge and extremely conservative uncertainties quoted in the pdg for the @xmath29 mass and width , of hundreds of mev .
the other waves are of less relevance for this conference and we comment them very briefly , since the details can be found in @xcite .
the best determination of threshold parameters is obtained by using the cdf set inside appropriate sum rules @xcite . in brief
, we agree with @xcite in the p - wave scattering length , but find disagreements of 2 to 3 standard deviations in the p - wave slope , and also in some @xmath47 wave parameters . in summary the cdf set provides a model independent and very precise description of the @xmath0 scattering data consistent with analyticity and crossing .
the upper row plots in fig.2 show that , despite the small uncertainties , the cdf set satisfies remarkably well the fdrs up to 1420 mev .
the same happens for the re , although now the uncertainties , small close to threshold , become , for the scalar channels , rather large around 800 mev .
this is due to the polynomial factors in the re that multiply the scalar scattering lengths , particularly @xmath48 . in order to avoid this ,
we have derived a modified set of once - subtracted roy - like equations ( gmkpy for brevity ) .
actually , the plots of the bottom row of fig.3 show our preliminary results for a constrained fit including also the gmpky .
note that close to threshold the resulting gmkpy uncertainties are larger than for the standard re , but the uncertainties are dramatically reduced at intermediate energies , the region of interest for scalar spectroscopy and this conference ( see the talk by r. garca - martn ) .
obviously , these tiny uncertainties will force us to improve the region around 850 to 1 gev with more flexible parametrizations , and in particular the matching of the low and intermediate energy regions for the scalar - isoscalar channel , not sufficiently smooth at the moment ( see fig 1 , bottom ) .
work is in progress in these directions .
j. gasser and h. leutwyler , annals phys .
* 158 * , 142 ( 1984 ) .
j. r. batley _ et al . _
[ na48/2 collaboration ] , eur .
j. c * 54 * ( 2008 ) 411 .
s. pislak _ et al .
_ [ bnl - e865 collaboration ] , phys . rev .
* 87 * ( 2001 ) 221801 [ arxiv : hep - ex/0106071 ] .
r. garcia - martin , r. kaminski , and j. r. pelez arxiv:0804.2643 [ hep - ph ] .
s. m. roy , phys .
b * 36 * , 353 ( 1971 ) .
g. colangelo , j. gasser and h. leutwyler , nucl .
b * 603 * ( 2001 ) 125 b. ananthanarayan , g. colangelo , j. gasser and h. leutwyler , phys .
* 353 * ( 2001 ) 207 j. r. pelaez and f. j. yndurain , nucl .
* 164 * ( 2007 ) 93 ; phys . rev .
d * 68 * ( 2003 ) 074005 r. kaminski , l. lesniak and b. loiseau , phys .
b * 551 * , 241 ( 2003 ) s. descotes - genon , n. h. fuchs , l. girlanda and j. stern , eur .
j. c * 24 * , 469 ( 2002 ) r. kaminski , j. r. pelaez and f. j. yndurain , phys . rev .
d * 77 * ( 2008 ) 054015 ; phys .
d * 74 * ( 2006 ) 014001 r. garcia - martin and j. r. pelaez , f. j. yndurain , phys .
d * 76 * ( 2007 ) 074034 j. r. pelaez and f. j. yndurain , phys .
d * 71 * ( 2005 ) 074016 ; phys .
d * 69 * ( 2004 ) 114001 | knowledge pion - pion scattering provides test chiral perturbation theory information about quark masses chiral condensate @xcite reaction elastic regime symmetric isospin crossing symmetries information @xmath0 scattering conflicting data at intermediate energies no data close to threshold region hard obtain conclusive results on scattering at low energies sigma region recent @xcite experiments on kaon decays renewed interest dispersive integral formalism model independent based on analyticity crossing relates @xmath0 amplitude at energy with integral over energy range increasing precision information on amplitude at energies data poor complex plane makes parametrization data irrelevant once included in integral relates scattering channels to study threshold region poles in complex plane associated resonances see h. leutwyler r. garca martn @xcite talks recent works two complementary dispersive approaches @xmath1 _ forward dispersion relations calculated at @xmath2 unknown large - t behavior amplitude not needed consider two symmetric one asymmetric isospin combinations basissymmetric @xmath3 @xmath4 have two subtractions written as @xmath5 @xmath6 for @xmath7 or amplitudes contributions to integrands positive precise antisymmetric isospin combination @xmath9 subtractions implemented up to @xmath11 @xmath1 _ roy equations infinite set coupled equations equivalent to nonforward dispersion relations @xmath12 crossing symmetry partial waves isospin angular momentum @xmath13 complicated left cut contribution rewritten as integrals over partial waves physical region @xmath14 constants kernels known calculation truncated at @xmath18 cutoff energy @xmath19 @xmath20 waves high energy treated as input poles resonances limited to @xmath21mev implemented up to @xmath22 re improve precision scattering data test chpt obtain subtraction constants at low energies precise predictions on @xmath0 scattering re analysis in @xcite using @xmath0 data parametrizations for @xmath25 waves above 800 mev rest regge input performed with without chpt constraintslatter provided @xmath26 and @xmath27 precise claim predictions for scattering lengths s and p wave phase shifts up to 800 mev input regge theory d waves questionable @xcite small influence in threshold region scalar waves krakow - paris @xcite and paris @xcite groups performed analysis former resolved ambiguity discarding `` up'' solution including study using polarized target data latter checked calculation in @xcite claimed small discrepancy in olsson sum rule approach followed our group summarized obtain simple fits to each @xmath0 channel `` unconstrained data fits '' udf all waves uncorrelated easily changed if new precise data available happened in recent included newest @xmath28 data precise fitted regge theory to @xmath0 high energy data precision improving improved udf fits with flexible parametrizations check udf satisfy dispersion relations some used parametrizations fail to satisfy fdrs select data parametrizations in better agreement with fdrs impose in fits dispersion relations provides precise data fits waves correlated constrained data fits ( cdf) consistent with analyticity unitarity crossing initially considered fdrs recent work included re fig.1 unconstrained fits data several waves details parametrizations in @xcite plot constrained fits cdf for isospin-2 waves indistinguishable from udf remarkable stability all cdf waves differ from udf waves less than one @xmath29 except @xmath30 @xmath31 1.5 standard deviations unconstrained case bottom row fig.1 ufd cfd results for s0 wave controversial relevant conference constrained fits indistinguishable except s2 d2 s0 waves shown quantify dispersion relations define @xmath32 difference between left right sides dispersion relation uncertainties call @xmath33 define average discrepancies @xmath34 values @xmath35 at intervals 25 mev similarity with averaged @xmath36 @xmath37 implies fulfillment dispersion relation table 1 average discrepancies udf for each fdr two different energies re up to @xmath38 total average discrepancy udf set one improved in high energy region antisymmetric fdr scalar isospin-2 re done in cdf set obtained by minimizing @xmath39 @xmath40 parameters different udf parametrization for each wave regge trajectory ensuring data description @xmath41 @xmath42 discrepancies for crossing sum rules weight @xmath43 estimated from typical degrees freedom shape dispersion relations.average discrepancies @xmath44 of udf cdf for each fdr re udf consistent with dispersion relations remarkable cdf consistency cdf set satisfies all dispersion relations within uncertainties used olsson sum rule precise determination data : @xmath45 @xmath46 agreement with predictions re chpt of @xcite agreement good up to 450 mev up to 800 mev predictions deviate from data analysis disagreement of few degrees determination sigma mass by tens of mev improvement compared with few years ago conservative uncertainties in pdg for @xmath29 mass width hundreds of mev other waves less relevance for conference comment briefly details found in @xcite determination threshold parameters using cdf set inside sum rules @xcite agree with @xcite in p - wave scattering length disagreements 2 to 3 standard deviations in p - wave slope @xmath47 wave parameters cdf set provides model independent precise description @xmath0 scattering data consistent with analyticity crossing plots.2 show small uncertainties cdf set satisfies fdrs up to 1420 mev same re uncertainties small close to threshold large around 800 mev due to polynomial factors in re scalar scattering lengths particularly @xmath48 avoid derived modified once - subtracted roy - like equations gmkpy plots fig.3 show preliminary results for constrained fit including gmpky close to threshold gmkpy uncertainties larger than standard re reduced at intermediate energies tiny uncertainties force improve region around 850 to 1 gev with flexible parametrizations matching low and intermediate energy regions for scalar - isoscalar channel work progress j. gasser h. leutwyler , annals phys . * 158 , 142 ( 1984 )batley na48/2 collaboration 54 2008 ) 411 pislak bnl - e865 collaboration phys . rev 87 2001 ) 221801 hep - ex/0106071 garcia - kaminski r pelez arxiv:0804.2643 hep - ph roy 36 353 ( 1971 ) colangelo j gasser h leutwyler 603 2001 ) 125 ananthanarayan colangelo gasser leutwyler. 353 2001 ) 207 r pelaez f j. yndurain nucl 164 2007 ) 93 phys rev 68 2003 ) 074005 kaminski lesniak loiseau phys 551 241 ( 2003 ) descotes - genon fuchs girlanda j stern c 24 469 ( 2002 ) kaminski pelaez f j. yndurain phys 77 2008 ) 054015 74 2006 ) 014001 garcia - martin. pelaez j. yndurain phys 76 2007 ) 074034 r. pelaez f j. yndurain phys 71 2005 ) 074016 69 2004 ) 114001 | 0.439136 | 2,973 | 1,505 | , Saving $0.1 in GPT-4. |
177,713 | many of the known solutions of the yang - baxter equation , which are related to solvable lattice models of vertex- and irf - type , yield representations of the birman - wenzl - murakami algebra . from these
, representations of a two - colour generalization of the birman - wenzl - murakami algebra can be constructed , which in turn are used to derive trigonometric solutions to the yang - baxter equation . in spirit
, this construction resembles the fusion procedure , in the sense that starting from known solutions of the yang - baxter equation new solutions can be obtained . | over the the past fifteen years , since baxter s famous book first appeared in print @xcite , the full relevance of the yang - baxter equation ( ybe ) in the theory of solvable two - dimensional lattice models has been realized .
moreover , many fruitful and partly unexpected connections to other branches of mathematics and physics , including for example quantum groups , knot and link invariants , and integrable quantum field theories with purely elastic ( factorized ) scattering , have been discovered . among the several algebraic techniques used to construct solutions to the ybe
, a particularly interesting approach is based on braid - monoid algebras ( bma ) @xcite . by a procedure called _ baxterization _
@xcite , one reduces the problem of finding representations of the yang - baxter algebra ( yba ) to the simpler task of constructing representations of a certain bma .
examples of bma for which a baxterization is known are the temperley - lieb algebra @xcite , the birman - wenzl - murakami ( bwm ) algebra @xcite and their recently introduced @xcite ` dilute ' @xcite and ` two - colour ' generalizations @xcite . in this short note
, we present explicit expressions for ` vertex - type ' representations of the two - colour bwm ( tcbwm ) algebra , by which we understand representations acting on a tensor product space .
this is complementary to the content of ref .
@xcite where exclusively rsos ( restricted solid - on - solid ) type representations were considered , for which the representation space is spanned by the set of all paths on certain graphs . here
, the representations are constructed by suitably combining representations of the ordinary bwm algebra underlying the solvable vertex models of bazhanov @xcite and jimbo @xcite . by the known baxterization of the tcbwm algebra @xcite ,
this yields solutions of the ybe and hence solvable lattice models of vertex type .
in contrast to the case of rsos type representations discussed in ref .
@xcite , we do not expect that the corresponding solvable vertex models are new .
instead , we expect a similar scenario as in the dilute case , where the @xmath0-matrices constructed from representations of the dilute bwm algebra are related by a suitable gauge transformation to other members of the lists of well - known @xmath0-matrices @xcite . for one series of @xmath0-matrices ,
this has been explicitly demonstrated in ref .
for lack of space , we can not repeat the complete definition of the tcbwm algebra here , thus we only give a rough sketch of the idea , see refs .
@xcite for details .
an ordinary bma @xcite is generated by two types of generators , the braids ( with their inverses ) and the monoids or temperley - lieb ( tl ) operators .
besides the usual braid relations and the defining relations of the tl algebra , a number of relations involving both types of generators are imposed .
all relations have a natural interpretation in terms of continuous moves in a diagrammatic presentation of the generators acting on arrays of strings @xcite , hinting at the connection to the theory of knot and link invariants .
the defining relations also involve two central elements customarily denoted by @xmath1 and @xmath2 , which in the graphical interpretation are associated to closed loops and so - called ` twists ' , respectively .
the bwm algebra is a certain quotient of a bma , in which the braid generators satisfy a cubic reduction relation @xcite .
the tcbwm algebra is a two - colour generalization of the bwm algebra , with two species ( colours ) of strings .
the generators now carry colour indices @xmath3 and one has additional generators @xmath4 which act as projection operators on certain colours . here , @xmath5 label the two colours , and @xmath6 denotes the complementary ( or opposite ) colour .
furthermore , the generators carry a label @xmath7 . in the graphical interpretation
, this means that the corresponding generator only acts on the two of the @xmath8 strings , namely on those labeled by @xmath9 and @xmath10 .
the central elements mentioned above now also carry colour indices , thus we have factors of @xmath11 and @xmath12 associated to closed loops and to ` twists ' in a string of colour @xmath13 , respectively .
the defining relations of the tcbwm algebra can most easily be understood directly from the diagrammatic interpretation : they are given by considering all possible colourings of the diagrams corresponding to the defining relations of the ordinary bwm algebra .
in addition , we demand that any products leading to diagrams with colour mismatches vanish in the algebra .
this statement basically reduces to the requirement that the generators @xmath4 are indeed orthogonal projectors .
before we move on to the representations , let us briefly present the relations defining the bwm quotient . for the coloured braids ,
one has the cubic reduction relations @xmath14 and the temperley - lieb generators @xmath15 are given by quadratic expressions in the braids as follows @xmath16 where @xmath17 is related to @xmath11 and @xmath12 by @xmath18 which is a consequence of eqs .
( [ cubic ] ) and ( [ quad ] ) and the defining relations of the tl algebra .
we now present explicit expressions for representations of the tcbwm algebra .
we build these representations on a pair of representations of the ordinary bwm algebra , which are labeled by affine lie algebras @xmath19 , @xmath20 , and @xmath21 @xcite and underlie the corresponding series of @xmath0-matrices of refs . @xcite . thus the tcbwm representations are labeled by pairs @xmath22 , where @xmath23 and the index @xmath13 refers to the colour . given @xmath24 and @xmath25 ,
the representation still contains two parameters @xmath17 , which determine @xmath12 through @xmath26 and thereby @xmath11 by eq .
( [ q ] ) .
the two - colour bwm representations we are interested in act on the tensor product space , where @xmath27 is given by @xmath28 and is the representation space for the single - colour representations . in these representations , the generators ( [ gen ] )
act non - trivially only in two factors ( labeled by @xmath9 and @xmath10 ) of the @xmath8-fold tensor product , wherefore it suffices to give the expressions for two factors which thus are @xmath29 matrices which we denote by the same symbols ( [ gen ] ) without the site index @xmath9 .
we introduce two index sets @xmath30 to keep apart labels referring to different colours . for @xmath31 , we define @xmath32 such that the possible values of @xmath33 for both colours lie symmetrically around zero . furthermore , we introduce symbols @xmath34 and @xmath35 which are defined as follows : @xmath36 note that @xmath37 can only occur for representations labeled by @xmath38 because these have an odd @xmath39 . with these notations ,
we obtain a two - colour representation by setting @xmath40\ ; e_{\alpha,\alpha } \otimes e_{\alpha,\alpha } \nonumber \\ * & & \ ; + \ ; \sum_{\alpha\ne\beta\in i_c}\ ; [ 1+(q_c^{}-1)\,\delta_{\alpha,\beta'}]\ ; e_{\alpha,\beta } \otimes e_{\beta,\alpha } \nonumber \\ * & & \ ; -\ ; ( q_c^{}-q_c^{-1})\ ; \sum_{\alpha<\beta\in i_c}\ ; e_{\alpha,\alpha } \otimes e_{\beta,\beta } \nonumber \\ * & & \ ; + \ ; ( q_c^{}-q_c^{-1})\ ; \sum_{\alpha>\beta\in i_c}\ ; \varepsilon_{\alpha}\,\varepsilon_{\beta}\ , q_c^{\bar{\alpha}-\bar{\beta}}\ ; e_{\alpha',\beta } \otimes e_{\alpha,\beta'}\\ { \ ; b\mbox{\raisebox{.68em}{$\!\scriptscriptstyle -$}}{}^{(c , c ) } } & = & \sum_{\alpha\in i_c}\ ; q_c^{}\ , [ 1+(q_c^{-1}-1)\,\delta_{\alpha,\alpha'}]\ ; e_{\alpha,\alpha } \otimes e_{\alpha,\alpha } \nonumber \\ * & & \ ; + \ ; \sum_{\alpha\ne\beta\in i_c}\ ; [ 1+(q_c^{-1}-1)\,\delta_{\alpha,\beta'}]\ ; e_{\alpha,\beta } \otimes e_{\beta,\alpha } \nonumber \\ * & & \ ; -\ ; ( q_c^{}-q_c^{-1})\ ; \sum_{\alpha>\beta\in i_c}\ ; e_{\alpha,\alpha } \otimes e_{\beta,\beta } \nonumber \\ * & & \ ; -\ ; ( q_c^{}-q_c^{-1})\ ; \sum_{\alpha<\beta\in i_c}\ ; \varepsilon_{\alpha}\,\varepsilon_{\beta}\ , q_c^{\bar{\alpha}-\bar{\beta}}\ ; e_{\alpha',\beta } \otimes e_{\alpha,\beta'}\\ { \ ; b^{(c,\bar{c } ) } } & = & \sum_{\alpha\in i_c}\ ; \sum_{\beta\in i_{\bar{c}}}\ ; e_{\beta,\alpha } \otimes e_{\alpha,\beta}\\ { \ ; e^{(c,\bar{c } ) } } & = & \sum_{\alpha\in i_c}\ ; \sum_{\beta\in i_{\bar{c}}}\ ; \varepsilon_{\alpha}^{}\,\varepsilon_{\beta}^{}\ , q_c^{\bar{\alpha}}\ , q_{\bar c}^{-\bar{\beta}}\ ; e_{\alpha',\beta } \otimes e_{\alpha,\beta'}\end{aligned}\ ] ] where , as above , @xmath5 and @xmath6 .
the @xmath41 matrices @xmath42 have matrix elements @xmath43 .
the expressions for the generators @xmath44 follow from eq .
( [ quad ] ) .
the boltzmann weights of solvable vertex models derived from representations of the tcbwm algebra are encoded in the matrix elements of the @xmath0-matrix .
the baxterization of ref .
@xcite yields a general expression for a trigonometric @xmath0-matrix for _ any _ representation given in the previous section , with the restriction that the parameters @xmath45 have to be equal , i.e. , @xmath46 .
let us introduce two parameters @xmath47 and @xmath48 by @xmath49 where of course the values of @xmath12 are determined by eq .
( [ omega ] ) . denoting the spectral parameter by @xmath50
, the @xmath0-matrix has the form @xcite @xmath51{0em}{2em}\!\ ! \sum_{c=1}^{2}\;\;\right\ { \!\!\!\!\!\ ! & { \ ; p^{(c , c ) } } \;+\;\frac{\sin(\eta\lambda - u)}{\sin(\eta\lambda)}\ ; { \ ; p^{(c,\bar{c } ) } } \nonumber\\ * & & \left.\rule[0em]{0em}{2em } \ ; -\;\frac{\sin(u)}{2i\,\sin(\lambda)\,\sin(\eta\lambda)}\:\left(\ , e^{i(\eta\lambda - u)}\ , { \ ; b\mbox{\raisebox{.68em}{$\!\scriptscriptstyle + $ } } { } ^{(c , c)}}\ : -\ : e^{-i(\eta\lambda - u)}\ , { \ ; b\mbox{\raisebox{.68em}{$\!\scriptscriptstyle -$}}{}^{(c , c)}}\ , \right)\right . \nonumber\\ * & & \left.\rule[0em]{0em}{2em}\ ; + \ ; \frac{\sin(u)\,\sin(\eta\lambda - u)}{\sin(\lambda)\,\sin(\eta\lambda ) } \;{\ ; b^{(c,\bar{c } ) } } \;+\;\frac{\sin(u)}{\sin(\eta\lambda)}\;{\ ; e^{(c,\bar{c})}}\;\right\ } \label{r}\end{aligned}\ ] ] which is completely symmetric in the two colours .
it satisfies the quantum ybe @xmath52 where @xmath53 denotes the @xmath41 unit matrix .
these solutions of the ybe are crossing - symmetric @xcite with crossing parameter @xmath54 , and satisfy the inversion relation @xmath55 where the function @xmath56 is given by @xmath57
we presented explicit representations of the tcbwm algebra acting on tensor product spaces .
these representations are built on a pair of two representations of the ordinary bwm algebra . from any representation of this type
, we construct an @xmath0-matrix via the baxterization derived in ref .
@xcite , and hence a solvable lattice model of vertex type .
although the construction shows some similarity to the fusion procedure , these models are different from those constructed by fusion , compare e.g. ref .
@xcite for the fusion of the dilute @xmath58 models .
note that the @xmath0-matrices related to the dilute bwm algebra @xcite and to the dilute and two - colour tl algebras @xcite can be recovered as special cases of the tcbwm result , see ref .
@xcite .
as already mentioned in the introduction , we expect that the @xmath0-matrices obtained in this way are not new , but related to other trigonometric @xmath0-matrices within the lists given in refs .
@xcite by gauge transformations .
it would be interesting to know the precise relationship between these @xmath0-matrices and to understand their construction on the level of the underlying quantum groups .
h. n. v. temperley and e. h. lieb , `` relations between the ` percolation ' and ` colouring ' problem and other graph - theoretical problems associated with regular planar lattices : some exact results for the ` percolation ' problem '' , _ proc .
( london ) _ * a 322 * ( 1971 ) 251280 . | past fifteen years since baxter s book relevance of yang - baxter equation in theory solvable two - dimensional lattice models realized connections to mathematics physics including quantum groups knot link invariants quantum field theories with elastic scattering discovered algebraic techniques solutions ybe interesting approach based on braid - monoid algebras procedure baxterization reduces problem finding representations yang - baxter algebra to constructing representations of bma examples are temperley - lieb algebra birman - wenzl - murakami algebra ` dilute ' two - colour ' generalizations present expressions for vertex - type ' representations of two - colour bwm algebra representations acting on tensor product space complementary to content of ref @xcite rsos restricted solid - on - solid ) type representations considered space spanned by paths graphs representations constructed by combining representations ordinary bwm algebra solvable vertex models of bazhanov jimbo baxterization of tcbwm algebra @xcite yields solutions of ybe solvable lattice models of vertex typecontrast to rsos representations in ref . @xcite expect solvable vertex models new expect similar scenario as dilute case @xmath0-matrices from dilute bwm algebra related by gauge transformation to known @xmath0-matrices demonstrated in ref . repeat complete definition tcbwm algebra rough sketch idea see refs. @xcite for details ordinary bma @xcite generated by two generators braids monoids operators braid relations defining relations tl algebra relations involving both generators imposed relations continuous moves in diagrammatic presentation generators acting on arrays strings connection to theory of knot and link invariants defining relations involve two central elements denoted by @xmath1 and @xmath2 associated to closed loops twists bwm algebra is quotient of bma braid generators satisfy cubic reduction relation tcbwm algebra two - colour generalization of bwm algebra two species colours of strings generators carry colour indices @xmath3 and additional generators @xmath4 act as projection operators on certain colours@xmath5 label two colours @xmath6 denotes complementary colour generators carry label @xmath7 generator acts on two @xmath8 strings labeled @xmath9 and @xmath10 central elements carry colour indices factors @xmath11 and @xmath12 associated to closed loops twists in string colour @xmath13 defining relations of tcbwm algebra understood from diagrammatic interpretation considering colourings diagrams ordinary bwm algebra demand products leading to diagrams with colour mismatches vanish in algebra generators @xmath4 orthogonal projectors relations defining bwm quotient coloured braids cubic reduction relations @xmath14 temperley - lieb generators @xmath15 given by quadratic expressions in braids @xmath16 @xmath17 related to @xmath11 and @xmath12 by @xmath18 consequence of eqs [ cubic [ quad ] ) defining relations tl algebra explicit expressions for representations tcbwm algebra build representations on representations ordinary bwm algebra labeled by algebras @xmath19 @xmath20 @xmath21 underlie @xmath0-matricestcbwm representations labeled by pairs @xmath22 @xmath23 @xmath13 colour @xmath24 @xmath25 representation contains two parameters @xmath17 determine @xmath12 @xmath26 @xmath11 by eq two - colour bwm representations act on tensor product space @xmath27 by @xmath28 representation for single - colour representations generators act in two factors labeled @xmath9 @xmath10 @xmath8-fold tensor product expressions for two factors @xmath29 matrices symbols without site index @xmath9 introduce two index sets @xmath30 labels colours @xmath31 @xmath32 possible values of @xmath33 for both colours around zero introduce symbols @xmath34 @xmath35 defined @xmath37 for representations labeled by @xmath38 odd @xmath39notations obtain two - colour representation setting @xmath40{\alpha, \nonumber\sum{\alpha\ne\beta\in i_c} 1+(q_c^{}-1)\delta{\alpha,\beta'}{\alpha,\beta\nonumber * q_c^{}-q_c^{-1})\sum{\alpha<\beta\in i_c} e_{\alpha,\alpha{\beta, \nonumber * q_c^{}-q_c^{-1})\sum{\alpha>\beta\in i_c} \varepsilon_{\alpha}_{\beta} q_c^{\bar{\alpha}-\bar{\beta}} e_{\alpha',\beta }{\alpha,\beta' b\mbox{\raisebox{.{$\\scriptscriptstyle -$}}{^{(c )\sum_{\alpha\in i_c} q_c^{} 1+(q_c^{-1}-1)\delta_{\alpha,'}_{\alpha,{\alpha\nonumber\sum_{\alpha\ne\beta\in i_c} 1+(q_c^{-1}-1)\delta_{\alpha,\beta'}]{\alpha,\beta{\beta\alpha \nonumber q_c^{}-q_c^{-1})\sum_{\alpha>\beta\in i_c}{\alpha,\alpha{\beta\nonumber q_c^{}-q_c^{-1})\sum_{\alpha<\beta\in i_c}\varepsilon_{\alpha}{\beta}{\bar{\alpha}-\bar{\beta}}_{\alpha',\beta e_{\alpha,\beta' b^{(c,\bar{c } ) \sum_{\alpha\in i_c}{\beta\in{\bar{c}}} e_{\beta,\alpha e_{\alpha,\beta e^{(c,\bar{c } )\sum_{alpha\in i_c}{\beta\in\bar{c}}}\varepsilon{\alpha}{\beta}^{ q_c^{\bar{\alpha}}{{-\bar{\beta}} e_{\alpha',\beta{\alpha,\beta'}\end{aligned} @xmath5 @xmath6 @xmath41 matrices @xmath42 have matrix elements @xmath43 expressions generators @xmath44 follow from eq boltzmann weights of vertex models tcbwm algebra encoded in matrix elements @xmath0-matrix baxterization ref @xcite yields general expression trigonometric @xmath0-matrix restriction parameters @xmath45 equal @xmath46 introduce parameters @xmath47 @xmath48 by @xmath49 values @xmath12 determined by eqspectral parameter @xmath50 @xmath0-matrix form @xcite @xmath51{0em}{2em}\sum_{c=1}^{2} p^{(c , c )\frac{\sin(\eta\lambda - u)\eta\lambda)}{(c{c\nonumber\\ \left.\rule[0em]{0em}{2em\frac{\sin(u)}\sin(\lambda)\lambda)}\left(\eta\lambda - u)} b\mbox{\raisebox{.68em}$\scriptscriptstyle + $(c , c)}}{-i(\eta\lambda - u)} b\mbox{\raisebox{.68em}\scriptscriptstyle -$}}{(c , c)}}\right\nonumber\\ *\left.\rule[0em]{2em}\frac{\sin(u(\eta\lambda b^{(c,\bar{c }\frac{\sin(u\lambda)} e^{(c,\bar{c}}\end{aligned} symmetric two colours satisfies quantum ybe @xmath52 @xmath53 @xmath41 unit matrix solutions crossing - @xcite parameter @xmath54 satisfy inversion relation @xmath55 function @xmath56 given @xmath57 representations tcbwm algebra tensor product spaces representations built on two representations ordinary bwm algebra construct @xmath0-matrix baxterization ref. @xcite solvable lattice model vertex type construction similarity fusion procedure models different from fusion compare ref . @xcite fusion dilute @xmath58 models @xmath0-matrices related to dilute bwm algebra @xcite two - colour tl algebras recovered as special cases tcbwm result see ref . @xcitementioned introduction expect @xmath0-matrices obtained not new related trigonometric-matrices lists refs gauge transformations interesting know precise relationship-matrices understand construction underlying quantum groups n v. temperley e. h. lieb relations percolation colouring ' problem - theoretical problems regular planar lattices exact results percolation ' problem '' proc. london ) 322 * ( 1971 ) 251280 . | 0.608224 | 3,288 | 1,780 | , Saving $0.1 in GPT-4. |
177,715 | we study and compare the discovery potential for heavy neutral gauge bosons ( @xmath0 ) at the various @xmath1 and @xmath2 colliders that have been proposed .
typical search limits for the @xmath1 colliders are @xmath3 with the large variation reflecting the model dependence of the limits .
the search limits for the @xmath2 colliders are slightly lower .
polarization and flavour tagging are important in realizing the highest discovery limits possible .
because the search limits are based on indirect inferences of deviations from standard model predictions , they are sensitive to systematic errors . | extended gauge symmetries and the associated heavy neutral gauge bosons , @xmath0 , are a feature of many extensions of the standard model such as grand unified theories , left - right symmetric models , and superstring theories .
if a @xmath0 were discovered it would have important implications for what lies beyond the standard model .
it is therefore important to study and compare the discovery reach for extra gauge bosons at the various facilities that are under consideration for the future . included in the list of proposed facilities considered at the snowmass96 workshop
are high energy @xmath1 and @xmath4 colliders . in this report
we update previous studies @xcite to include these high energy lepton colliders .
quite a few models predicting extra gauge bosons exist in the literature . we will present search limits for several of these models which , although far from exhaustive , form a representative set for the purposes of comparison .
the models chosen for study are listed below but for details we direct the interested reader to ref . @xcite and references therein .
( i ) : : effective rank-5 models originating from @xmath5 grand unified theories are conveniently labelled in terms of the decay chain @xmath6 .
thus , the @xmath0 charges are given by linear combinations of the @xmath7 and @xmath8 charges resulting in the @xmath0-fermion couplings : @xmath9 where @xmath10 is a free parameter which lies in the range @xmath11 .
specific models of interest are model @xmath12 ( @xmath13 ) corresponding to the extra @xmath0 of @xmath14 , model @xmath15 ( @xmath16 ) corresponding to the extra @xmath0 of @xmath5 , and model @xmath17 ( @xmath18 ) corresponding to the extra @xmath0 arising in some superstring theories .
( ii ) : : the left - right symmetric model ( lrm ) extends the standard model gauge group to @xmath19 .
it has a parameter @xmath20 with @xmath21 .
we assume @xmath22 in our analysis which corresponds to strict left - right symmetry .
( iii ) : : the alternative left - right symmetric model ( alrm ) originates from @xmath5 gut s and is also based on the electroweak gauge group @xmath19 . here
the assignments for @xmath23 differ from that of the usual lrm due to an ambiguity in how the fermions are embedded in the 27 representation of @xmath5 .
( iv ) : : the `` sequential '' standard model ( ssm ) consists of a @xmath0 which is just a heavy version of the sm @xmath24 boson with identical couplings .
although it is not a realistic model it is often used as a benchmark and for purposes of comparison .
( v ) : : the harvard model ( harv ) is based on the gauge group @xmath25 , i.e. , left - handed leptons ( quarks ) transform as doublets under @xmath26 ( @xmath27 ) and singlets under @xmath27 ( @xmath26 ) , and right - handed fields are singlets under both groups .
this model has a parameter @xmath28 which lies in the range @xmath29 .
we take @xmath30 in our calculations .
the @xmath0 is purely left handed in this model .
there are numerous other models in the literature predicting @xmath0 s but the subset described above have properties reasonably representative of broad classes of models , at least for the purposes of comparing search limits of high energy colliders .
at @xmath1 colliders searches for @xmath0 s are indirect , being inferred from deviations from the standard model predictions due to interference between the @xmath0 propagator and the @xmath31 and @xmath24 propagators @xcite .
this is similar to pep / petra seeing the standard model @xmath24 as deviations from the predictions of qed .
the basic process is @xmath32 where @xmath33 could be leptons @xmath34 or quarks @xmath35 and @xmath36 denotes the @xmath37 polarization .
the cross section for the basic process is given by : @xmath38 where @xmath39 where the @xmath40 are the sm @xmath24 couplings and the @xmath41 are the @xmath0 couplings .
for right - handed electrons make the substitutions @xmath42 and @xmath43 . from these expressions
we can obtain everything else we need .
for example , the total differential cross section is just @xmath44\ ] ] and the various polarized or unpolarized total cross sections can be obtained by integrating these expressions .
the spin averaged ( unpolarized ) cross section is given by : @xmath45\ ] ] from the basic reactions a number of observables can be used to search for the effects of @xmath0 s : * @xmath46 the cross section to specific final state fermions * @xmath47 the ratio of the hadronic to the qed point cross section * @xmath48 the forward - backward asymmetry to fermion @xmath33 in the final state ; @xmath49 d\cos\theta { { d\sigma}\over{d\cos\theta } } } \over { \left [ { \int^1_0 -\int^0_{-1 } } \right ] d\cos\theta { { d\sigma}\over{d\cos\theta } } } } \ ] ] * @xmath50 the left - right asymmetry with fermion @xmath33 in the final state ; @xmath51 * @xmath52 the left - right asymmetry with hadrons in the final state * @xmath53 the polarized forward - backward double asymmetry with fermion @xmath33 in the final state ; @xmath54 * @xmath55 the tau polarization in these expressions we generally take the index @xmath56 and @xmath57`sum over all hadrons ' to indicate the final state fermions . to obtain discovery limits for new physics , which in this case means evidence for extra neutral gauge bosons , we look for statistically significant deviations from standard model expectations . in fig . 1 a number of observables
are shown with their standard model values and for various @xmath0 s as a function of the @xmath0 mass .
the error bars are based on the statistics expected in the standard model .
( note that there is a cheat here in that i did not include @xmath58 and @xmath59-quark tagging efficiences in the statistical errors so that in reality some of the error bars should really be bigger than what is shown . )
what is important to note here is that the different observables have different sensitivities to the different models .
for example , of the models shown , @xmath60 is most sensitive to @xmath61 while @xmath62 is most sensitive to @xmath63 .
similarly , @xmath64 is most sensitive to @xmath61 while @xmath65 is most sensitive to @xmath63 . therefore to have the highest possible reach for the largest number of possible models it is important to include all possible observables .
we quantify the sensitivity to an extra gauge boson by comparing the predictions for various observables assuming the presence of a @xmath0 to the predictions of the standard model and constructing the @xmath66 figure of merit ; @xmath67 where the sum is over observables included in the @xmath66 and @xmath68 is the experimental error of the observable .
the contributions of some observables to the @xmath66 for @xmath63 , @xmath69 , @xmath70 , and @xmath61 are shown in figure 2 .
there are several points to note from this figure .
first , we again see that the observables have different sensitivities to the various models so that it is important to consider as many observables as possible .
second , the polarization asymmetries are in many cases the most sensitive observables so that polarization is potentially very important for searches for @xmath0 s . and finally , flavour tagging can contribute a considerable amount of information . to obtain `` discovery '' limits for the @xmath1 case
we include the eighteen observables : @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath62 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , and @xmath55 . for the @xmath2 case we included the ten observables that did not involve polarized electrons .
in calculating the @xmath66 we assumed 90% electron polarization , 35% @xmath58-tagging efficiency and 60% @xmath59-tagging efficiency . the 99% c.l .
discovery limits are shown in figure 3 .
only statistical errors are considered in obtaining the limits shown . because search limits obtained at @xmath1 colliders are indirect , based on deviations from the standard model in precision measurements , they are sensitive to both statistical and systematic errors .
if , for example , the nlc integrated luminosity is reduced from 50 fb@xmath87 to 10 fb@xmath87 ( 200 fb@xmath87 to 50 fb@xmath87 ) for the 500 gev ( 1 tev ) case , the search limits are reduced by about 33% . including a 5% systematic error in cross section measurements due to , for example , luminosity uncertainties , and a 2% systematic error in asymmetries where systematic errors partially cancel
, typically reduces these numbers by 30% although in some cases ( the ssm ) it can reduce the limits by up to 50% .
clearly , systematic errors will have to be kept under control for high precision measurements .
finally , we did not include initial rate radiation ( isr ) .
rizzo has found that including isr can lower the search reach by 1520% @xcite .
one sees that the discovery limits obtained at @xmath1 colliders can be quite substantial although they can be quite model dependent .
for example , for the @xmath88 , @xmath89 so that either @xmath90 or @xmath91 . for @xmath92
sufficiently far away from the @xmath24 pole deviations are dominated by @xmath93 and @xmath94 interference which is proportional to @xmath95 . since for the photon @xmath96 ,
when @xmath90 is also equal to 0 deviations from the standard model become small .
in this report we have shown that high luminosity lepton colliders can put limits on the existence of extra gauge bosons via deviations of measurements from their standard model values .
the so called discovery limits can be many times the centre of mass energy of the collider . polarization and flavour
tagging are important in realizing the highest discovery limits possible . since bounds on @xmath0 s are based on precision measurements they are quite sensitive to measurement errors . given the high statistics , to achieve the highest possible discovery limits , it will be crucial to minimize systematic errors .
finally , we note that while the non - observation of deviations from the sm will put constraints on the existence of @xmath0 s , if deviations are found , disentangling the underlying physics will be far from trivial .
the author gratefully acknowledges helpful conversations and communications with dean karlen and joanne hewett and the encouragement of tom rizzo and linda hopson .
hewett , _ proceedings of the workshop on physics and experiments with linear @xmath1 colliders _ , ed . f.a .
harris , s.l .
olsen , s. pakvasa , and x. tata , april 26 - 30 , 1993 , waikoloa , hawaii ( world scientific , singapore , 1993 ) p 246 ; f. del aguila , m. cveti , and p. langacker , _ ibid _ , p. 490 .
f. boudjema , b.w .
lynn , f.m .
renard , c. verzegnassi , z. phys . * c48 * , 595 ( 1990 ) ; a. blondel , f.m .
renard , p. taxil , and c. verzegnassi , nucl . phys . *
b331 * , 293 ( 1990 ) ; g. belanger and s. godfrey , phys . rev . * d34 * , 1309 ( 1986 ) ; * d35 * , 378 ( 1987 ) ; p.j . franzini and f.j .
gilman , phys .
* d35 * , 855 ( 1987 ) ; m. cveti and b. lynn , phys . rev . *
d35 * , 1 ( 1987 ) ; b.w .
lynn and c. verzegnassi , phys . rev .
* d35 * , 3326 ( 1987 ) ; t.g .
ibid _ , * 36 * , 713(1987 ) ; a. bagneid , t.k .
kuo , and g.t .
ibid _ , * 44 * , 2188 ( 1991 ) ; a. djouadi , a. leike , t. riemann , d. schaile and c. verzegnassi , z. phys .
* c56 * 289 ( 1992 ) ; a. leike , z. phys . *
c62 * , 265 ( 1994 ) ; j.l . hewett and t.g .
rizzo , _ proceedings of physics and experiments with linear colliders _ , ed .
r. orava _ et al .
_ , saariselk , finland ( world scientific , 1992 ) ; t.g .
rizzo , _ ibid _ ; j.l .
hewett and t.g .
rizzo , phys .
* d45 * , 161 ( 1992 ) ; f. cuypers , nucl . phys . *
b474 * , 72 ( 1996 ) ; a. leike and s. riemann , hep - ph/9604321 ; s. riemann , these proceedings ( hep - ph/9610513 ) . | extended gauge symmetries heavy neutral gauge bosons @xmath0 feature extensions standard model grand unified theories left - right symmetric models superstring theories if @xmath0 discovered implications beyond standard model important to study compare discovery reach extra gauge bosons at facilities future proposed facilities snowmass96 workshop high energy @xmath1 @xmath4 colliders report update studies @xcite include high energy lepton colliders models predicting extra gauge bosons exist present search limits for models representative set for comparison models chosen study listed below details direct @xcite references effective rank-5 models from @xmath5 grand unified theories labelled decay chain @xmath6 @xmath0 charges given by linear combinations @xmath7 @xmath8 charges @xmath0-fermion couplings @xmath9 @xmath10 free parameter in range @xmath11 models interest @xmath12 ) extra @xmath0 @xmath14 @xmath15 extra @xmath5 @xmath17 extra @xmath0 superstring theories left - right symmetric model extends standard model gauge group to @xmath19 parameter @xmath20 with @xmath21assume @xmath22 corresponds left - right symmetry alternative left - right symmetric model originates from @xmath5 gut based on electroweak gauge group @xmath19 assignments for @xmath23 differ due ambiguity fermions 27 representation @xmath5 sequential standard model consists @xmath0 heavy version of @xmath24 boson identical couplings not realistic used as benchmark comparison harvard model based on gauge group @xmath25 left - handed leptons transform as doublets under @xmath26 singlets right - handed fields are singlets under both groups model parameter @xmath28 in range @xmath29 take @xmath30 calculations @xmath0 left handed numerous other models predicting @xmath0 s properties representative of broad classes models comparing search limits high energy colliders @xmath1 colliders searches for @xmath0 s indirect deviations standard model predictions interference between @xmath0 propagator @xmath31 @xmath24 propagators similar to pep / petra standard model @xmath24 deviations from predictionsbasic process @xmath32 @xmath33 leptons @xmath34 quarks @xmath35 @xmath36 @xmath37 polarization cross section @xmath38 @xmath39 @xmath40 sm @xmath24 couplings @xmath41 @xmath0 couplings right - handed electrons substitutions @xmath42 @xmath43 expressions obtain everything total differential cross section @xmath44\ polarized unpolarized total cross sections integrating expressions spin averaged unpolarized cross section @xmath45\ reactions observables effects @xmath0 @xmath46 cross section final state fermions @xmath47 ratio hadronic qed point cross section @xmath48 forward - backward asymmetry fermion @xmath33 final state @xmath49\cos\theta @xmath50 left - right asymmetry with fermion @xmath33 final state @xmath51 @xmath52 left - right asymmetry hadrons final state @xmath53 polarized forward - backward double asymmetry fermion @xmath33 final state @xmath54 @xmath55 tau polarization index @xmath56 @xmath57`sum over hadrons final state fermionsobtain discovery limits for new physics evidence for extra neutral gauge bosons look for significant deviations from standard model expectations in fig . 1 observables shown with standard model values for @xmath0 s @xmath0 mass error bars based on statistics standard model include @xmath58 and @xmath59-quark tagging efficiences in statistical errors error bars should be bigger different observables have different sensitivities to models @xmath60 sensitive to @xmath61 @xmath62 sensitive to @xmath63 @xmath64 sensitive to @xmath61 @xmath65 sensitive to @xmath63 highest reach for largest number models important to include all observables quantify sensitivity to extra gauge boson comparing predictions for observables to standard model @xmath66 figure of merit @xmath67 sum over observables @xmath66 @xmath68 is experimental error of observable contributions of observables to @xmath66 for @xmath63 @xmath69 @xmath70 @xmath61 shown in figure 2 observables different sensitivities to models important to consider many observables polarization asymmetries most sensitive polarization important for searches for @xmath0 s flavour tagging informationobtain discovery limits for @xmath1 case include eighteen observables : @xmath71 @xmath86 @xmath55 @xmath2 case included ten observables not involve polarized electrons calculating @xmath66 assumed 90% electron polarization 35% @xmath58-tagging efficiency 60% @xmath59-tagging efficiency 99% c.l . discovery limits shown in figure 3 statistical errors considered limits search limits at @xmath1 colliders indirect based on deviations from standard model sensitive to statistical systematic errors nlc integrated luminosity reduced from 50 fb@xmath87 to 10 fb@xmath87 search limits reduced by about 33% 5% systematic error in cross section measurements to luminosity uncertainties 2% systematic error in asymmetries reduces numbers by 30% some cases can reduce limits by up to 50% systematic errors kept under control for high precision measurements include initial rate radiation ( isr ) including isr lower search reach by 1520% @xcite discovery limits at @xmath1 colliders substantial model dependent@xmath88 @xmath89 @xmath90 or @xmath92 far from @xmath24 pole deviations dominated by @xmath93 @xmath94 interference proportional to @xmath95 photon @xmath96 @xmath90 equal to 0 deviations from standard model small high luminosity lepton colliders extra gauge bosons deviations from standard model values discovery limits centre of mass energy collider polarization flavour tagging important highest discovery limits bounds @xmath0 s based precision measurements sensitive to errors high statistics highest discovery limits crucial minimize systematic errors non - observation of deviations sm @xmath0 s if deviations found disentangling physics trivial author acknowledges conversations dean karlen joanne hewett tom rizzo linda hopson hewett proceedings workshop on physics experiments with linear @xmath1 colliders harris olsen pakvasa tata april 26 - 30 1993 waikoloa hawaii world scientific singapore p 246 f del aguila cveti p langacker ibid p. 490 boudjema lynn renardverzegnassi phys c48 595 ( 1990 ) blondel renard taxil c. verzegnassi phys b331 293 ( 1990 ) g. belanger s godfrey phys . d34 1309 ( 1986 ) d35 378 ( 1987 ) franzini f gilman d35 855 ( 1987 ) cveti b. lynn d35 1 ( 1987 ) lynn c verzegnassi d35 3326 ( 1987 ) ibid 36 713(1987 ) bagneid kuo ibid 44 2188 ( 1991 ). djouadi leike riemann d. schaile c. verzegnassi phys c56 289 ( 1992 ) leike c62 265 ( 1994 ) j.l hewett t.g rizzo proceedings physics experiments linear colliders orava saariselk finland scientific 1992 ) t rizzo ibid phys . d45 161 ( 1992 ) f. cuypers . b474 72 ( 1996 ) a. leike s. riemann hep - ph/9604321riemann proceedings ph/9610513 | 0.500716 | 3,183 | 1,644 | , Saving $0.1 in GPT-4. |
177,717 | we report on the superconducting properties of amorphous nb@xmath0si@xmath1 thin films .
the normal - state resistance and critical temperatures can be separately adjusted to suit the desired application .
notably , the relatively low electron - phonon coupling of these films makes them good candidates for an `` all electron bolometer '' for cosmological microwave background radiation detection .
moreover , this device can be made to suit both high and low impedance readouts . | one of the long - standing challenges of cosmology is the comprehension of the forming of the universe . in this field ,
the study of the cosmological microwave background ( cmb ) is a precious source of information .
however , the measurement of this primordial radiation is now reaching the sensitivity limit of single pixel bolometer - type detectors .
numerous observations related to the millimetric domain are indeed limited by the photon noise and further improvement of the detectors therefore implies optimizing the focal plane filling factor by the use of large bolometric matrices . amongst all cmb measurements that the scientific community has yet to perform ,
the cmb b - mode polarization is probably the most constraining from the instrumental point of view .
detecting the signature of primordial gravitational waves yielding a b - type polarization is one of the utmost goals in today s cosmology and amongst the first objectives in the field . in order to meet this scientific challenge ,
the current trend followed by different collaborations is to develop a `` bolometric camera '' similar to ccds .
each pixel of this camera will consist in a single detector which sensitivity - measured through its noise equivalent power sensitivity ( nep ) - will be at the level of the experimental photon noise : about 10@xmath2 w/@xmath3 for terrestrial use and ten times lower for space - based use . by filling the telescope focal plane with thousands of such pixels , the instrument overall performance will thus be considerably increased .
the base structure of a single pixel for submillimetric measurements requires three components : a radiation absorber , a thermometer and a thermally isolated holder enabling a very weak coupling of the former two with respect to the cold bath .
the problem of the thermal decoupling is extremely critical since the sensitivity of a bolometer to an incident power @xmath4 is limited by its nep : nep@xmath5 and therefore directly depends on the thermal decoupling @xmath6 between the absorber and the thermometer on the one hand and the cold bath on the other . to date , all devices designed for this application are based on using phonons as vectors for energy transport between the different parts of the bolometer .
this implies that the thermal decoupling is achieved through very thin mechanical structures involving delicate clean room processes that are difficult to implement on macroscopic lengths such as those needed for this particular application .
we have recently proposed an innovative solution to this problem @xcite where these caveats are lifted : in the envisioned `` all electron bolometer '' , the incoming energy will be directly captured and measured in the sensor electron bath and the thermal decoupling will be ensured via electron - phonon decoupling at low temperature .
the idea of using electron - phonon decoupling has been introduced earlier @xcite , but in our imagined device , the absorption will also take place within the superconducting film , so that complicated antenna designs are no longer needed .
one possible device consists in a single superconducting thin film combining the absorber and the thermometer functions .
the incoming photons having energies larger than the superconducting gap ( @xmath7 ) @xcite , they will see the film normal resistance and the electromagnetic wave will be absorbed by the electrons .
the radiation - induced electronic temperature increase will be measured by the resistivity variation of the superconducting film , similarly to a transition - edge - sensor @xcite .
the superconducting film then has to meet four requirements : 1 . in order to be suitable for matrix fabrication
, it must have a size of the order of the wavelength ( @xmath8 1 mm ) ; 2 .
for the wave absorption to be optimum , the film must meet the vacuum impedance ( i.e. a sheet resistance of 377 @xmath9 per square ) ; 3 . it must have a competitive electron - phonon decoupling in order to ensure a small enough nep ; 4 .
its response time must be suitable for experimental observation ( typically , a few ms is acceptable ) . the first condition is easily met thanks to standard evaporation or lithography techniques . in the present paper , we will focus on the three other pre - requisites . the present work centers on superconducting amorphous nb@xmath0si@xmath1 thin films that have already shown interesting properties as thermometric thin films @xcite . by carefully adjusting the films composition @xmath10 and thickness @xmath11
, we show that the superconducting properties , namely the normal sheet resistance @xmath12 and the critical temperature @xmath13 can be independently adjusted to build a detector suitable for submillimetric measurements .
the amorphous nbsi thin films have been synthesized under ultrahigh vacuum ( 10@xmath14 to 10@xmath15 mbar ) by electron beam ( e - beam ) co - deposition of nb and si onto sapphire substrates coated with a 50-nm - thick sio underlayer .
each evaporation was controlled in situ by a dedicated set of piezoelectric quartz in order to precisely monitor the composition and the thickness of the deposition .
its homogeneity throughout the film volume was also carefully checked .
the in situ measurements were then systematically and successfully compared to ex situ rutherford back scattering ( rbs ) measurements .
the compositions ranged from 14% to 18% and thicknesses @xmath11 ranging from 45 to 1000 .
the films size was of a few mm@xmath16 and the samples have all been annealed at @xmath17 .
the electrical characteristics of the films have been measured down to below 10 mk using a dilution refrigerator .
resistance measurements were performed using a standard ac lock - in detection technique and all electrical leads were filtered from radio frequency at room temperature .
the results for the studied superconducting thin films are summarized in table [ tab:1 ] .
the superconducting critical temperature strongly depends on the niobium composition : for the studied samples , it has been varied from 34 mk ( for the 105 thick nb@xmath18si@xmath19 sample ) to 840 mk ( for the 125 thick nb@xmath20si@xmath21 sample ) .
`` thick '' films ( @xmath22 ) have a constant resistivity , and hence a normal sheet resistance @xmath12 inversely proportional to the thickness .
thinner films have a resistivity larger than the bulk value due to prominent 2d effects . as can be seen from figure [ fig:1].a .
, @xmath13 and @xmath12 can thus be separately tuned and adapted to the desired application . regarding sub - millimetric measurements ,
the 500 thick nb@xmath18si@xmath19 sample is an optimal candidate for wave absorption , while the 175 thick nb@xmath18si@xmath19 sample has a @xmath13 suitable for an operating temperature of 100 mk ( the working temperature for the planck satellite experiment @xcite ) ( figure [ fig:1].b . ) .
furthermore , it must be noted that the characteristics are here given for films annealed at @xmath17 .
a heat treatment at higher temperature renders the sample more insulating : the critical temperature is lowered and the normal resistance enhanced .
the annealing can therefore be used to fine - tune the desired @xmath13 and film resistance ( see for example @xcite ) .
it has been shown elsewhere @xcite that , for a 500 - thick nb@xmath23si@xmath24 sample , the resistivity and the superconducting critical temperature can be linearly tuned by the annealing temperature . when the annealing temperature is varied from @xmath17 to @xmath25 ,
the resistivity increases by 17% while the @xmath13 drops by 61% .
it is therefore conceivable that the annealing be used at the end of the detector fabrication process to fine - tune the adaptation to the operating temperature , without much affecting the normal resistance .
llll @xmath10 ( % ) & @xmath11 ( ) & @xmath12 ( @xmath9 ) & @xmath13 ( mk ) + 14 & 105 & 1896 & 34 + 14 & 150 & 1128 & 73 + 14 & 175 & 1065 & 92 + 14 & 500 & 349 & 236 + 15 & 75 & 2256 & 200 + 15 & 100 & 1476 & 250 + 15 & 250 & 630 & 335 + 15 & 500 & 287 & 470 + 15 & 1000 & 152 & 525 + 17 & 75 & 1615 & 375 + 17 & 125 & 831 & 600 + 18 & 45 & 2637 & 225 + 18 & 75 & 1291 & 604 + 18 & 125 & 673 & 840 + si@xmath19 samples.,scaledwidth=80.0% ] in order to test the competitiveness of these films as detectors , electron - phonon coupling measurements have been performed using i - v characteristics , following the procedure described in @xcite .
note that the films were current - biased , so that no phase separation ( such as reported in @xcite ) was observed .
for the considered samples , the electron - phonon coupling weakly depends on the niobium concentration or the film thickness . as is usual for disordered materials , the relation between the applied power @xmath4 on a film of volume @xmath9 and the electron and phonon temperatures is given by @xcite : @xmath26 with @xmath27 the electron - phonon coupling constant .
the thermal conductivity is then given by @xmath28 where @xmath29 is the thin film volume .
the response time @xmath30 then has been evaluated considering the relevant temperature dependance of the specific heat @xmath31 @xcite .
typical results are given in table [ tab:2 ] for the 175 thick nb@xmath18si@xmath19 film at 50 and 100 mk .
at 100 mk , this film is still close to mid - transition and has both a nep suitable for terrestrial telescope applications and a sufficiently short response time .
the nep can be improved down to a few 10@xmath32 @xmath33 by lowering the operating temperature to 50 mk , however at the cost of a slightly longer - but still acceptable - response time .
these films thus have performances comparable to currently used state - of - the - art bolometers and present the advantage of a relatively simple fabrication process .
they indeed can be synthesized on a regular substrate and do not require the use of membranes .
this considerably simplifies the fabrication of bolometer matrices with a large number of pixels .
lllll t ( mk ) & @xmath27 ( w.k@xmath34.@xmath35 ) & @xmath36 ( w.k@xmath37 ) & @xmath38 ( ms ) & nep ( @xmath39 ) + 50 & 26.8 & 10@xmath40 & 0.97 & 2.3@xmath41 + 100 & 41.8 & 9.8@xmath42 & 8.4 & 3.7@xmath43 +
regarding possible applications for radiation detection , the tunable superconducting properties of amorphous nb@xmath0si@xmath1 thin films allow the tuning of the operating temperature while maintaining an optimal incoming wave absorption .
moreover , the low electron - phonon coupling in these films enable us to take advantage from the electron - phonon decoupling to build an efficient bolometer with state - of - the - art nep and which does not require elaborate fabrication process .
another concern for these bolometer matrices fabrication is the adaptation of the pixel to the readout electronics . indeed ,
if thousands of pixels are to be implemented , the readout multiplexing is an important technological issue .
currently , the tes are mainly measured by a squid - based electronics . however a jfet - based electronics could be an interesting alternative .
an notable feature of the a - nb@xmath0si@xmath1 films concerns their adaptability to both considered electronic readouts .
indeed , the important parameter for radiation absorption is the sheet resistance .
however , the films can be designed to yield a high resistance sample ( typically a few m@xmath9 ) suitable for jfet readout through a large length / width ratio . on the contrary
, a small value of such a ratio will give a low resistance ( typically less than 1 @xmath9 ) sample that could be probed by a squid readout .
we have fabricated two such geometries ( figure [ fig:2 ] ) , based on 50 nm - thick a - nb@xmath18si@xmath19 films which superconducting temperatures have been tuned around 100 mk by a specific annealing at @xmath44 . the flexibility of these a - nb@xmath0si@xmath1 superconducting films and the versatility of the readout systems that could be used make them particularly interesting for applications in detection that could extend beyond cmb measurements .
s. marnieros , l. dumoulin , a. benoit , l. berge , p. camus , s. collin , a. juillard , and c.a .
marrache - kikuchi , `` all electron bolometer for radiation detection '' , j. phys : conf .
series , 150 , 012027 ( 2009 ) y.p .
gousev , g.n .
goltsman , b.s .
karasik , e.m .
gershenzon , a.d .
semenov , h.s .
barowski , r.s .
nebosis , ans k.f .
renk , `` quasioptical superconducting hot electron bolometer for submillimeter wave '' , int .
j. infrared millim .
waves , 17 , 317 ( 1996 ) s. marnieros , l. berge , a. juillard , and l. dumoulin , `` dynamical properties near the metal - insulator transition : evidence for electron - assisted variable range hopping '' , phys .
lett . , 84 , 2469 ( 2000 ) g. morgante , d. pearson , f. melot , p. stassi , l. terenzi , p. wilson , b. hernandez , l. wade , a. gregorio , m. bersanelli , c. butler , and n. mandolesi , `` cryogenic characterization of the planck sorption cooler system flight model '' , j. inst . , 4 , t12016 ( 2009 ) o. crauste , `` tude des transitions de phase quantiques supraconducteur - isolant , mtal - isolant dans des matriaux amorphes dsordonns proches de la dimension 2 '' , ph.d dissertation , universit paris - sud xi ( 2010 ) | cosmology is comprehension forming universe study of cosmological microwave background ( cmb ) is precious source information measurement primordial radiation reaching sensitivity limit of single pixel bolometer detectors observations millimetric domain limited by photon noise improvement implies optimizing focal plane filling factor large bolometric matrices cmb measurements cmb b - mode polarization most constraining detecting primordial gravitational waves b - type polarization is utmost goals cosmology first objectives field current trend to develop `` bolometric camera '' similar to ccds each pixel single detector sensitivity noise equivalent power sensitivity at level of experimental photon noise about 10@xmath2 w/@xmath3 for terrestrial use ten times lower for space - based use filling telescope focal plane with thousands pixels instrument performance base structure single pixel for submillimetric measurements requires three components radiation absorber thermometer thermally isolated holder weak coupling problem thermal decoupling critical sensitivity bolometer to incident power @xmath4 limited by nep : nep@xmath5 depends on thermal decoupling @xmath6 between absorber thermometer cold bathdevices for this application based phonons vectors for energy transport between parts bolometer implies thermal decoupling through thin mechanical structures delicate clean room processes difficult to on macroscopic lengths for proposed innovative solution @xcite caveats lifted in envisioned all electron bolometer incoming energy captured measured in sensor electron bath thermal decoupling via electron - phonon decoupling at low temperature idea of electron - phonon decoupling introduced earlier @xcite in imagined device absorption within superconducting film complicated antenna designs no needed possible device in single superconducting thin film combining absorber thermometer functions incoming photons energies larger than superconducting gap see film normal resistance electromagnetic wave absorbed by electrons radiation - induced electronic temperature increase measured by resistivity variation of superconducting film sensor superconducting film four requirements suitable for matrix fabrication size of wavelength ( 1 mm ) for film meet vacuum impedance 377 per square competitive electron - phonon decoupling small nep response time suitable for experimental observation few ms acceptablefirst condition met standard evaporation lithography techniques present paper focus on three pre - requisites work centers on superconducting amorphous nb@xmath0si@xmath1 thin films properties thermometric thin films adjusting composition thickness @xmath11 superconducting properties normal sheet resistance critical temperature adjusted detector for submillimetric measurements amorphous nbsi thin films synthesized under ultrahigh vacuum by electron beam co - deposition of nb si sapphire substrates 50-nm - thick sio underlayer evaporation controlled by piezoelectric quartz composition thickness homogeneity checked measurements compared to ex situ rutherford back scattering measurements compositions ranged 14% to 18% thicknesses 45 to 1000 films size few mm@xmath16 samples annealed at @xmath17 electrical characteristics measured below 10 mk dilution refrigerator resistance measurements standard ac lock - in detection technique electrical leads filtered from radio frequency at room temperature results superconducting films summarized in tablesuperconducting critical temperature depends on niobium composition samples varied from 34 mk to 840 mk 125 thick films ( @xmath22 ) have constant resistivity normal sheet resistance @xmath12 inversely proportional to thickness thinner films have resistivity larger than bulk due to 2d effects @xmath13 and @xmath12 can be separately tuned adapted to application 500 thick@xmath18si sample optimal for wave absorption 175 thick has @xmath13 suitable for operating temperature 100 mk planck experiment characteristics for films annealed at @xmath17 heat treatment at higher temperature renders sample more insulating critical temperature lowered normal resistance enhanced annealing can used to fine - tune @xmath13 and film resistance for 500 - thick sample resistivity superconducting critical temperature can be linearly tuned by annealing temperature annealing temperature varied from to @xmath25 resistivity increases by 17% @xmath13 drops by 61%conceivable annealing used end detector fabrication process tune adaptation operating temperature without affecting resistance @xmath10 @xmath11 @xmath12 mk ) + 14 105 1896 34 + 14 150 1128 73 14 175 1065 92 + 14 500 349 236 + 15 75 2256 200 15 100 1476 250 630 335 + 15 500 287 470 + 15 1000 152 525 17 75 1615 375 + 17 125 831 600 + 18 45 2637 225 + 18 75 1291 604 + 18 125 673 840@xmath19 samples=80.0% test competitiveness films electron - phonon coupling measurements performed i - v characteristics procedure films current - biased no phase separation observed electron - phonon coupling depends on niobium concentration film thickness relation applied power @xmath4 film volume electron phonon temperatures given @xcite @xmath26 @xmath27 constant thermal conductivity by @xmath28 @xmath29 thin film volume response time @xmath30 evaluated considering temperature dependance heat typical results table:2 for 175 thick nb@xmath18si@xmath19 film at 50 100 mk 100 mk film close to mid - transition nep suitable for terrestrial telescope short response time nep improved lowering operating temperature to 50 mk longer response time films performances comparable to state bolometers simple fabrication process synthesized on regular substrate require membranes simplifies fabrication bolometer matrices with large pixels t ( mk ) @xmath27@xmath34@xmath35 @xmath36 @xmath38 ms ) nep @xmath39 ) + 50 & 26.8 & 10@xmath40 0.97 & 2.3@xmath41 100 & 41.8 9.8@xmath42 8.4 & 3.7@xmath43 applications for radiation detection tunable superconducting properties of amorphous nb@xmath0si@xmath1 thin films allow tuning operating temperature optimal incoming wave absorption low electron - phonon coupling films efficient bolometer with state - - art nep elaborate fabrication process concern adaptation pixel to readout electronics if thousands of pixels readout multiplexing important issue measured by squid - based electronics jfet - based electronics could alternative a - nb@xmath0si@xmath1 films adaptability to electronic readoutsimportant parameter radiation absorption sheet resistance films high resistance sample m@xmath9 jfet readout large length width ratio small value ratio low resistance less than 1 @xmath9 sample probed squid readout fabricated two geometries 50 nm - thick a - nb@xmath18si@xmath19 films superconducting temperatures tuned around 100 mk annealing @xmath44 flexibility a - nb@xmath1 superconducting films versatility readout systems interesting for applications detection beyond cmb measurements marnieros dumoulin benoit berge camus collin juillard marrache - kikuchi electron bolometer for radiation detection j. : conf . series 150 , 012027 2009 gousev goltsman karasik gershenzon semenov barowski nebosis renk quasioptical superconducting hot electron bolometer submillimeter wave j infrared millim waves , 17 , 317 ( 1996 marnieros dumoulin dynamical properties near metal - insulator transition evidence electron - assisted variable range hopping phys84 2469 2000 ) morgante pearson melot stassi terenzi wilson hernandez wade gregorio bersanelli butler mandolesi cryogenic characterization planck sorption cooler system flight model j inst . 4 t12016 2009 transitions phase quantiques supraconducteur isolant isolant matriaux amorphes 2 ph.d dissertation universit paris - sud 2010 | 0.505406 | 3,220 | 1,642 | , Saving $0.1 in GPT-4. |
177,718 | the application of optimization techniques derived from the study of euclidean full steiner trees to macromolecules like proteins is reported in the present work .
we shall use the concept of euclidean steiner ratio function ( srf ) as a good lyapunov function in order to perform an elementary stability analysis .
* keywords : * steiner , biomacromolecular structure , full trees , geometric chirality . | nature has followed mathematical principles of structural organization in the construction of macromolecular configurations .
our proposal in the present work is the modelling of the folded stage of proteins by some combinatorial optimization techniques associated to euclidean full steiner trees @xcite .
this means that henceforth we take the 3-dimensional euclidean space @xmath0 as our metric manifold .
the analysis to be undertaken can be summarized by the trial of obtaining the potential energy minimization of a protein structure through the problem of length minimization of an euclidean steiner tree @xcite .
our fundamental pattern of input points will be given by sets of evenly spaced points along a right circular helix of unit radius .
we have , @xmath1 where @xmath2 is the angular coordinate and @xmath3 stands for the pitch of the helix .
we also use the result of steiner points belonging to another helix of the same pitch and smaller radius or @xmath4 where @xmath5 the function @xmath6 above is easily obtained from the requirement of meeting edges at an angle of @xmath7 on each steiner point . to be rigorous ,
we should write , @xmath8 where the max above should be understood as a piecewise choice of the maple@xmath9 software .
we now introduce a generalization of the formulae above by thinking on subsequences of input and steiner points , corresponding to non - consecutive points . these subsequences are of the form : @xmath10 @xmath11 where @xmath12 , @xmath13 are the number of intervals of skipped points before the present point on each subsequence and @xmath14 is the number of skipped points .
we also have : @xmath15;\,\,\,\,\,\,\,\,\,1\leq l_s\leq l_{s max } = \left[\frac{n - k-2}{m}\right]\\ \nonumber\\0\leq j\leq m-1 , \,\,\,\,\,\,\,1\leq k \leq m \nonumber\end{aligned}\ ] ] and the square brackets stand for the greatest integer value .
the sequences corresponding to eqs.(1.1 ) and ( 1.2 ) are of course included in the scheme above .
they are @xmath16 and @xmath17 , respectively . in the general case
, we can define new sequences of @xmath18 and @xmath19 points instead those given by eqs.(1.1 ) and ( 1.2 ) .
we shall have respectively , @xmath20 the present development is independent of a specific coordinate representation of the points .
if we now assume helical point sets whose points are evenly spaced along right circular helices , we get @xmath21 @xmath22 the function @xmath23 is obtained through the same requirement of meeting edges at @xmath7 on each steiner point .
we have , analogously , @xmath24 where @xmath25 in figure ( 1 ) below we show some sequences of input points for @xmath26 .
[ hp ] from eqs.(2.5 ) and ( 2.6 ) and figure ( 1 ) , we can write for the length of the spanning trees @xmath27+(m-1)\sqrt{\alpha^2\omega^2+a_1 + 1}.\ ] ] the length of the steiner trees is then @xmath28\right)\\ & + & \sqrt{m^2\alpha^2\omega^2+r_m(\omega,\alpha)(a_m+1)}\,\,\,.\,\,\sum_{k=1}^{m}\left[\frac{n - k-2}{m}\right]\nonumber\\ & + & 2\sqrt{m^2\alpha^2\omega^2+(1-r_m(\omega,\alpha))^2+r_m(\omega,\alpha)(a_m+1)}\nonumber.\end{aligned}\ ] ] after using some useful relations like @xmath29=n - m\ ] ] @xmath30=n - m-2\ ] ] and taking the limit for @xmath31 , we get @xmath32 @xmath33 by following the prescriptions for writing the steiner ratio , we can write for the steiner ratio function of very large helical point sets with points evenly spaced along right circular helices @xmath34 where the @xmath35 process above should be understood in the sense of a piecewise function formed by the functions corresponding to the values @xmath36
eq.(2.15 ) is our proposal for a steiner ratio function ( srf ) @xcite .
it allows for an analytic formulation of the search of the steiner ratio which is then defined as the minimum of the srf function , eq.(2.15 ) .
actually , there is a further restriction to be imposed on function ( 2.15 ) in order to characterize it as an useful srf function .
this restriction is that we should consider only full steiner trees , i.e. , non - degenerated steiner trees in which there are exactly three edges meeting at each steiner point .
this restriction can be imposed on the spanning trees , by requesting that the angle @xmath37 between consecutive edges formed with the points @xmath38 as vertices should be lesser than @xmath7 .
we have @xmath39 in figure ( 2 ) below we can see the restrictions corresponding to eq.(2.16 ) , for @xmath40 .
the horizontal line is @xmath41 .
[ hp ] the @xmath42 spanning tree is the only one which corresponds to full steiner trees in a large region of the @xmath2-interval convenient for our work .
the other trees , @xmath43 correspond to forbidden regions in the same @xmath2-interval .
the corresponding steiner trees to be obtained from the positions of the points @xmath44 and @xmath38 are necessarily degenerate and should not be taken into consideration .
thus , the prescription ( 2.15 ) for the srf function turns into @xmath45 where @xmath46 the function ( 2.17 ) has a global minimum in the point @xmath47 and @xmath48 for a proof see @xcite .
the last value corresponds to the famous main conjecture of ref .
@xcite about the value of the steiner ratio in 3-dimensional euclidean space .
it lead us also to think that nature has solved the problem of energy minimization in the organization of intramolecular structure by choosing steiner trees as an intrinsic part of this structure @xcite .
in the following we continue to work in a @xmath49 manifold with an euclidean distance definition .
let us now introduce a tree as that of figure ( 3 ) [ hp ] there are @xmath18 input points ( position vectors @xmath50 ) and @xmath51 steiner points ( position vectors @xmath52 ) .
if @xmath53 is not an integer number , there is not a tree with these @xmath18 , @xmath54 values @xcite . in figure
( 3 ) with @xmath55 , we assume @xmath18 to be a feasible value .
the knowledge of the steiner problem tell us that this tree structure is not stable since its total length can be reduced by decreasing the number @xmath54 @xcite .
the usual stable steiner problem corresponds to @xmath56 . in this section
we shall give another proof of this fact by exploring the concept of a steiner network with physical interaction among its vertices .
the structure depicted at figure ( 3 ) is a representative of the network which models the fundamental interactions inside a biomacromolecule .
let us consider the interaction of this structure with similar structures .
let the resulting interaction forces as applied to input and steiner points be @xmath57 , @xmath58 , respectively and let @xmath59 be the length of an edge between a steiner point and an input point on its neighbourhood .
we have the following identities : @xmath60 @xmath61 where @xmath62 , @xmath63 , @xmath64 , @xmath65 stand for the modulus of the parallel components to the edges of the resulting forces @xmath57 , @xmath58 , respectively .
the total length of the tree above is @xmath66 from eqs.(2.20 ) , ( 3.1 ) , we can write the total length in the form @xmath67 we now specialize this set of applied forces at the vertices as being collinear with the edges joining them , or @xmath68 is the elastic constant .
the assumption of local equilibrium of these forces lead to the conditions : @xmath69 @xmath70 @xmath71 this is a set of generalized fermat problems or steiner problems @xcite . for this equilibrium configuration , eq.(3.3 )
turns into @xmath72 the stability of this equilibrium configuration under a variation of the applied forces can be tested by latexmath:[\[\delta l=\sum_{j=1}^{n}\mathbf{r}_j\,\,\cdot\,\,\delta\hat{f}_{j we take cartesian coordinates for the @xmath49 vectors @xmath74 , @xmath75 and we consider the three independent variations @xmath76 , @xmath77 , @xmath78 in the coordinates of the steiner points .
the corresponding variations in the length of the tree are of the form @xmath79(x_j - x_{s_k})}{||\mathbf{r}_j-\mathbf{s}_k||^3}=0\ ] ] and two other analogous expressions for the variations @xmath77 , @xmath78 . from the arbitrariness of these variations we can write , @xmath80}{||\mathbf{r}_j-\mathbf{s}_k||^3}=0\,.\ ] ] we can also write @xmath81 we now write the position vectors @xmath50 , @xmath52 for the configuration depicted at figure ( 3 ) .
the points can be taken as evenly spaced along right circular helices which radii are 1 and @xmath82 , respectively .
we have , @xmath83 @xmath84 @xmath85 is a function which can be derived from the equilibrium conditions in eqs.(3.7)(3.9 ) . for @xmath56
there is only one solution given by @xmath86 this solution coincides with eq.(1.3 ) . for @xmath87
, another useful solution could be obtained from the equations : @xmath88 curiously , nature has chosen this solution for @xmath89 to keep sure of partial equilibrium of side chains between the amide plane conformation in proteins @xcite .
for the configuration given by eqs.(3.15)(3.16 ) ,
eq.(3.15 ) can be written as @xmath90 where the geometrical object @xmath91 can be written in the coordinates of eqs.(3.16 ) and ( 3.17 ) as @xmath92 @xmath93r_p^2+\alpha^2\omega^2[2r_p(j\!\!-\!\!1)k\cos(j\!\!-\!\!1\!\!-\!\!k)\omega\!\!-\!\!k^2]}{[1+r_p^2\!\!-\!\!2r_p\cos(j\!\!-\!\!1\!\!-\!\!k)\omega+\alpha^2\omega^2 ( j\!\!-\!\!1\!\!-\!\!k)^2]^{3/2}}.\ ] ] @xmath94 to each @xmath95-value , there will be a term @xmath96 which dominates the sum above . however , we can not have @xmath96 for @xmath87 .
this can be seen from the fact for a vertex @xmath97 ( @xmath98 ) there are ( @xmath99 ) nearest external vertices @xmath100 .
the sequence of their consecutive position vectors is @xmath101 and the requirement @xmath96 corresponds to an integer @xmath54-value only for @xmath56 .
this @xmath56 case which is known to correspond to the most stable problem @xcite has as a possible configuration the figure ( 4 ) below [ hp ]
we have stressed on some past publications that there is a self - consistent treatment of the intramolecular organization of biomacromolecules in terms of steiner networks .
this representation is able at deriving information concerning its stability and evolution .
the supporting facts for stability are now well - established and the ideas related to the evolution of macromolecules are in their way to be developed and accepted as a preliminary theory of molecular evolution .
the missing subject is a full description of geometric chirality and in order to unveil some of its properties , we have proposed to study the influence of some proposals for chirality measure on the dynamics of optimization problems .
these are aimed at studying the structures which energy is around the assumed energy of the minimum solution and the variation process of the chiral properties in the neighbourhood of this minimum .
we think that this research line is worth of serious scientific work and should take advantage of the best efforts of very good scientific researchers for some years .
mondaini , r.p . : the disproof of a conjecture on the steiner ratio in @xmath102 and its consequences for a full geometric description of macromolecular chirality .
biomat symp .
* 2 * , 101177 ( 2003 ) . | nature followed mathematical principles macromolecular configurations proposal modelling folded stage proteins by combinatorial optimization techniques euclidean full steiner trees take 3-dimensional euclidean space @xmath0 as metric manifold analysis summarized potential energy minimization protein structure through length minimization euclidean steiner tree fundamental pattern input points given by evenly spaced points along circular helix , @xmath1 @xmath2 angular coordinate @xmath3 pitch helix use result of steiner points helix same pitch smaller radius @xmath4 @xmath5 function @xmath6 obtained from meeting edges at angle @xmath7 on each steiner point, @xmath8 max piecewise choice of maple@xmath9 software generalization formulae subsequences of input steiner points non - consecutive points subsequences @xmath10 @xmath11 @xmath12 @xmath13 number intervals skipped points before present point @xmath14 number skipped points@xmath15\leq l_s max = \left[{n - k-2}{m}\right\nonumber\\0\leq j\leq m-1\leq k \leq m \nonumber\end{aligned} brackets greatest integer value sequences eqs.(1.1 ) ( 1.2 ) included scheme @xmath16 @xmath17 define new sequences @xmath18 @xmath19 points eqs.(1.1 1.2 ) @xmath20 development independent coordinate representation assume helical point sets evenly spaced circular helices @xmath21 @xmath22 function @xmath23 meeting edges at @xmath7 each steiner point @xmath24 @xmath25 figure ( 1 ) sequences input points @xmath26 eqs.(2.5 ) ( 2.6 ) figure 1 ) length spanning trees @xmath27+(m-1)\sqrt{\alpha^2\omega^2+a_1 + 1} length steiner trees @xmath28\right)\\\sqrt{m^2\alpha^2\omega^2+r_m(_m+1)}\\sum_{k=1}^{m}\left[\frac{n - k-2}{m}\right]\nonumber\\ + 2\sqrt{m^2\alpha^2\omega^2+(1-r_m(\omega,\alpha))^2+r_m(_m+1)}\nonumber.\end{aligned} useful relations @xmath29=n - m @xmath30=n - m-2 limit @xmath31 @xmath32 @xmath33 prescriptions steiner ratio function large helical point sets evenly spaced circular helices @xmath34 @xmath35 process piecewise function functions values @xmath36 eq.(2.15 ) proposal steiner ratio function allows formulation search steiner ratio defined minimum srf function further restriction on function ( 2.15 ) useful srf function consider full steiner trees non - degenerated three edges meeting at each steiner point restriction spanning trees angle @xmath37 between consecutive edges points @xmath38 vertices lesser than @xmath7 @xmath39 figure ( 2 ) restrictions eq.(2.16 ) @xmath40 horizontal line @xmath41 @xmath42 tree corresponds to full steiner trees in large region @xmath2-interval for work other trees @xmath43 correspond to forbidden regions steiner trees from points @xmath44 and @xmath38 degenerate prescription ( 2.15 ) for srf function turns into @xmath45 @xmath46 function ( 2.17 ) global minimum in point @xmath47 and @xmath48 see @xcite last value corresponds to conjecture about steiner ratio in 3-dimensional euclidean space nature solved energy minimization intramolecular structure choosing steiner trees continue work in @xmath49 manifold with euclidean distance definition introduce tree figure ( 3 ) @xmath18 input points @xmath51 steiner points ) if @xmath53 not integer number tree with @xmath18 @xmath54 values figure ( 3 ) with @xmath55 assume @xmath18 feasible value tree structure not stable total length reduced by decreasing number @xmath54 @xcite usual stable steiner problem corresponds to @xmath56 steiner network with physical interaction among vertices structure at figure ( 3 ) representative network models fundamental interactions inside biomacromoleculeconsider interaction structure with similar structures forces to input steiner points @xmath57 @xmath58 @xmath59 length edge between steiner point input point identities : @xmath60 @xmath61 @xmath62 for modulus parallel components to edges forces @xmath57 total length tree is @xmath66 from eqs.(2.20 3.1 total length @xmath67 specialize applied forces at vertices collinear edges joining @xmath68 elastic constant assumption local equilibrium to conditions : @xmath69 @xmath70 @xmath71 generalized fermat problems steiner problems configuration eq.(3.3 ) turns into @xmath72 stability under variation forces tested by latexmath:\delta l=\sum_{j=1}\mathbf{r cartesian coordinates for @xmath49 vectors @xmath74 @xmath75 three variations @xmath76 @xmath77 @xmath78 in coordinates steiner points variations in length tree @xmath79(x_j - x_{s_k})}\mathbf{r_j-{s_k^3}=0 two analogous expressions for variations @xmath77 @xmath78arbitrariness variations write @xmath80\mathbf{r}_j-\mathbf{s}_k^3}=0\ write @xmath81 position vectors @xmath50 @xmath52 configuration figure ( 3 ) points evenly spaced circular helices radii 1 @xmath82 @xmath83 @xmath84 @xmath85 function derived conditions eqs.(3.7)(3.9 ) @xmath56 one solution @xmath86 coincides eq.(1.3 ) @xmath87 solution equations @xmath88 nature chosen solution @xmath89 partial equilibrium side chains amide plane conformation proteins @xcite configuration eqs.(3.15)(3.16 ) eq.(3.15 ) written @xmath90 geometrical object @xmath91 written coordinates eqs.(3.16 ) 3.17 ) @xmath92 @xmath93r_p^2+\alpha^2\omega^2[2r_p(j-\cos^2]+r_p^2_p\cos\omega+\alpha^2\omega^2^2]^{3/2}}.@xmath94 each @xmath95-value term @xmath96 dominates sum have @xmath96 for @xmath87 vertex @xmath97 ( @xmath98 ) ( @xmath99 ) nearest external vertices @xmath100 sequence consecutive position vectors is @xmath101 requirement @xmath96 corresponds integer @xmath54-value for @xmath56 @xmath56 case stable problem @xcite possible configuration figure ( 4 ) self - consistent treatment intramolecular organization biomacromolecules steiner networks representation deriving information stability evolution supporting facts for stability established ideas evolution macromolecules preliminary theory molecular evolution missing subject full description of geometric chirality proposed study influence proposals chirality measure on dynamics optimization problems studying structures energy around assumed energy minimum solution variation process chiral properties minimum research line worth serious scientific work best efforts scientific researchers mondaini , r.p disproof conjecture steiner ratio in @xmath102 consequences full geometric description of macromolecular chirality biomat symp . * 2 , 101177 ( 2003 ) | 0.436227 | 2,946 | 1,607 | , Saving $0.1 in GPT-4. |
177,731 | to study the structure of emission beam , we have analysed the single pulse data of psr b0329 + 54 at 325 and 606 mhz . in order to unambiguously detect the weak emission components ,
we have developed a new data analysis technique , which we term `` window thresholding '' . by applying this technique to the data , we have detected three new emission components , and also confirmed the presence of a component which was proposed earlier .
hence our analysis indicates that psr b0329 + 54 has nine components , which is among the highest of all the known pulsars .
the symmetric distribution of pulse components about the pulse centre , indicates that the emission beam is conal .
epsf.sty # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | for the strong pulsar psr b0329 + 54 , the high - quality single pulses are easily observable .
we obtained the data at 325 mhz in march 1999 using the giant metrewave radio telescope near pune , and the data at 606 mhz was obtained in august 1996 using the 76-m lovell telescope at jodrell bank . for the analysis , we considered about 2500 single pulses at each frequency .
the time resolutions of data at 325 and 606 mhz were 0.516 and 0.250 ms , respectively . to estimate the pulse profiles which clearly show the presence of weaker components we developed a ` window - threshold ' technique . in this technique , we set a window in longitude and employ an intensity threshold while considering the single pulses for making average profiles , i.e. we consider all those pulses , which have intensity levels above the threshold within the window . for threshold
the _ rms _ in intensity was computed from the off - pulse region . in figs .
1a&b we have plotted the average pulse profiles which were obtained by applying this technique to nine component windows . the average profile obtained from all those pulses is plotted in figs .
1c&d , which clearly shows the presence of 9 emission components ( i , ii , iii , iv , v , vi , vii , viii , ix ) in psr b0329 + 54 . the components on trailing side of the profile are closely spaced compared to the leading side
. 8.80 truecm
we have developed a technique based on windowing and thresholding , to detect the weak emission components in pulsar profiles . by applying it to the single pulse data of psr b0329 + 54 we have detected three new emission components ( vii , viii and ix ) of this pulsar , and also confirmed the presence of a component ( vi ) proposed by kuzmin & izvekova ( 1996 ) .
the near - symmetric distribution of components around the core / centre of the profile , favours the idea that the pulsar emission beam is annular or conal ( oster & sieber 1977 ; rankin 1983 ) .
we thank a. g. lyne for providing the jodrell bank data , and j. m. rankin for useful discussions .
kuzmin , a. d. , & izvekova , v. a. 1996 , , vol .
105 , pulsars : problems & progress , ed .
s. johnston , m.a .
walker & m. bailes ( san francisco : asp ) , 217 oster , l. , & sieber , w. 1977 , , 58 , 303 rankin , j. m. 1983 , , 274 , 333 | strong pulsar psr b0329 + 54 high - quality single pulses observable obtained data 325 mhz march 1999 giant metrewave radio telescope pune 606 mhz august 1996 76-m lovell telescope jodrell bank considered 2500 single pulses each frequency time resolutions 325 606 mhz 0.516 0.250 ms estimate pulse profiles weaker components developed window - threshold technique set window longitude intensity threshold single pulses intensity levels above threshold rms intensity computed off - pulse region figs . 1a&b plotted average pulse profiles nine component windows average profile plotted figs . 1c&d shows 9 emission components i ii iii iv v vi vii viii ix ) psr b0329 + 54 components trailing side closely spaced leading side developed technique windowing thresholding detect weak emission components pulsar profiles single pulse data psr b0329 + 54 detected three new emission components vii viii ix ) confirmed presence component ( vi ) proposed kuzmin & izvekova 1996 )near - symmetric distribution components core centre profile favours pulsar emission beam annular oster sieber 1977 rankin 1983 thank lyne jodrell bank data j. m. rankin discussions kuzmin izvekova 1996 vol . 105 pulsars problems progress s. johnston walker m. bailes san francisco asp ) 217 oster sieber 1977 58 , 303 rankin 1983 274 , 333 | 0.130978 | 608 | 320 | , Saving $0.0 in GPT-4. |
177,733 | it is well known that inclusion of torsion in the gravitational formalism , leads to four - fermion interactions .
although the coupling constant of this interaction is strongly suppressed in four dimensions , its value is enhanced in models with @xmath0 extra dimensions . in this context
, we reinterpret the recent limits established by lhc experiments to four - fermion contact interactions , to set bounds on the size of the extra dimensions . for @xmath1 ,
the limits are comparable to those in the literature , while for @xmath2 the volume of the extra dimensions is strongly constrained . | the standard model ( sm ) of particle physics has proved to be a very successful theory , but still has problems . among these ,
one that is considered fundamental is the vast difference between the electroweak and gravitational scales , known as the _ hierarchy problem _ , which might be solved introducing extra dimensions .
the gravitational theory of einstein can be generalized in several ways , specially in higher dimensional spacetimes .
one of the simplest generalizations is due to cartan , who proposed that the torsion does not necessarily vanishes.@xcite cartan s extension of general relativity , can be naturally written in the first order formalism , using two independent kind of fields : ( a ) the vielbeins , which encode the information carried by the metric , and ( b ) the spin connection , which encipher the way the parallel transport is performed . however , in general the spin connection possesses an antisymmetric part , that is not present in the levi - civita one , which introduces torsion to the picture . in this formalism
, gravity can be coupled to fermionic matter naturally .
since the field equation for the spin connection is algebraic , it can be used to substitute the torsion in terms of the spin of fermion fields .
the new action contains standard general relativity and matter fields with an additional contact four - fermion interaction.@xcite in four dimensions , within the most straightforward torsional extension of einstein - hilbert gravity , the effective four - fermion interaction term has a coupling constant proportional to newton s gravitational constant , @xmath3 , where @xmath4 is the planck s mass ; although in the most general torsional generalization of einstein gravity , the effective four - fermion interaction term has a coupling constant proportional to a yet undetermined constant .
therefore , at first glimpse this interaction is highly suppressed .
nevertheless , in the last twenty years diverse scenarios have proposed that the existence of extra dimensions could solve the hierarchy problem , due to the introduction of a ( higher dimensional ) fundamental gravity scale of roughly @xmath5.@xcite recently , limits to the fundamental scale of gravity have been set up by direct searches of quantum black holes@xcite and the influence of the exchange of virtual gravitons on dilepton events@xcite .
these might be compared with gravitational theories with torsion through their observables torsional observables@xcite , including some generalisations to higher dimensional spacetimes@xcite or lorentz violating models.@xcite on the other hand , atlas and cms collaborations have presented experimental limits for the coupling constant of four - fermion contact interaction.@xcite adequately interpreted , these results are useful for imposing bounds on the value of the fundamental gravity scale , @xmath6 , in the context of theories with torsion . by extension , it is possible to find limits on the size of the eventual extra dimensions .
the aim of this work is to obtain updated bounds and limits on the typical size of the extra dimensions using data coming from the lhc experiment .
the paper is organised as follows , in sec .
[ sec : cef ] , for the sake of completeness , a brief presentation of cartan s generalisation of gravity coupled with fermions is presented . in the next sections constraints
are found for different higher dimensional scenarios , such as those proposed by arkani - hamed , dimopoulos and dvali@xcite in sec .
[ sec : add ] , and by randall and sundrum@xcite in sec .
[ sec : rs ] . in sec .
[ sec : res ] a summary of results and conclusions is presented .
it is a well - known fact that einstein s gravity is a field theory for the metric , and the spacetime connection is required to be torsion - free ( see for example ref . ) .
interestingly , when the first order formalism of pure gravity is considered , _
i.e. _ with all matter fields turned off , whose fields are the vielbeins and spin connection , the torsion - free imposition is nothing but the equation of motion for the spin connection.@xcite however , for gravity coupled with fermionic matter , this condition changes and introduces a four - fermion contact interaction .
the modification of the fermionic lagrangian due to the presence of torsion is presented below , using the notation in ref . .
the cartan - einstein action in @xmath7-dimensions is , @xmath8 where @xmath9 and @xmath10 are the vielbeins 1-form and curvature 2-forms respectively , and these are related with the spin connection 1-form , @xmath11 , and torsion 2-form , @xmath12 , , and @xmath13 . ] through the so called structural equations , @xmath14 additionally , the dirac action in arbitrary dimension is @xmath15 with @xmath16 the exterior derivative twisted by the spin connection , @xmath17 with @xmath18 the generators of the spinorial representation of the lorentz group . since the total action is @xmath19 the equations of motion for the whole system are , s denote the anti - symmetric product of @xmath20-matrices . ]
@xmath21}\psi , \label{ed - eom}\\ \frac{1}{2}{\left(}\hat{{\mathcal{t}}}_{\hat{b } \hat{c } \hat{a } } + \hat{{\mathcal{t}}}_{\hat{b } \hat{a } \hat{c } } + \hat{{\mathcal{t}}}_{\hat{a } \hat{b } \hat{c } } { \right)}&= -\frac{\kappa^2}{4 } \bar{\psi}{\gamma}_{\hat{a } \hat{b } \hat{c}}\psi , \label{cont - constraint}\end{aligned}\ ] ] the lhs of eq .
is generally denoted by @xmath22 , and is called contorsion .
notice that in the case of completely antisymmetric spin density , as in the case of dirac fields above , it is enough to consider the completely antisymmetric part of the torsion ( as in eq . ) in what will be the single squared - torsion contribution.@xcite since eq .
is a constraint , it can be substituted back into the action .
it is done by splitting the spin connection into a torsion - free part , @xmath23 , and the contorsion . in differential forms ,
it is expressed as . ]
@xmath24 the new action , expressed in terms of torsion - free quantities , is @xmath25},\end{aligned}\ ] ] which is the usual torsion - free theory of gravity coupled to a fermion field with a four - fermion contact interaction . in the next sections
the gravitational effects , other than the generation of a four - fermion interaction , are neglected .
hereon , our objective will be to compare the four - fermion interaction coming from this gravity model with the four - fermion interaction limits found at the lhc experiments .
early proposals of contact four - fermion interaction signals in colliders are found in @xcite .
these works inspired searches at the lhc experiments .
in particular , atlas and cms collaborations have found limits on the scale of four - fermion contact interaction , by analyzing the invariant mass and angular distribution of dijets@xcite . the strongest constraint comes from an interaction of chiral fermions in the form @xmath26 which is experimentally excluded for @xmath27 , assuming that the coupling constant @xmath28 . in this section
the above limit is compared with the four - fermion interaction coming from the cartan - einstein gravity action , , while the fermion in arbitrary dimension is denoted with the capitalized one , @xmath29 . ]
@xmath30 within the context of the straightforward torsional generalization of einstein s gravity in four dimensions , . ] where @xmath31 , @xmath32 and @xmath33 , the comparison implies @xmath34 , which is tautological . nonetheless , there exist models where the fundamental scale of gravity is not @xmath35 , but rather a much lower one , @xmath6 , which could be of order of the electro - weak scale , i.e. @xmath36 , giving a natural solution to the _ hierarchy problem_. these models , however , require extra dimensions .
therefore , in the following the spacetime will be considered to be @xmath37-dimensional , where the @xmath0 extra dimensions are either compact or extended , and differentiation between models is made according to this characteristic . in order to achieve the goal of comparison , it is necessary to reduce the dimension of the spacetime from @xmath7 down to four .
although the dimensional reduction could be a complex procedure , we will sketch how a term like eq . appears .
first , in the contraction @xmath38 one splits the indices to consider the four - dimensional contribution,@xcite i.e. , @xmath39where the ellipsis stand for additional terms , but the focus will be on the first one , because it will rise the kind of interaction comparable with the experimental data .
second , the antisymmetric product of three elements of the four - dimensional clifford algebra is equal to the product of the missing element of the clifford algebra times the chiral element , i.e. , @xmath40 .
next , the higher dimensional spinor @xmath29 can be decomposed ( or compactified ) as the product of the four - dimensional times @xmath0-dimensional spinors,@xcite @xmath41 where the standard kaluza - klein decomposition for fermionic fields has been used , and @xmath42 ) . nonetheless , for the purpose of comparison with the term in eq .
( where all spinors are left - handed ) , we restrict ourselves to the case of equal profiles . ]
@xmath43 denotes the coordinates on the four - dimensional spacetime , @xmath44 denotes the extra dimensional coordinates , and also dirac spinors ( in four dimensions ) have been decomposed in terms of weyl spinors .
therefore , after integration of the extra dimensions , the effective four - dimensional theory would have a term of the desired form @xmath45 dimensional analysis gives two possibilities .
either the effective coupling constant is directly related to the fundamental gravitational scale , @xmath46 , or is related to the effective four - dimensional one @xmath47 , where @xmath48 is the redefined planck mass after the dimensional reduction .
these two interpretations originate different limits , which rise the bound limits presented below . in the following
, the limits corresponding to the first interpretation will be called @xmath49 , while the limit given by the second interpretation will be called @xmath50 add models@xcite consist of a four - dimensional spacetime with a set of @xmath0 compact extra dimensions , with typical length @xmath51 .
matter is confined to the four - dimensional spacetime for energies below @xmath52 , while gravity propagates through the whole spacetime .
this configuration allows to solve the hierarchy problem , because the natural scale for gravity is not the effective four - dimensional one , but rather the @xmath37-dimensional .
the relation between the fundamental gravitational scale , @xmath6 , and the four - dimensional effective one , @xmath4 is given by @xmath53 additionally , the coupling constant of the einstein action is @xmath54 .
it is worthwhile to mention that add scenarios are restricted to @xmath55 , because of gravitational phenomenology.@xcite from eq .
, it follows that the typical radii of the extra dimensions are @xmath56 assuming that the scale of the four - fermion interaction , @xmath57 , is essentially the fundamental scale of gravity , @xmath6 , one finds sizes of the extra dimensions from roughly a few micrometers down to a few tens femtometers , depending on the number of extra dimensions considered , as shown in the second column of table [ tab : add - r ] .
these results are a refinement of the original add claim . on the other hand , the effective coupling constant in four dimensions , @xmath58 , should be related directly with the fundamental scale of gravity , i.e. , @xmath59 , or in a similar way as before @xmath60 hence , one finds another limit for the typical size of the extra dimensions , as shown on the third column of table [ tab : add - r ] .
when considering randall - sundrum brane - worlds scenarios,@xcite with metric @xmath61 the relation between the four - dimensional planck mass , @xmath4 , and the fundamental ( five - dimensional ) planck scale , @xmath6 , is given by @xmath62 a well - known modulus stabilization method would ensure that the product @xmath63@xcite , the relation between the gravitational scales and the length of the extra dimension is found to be @xmath64 analyzing both limits as in the previous section , the limit on the extra dimension size is @xmath65 the range is particularly wide because there is a single extra dimension .
although brane - worlds of codimension higher than one have been considered@xcite , without a carefully thought - out moduli stabilization process the bounds on the extra dimensions sizes are equal to those found in sec .
[ sec : add ] ( shown in table [ tab : add - r ] ) .
in this work we have used the limits on four - fermion chiral contact interactions obtained by the lhc collaborations in order to constrain the typical size of eventual extra dimensions in models where gravity admits torsion in the bulk .
depending of the approach , there are two types of interpretations .
these approaches , that we refer to as type a and type b , yield to constraints differ by at least two orders of magnitude . according to the table [ tab : add - r ] , and the limits of eq . , it is more likely to rule out the type b interpretation . for a codimension 2
we have found an upper bound for the radius of the extra dimensions of the order of @xmath66 , which is comparable to the limits obtained from direct search of the kaluza - klein excitations of the graviton@xcite . nevertheless , for higher numbers of extra dimensions we found that the constrains are a lot more stringent .
this is due to the fact that the fundamental gravitational scale , @xmath6 , is related with the effective one , say @xmath48 , through higher powers , reducing the dependence of the model on the size of the extra dimensions .
this result provides an example of the possibility of testing non - trivial extensions of general relativity using collider data .
a refining of the analysis can be achieved by introducing different profile functions to left and right chiralities in eq . , and performing the dimensional reduction analysis.@xcite this model illustrates that in the context of extra dimensions , extensions of the gravitational sector might produce interesting effects in collider phenomenology .
for example , even the physics underlying quark masses might have a gravitational origin in the bulk.@xcite moreover , the same type of contact interaction , generates corrections to very precisely measured electroweak observables , such as the @xmath67 decay , and their comparison to experimental values provide additional constraints on the scales of the effective coupling constant.@xcite additionally , the structure of the contact interactions and their universality allows a new contact interaction among neutrinos , which may be important in cosmological contexts .
we would like to thank to c. dib , n. neills , j.c . helo and a. carcamo for fruitful discussions and
encourage through the realisation of this work . | standard model of particle physics successful theory but has problems fundamental is difference between electroweak and gravitational scales _ hierarchy problem _ solved introducing extra dimensions gravitational theory of einstein generalized higher dimensional spacetimes simplest generalizations due to cartan proposed torsion not vanishes cartan s extension of general relativity written in first order formalism using two independent fields vielbeins metric spin connection parallel transport spin connection antisymmetric part not present in levi - civita one introduces torsion picture gravity coupled to fermionic matter field equation for spin connection algebraic can substitute torsion in spin of fermion fields new action contains standard general relativity matter fields with additional contact four - fermion interaction in four dimensions effective four - fermion interaction term has coupling constant proportional to newton s gravitational constant @xmath3 @xmath4 is planck s mass four fermion interaction coupling constant proportional to undetermined constant interaction highly suppressed last twenty years scenarios proposed extra dimensions solve hierarchy problem due to introduction higher dimensional fundamental gravity scale @xmath5. limits to fundamental scale gravity set by searches quantum black holes influence exchange virtual gravitons on dilepton events compared with gravitational theories with torsion generalisations to higher dimensional spacetimes lorentz violating models. atlas cms collaborations presented experimental limits for coupling constant of four - fermion contact interaction results useful for imposing bounds on value fundamental gravity scale context of theories with torsion possible to find limits on size of eventual extra dimensions aim work to obtain updated bounds limits on typical size extra dimensions using data from lhc experiment paper organised brief presentation of cartan s generalisation of gravity coupled with fermions presented next sections constraints found for higher dimensional scenarios proposed by arkani - hamed dimopoulos dvali randall sundrum summary of results conclusions presented einstein s gravity is field theory for metric spacetime connection required to be torsion - free first order formalism of pure gravity consideredmatter fields turned off vielbeins spin connection torsion - free imposition equation of motion for spin connection gravity fermionic matter condition changes introduces four - fermion contact interaction modification fermionic lagrangian torsion presented cartan - einstein action in @xmath7-dimensions @xmath8 vielbeins 1-form curvature 2-forms related with spin connection 1-form torsion 2-form structural equations dirac action arbitrary dimension @xmath15 @xmath16 exterior derivative twisted by spin connection @xmath17 @xmath18 generators spinorial representation lorentz group total action @xmath19 equations of motion system anti - symmetric product of @xmath20-matrices@xmath21}\psi \label{ed - eom}\\ \frac{1}{2}{\left(}\hat{{\mathcal{t}}}_{\hat{b } \hat{c } \hat{a }{b{c \right)} -\frac{\kappa^2}{4 } \bar{\psi}{\gamma}_{\hat{a } \hat{b } \hat{c}}\psi \label{cont - constraint}\end{aligned} lhs of eq . denoted by @xmath22 called contorsion . antisymmetric spin density consider antisymmetric part torsion single squared - torsion contribution eq . constraint substituted into action splitting spin connection into torsion - free part @xmath23 contorsion expressed as. @xmath24 new action torsion - free quantities @xmath25},\end{aligned} usual torsion - free theory of gravity fermion field four - fermion contact interaction next sections gravitational effects neglectedobjective compare four - fermion interaction from gravity model with limits lhc experiments early proposals four - fermion interaction signals in colliders in @xcite inspired searches lhc experiments atlas cms collaborations found limits four - fermion contact interaction analyzing invariant mass angular distribution of dijets@xcite strongest constraint from interaction chiral fermions form @xmath26 experimentally excluded for @xmath27 coupling constant @xmath28 limit compared with four - fermion interaction from cartan - einstein gravity action fermion in arbitrary dimension denoted with capitalized one @xmath29 @xmath30 einstein s gravity four dimensions @xmath31 comparison implies @xmath34 tautological models where fundamental scale of gravity not @xmath35 lower one @xmath6 electro - weak scale @xmath36 solution to hierarchy problem models require extra dimensions spacetime @xmath37-dimensional @xmath0 extra dimensions compact or extended differentiation between models comparison necessary reduce dimension spacetime from @xmath7 to four dimensional reduction complex sketchcontraction @xmath38 splits indices four - dimensional contribution @xmath39where ellipsis for additional terms focus on first one interaction comparable with experimental data antisymmetric product of three elements four - dimensional clifford algebra equal to missing element chiral element @xmath40 higher dimensional spinor @xmath29 decomposed as product four - times @xmath0-dimensional spinors @xmath41 standard kaluza - klein decomposition for fermionic fields @xmath42 comparison restrict to equal profiles @xmath43 denotes coordinates four - dimensional spacetime @xmath44 extra dimensional coordinates dirac spinors four decomposed weyl spinors after integration extra dimensions effective four - dimensional theory desired form @xmath45 dimensional analysis two possibilities effective coupling constant related to fundamental gravitational scale @xmath46 or effective four - dimensional one @xmath47 @xmath48 redefined planck mass after dimensional reduction two interpretations originate different limits bound limitslimits first interpretation called @xmath49 second interpretation @xmath50 models four - dimensional spacetime @xmath0 compact extra dimensions typical length @xmath51 matter confined four - dimensional spacetime for energies below @xmath52 gravity propagates through whole spacetime configuration hierarchy problem natural scale gravity not effective four - dimensional @xmath37-dimensional relation between fundamental gravitational scale @xmath6 four - effective one @xmath4 given by @xmath53 coupling constant einstein action @xmath54 add scenarios restricted to @xmath55 gravitational phenomenology. typical radii extra dimensions @xmath56 scale four - fermion interaction @xmath57 fundamental scale gravity @xmath6 sizes extra dimensions few micrometers to tens femtometers extra dimensions second column table [ tab add - r ] results refinement of original add claim effective coupling constant in four dimensions @xmath58 related with fundamental scale gravity @xmath59 @xmath60 another limit for typical size extra dimensions third column table add - rconsidering randall - sundrum brane - worlds scenarios@xcite metric @xmath61 relation between four - dimensional planck mass @xmath4 fundamental five - dimensional planck scale @xmath6 given by @xmath62 modulus stabilization method product @xmath63@xcite relation between gravitational scales length extra dimension @xmath64 analyzing limits limit extra dimension size @xmath65 range wide single extra dimension brane - worlds codimension higher one considered without moduli stabilization process bounds on extra dimensions sizes equal to sec . add shown table tab add - r used limits on four - fermion chiral contact interactions lhc collaborations constrain size extra dimensions in models gravity admits torsion bulk two types of interpretations type a and b constraints differ two orders of magnitude table limits likely to rule out type b interpretation codimension 2 found upper bound for radius extra dimensions @xmath66 comparable to limits from direct search kaluza - klein excitations graviton higher numbers extra dimensions constrains more stringentdue fundamental gravitational scale @xmath6 related with effective @xmath48 higher powers reducing dependence model on size extra dimensions result testing non - trivial extensions general relativity using collider data refining analysis introducing profile functions to left right chiralities performing dimensional reduction analysis model illustrates extra extensions gravitational sector produce effects in collider phenomenology quark masses have gravitational origin in bulk contact interaction generates corrections to precisely electroweak observables @xmath67 decay comparison experimental values constraints on scales effective coupling constant structure contact interactions universality allows new contact interaction among neutrinos important in cosmological contexts c. dib n. neills j.c . helo a. carcamo for discussions realisation work | 0.567034 | 3,334 | 1,688 | , Saving $0.1 in GPT-4. |
177,741 | we present the main features of citius , a new light source for ultrafast science , generating tunable , intense , femtosecond pulses in the spectral range from ir to xuv .
the xuv pulses ( about @xmath0-@xmath1 photons / pulse in the range 14 - 80 ev ) are produced by laser - induced high - order harmonic generation in gas .
this radiation is monochromatized by a time - preserving monochromator , allowing also to work with high - resolution bandwidth selection .
the tunable ir - uv pulses ( @xmath2-@xmath3 photons / pulse in the range 0.4 - 5.6 ev ) are generated by an optical parametric amplifier , which is driven by a fraction of the same laser pulse that generates high order harmonics .
the ir - uv and xuv pulses follow different optical paths and are eventually recombined on the sample for pump - probe experiments .
the new light source will become the fulcrum of a new center located at the university of nova gorica , active in a wide range of scientific fields , including materials science , catalysis , biochemistry and magnetism .
we also present the results of two pump - probe experiments : with the first one , we fully characterized the temporal duration of harmonic pulses in the time - preserving configuration ; with the second one , we demonstrated the possibility of using citius for studying of ultra - fast dynamics . | thanks to advances in laser science occurred over the last two decades , it is now possible to produce soft x - ray pulses in the femtosecond range , using different approaches : laser - induced high - order harmonic generation in gas ( hhg ) , free - electron lasers , and laser slicing at synchrotron beamlines @xcite . in hhg , noble - gas atoms
interact with an intense laser field and emit light at the harmonics of the latter .
the process can be described by means of a semiclassical `` three step model '' @xcite . in the first step ,
a bound electron is extracted from a gas atom under the action of the laser field ; in the second step , the electron undergoes an oscillating motion imposed by the laser field ; in the third step , the electron collides back with the parent ion . as a result
, laser harmonics are emitted up to wavelengths extending into the soft x - ray regime , and droping quickly at some cutoff energy @xcite .
+ since the electron dynamics in atomic , molecular or more complex systems occurs at timescales ranging from hundreds of femtoseconds down to tens of attoseconds , hhg sources provide an ideal tool to study , and possibly control , the interactions between charge , lattice , orbital , and spin dynamics in a chemical reaction . + in 2008 , a new project has been funded in the framework of the `` cross - border cooperation program between italy and slovenia 2007 - 2013 '' @xcite .
project s main objective is to set up a state - of - the - art light source at university of nova gorica ( slovenia ) .
the source , named citius , has been assembled and commissioned at elettra - sincrotrone trieste ( italy ) and is going to be moved to slovenia .
the mission of the new slovenian laboratory is to carry out cutting - edge experiments in a wide range of both applied and fundamental research fields , including materials science , catalysis , biochemistry and magnetism .
part of the envisaged scientific program will be connected to that developed at the low - density matter beamline at the fermi@elettra free - electron laser @xcite .
+ in this paper , we first provide a technical description of citius , and then report the results of two proof - of - principle pump - probe experiments , namely the characterization of the temporal structure of hhg pulses , and the selective investigation of the ultra - fast dynamics of different elements in a magnetic compound .
the citius laser system and the beamline are schematically shown in fig.[fig1 ] .
the laser system comprises a commercial amplifier `` legend elite duo '' and an optical parametric amplifier `` opera '' ( opa ) , both produced by coherent inc .
the amplifier is seeded by a micra oscillator ( wavelength : 800 nm , spectral bandwidth about 100 nm , power about 380 mw ) and includes two amplification stages : a regenerative amplifier ( evo 30 ) , pumped by an evolution 30 laser ( frequency - doubled q - switched nd : ylf laser ) , and a single - pass amplifier ( evo he ) , pumped by an evolution he laser ( frequency - doubled q - switched nd : ylf laser ) .
the system can be operated at four different repetition rates : 100 hz , 1 khz , 5 khz and 10 khz . at 5 khz , which is the currently adopted repetition rate
, it generates pulses carrying about 3.1 mj , with a duration of about 35 fs , centered at 805 nm .
two thirds of the energy is used for generating xuv pulses through hhg , one third as a pump for the opa , or directly in combination with the hhg beam for pump - probe experiments .
the opa produces tunable radiation in the range between 0.4 ev and 5.6 ev , with variable energy per pulse ( from few to hundreds of microjoules ) .
+ the beamline includes a high - vacuum section , through which the xuv beam propagates , and a section in air , used to transport the ir - uv light generated by the opa .
the xuv part comprises a hhg generation chamber and a monochromator .
the laser is focused in the generation chamber ( using the lens l1 , see fig .
[ fig1 ] ) , where it interacts with a noble gas of choice ( ar or ne , for the reported experiments ) contained in a cell , generating the high - order harmonics .
the harmonic generation efficiency can be optimized by varying the laser intensity , the gas pressure ( typically , of the order of @xmath4 mbar ) and the relative position of the laser focus within the cell .
+ the monochromator , whose scheme is shown in fig .
[ fig2 ] , was developed by the c.n.r .- i.f.n .
padova ( italy ) and will be described in detail in a different paper @xcite .
it was designed to select one single harmonic , or a sub - band of it , in the spectral range between 250 and 25 nm ( 5 - 50 ev ) .
however , the transmitted energy range could be further extended at the expense of efficiency , up to 100 ev .
the citius monochromator is superior to similar systems @xcite , in that it adopts a single - grating design with a double configuration @xcite : an off - plane geometry ( opg ) for ultrafast response ( few tens of femtoseconds ) at the expense of energy resolution , and a classical geometry ( cg ) for high - energy resolution , albeit with long temporal response ( several hundreds of femtoseconds ) .
a first toroidal mirror collimates the light coming from the generation chamber and deflects it onto a planar ( opg or cg ) diffraction grating .
the diffracted light is then refocused by a second toroidal mirror onto the exit slit .
the characteristics of the opg and cg gratings are reported in table [ table1 ] .
.characteristics of the monochromator gratings .
the energy resolution has been calculated on a 100-@xmath5m - wide exit slit . [ cols="^,^,^,^ " , ] [ table3 ] in this section , we report the results of a pump - probe experiment demonstrating the possibility of using citius to study the ultra - fast dynamics of selected chemical species , with a temporal resolution of few tens of femtoseconds .
+ the spectral range covered by citius includes the @xmath6 thresholds of transition metals ( 54.25 ev , i.e. , the 35th harmonic of the fundamental , see table [ table2 ] ) , making citius an appealing source for the study of magnetic materials . as a proof - of - principle application
, we measured the demagnetization of a 100 mm thick permalloy film ( @xmath7 ) , using x - ray resonant magnetic scattering @xcite in pump - probe configuration .
+ the experiment was performed using the irma reflectometer @xcite .
the instrument permits to precisely adjust the angles between the sample surface and the incoming / reflected xuv beam .
three photodiodes and one channel electron multiplier are mounted on the detector arm . by means of a horseshoe electromagnet
, it is possible to apply a variable magnetic field up to @xmath8 oe , parallel to the sample surface .
the scattering experiment was conducted in a transverse moke - like configuration @xcite . in this configuration ,
using p - polarized xuv light , one is sensitive to the magnetization component perpendicular to the scattering plane .
moreover , choosing the incoming / scattering angle close to the brewster condition ( about 45 deg at 54 ev ) , the magnetization contrast is maximized with respect to the ( strongly suppressed ) non - magnetic scattered background , thus improving the signal - to - noise ratio of the measurement .
+ we have measured the magnetic hysteresis loop at the fe @xmath6 edge , as a function of the delay between the transmitted ir beam ( pump ) and the xuv pulse ( probe ) .
figure [ fig5]a shows the hysteresis loop when the xuv pulse arrives on the sample 1 ps before ( blue curve ) and 1 ps after ( red curve ) the ir pulse .
the plots are normalized to the average scattered intensity .
the magnetic asymmetry ratio is defined as @xmath9 , where @xmath10 and @xmath11 denote the intensities of the reflected light as the externally applied magnetic field is reversed in direction . in figure
[ fig5]b , the asymmetry ratio at saturation is plotted as a function of the pump - probe delay .
data show the ultrafast demagnetization of the fe atoms of the sample .
almost 80 % of the magnetization is dissipated 0.5 ps after the sample has been illuminated with the ir pulse .
the data have been fitted with an exponential decay curve , indicating a characteristic demagnetization time of 0.21 @xmath12 0.02 ps . despite the modest scientific interest of this particular sample , which has been already investigated in detail by other groups @xcite , these results demonstrate that our apparatus is capable of measuring the ultrafast magnetization dynamics of selected chemical species , with a resolution of few tens of femtoseconds .
we have presented the characteristics of citius , a new ir - xuv , femtosecond , italo - slovenian light source , based on laser - induced high - harmonic generation in gas . as a next step
, we will carry out the commissioning of the experimental equipment for time resolved photoemission investigations .
the apparatus consists of two vacuum chambers equipped with two different electron spectrometers : an angle - resolved time of flight ( vg - scienta artof 10k ) for pump - probe experiments and an hemispherical spectrometer ( vg - scienta r3000 ) , which will be used in combination with a monochromatized twin anode x - ray source for conventional x - ray photoemission spectroscopy measurements .
a preparation chamber , equipped with evaporators and standard sample preparation tools , such as sample heating , ion sputtering and leed , will connect the two chambers .
the sample will be hosted on a cryomanipulator ( able to reach a temperature below 10 k ) , operated with liquid helium , with five degrees of freedom .
+ starting from 2014 , the source will be open to external users , interested in developing time - resolved applications in different domains of both fundamental and applied science
. envisaged experiments will be in particular related to photophysics and photochemistry , including materials science , catalysis , magnetism and biochemistry .
part of the activity will be carried out in close connection with that developed at the low - density matter beamline of the fermi free - electron laser .
the citius project is funded by the program for cross - border cooperation between italy and slovenia 2007 - 2013 . c. grazioli gratefully acknowledges financial support through the talents programme ( 7th framework programme , specific programme : people - marie curie actions - cofund ) .
we aknowledge the department of physics and astronomy of uppsala university for providing the vg - scienta ses200 , in the framework of the collaboration with cnr - iom ( tasc laboratory , gas phase beamline , trieste ) .
we also thank carla puglia and monica de simone for technical assistance in setting up the photoemission experimental set - up , and sandra gardonio for valuable suggestions about the source desing .
is particularly thankful to gvido bratina for technical and moral support during the preparation and implementation of the project citius .
t. pfeifer , c. spielmann and g. gerber , _ rep .
phys . _ * 69 * , 604 ( 2006 ) .
p. b. corkum , _ phys . rev
. lett . _ * 71 * , 143903 ( 1993 ) .
z.h . chang et al . , _ phys .
lett . _ * 79 * , 2967 ( 1997 ) . c. spielmann et al . , _ science _ * 278 * , 661 ( 1997 ) .
e. seres et al .
_ * 92 * , 163002 ( 2004 ) .
l. poletto et al . , in preparation .
l. poletto et al . , _ rev .
inst . _ * 80 * , 123109 ( 2009 ) .
f. frassetto et al . , _ opt .
express _ * 19 * , 19169 ( 2011 ) .
l. poletto and f. frassetto , _ appl . opt . _ * 49 * , 5465 ( 2010 ) .
l. poletto and f. frassetto , _ applied sciences _ * 3 * , 1 ( 2013 ) .
t. e. glover et al . , _ phys .
_ * 76 * , 2468 ( 1996 )
. e. s. toma et al . , _ phys .
a _ * 62 * , 061801 ( 2000 ) . n. mrtensson et al . , _ j. electron spectrosc .
relat . phenom .
_ * 70 * , 117 ( 1994 ) .
d. gibbs et al .
lett . _ * 61 * , 1241 ( 1998 ) .
j. p. hannon et al . , _ phys .
lett . _ * 61 * , 1245 ( 1998 ) .
maurizio sacchi et al . , _ rev .
instrum . _ * 74 * , 2791 ( 2003 ) .
n. argyres , _ phys
_ * 97 * , 334 ( 1955 ) .
e. beaurepaire et al .
* 76 * , 4250 ( 1996 ) . | advances in laser science produce soft x - ray pulses in femtosecond range approaches laser - induced high - order harmonic generation in gas free - electron lasers laser slicing at synchrotron beamlines in noble - gas atoms interact with laser field emit light at harmonics process three step model first bound electron extracted from gas atom laser field second oscillating motion third electron collides with parent ion laser harmonics emitted to wavelengths extending into soft x - ray regime droping quickly at cutoff energy electron dynamics in at femtoseconds tens attoseconds hhg sources provide tool to study interactions between charge lattice orbital spin dynamics in chemical reaction 2008 new project funded cross - border cooperation program between italy slovenia 2007 - 2013 state - of - - art light source at university of nova gorica slovenia source citius assembled commissioned at elettra - sincrotrone trieste italy moved to slovenia mission laboratory cutting - edge experiments in applied fundamental research fields materials science catalysis biochemistry magnetismenvisaged scientific program connected to low - density matter beamline fermi@elettra free - electron laser @xcite technical description citius report results proof pump - probe experiments temporal structure hhg pulses ultra - fast dynamics elements in magnetic compound citius laser system beamline shown in. laser system comprises commercial amplifier legend elite duo optical parametric amplifier opera produced by coherent inc amplifier seeded by micra oscillator wavelength 800 nm spectral bandwidth 100 nm power 380 mw two amplification stages regenerative amplifier evo 30 pumped evolution 30 laser single - pass amplifier evo he ) pumped laser operated at four repetition rates 100 hz 1 khz 5 10 khz 5 khz generates pulses 3.1 mj duration 35 fs centered at 805 nm two thirds energy for generating xuv pulses hhg one third pump for opa combination with hhg beam for pump - probe experiments opa produces tunable radiation 0.4 and 5.6 ev variable energy per pulse microjoulesbeamline includes high - vacuum section xuv beam propagates section in air ir - uv light opa xuv part comprises hhg generation chamber monochromator laser focused in generation chamber lens l1 interacts with noble gas ( ar or ne in cell generating high - order harmonics harmonic generation efficiency optimized by varying laser intensity gas pressure @xmath4 mbar relative position laser focus within cell monochromator scheme in fig fig2 developed by c.n.r .- i.f.n . padova described in designed to select one single harmonic in spectral range between 250 and 25 nm ( 5 - 50 ev ) transmitted energy range could extended at expense efficiency up to 100 ev citius monochromator superior adopts single - grating design double configuration off - plane geometry for ultrafast response classical geometry for high - energy resolution long temporal response first toroidal mirror collimates light from generation chamber deflects onto planar diffraction grating diffracted light refocused by second toroidal mirror onto exit slitcharacteristics opg cg gratings reported table monochromator gratings energy resolution calculated on 100-@xmath5m - wide exit slit table3 results pump - probe experiment citius study ultra - fast dynamics chemical species temporal resolution tens femtoseconds spectral range citius includes @xmath6 thresholds transition metals 54.25 ev 35th harmonic fundamental citius appealing study magnetic materials measured demagnetization 100 mm thick permalloy film @xmath7 using x - ray resonant magnetic scattering @xcite pump - probe configuration experiment irma reflectometer @xcite instrument angles sample surface incoming reflected xuv beam three photodiodes one channel electron multiplier on detector arm horseshoe electromagnet variable magnetic field up to @xmath8 oe parallel sample surface scattering experiment transverse moke - like configuration @xcite p - polarized xuv light sensitive magnetization component perpendicular scattering plane incoming scattering angle close brewster condition 45 deg at 54 ev ) magnetization contrast maximized non - magnetic background improving signal - to - noise ratiomeasured magnetic hysteresis loop at fe @xmath6 edge function delay between ir beam xuv pulse probe figure shows hysteresis loop xuv pulse arrives before after ir pulse plots normalized to average scattered intensity magnetic asymmetry ratio defined as @xmath9 @xmath10 @xmath11 denote intensities reflected light magnetic field reversed asymmetry ratio at saturation function pump - probe delay data show ultrafast demagnetization of fe atoms 80 % magnetization dissipated 0.5 ps after illuminated ir pulse data fitted with exponential decay curve demagnetization time 0.21 0.02 ps modest scientific interest sample results demonstrate apparatus ultrafast magnetization dynamics chemical species resolution tens femtoseconds presented characteristics of citius new ir - xuv femtosecond italo - slovenian light source laser - induced high - harmonic generation gas next step experimental equipment for time resolved photoemission investigationsapparatus consists two vacuum chambers two electron spectrometers angle - resolved time of flight ( vg - scienta artof 10k ) for experiments hemispherical spectrometer vg - scienta r3000 ) used with monochromatized twin anode x - ray source for photoemission spectroscopy measurements preparation chamber evaporators standard sample preparation tools chambers sample hosted on cryomanipulator operated with liquid helium five degrees of freedom starting 2014 source open to external users time - resolved applications in fundamental applied science experiments related to photophysics photochemistry materials science catalysis magnetism biochemistry activity with low - density matter beamline of fermi free - electron laser citius project funded by program for cross - border cooperation between italy slovenia 2007 - 2013 financial support talents programme department of physics and astronomy of uppsala university vg - scienta ses200 collaboration with cnr - iom thank carla puglia monica de simone for technical assistance sandra gardonio for suggestions source thankful to gvido bratina for technical moral support during project citiusspielmann gerber rep 69 604 ( 2006 ) corkum phys . rev 71 143903 ( 1993 ) chang phys . 79 2967 ( 1997 ) spielmann science 278 661 ( 1997 ) seres al 92 163002 ( 2004 ) poletto al preparation rev . inst . 80 * 123109 ( 2009 ) frassetto opt . express 19 19169 ( 2011 ) poletto appl . opt . 49 * 5465 ( 2010 ) applied sciences 3 * 1 ( 2013 ) glover al 76 * 2468 ( 1996 ) toma phys 62 * 061801 ( 2000 ) mrtensson al electron spectrosc. phenom 70 * , 117 ( 1994 ) gibbs al 61 * , 1241 ( 1998 ) hannon al phys . 61 1245 ( 1998 ) maurizio sacchi al rev . instrum. 74 * , 2791 ( 2003 ) argyres phys 97 * , 334 ( 1955 ) beaurepaire al 76 4250 ( 1996 | 0.503157 | 3,017 | 1,538 | , Saving $0.1 in GPT-4. |
177,751 | we have investigated the implications of the neutron halo configuration , observed in the ground - state of some neutron - rich light nuclei , on neutron radiative transition processes .
in particular , we have studied the influence of the neutron halo on the direct radiative capture ( drc ) process .
the energy dependence as well as the strength of e1 emission due to incident _ p_-wave neutrons is strongly influenced by the halo configuration of the residual nucleus capturing state .
we have compared the calculated @xmath0be@xmath1be drc cross section with that derived from the experiment in the inverse kinematics ( coulomb dissociation of @xmath2be ) .
we show from the comparison that some important information on the structure of the halo nucleus @xmath2be can be derived .
= 16.0 cm = 24.0 cm = -2.5 cm = -2.5 cm * exotic properties of light nuclei * * and their neutron capture cross sections * a. mengoni@xmath3 , t. otsuka@xmath4 , t. nakamura@xmath5 , and m. ishihara@xmath4 _ @xmath6riken , radiation laboratory , 2 - 1 hirosawa , wako , saitama 351 - 02 , japan _ _
@xmath7enea , applied physics section , v. don fiammelli 2 , 40128 bologna , italy _ _ @xmath8the university of tokyo , department of physics , hongo , bunkyo - ku , tokyo 113 , japan _ | in the category of `` exotic '' nuclear structure properties we can include the neutron skin and the neutron halo observed in ground - state configurations of light neutron rich nuclei @xcite .
a rigorous definition of neutron skin and of neutron halo can not be given .
therefore , we will show here an example of both structures to elucidate their features and to show their basic properties .
the effect of neutron skin structure has been recently investigated @xcite and has been shown to be responsible for a low energy excitation mode , decoupled from the isoscalar giant quadrupole resonance @xcite . here
we will concentrate on the effects of the neutron halo structure .
in particular , we will show its importance in relation to the low energy electric dipole transition mechanism in @xmath2be .
an example of neutron skin structure is shown in figure [ fig : o_potential ] . there , the result of a mean - field calculation using hartree - fock method @xcite is shown for the two nuclei @xmath9o ( in the upper part ) and @xmath10o ( in the lower part ) .
the calculations were performed using the common skyrme - s3 interaction , but the general features of the result can be deduced with any other kind of effective interaction . in the upper part of the figure , the mean - field potential for protons ( left ) and neutrons ( right ) is shown together with the bound single - particle energy spectrum . in this double - magic nucleus ,
neutrons and protons occupy analogous single - particle orbits located at comparable energies . in the lower part of the figure ,
the same type of calculation is shown for the very neutron rich @xmath10o ( n=20 ) nucleus , a double - magic nucleus too . in this case
the neutrons occupy single - particle states with energies very different compared to the proton case . moreover , the fermi energies of protons and neutrons are very different here due to the much larger neutron number
. a plot of the densities for these two nuclei would reveal a thick neutron configuration `` covering '' the proton density distribution .
a typical shell model configuration can be seen in the mean - field calculation shown in figure [ fig : o_potential ] . in the case of @xmath10o ,
the _ s _ and _ d _ shells are both occupied and the single - particle ordering of the states is the usual @xmath11 . the last occupied orbit is therefore the @xmath12 orbit .
a different situation is encountered when similar calculations are performed for the neutron rich @xmath2be nucleus .
it is well known @xcite that its ground state is a @xmath13 state with a dominant configuration .
this is in contrast with the normal single - particle ordering , where the @xmath14 orbit is occupied before the @xmath15 orbit .
this peculiar configuration , together with the very low binding energy , 0.505 mev , builds up the neutron halo structure of the @xmath2be ground state .
the mean - field potential for @xmath2be is shown in figure [ fig:11be_potential ] .
the tail component of the corresponding neutron density , extending well outside the nuclear surface ( defined by a root - mean - square radius of @xmath16 fm @xcite ) can be seen in figure [ fig:11be_density ] . there
, the neutron and proton densities are shown as calculated by the hartree - fock method with a skyrme - m * interaction
. the standard hartree - fock calculation would produce a ground - state with a @xmath14 configuration . to simulate the experimental situation ,
the results shown in the figure are obtained forcing the ground - state to be a @xmath15 state .
also shown in the figure is a neutron density with an added yukawa - tail determined by fixing the binding energy of the ground - state . in this way ,
the resulting neutron density compares well with the experimentally derived density .
it should be noted that the wave function corresponding to the @xmath15 orbit in a woods - saxon potential itself can reproduce the external part of the neutron density . in the calculations of the capture cross section shown below
we have used this wave function to evaluate the electric dipole matrix elements .
the peculiar structure properties briefly described in the previous section have been shown to exhibit strong influence on nuclear excitation modes @xcite . in what follows
, we will show the results of the effect of the neutron halo structure on electric dipole transitions .
these can be investigated , in principle , in both the reaction directions : in the neutron capture as well as in the coulomb dissociation process .
our group has recently investigated the effect of the nuclear wave function component in the external region on the neutron capture process @xcite .
we have calculated several neutron capture cross sections of light nuclei using the direct radiative capture ( drc ) model .
the calculations for @xmath17c@xmath18 and @xmath19o@xmath18 reactions have been compared with recent experimental results from direct measurements @xcite .
the drc reaction mechanism may be responsible for the most part of the capture reaction cross section in the particular condition in which the density of states is low enough to hinder the compound nucleus formation .
furthermore , because the halo structure arises mainly from loosely bound _ s
_ orbits ( see the discussion above ) , electric dipole @xmath20-ray emissions can only be induced by incident _
p_-wave neutrons .
in fact , the drc process is essentially determined by the e1 transition matrix elements @xmath21 in the case of incident _ p_-wave neutrons , these matrix elements are very much sensitive to the tail component of the final capturing state wave function @xmath22 and very little sensitive to the treatment of the incident channel neutron scattering state @xmath23 @xcite . the energy dependence as well as the strength of e1 emission due to incident _ p_-wave neutrons is therefore strongly influenced by the neutron halo structure of the residual nucleus capturing state .
whether this state is the ground or an excited nuclear state makes no difference in this scheme . as mentioned above
, we can compare the @xmath0be@xmath1be drc cross section with that derived from the experiment in the inverse kinematics : the coulomb dissociation of @xmath2be .
a recent coulomb dissociation experiment has been performed using the radioactive beam line rips at riken @xcite . a schematic view of the process is shown in figure [ fig:11be_diss ] .
note that in the actual experiment , the dissociation from the first excited state at @xmath24 mev ( @xmath25 ) does not take place because all the nuclei of the incident beam are in their ground state .
the electric dipole strength distribution @xmath26 , as measured in the coulomb dissociation experiment , can be related to the neutron capture cross section , @xmath27 , by @xmath28 where @xmath29 is the excitation energy and @xmath30 the emitted neutron wave number .
this relation is only valid for the present combination of angular momentum values .
the results of the drc calculation in comparison to the experimental values derived from the coulomb dissociation experiment and converted into the neutron capture channel are shown in figure [ fig:10be_ng ] .
as a first comment to this result , one has to note that the experimental values have to be compared only to the calculation including the @xmath31 transition because , as noted above , the experiment only includes incident @xmath2be nuclei in their ground - state .
the next feature which can be seen in the figure is that the low energy part , where the experimental uncertainty is lowest , is quite well reproduced by the calculation .
this can be interpreted as a confirmation of the halo structure of the ground - state of @xmath2be .
in fact , the matrix elements which enter in the expression for the drc capture given above can be decomposed into a product of three factors @xmath32 where @xmath33 is the spectroscopic factor of the bound state , @xmath34 a geometrical factor containing only angular momentum coupling constants and @xmath35 is the radial part of the overlap between the initial and final state . now ,
if the initial state only consists of _ p_-wave neutrons in the continuum , its wave function is simply given by @xmath36 with @xmath37 ( @xmath38 is a spherical bessel function ) .
the drc cross section is proportional to the square of the overlap integral @xmath39 and is , therefore , nothing but a fourier - bessel transformation of the bound state radial wave function @xmath40 .
a measure of the capture cross section ( or of the coulomb dissociation cross section ) is a measure of the wave function in the fourier space .
then , the low energy part of the capture cross section plot corresponds to large values of the radial coordinate which is influenced by the halo configuration .
finally , the fluctuations in the measured values at @xmath41 mev , though still masked by a large experimental uncertainty , may contain information on a @xmath2be excited state .
in fact , the first state above the neutron emission threshold is at @xmath42 mev ( corresponding to @xmath43 mev ) @xcite and has @xmath44 or @xmath45 .
it can be populated by electric quadrupole excitation with the emission of _ d_-wave neutrons .
should the experimental uncertainty be reduced to the level of the low energy part , the structure of this @xmath2be excited state could be deduced .
we have shown that the halo structure of light neutron - rich nuclei can be investigated in the frame of a simple drc model .
we have shown this in the particular case in which the cross section was not obtained in a direct measurement but derived from an experiment in the inverse kinematics .
this technique can be very useful to study `` exotic '' nuclear structure properties .
in addition , it can be utilized to obtain neutron capture cross sections for unstable nuclei where direct measurements are not possible .
this possibility is particularly important for applications in nuclear astrophysics @xcite where neutron capture reaction rates are needed to investigate primordial ( big - bang ) as well as stellar nucleosynthesis .
we thank y. nagai , t. shima and n. fukunishi for many useful discussions that greatly helped to generate the present contribution . this work has been supported by the japan science and technology agency through the sta fellowship program ( i d 194102 ) .
9 i. tanihata , nucl . phys . * a 552 * ( 1991 ) 275c
. i. tanihata _
et al . _ ,
b * 289 * ( 1992 ) 261 .
n. fukunishi , t. otsuka , and i. tanihata , phys .
c * 48 * ( 1993 ) 1648 .
m. yokoyama , t. otsuka and n. fukunishi , phys .
c * 52 * ( 1995 ) 1122 .
reinhard , in _ computational nuclear physics _
, pag . 28 .
k. langanke , j. a. maruhn , and s. e. koonin eds .
( springer - verlag , berlin 1991 ) .
f. ajzenberg - selove , nucl .
phys . * a 506 * ( 1990 ) 1 .
t. nakamura _
et al . _ ,
b * 331 * ( 1994 ) 296 .
t. otsuka , m. ishihara , n. fukunishi , t. nakamura , and m. yokoyama , phys .
c * 49 * ( 1994 ) r2289 .
a. mengoni , t. otsuka , and m. ishihara , phys .
c * 52 * ( 1995 ) r2334 .
a. mengoni , t. otsuka and m. ishihara , proceedings of a _ workshop on experimental perspectives with radioactive nuclear beams_. c. signorini and a. r. spalla eds .
lnl - infn ( rep ) 96/95 , pag
a. mengoni , proceedings of the _ riken - infn joint symposium _ , riken , wako , japan , may 22 - 26 , 1995 . a. yoshida and t. fukuda eds .
( world scientific , singapore ) , in press ( 1996 ) .
a. mengoni , t. otsuka and m. ishihara , proceedings of the _ international symposium on origin of matter and evolution of galaxies_. atami , japan , 18 - 20 january , 1996 .
t. kajino et al .
eds , world scientific , in press ( 1996 ) . | exotic nuclear structure properties include neutron skin neutron halo in light neutron rich nuclei @xcite rigorous definition of neutron skin neutron halo show example both structures features properties neutron skin structure investigated @xcite responsible for low energy excitation mode decoupled from isoscalar giant quadrupole resonance concentrate on effects neutron halo structure importance low energy electric dipole transition mechanism in @xmath2be example neutron skin structure shown in figure result of mean - field calculation using hartree - fock method @xcite for nuclei @xmath9o @xmath10o calculations performed using common skyrme - s3 interaction general features deduced with other interaction mean - field potential for protons neutrons shown with bound single - particle energy spectrum double - magic nucleus neutrons protons occupy analogous single - particle orbits comparable energies lower same calculation for neutron rich @xmath10o ( n=20 ) nucleus double neutrons occupy single - particle states energies different fermi energies of protons neutrons different due to larger neutron numberplot densities two nuclei thick neutron configuration covering proton density distribution typical shell model configuration in mean - field calculation figure_potential @xmath10o s d _ shells occupied single - particle ordering usual @xmath11 last occupied orbit @xmath12 orbit different situation neutron rich @xmath2be nucleus ground state @xmath13 state dominant configuration contrast normal - particle ordering @xmath14 orbit occupied before @xmath15 orbit configuration low binding energy 0.505 mev builds neutron halo structure @xmath2be ground state mean - field potential @xmath2be in figure fig:11be_potential tail component neutron density extending outside nuclear surface root - mean - square radius @xmath16 @xcite in figure_density neutron proton densities calculated by hartree - fock method interaction standard hartree - fock calculation ground - state @xmath14 configuration experimental results forcing ground - state @xmath15 state neutron density with added yukawa - tail determined fixing binding energy ground - state resulting neutron density compares with experimentally derived densitywave function @xmath15 orbit in woods - saxon potential external neutron density calculations capture cross section used wave function electric dipole matrix elements peculiar structure properties nuclear excitation modes @xcite results effect neutron halo structure on electric dipole transitions investigated in reaction neutron capture coulomb dissociation process group investigated effect nuclear wave function external region on neutron capture process calculated neutron capture cross sections using direct radiative capture ( drc ) model calculations for @xmath17c@xmath18 @xmath19o@xmath18 reactions compared with results direct measurements drc reaction mechanism may responsible for capture reaction cross section density of states low compound nucleus formation halo structure arises from loosely bound _ s orbits electric dipole @xmath20-ray emissions induced by incident _ p_-wave neutrons drc process determined by e1 transition matrix elements @xmath21 incident-wave neutrons matrix elements sensitive to tail component final capturing state wave function @xmath22 to incident channel neutron scattering state @xmath23energy dependence strength of e1 emission due to incident p-wave neutrons influenced by neutron halo structure of residual nucleus capturing state state ground or excited nuclear state no difference compare @xmath0be@xmath1be drc cross section with inverse kinematics coulomb dissociation of @xmath2be recent coulomb dissociation experiment using radioactive beam line rips at riken @xcite process in figure fig:11be_diss dissociation from first excited state at @xmath24 @xmath25 nuclei in ground state electric dipole strength distribution @xmath26 related to neutron capture cross section @xmath27 @xmath28 @xmath29 excitation energy @xmath30 emitted neutron wave number relation valid for angular momentum values results of drc calculation to experimental values from coulomb dissociation experiment shown in figure fig:10be_ng experimental values compared only to calculation including @xmath31 transition experiment includes @xmath2be nuclei in ground - state low energy part experimental uncertainty lowest reproduced by calculation confirmation of halo structure of ground - state of @xmath2bematrix elements drc capture decomposed into three factors @xmath32 @xmath33 spectroscopic factor bound state geometrical factor angular momentum coupling constants @xmath35 radial part overlap between initial final state if initial state p_-wave neutrons wave function by @xmath36 with @xmath37 @xmath38 spherical bessel function drc cross section proportional to overlap integral @xmath39 fourier - bessel transformation of bound state radial wave function @xmath40 measure capture cross section wave function in fourier space low energy part capture cross section corresponds to large values radial coordinate influenced by halo configuration fluctuations in values at @xmath41 mev may contain information on @xmath2be excited state first state above neutron emission threshold at @xmath42 mev @xmath43 @xmath44 or @xmath45 populated by electric quadrupole excitation emission d_-wave neutrons experimental uncertainty reduced to low energy structure of @xmath2be excited state deduced halo structure of light neutron - rich nuclei investigated in simple drc model cross section not from experiment inverse kinematicstechnique useful study exotic nuclear structure properties obtain neutron capture cross sections unstable nuclei direct measurements important nuclear astrophysics neutron capture reaction rates primordial stellar nucleosynthesis thank y. nagai t. shima n. fukunishi discussions contribution work supported japan science technology agency sta fellowship program tanihata nucl . . 552 * 1991 ) 275c tanihata 289 * 1992 ) 261 n fukunishi t. otsuka tanihata phys . 48 * 1993 ) 1648 yokoyama t. otsuka n fukunishi phys 52 * 1995 ) 1122 reinhard computational nuclear physics pag. 28 langanke. maruhn s e. koonin springer - verlag berlin 1991 ajzenberg - selove nucl . phys . 506 * 1990 ) t nakamura al 331 * 1994 ) 296 t otsuka ishihara fukunishi yokoyama phys . 49 * 1994 ) r2289 mengoni t otsuka ishihara phys . * 52 * 1995 ) r2334ishihara proceedings workshop radioactive nuclear beams. signorini spalla eds lnl infn 96/95 mengoni proceedings riken - infn symposium riken japan may 22 - 26 1995 yoshida t fukuda eds world scientific singapore press 1996 mengoni otsuka ishihara proceedings international symposium origin matter evolution galaxies atami japan 18 - 20 january 1996 kajino eds world scientific press 1996 | 0.436539 | 2,766 | 1,422 | , Saving $0.1 in GPT-4. |
177,755 | accretion , accretion discs mhd methods : numerical gamma - rays : bursts i present results from magnetohydrodynamic ( mhd ) simulations of a gaseous envelope collapsing onto a black hole .
these results support the notion that the collapsar model is one of most promising scenarios to explain the huge release of energy in a matter of seconds associated with gamma ray bursts ( grb ) .
additionally , the mhd simulations show that at late times , when the mass supply rate is expected to decrease , the region in the vicinity of the black hole can play an important role in determining the rate of accretion , its time behaviour , and ultimately the energy output . in particular , the magnetic flux accumulated around the black hole can repeatedly stop and then restart the energy release .
as proposed by proga and zhang , the episode or episodes of reoccurring of accretion processes can correspond to x - ray flares discovered recently in a number of grbs .
= = = 1=1=0pt = 2=2=0pt [ firstpage ] | the collapsar model proposes that long duration gamma - ray bursts ( grbs ) are powered by accretion onto the collapsed iron core of a massive star ( woosley 1993 and paczyski 1998 ) . as gas accretes at a very high rate ( @xmath0 ) , large neutrino and magnetic fluxes , a powerful outflow , and a grb are produced .
this model is strongly supported by the association of long - duration grbs with stellar collapse ( e.g. , hjorth et al .
2003 ; stanek et al . 2003 ) and by theoretical studies of stellar collapse ( macfadyen & woosley 1999 ; popham , woosley & fryer 1999 ; proga et al .
2003 ; mizuno et al .
2004 ; fujimoto et al .
2006 ) .
recent grb observations obtained with _ swift _
provide new challenges to the model as early x - ray afterglow lightcurves of nearly half of the long - duration grbs show x - ray flares ( burrows et al .
2005 ; romano et al .
2006 ; falcone et al .
the model must then account for an initial powerful burst of energy and for an extended activity in a form of x - ray flares .
i present results from direction simulations of early evolution of stellar collapse and explore the implications of these and other simulations to infer the physical conditions in the vicinity of a black hole ( bh ) during the late phase of evolution , i.e. , when most of the stellar mass is accreted . to study the extended grb activity
, one would need to follow the collapse of the entire star . however , such studies are beyond current computer and model limits .
proga et al .
( 2003 ) performed the first mhd simulation of a collapsar model .
their simulation begins after the @xmath1 iron core of a 25 @xmath2 presupernova star has collapsed and follows the ensuing accretion of the @xmath3 helium envelope onto the central bh formed by the collapsed iron core .
a spherically symmetric progenitor model is assumed , but the symmetry is broken by the introduction of a small , latitude - dependent angular momentum and a weak split - monopole magnetic field .
the simulation includes pseudo - newtonian potential , a realistic equation of state , photodisintegration of bound nuclei and cooling due to neutrino emission .
the simulation shows that the early phase of evolution starts with a transient episode of infall .
then the rotating gas starts to pile up outside the bh and forms a thick accretion torus bounded by a centrifugal barrier near the rotation axis .
soon after the torus forms ( i.e. , within a couple of orbital times at the inner edge ) , the magnetic field is amplified by the magnetorotational instability ( mri , e.g. , balbus & hawley 1998 ) and shear .
mri facilitates outward transport of angular momentum in the torus and enables torus accretion onto a bh .
another important effect of magnetic fields is that the torus produces a magnetized corona and an outflow .
the presence of the corona and outflow is essential to the evolution of the inner flow at all times and to the entire flow close to the rotational axis during the latter phase of the evolution .
the outflow very quickly becomes sufficiently strong to overcome supersonically infalling gas and to escape the star .
the top left panel in fig .
1 shows a schematic picture on the inner most part of the flow .
the main conclusion from the simulation is that , within the collapsar model , mhd effects alone are able to launch , accelerate and sustain a strong polar outflow .
other important insights gained from the mhd simulation include : 1 ) in the accretion torus , the toroidal field dominates over the poloidal field and the gradient of the former drives a polar outflow ; 2 ) the polar outflow can be poynting flux - dominated ; 3 ) the polarity of the toroidal field can change with time ; 4 ) the polar outflow reaches the outer boundary of the computational domain ( @xmath4 cm ) with an expansion velocity of 0.2 c ; 5 ) the polar outflow is in a form of a relatively narrow jet ( when the jet breaks through the outer boundary its half opening angle is @xmath5 ) ; 6 ) most of the energy released during the accretion is in neutrinos , @xmath6 .
neutrino driving will increase the outflow energy ( e.g. , fryer & mszros 2003 and references therein ) , but could also increase the mass loading of the outflow if the energy is deposited in the torus . due to limited computing time
, proga et al.s ( 2003 ) simulation was stopped at @xmath7 s , which corresponds to 6705 orbits of the flow near the inner boundary .
the total mass and angular momentum accreted onto the bh during the simulation are @xmath8 and @xmath9 , respectively .
thus , the simulation captured just the beginning of the accretion and one expects the accretion to continue much longer , roughly the collapse timescale of the helium core ( @xmath10 s ) .
however , one also expects a decrease of the mass supply rate with time , especially in the late phase of activity , because the stellar mass density decreases with increasing radius .
the long time evolution of mhd accretion flows was studied by proga & begelman ( 2003 ) who explored simulations very similar to the one in proga et al .
( 2003 ) but with much simpler physics ( i.e. , an adiabatic equation of state , no neutrino cooling or photodisintegration of helium ) .
next , i present some key results from proga & begelman ( 2003 ) as they are relevant to late evolution of a collapsing star . the early phase of the time evolution and the dynamics of the innermost flow , are very similar in simulations present in proga et al .
( 2003 ) and proga & begelman ( 2003 ) .
in particular , after an initial transient behaviour , the flow settles into a complex convolution of several distinct , time - dependent flow components including an accretion torus , its corona and outflow , and an inflow and outflow in the polar funnel ( see the bottom left panel in fig .
however , the late evolution of proga & begelman s ( 2003 ) simulations shows that the torus accretion can be interrupted by a strong poloidal magnetic field in the vicinity of a bh ( compare the bottom left and right panels in fig .
thus the region in the vicinity of the bh and the bh itself can play an important role in determining the rate and time behaviour of the accretion and the energy output .
motivated by this result , proga & zhang ( 2006 ) conjectured that the energy release can be repeatedly stopped and then restarted , provided the mass supply rate decreases with time even if the decrease is smooth ( see fig .
proga & zhang appealed to the fact that , as mass is being accreted onto a bh , the magnetic flux is accumulating in the vicinity of the bh . eventually , this magnetic flux must become dynamically important and affect the inner accretion flow , unless the magnetic field is very rapidly diffused .
analytic estimates derived by proga & zhang show that the model can account for the observed features of the x - ray flares .
both numerical and theoretical models of magnetized accretion flows support the notion that long duration grbs can be explained by the collapsar model .
the models also show that the innermost part of the flow and accretor can respond _ actively _ to changes of the accretion flow at larger radii .
in particular , the innermost accretion flow can be halted for a very long time as to account for an extended activity of grbs with x - ray flares . | collapsar model proposes long gamma - ray bursts powered by accretion collapsed iron core of massive star woosley 1993 paczyski 1998 gas accretes at high rate large neutrino magnetic fluxes powerful outflow grb produced model supported by association grbs with stellar collapse hjorth 2003 stanek theoretical studies macfadyen woosley 1999 mizuno 2004 fujimoto 2006 recent grb observations provide challenges early x - ray afterglow lightcurves of half grbs show x - ray flares model must account for initial burst energy extended activity x - ray flares present results from simulations of early evolution of stellar collapse explore implications to infer physical conditions in vicinity black hole during late phase evolution stellar mass accreted to study extended grb activity follow collapse of entire star studies beyond current computer model limits proga et al . 2003 performed first simulation of collapsar model simulation begins after iron core of 25 @xmath2 presupernova star collapsed follows accretion of @xmath3 helium envelope onto central bh collapsed iron corespherically symmetric progenitor model assumed symmetry broken by small latitude - dependent angular momentum weak split - monopole magnetic field simulation includes pseudo - newtonian potential realistic equation of state photodisintegration nuclei cooling due to neutrino emission early phase evolution starts with transient episode infall rotating gas up outside bh forms thick accretion torus centrifugal barrier near rotation axis torus magnetic field amplified by magnetorotational instability ( mri shear mri facilitates transport angular momentum enables torus accretion onto bh torus produces magnetized corona outflow corona outflow essential to evolution inner flow flow close to rotational axis during latter phase outflow strong to overcome infalling gas escape star top left panel in fig . 1 shows schematic picture inner part flow conclusion collapsar model mhd effects launch accelerate sustain strong polar outflowinsights from mhd simulation include in accretion torus toroidal field dominates over poloidal field gradient drives polar outflow polar outflow can be poynting flux - dominated polarity with time polar outflow reaches outer boundary computational domain with expansion velocity 0.2 c polar outflow narrow jet half opening angle is @xmath5 ) energy released during accretion in neutrinos neutrino driving outflow energy increase mass loading if energy deposited in torus limited time proga et al.s 2003 simulation stopped at @xmath7 s to 6705 orbits near inner boundary total mass and angular momentum accreted onto bh are @xmath8 and @xmath9 simulation captured beginning of accretion expects to continue longer collapse timescale of helium core @xmath10 s ) expects decrease of mass supply rate with time late phase activity stellar mass density decreases with radius long time evolution of mhd accretion flows studied by proga & begelman ( 2003 ) simulations similar simpler physicsadiabatic equation no neutrino cooling photodisintegration of helium present key results from proga & begelman ( 2003 ) relevant to late evolution of collapsing star early phase evolution dynamics of innermost flow similar in simulations in proga et al 2003 & begelman after initial transient behaviour flow settles into complex convolution of time - dependent components including accretion torus corona and outflow inflow and outflow in polar funnel late evolution of proga begelman simulations shows torus accretion interrupted by strong poloidal magnetic field vicinity of bh region vicinity of bh rate time behaviour of accretion energy output proga & zhang ( 2006 ) conjectured energy release can be repeatedly stopped restarted mass supply rate decreases with time mass accreted onto bh magnetic flux in vicinity affect inner accretion flow unless magnetic field rapidly diffused analytic estimates by proga & zhang show model can account for features x - ray flares numerical models of magnetized accretion support long duration grbs be explained by collapsar modelmodels show innermost part flow accretor respond to changes at larger radii innermost accretion flow can halted long extended activity of grbs with x - ray flares | 0.312228 | 1,726 | 852 | , Saving $0.1 in GPT-4. |
177,763 | we briefly discuss an arch - like structure that forms and grows during the rapid straightening of a chain lying on a table .
this short note accompanies a fluid dynamics video submission ( v029 ) to the aps dfd gallery of fluid motion 2011 .
the accompanying video shows fifty feet of chain ( silver - plated base metal , @xmath02 mm links , ) arranged on about two feet of table top ; off camera , a power drill pulls one end at a velocity @xmath1 8 m / s . the relevant forces are those due to the chain s inertia , and a line tension whose role is identical to that of pressure in a one - dimensional incompressible fluid .
gradients in this tension amplify and advect small of out - of - plane disturbances , presumably with some upward rectification by the table .
the advection velocity drops to zero as the chain approaches its terminal velocity @xmath2 .
there , the tension is uniform , structures are nearly `` frozen '' in the laboratory frame , and the vertical slack accumulates behind and within a slowly growing arch . by the end of the video , this arch has grown to about ten centimeters high .
however , an analysis that takes gravity into account suggests that such a structure may remain stable when grown to a height on the order of a meter , the natural length scale of the system with gravity included ( @xmath3 ) .
structures of such a size are difficult to access experimentally , as they do not evolve within the corresponding natural time scale of about one second ( @xmath4 ) .
this is because the system is oblivious to gravity during the observation period of the experiment ; in a region with local curvature @xmath5 on the order of a few tens of inverse meters , an element of chain sees a large ratio of inertial to gravitational accelerations @xmath6 . | we thank n. menon and c. d. santangelo for support and discussions , and acknowledge funding from national science foundation grants dmr 0907245 and dmr 0846582 . | thank n menon d santangelo support acknowledge national science foundation grants 0907245 0846582 | 0.065849 | 40 | 21 | , Saving $0.0 in GPT-4. |
177,788 | diffuse x - ray emission has been detected in a small number of planetary nebulae ( pne ) , indicating the existence of shocked fast stellar winds and providing support for the interacting - stellar - winds formation scenario of pne .
however , the observed x - ray luminosities are much lower than expected , similar to the situation seen in bubbles or superbubbles blown by massive stars .
ad hoc assumptions have been made to reconcile the discrepancy between observations and theoretical expectations .
we have initiated _ fuse _ programs to observe ovi absorption and emission from pne , and our preliminary results indicate that ovi emission provides an effective diagnostic for hot gas in pn interiors . | the formation of planetary nebulae ( pne ) is a complex issue .
the current fast stellar wind from the central star of a pn must interact with the previous slow agb wind ; it has thus been suggested that pne are formed by interacting stellar winds ( kwok , purton , & fitzgerald 1978 ) . the frequently observed multi - polar and point - symmetric structures of young pne suggest that collimated outflows are responsible for shaping pne in early evolutionary stages ( sahai & trauger 1998 ) .
the fast stellar winds and high - velocity collimated outflows , upon impacting the ambient medium , will be shocked to a temperature of @xmath0 , where @xmath1 is the average mass per particle , @xmath2 is the shock velocity , and @xmath3 is the boltzmann constant
. hot gas at x - ray - emitting temperatures , i.e. , @xmath4 k , can be generated for shock velocities greater than @xmath5300 km s@xmath6 . as reviewed by guerrero et al .
( in this volume ) , x - ray observations have revealed hot gas in a number of young pne such as ngc6543 and ngc7027 , and established stringent upper limits on the hot gas content in evolved pne such as ngc7293 ( the helix ) .
the detection of diffuse x - ray emission from pne provides a strong support for the interacting - stellar - winds scenario of pn formation .
this formation mechanism is in fact similar to that for circumstellar bubbles blown by wolf - rayet ( wr ) stars .
massive wr stars are descendants of red supergiants or luminous blue variables which lose mass via copious slow winds ; the fast wr wind sweeps up the previous slow wind and forms a circumstellar bubble . comparisons between pne and wr bubbles will help us understand both types of objects . in the first part of this paper , we will discuss the physical properties of the 10@xmath7 k hot gas in pn interiors and make comparisons with shocked stellar winds in wr bubbles .
in the second part of this paper , we will discuss hot gas at cooler temperatures , i.e. , a few @xmath8 k , in order to : ( 1 ) determine whether some evolved pne are filled with such cooler hot gas , and ( 2 ) study the interface layer between the 10@xmath7 k interior gas and the 10@xmath9 k nebular shell in young pne .
we will present far - uv observations of the 10@xmath10 k gas in pne and use the results to make predictions about the presence or absence of 10@xmath7 k hot gas in pn interiors .
_ chandra _ and _ xmm - newton _ observations of diffuse x - ray emission from pne allow us to examine the spatial distribution and temperatures of the hot gas . among the six pne with diffuse x - ray emission ,
mz3 and ngc6543 are the only two that are adequately resolved by the instrumental point spread function , and both of them show a limb - brightened x - ray morphology , indicating that the dense x - ray - emitting gas is concentrated near the inner walls of their nebular shells .
this is similar to what we have observed in the wr bubbles ngc6888 and s308 ( chu et al .
2003 ; gruendl et al .
2004 ) ; see fig . 1 for an _ xmm - newton _ epic image of s308 and a comparison with an optical image
. the large angular size of s308 makes it possible to see the detailed relationship between the hot interior gas and the cooler nebular shell .
x - ray spectra of the hot gas in pn interiors ( guerrero et al . in this volume ) show a narrow range of temperatures , @xmath12 k. these temperatures are much lower than the @xmath13 k gas detected in superbubbles around young clusters , where colliding stellar winds may have produced the hot gas and the high temperatures ( townsley et al .
2003 ) . on the other hand ,
the temperatures of the x - ray - emitting gas in pne are similar to the @xmath14 k gas detected in the wr bubbles ngc6888 and s308 ( see their spectra in fig . 1 ) .
both pne and wr bubbles have been modeled hydrodynamically ( zhekov & perinotto 1996 , 1998 ; garca - segura et al.1996a , b ) , assuming that the fast stellar wind is shock - heated to high temperatures and forms a contact discontinuity with the photoionized 10@xmath9 k nebular shell .
it is further assumed that thermal conduction and mass evaporation take place at the hot / cold gas interface , which lowers the temperature and raises the density of the hot interior gas .
in fact , the hot interior gas is dominated by the evaporated nebular material . as the x - ray emission is proportional to the square of the density
, the integrated x - ray spectrum of the hot gas will be dominated by the emission from the densest gas which is at the lowest temperatures .
therefore , the low temperatures derived from model fits to the x - ray spectra are expected .
the x - ray luminosities of the diffuse emission from pne and wr bubbles are , however , at least an order of magnitude lower than those expected from models ( chu et al . 2001 ; wrigge et al .
1994 ) , implying that the amount of hot gas is significantly less than that predicted by models with thermal conduction .
several possibilities have been suggested to reconcile the discrepancies between observation and theory ( arnaud et al.1996 ; soker & kastner 2003 ) : ( 1 ) the fast stellar wind may have evolved with time , with a lower velocity or mass loss rate in the past ; ( 2 ) the fast stellar wind is anisotropic , with a bipolar structure ; and ( 3 ) the thermal conduction at the hot / cold gas interface may be inhibited by the presence of magnetic fields . unfortunately , diffuse x - ray emission has been detected in only six pn interiors ; two pne are not adequately resolved spatially , and another two are detected with less than 100 counts .
the current x - ray observations are clearly inadequate to constrain models and to discern the above possibilities .
high - quality x - ray observations of pne are badly needed .
hot gas at 10@xmath10 k is traditionally studied with spectral lines of highly ionized species , such as civ , nv , and ovi . if these ions are produced by thermal collisions , civ , nv , and ovi will be the dominant ionization stages for c , n , and o at temperatures of @xmath15 , @xmath16 , and @xmath17 k , respectively . as the ionization potentials of ciii , niv , and ov are 47.9 , 77.5 , and 113.9 ev , photoionization of these ions becomes significant in pne with central stars at stellar effective temperatures greater than @xmath5 35,000 , 75,000 , and 125,000 k , respectively .
therefore , ovi provides the best diagnostic for 10@xmath10 k gas , as it is the least easily confused by photoionization .
the ovi @xmath181032,1037 lines can be observed with the recently launched _
far ultraviolet spectroscopic explorer ( fuse)_. four _ fuse _ observing programs were carried out to search for 10@xmath10 k gas in pn interiors .
some observations use spectra of the central stars to search for nebular ovi absorption , while others use spectra of the nebula ( avoiding the central star ) to search for ovi emission .
these four programs are : * ovi absorption observation of 10 pne ( pi : gruendl ; ao2 ) * ovi emission observation of ngc6543 ( pi : guerrero ; ao3 ) * ovi emission observation of ngc7009 ( pi : iping ; ao3 ) * ovi emission observation of 9 pne ( pi : guerrero ; ao4 ) these observations have produced many exciting results .
first of all , _ fuse _ observations detected ovi in ngc6543 and ngc7009 , two pne with known diffuse x - ray emission .
the top panels of fig .
2 display _ fuse _ spectra of their central stars ; both show pronounced p cygni profiles in the stellar ovi lines , indicating the presence of fast stellar wind .
ngc6543 shows a stellar spectrum with nebular ovi absorption at a velocity blue - shifted from the nebular systemic velocity , indicating that it is associated with the interface layer on the approaching side of the nebular shell .
ngc7009 does not show obvious nebular ovi absorption in its stellar spectrum .
the middle panels of fig .
2 display _ fuse _ spectra of apertures centered inside the nebulae .
both nebulae show obvious ovi emission lines .
the bottom panels of fig .
2 display _ fuse _ spectra of apertures centered outside the nebulae , where only airglow lines are seen .
_ fuse _ observations of ngc6543 and ngc7009 demonstrate clearly that the ovi emission from interfaces between the 10@xmath7 k interior and the 10@xmath9 k nebular shell can be detected .
quantitative analysis of the ovi emission and absorption lines , complicated by the interstellar and circumstellar h@xmath19 absorption along the line of sight , is in progress .
_ fuse _ observations of the helix nebula ( fig .
3 ) , a pn confirmed to lack 10@xmath7 k gas , show contrasting spectra in both the central star and the nebula .
the spectrum of the central star shows stellar ovi absorption , but not p cygni profiles , indicating a lack of fast stellar wind , which is consistent with the conclusion derived from _ iue _ observations ( cerruti - sola & perinotto 1985 ) . neither ovi absorption nor ovi emission from the nebula
is detected .
it is thus reasonable to expect that a lack of both stellar p cygni profile in ovi and nebular ovi is a good indication of an absence of hot gas in the pn interior .
ngc3132 is an awarded _
target in cycle 5
. the _ fuse _ observations of its central star and the nebula , displayed in fig .
3 , show neither stellar p cygni ovi nor nebular ovi .
we predict that no 10@xmath7 k hot gas exists in ngc3132 and that _ chandra _ observations of ngc3132 will not detect any diffuse x - ray emission .
x - ray observations have shown that hot gas exists in some pn interiors , and this interior hot gas may play an important role in the formation and evolution of pne . to understand the interacting stellar winds in pne , high s / n and
high spatial resolution x - ray observations of more pne are needed .
_ fuse _ observations of ovi emission offer an effective diagnostic for the presence of hot gas in pne , and should be used to guide the target selection for x - ray observations .
future models of interacting stellar winds need to consider the plasma micro - physics and dynamical mixing of nebular material , in addition to a critical assessment of the degree of thermal conduction . when high - quality x - ray spectra are available , they need to be modeled with considerations of non - equilibrium ionization and temperature structure in the hot gas .
arnaud , k. , borkowski , k. j. , & harrington , j. p. 1996
, , 462 , l75 cerruti - sola , s. , & perinotto m. 1985 , , 291 , 237 chu , y .- h . , guerrero , m. a. , gruendl , r. a. , et al .
2003 , , in press chu , y .- h . ,
guerrero , m. a. , gruendl , r. a. , et al .
2001 , , 553 , l69 garca - segura , g. , langer , n. , & mac low , m .-
1996a , a&a , 316 , 133 garca - segura , g. , mac low , m .-
m . , & langer , n. 1996b , a&a , 305 , 229 gruendl , r. a. , guerrero , m. a. , & chu , y .- h .
2004 , in preparation kwok , s. , purton , c. r. , & fitzgerald , p. m. 1978 , , 219 , l125 sahai , r. & trauger , j. t. 1998 , , 116 , 1357 soker , n. , & kastner , j. h. 2003 , , 583 , 368 townsley , l. k. , feigelson , e. d. , montmerle , t. , et al .
2003 , , 593 , 874 wrigge , m. , wendker , h. j. , & witsotzki , l. 1994 , , 286 , 219 zhekov , s. a. , & perinotto , m. p. 1996
, , 309 , 648 zhekov , s. a. , & perinotto , m. p. 1998
, , 334 , 239 | formation planetary nebulae complex fast stellar wind central star interact slow agb wind suggested pne formed by interacting stellar winds fitzgerald 1978 multi - polar point - structures young pne suggest collimated outflows shaping pne early trauger 1998 fast stellar winds high - velocity collimated outflows medium shocked to temperature @xmath0 average mass per particle shock velocity boltzmann constant hot gas at x - ray - emitting temperatures @xmath4 k generated for shock velocities greater than @xmath5300 km s@xmath6 x - ray observations revealed hot gas in young pne ngc6543 ngc7027 upper limits hot gas content in evolved pne ngc7293 detection diffuse x - ray emission from pne interacting - stellar - winds scenario pn formation formation mechanism similar circumstellar bubbles wolf - rayet stars stars descendants red lose mass via slow winds fast wr wind sweeps slow wind forms circumstellar bubble comparisons between pne wr bubbles understandfirst part paper discuss physical properties 10@xmath7 k hot gas pn interiors comparisons shocked stellar winds wr bubbles second part discuss hot gas cooler temperatures few @xmath8 k determine evolved pne filled cooler hot gas study interface layer 10@xmath7 k interior gas 10@xmath9 k nebular shell young pne present far - uv observations 10@xmath10 k gas pne predictions presence absence 10@xmath7 k hot gas pn interiors chandra xmm - newton observations diffuse x - ray emission pne examine spatial distribution temperatures hot gas six pne diffuse x - ray emission mz3 ngc6543 resolved instrumental point spread function show limb - brightened x - ray morphology dense x gas concentrated inner walls nebular shells similar wr bubbles ngc6888 s308 see fig . 1 image s308 comparison optical image large angular size s308 detailed relationship hot interior gas cooler nebular shell x - ray spectra hot gas pn interiors show narrow range temperatures @xmath12 ktemperatures lower than @xmath13 k gas in superbubbles around young clusters colliding stellar winds produced hot gas high temperatures townsley 2003 temperatures x - ray - gas in pne similar to @xmath14 k gas in wr bubbles ngc6888 s308 spectra fig . 1 pne wr bubbles modeled hydrodynamically zhekov perinotto 1996 - fast stellar wind shock - heated to high temperatures contact discontinuity with photoionized 10@xmath9 k nebular shell thermal conduction mass evaporation at hot / cold gas interface lowers temperature raises density hot interior gas hot gas dominated by evaporated nebular material x - ray emission proportional to density - spectrum dominated by densest gas lowest temperatures low temperatures model x - ray spectra expected x - ray luminosities diffuse emission from pne wr bubbles lower than chu 2001 wrigge 1994 hot gas less than predicted models thermal conduction possibilities suggested discrepancies observation theory arnaud1996 soker kastner 2003 fast stellar wind evolved lower velocity mass loss rate anisotropic bipolar structure thermal conduction at hot cold gas interface inhibited magnetic fields diffuse x - ray emission detected in six pn interiors two not resolved two less than 100 counts current x observations inadequate discern possibilities high - quality x - ray observations needed hot gas at 10@xmath10 k studied with highly ionized species civ nv ovi thermal collisions dominant ionization stages for c n o at temperatures @xmath15 @xmath17 k ionization potentials of ciii niv ov 47.9 77.5 113.9 ev photoionization significant in pne stars at stellar temperatures greater than @xmath5 35,000 75,000 125,000 k ovi best diagnostic for 10@xmath10 k gas least confused by photoionization ovi @xmath181032,1037 lines observed with far ultraviolet spectroscopic explorer fuse) four observing programs 10@xmath10 k gas in pn interiorsobservations use spectra central stars nebular ovi absorption others use spectra nebula ovi emission four programs are ovi absorption observation 10 pne gruendl ao2 emission ngc6543 guerrero ngc7009 iping 9 pne guerrero ao4 observations produced exciting results observations detected ovi in ngc6543 ngc7009 two pne with diffuse x - ray emission top panels 2 both show pronounced p cygni profiles stellar ovi lines fast stellar wind ngc6543 shows spectrum nebular ovi absorption velocity blue - shifted from nebular systemic velocity associated with interface layer side nebular shell ngc7009 show obvious nebular ovi absorption middle panels display apertures inside nebulae both nebulae show obvious ovi emission lines bottom panels outside nebulae only airglow lines seen observations of ngc6543 ngc7009 demonstrate ovi emission from interfaces between 10@xmath7 k interior 10@xmath9 k nebular shell detected quantitative analysis of ovi emission absorption lines interstellar circumstellar h@xmath19 absorption in progress observations helix nebula figpn lack 10@xmath7 k gas contrasting spectra central star nebula spectrum central star shows stellar ovi absorption not p cygni profiles lack fast stellar wind consistent with conclusion iue observations cerruti - perinotto 1985 neither ovi absorption nor ovi emission from nebula detected lack stellar p cygni profile ovi nebular ovi absence hot gas pn interior ngc3132 awarded target cycle 5 fuse observations central star nebula show neither stellar p cygni ovi nor nebular ovi predict no 10@xmath7 k hot gas in ngc3132 chandra observations detect diffuse x - ray emission x - ray observations hot gas in pn interiors may formation evolution pne stellar high s / n spatial resolution x - ray observations more pne needed fuse observations ovi emission diagnostic hot gas guide target selection x - ray observations future models stellar winds consider plasma micro - physics dynamical mixing nebular material thermal conduction high - quality x - ray spectra non - equilibrium ionization temperature structure hot gas arnaud borkowski harrington 1996 462 , l75 cerruti - sola perinotto m. 1985 291 , 237 chu guerrero m. a. gruendl r. a.. 2003 press chu guerrero gruendl 2001 553 , l69 garca - segura langer mac low 1996a 316 , 133 garca - segura mac low langer 1996b 305 , 229 gruendl r.. guerrero m.. chu 2004 preparation kwok purton fitzgerald . 1978 219 , l125 sahai trauger . 1998 116 , 1357 soker kastner j.. 2003 583 , 368 townsley feigelson montmerle . 2003 593 , 874 wrigge wendker witsotzki 1994 286 , 219 zhekov s. perinotto m. 1996 309 , 648 zhekov perinotto. 1998 334 , 239 | 0.439538 | 2,946 | 1,492 | , Saving $0.1 in GPT-4. |
177,796 | we investigate the fully general class of non - expanding , non - twisting and shear - free @xmath0-dimensional geometries using the invariant form of geodesic deviation equation which describes the relative motion of free test particles .
we show that the local effect of such gravitational fields on the particles basically consists of isotropic motion caused by the cosmological constant @xmath1 , newtonian - type tidal deformations typical for spacetimes of algebraic type d or ii , longitudinal motion characteristic for spacetimes of type iii , and type n purely transverse effects of exact gravitational waves with @xmath2 polarizations .
we explicitly discuss the canonical forms of the geodesic deviation motion in all algebraically special subtypes of the kundt family for which the optically privileged direction is a multiple weyl aligned null direction ( wand ) , namely d(a ) , d(b ) , d(c ) , d(d ) , iii(a ) , iii(b ) , iii@xmath3 , ii@xmath3 , ii(a ) , ii(b ) , ii(c ) and ii(d ) .
we demonstrate that the key invariant quantities determining these algebraic types and subtypes also directly determine the specific local motion of test particles , and are thus measurable by gravitational detectors . as an example , we analyze an interesting class of type n or ii gravitational waves which propagate on backgrounds of type o or d , including minkowski , bertotti robinson , nariai and plebaski hacyan universes . | spacetimes of the kundt class are defined by a purely geometric property , namely that they admit a geodesic null congruence which is non - expanding , non - twisting and shear - free . in the context of four - dimensional general relativity ,
such vacuum and pure radiation spacetimes of type n , iii , or o were introduced and initially studied 50 years ago by wolfgang kundt @xcite .
the whole kundt class is , in fact , much wider .
it admits a cosmological constant , electromagnetic field , other matter fields and supersymmetry .
the solutions may be of various algebraic types and can be extended to any number @xmath0 of dimensions .
all kundt spacetimes ( without assuming field equations ) can be written as @xmath4 see @xcite . in this metric
, the coordinate @xmath5 is an affine parameter along the optically privileged null congruence @xmath6 ( with vanishing expansion , twist and shear ) , @xmath7const .
label null ( wave)surfaces , and @xmath8 are @xmath9 spatial coordinates in the transverse riemannian space
. the spatial part @xmath10 of the metric must be independent of @xmath5 , all other metric components @xmath11 and @xmath12 can be functions of all the coordinates @xmath13 .
the kundt class of spacetimes is one of the most important families of exact solutions in einstein s general relativity theory , see chapter 31 of the monograph @xcite or chapter 18 of @xcite for reviews of the standard @xmath14 case .
it contains several famous subclasses , both in four and higher number of dimensions , with interesting mathematical and physical properties .
the best - known of these are pp - waves ( see @xcite and references therein ) which admit a covariantly constant null vector field .
there are also vsi and csi spacetimes @xcite for which all polynomial scalar invariants constructed from the riemann tensor and its derivatives vanish and are constant , respectively . moreover , all the relativistic gyratons known so far @xcite , representing the fields of localised spinning sources that propagate with the speed of light , are also specific members of the kundt class .
vacuum and conformally flat pure radiation kundt spacetimes provide an exceptional case for the invariant classification of exact solutions @xcite , and all type d pure radiation solutions are also known @xcite .
all vacuum kundt solutions of type d were found and classified a long time ago @xcite and generalized to electrovacuum and any value of the cosmological constant @xcite .
these contain a subfamily of direct - product spacetimes , namely the bertotti
robinson , ( anti-)nariai and plebaski hacyan spacetimes of type o and d ( see chapter 7 of @xcite , @xcite and @xcite for higher - dimensional generalizations ) representing , e.g. , extremal limits and near - horizon geometries . with minkowski and ( anti-)de sitter spaces they form the natural backgrounds for non - expanding gravitational waves of types n and ii @xcite . in our studies here
we consider the fully general class of kundt spacetimes of an arbitrary dimension @xmath15 ( results for the standard general relativity are obtained by simply setting @xmath14 ) . taking the spacetime dimension as a free parameter @xmath0
, we can investigate whether the extension of the kundt family to @xmath16 exhibits some qualitatively different features and unexpected properties .
our paper is thus also a contribution to the contemporary research analyzing various aspects of einstein s gravity extended to higher dimensions .
explicit kundt solutions help us illustrate specific physical properties and general mathematical features of such theories
. specifically , we systematically investigate the complete @xmath15 kundt class of solutions using geodesic deviation and discuss the corresponding effects on free test particles .
in section [ geodevkundt ] we summarize the equation of geodesic deviation , introduce invariant amplitudes of the gravitational field , and we discuss them for the fully general kundt family of geometries . in section [ specfrmes ]
we derive expressions for these amplitudes , and in section [ claasifkundt ] we evaluate them explicitly for all algebraically special kundt spacetimes for which the optically privileged congruence is generated by a multiple wand .
the main results are presented in section [ subclasses ] where we discuss the specific structure of relative motion of test particles for all possible algebraic types and subtypes of such kundt geometries , see subsections [ typo][typii ] . in the final section [ example ] we present a particular example , namely an interesting class of type ii and n non - expanding gravitational waves on d and o backgrounds of any dimension .
relative motion of nearby free test particles ( without charge and spin ) is described by the equation of geodesic deviation @xcite . it has long been used as an important tool for studies of four - dimensional general relativity , in particular to analyze fields representing gravitational waves and black hole spacetimes ( see @xcite for more details and references ) . in our recent work
@xcite generalizing @xcite we demonstrated that the equation of geodesic deviation in _ any _ @xmath0-dimensional spacetime can be expressed in the invariant form ( using einstein s field equations @xmath17 ) @xmath18\ ! , \label{invariant form of eq of geodesic deviationl}\\ & & \hspace{-13.8mm}\ddot{z}^{(i ) } = \frac{2\lambda}{(d-2)(d-1)}\,z^{(i)}-\psi_{2t^{(ij)}}\,z^{(j ) } + \frac{1}{\sqrt{2}}\,(\,\psi_{1t^i}-\psi_{3t^i})\,z^{(1)}\nonumber \\ & & \hspace{+55.8 mm } -\frac{1}{2}\,(\,\psi_{0^{ij}}+\psi_{4^{ij}})\,z^{(j)}\nonumber\\ & & \hspace{-2.5 mm } + \ , \frac{8\pi}{d-2}\!\left[t_{(i)(1)}\,z^{(1)}+t_{(i)(j)}\,z^{(j)}-\big(t_{(0)(0)}+\frac{2\,t}{d-1}\big)z^{(i)}\right]\ ! , \label{invariant form of eq of geodesic deviationt}\end{aligned}\ ] ] @xmath19 . here
@xmath20 are spatial components @xmath21 of the separation vector @xmath22 between the two test particles in a natural interpretation orthonormal frame @xmath23 , @xmath24 , where @xmath25 is the velocity vector of the fiducial test particle , and @xmath26 are the corresponding relative physical accelerations @xmath27 .
the coefficients @xmath28 denote frame components of the energy - momentum tensor ( @xmath29 is its trace ) , and the scalars @xmath30 with indices @xmath31 , are components of the weyl tensor with respect to the null frame @xmath32 associated with @xmath23 via relations @xmath33 , @xmath34 , @xmath35 , see figure [ figure1 ] .
evolution of the separation vector @xmath36 that connects particles moving along geodesics @xmath37 , @xmath38 is given by the equation of geodesic deviation ( [ invariant form of eq of geodesic deviationl ] ) and ( [ invariant form of eq of geodesic deviationt ] ) .
its components are expressed in the orthonormal frame @xmath23 with @xmath25 .
the associated null frame @xmath32 is also indicated.,scaledwidth=61.0% ] the weyl tensor components ( [ defpsicoef ] ) are listed by their boost weight and directly generalize the standard newman
penrose complex scalars @xmath39 known from the @xmath14 case @xcite . in equations ( [ invariant form of eq of geodesic deviationl ] ) , ( [ invariant form of eq of geodesic deviationt ] ) , only the `` electric part '' of the weyl tensor represented by the scalars in the left column of ( [ defpsicoef ] ) occurs . all these scalars respect the standard symmetries of the weyl tensor , for example @xmath40 } } \ , .\label{symmetries}\ ] ] moreover , there are relations between the left and right columns of ( [ defpsicoef ] ) , namely @xmath41}}={\textstyle\frac{1}{2}}\psi_{2^{ij } } \ , , \quad \psi_{2t^{(ij)}}={\textstyle\frac{1}{2}}\psi_{2^{ikj}}{}^{_k}\,.\label{constraints}\end{aligned}\ ] ] finally , let us remark that our notation which uses the symbols @xmath42 in any dimension is related to the notations employed elsewhere , namely in @xcite , @xcite , and @xcite .
identifications for the components present in the invariant form of the equation of geodesic deviation are summarized in table [ notationcomp ] ( more details can be found in @xcite ) .
.[notationcomp]different equivalent notations used in the literature for those weyl scalars that occur in the equation of geodesic deviation ( [ invariant form of eq of geodesic deviationl ] ) , ( [ invariant form of eq of geodesic deviationt ] ) .
[ cols="^,^,^,^,^,^,<",options="header " , ] relative motion of free test particles in these spacetimes is described by equations ( [ kundt geodesic deviationa ] ) , ( [ kundt geodesic deviationb ] ) where the scalars ( [ psinatn ] ) take the form @xmath43\big\ } \ , . \nonumber\label{psiexample}\end{aligned}\ ] ] notice that for the subtype ii(ab)@xmath44ii(abd ) , this simplifies considerably to @xmath45 when , in addition , @xmath46 , this becomes type ii(abcd)@xmath44n .
if , and only if , @xmath47 , the spacetimes are of type d or type o. when @xmath48 , these belong to the important family of _ direct - product spacetimes _ , see section 11 of @xcite , for which the first term of the metric ( [ wave on bert ] ) is a @xmath49-dimensional riemannian space with metric @xmath50 , while the second part is a 2-dimensional lorentzian spacetime of constant gaussian curvature @xmath51 . in general , @xmath50 need not be of constant curvature , but for the subtype d(a ) , @xmath51 is uniquely related to the _ constant _ ricci scalar @xmath52 of the transverse @xmath53-dimensional space .
such metrics represent natural higher - dimensional generalizations of the ( anti-)nariai , plebaski
hacyan , bertotti robinson and minkowski spacetimes of types d or o , see @xcite . for a non - trivial @xmath54 ,
the spacetimes ( [ wave on bert ] ) are of type ii or of type n. these can be naturally interpreted as the class of _ exact kundt gravitational waves _ with the profile @xmath55 propagating in various direct - product background universes of algebraic types d or o mentioned above ( and listed in table 6 of @xcite ; see also @xcite ) .
the class of metrics ( [ wave on bert ] ) clearly contains _ pp - waves _ ( without gyratonic sources ) propagating in flat space when @xmath56 .
these are of type n if , and only if , @xmath57 ( in which case they belong to the class of vsi spacetimes , see @xcite ) .
finally , let us observe that in the classic @xmath14 case , the scalars ( [ psiexample ] ) read @xmath58 the corresponding kundt geometries ( [ wave on bert ] ) are thus generally of type ii@xmath44ii(bcd ) .
they are of type n@xmath44ii(abcd ) if , and only if , @xmath59 with the only non - vanishing weyl scalar @xmath60 .
in fact , this is the subfamily @xmath61 of spacetimes discussed in @xcite and in sections 18.618.7 of @xcite ( with the identification @xmath62 , @xmath63 and @xmath64 ) which was interpreted as exact kundt gravitational waves of type ii propagating on type d backgrounds , and type n waves propagating on conformally flat type o backgrounds , respectively .
these background universes with the geometry of a direct product of two constant - curvature 2-spaces involve the standard minkowski , bertotti
robinson , ( anti-)nariai and plebaski hacyan spacetimes , cf .
we systematically analyzed the general class of kundt geometries in an arbitrary dimension @xmath15 using the geodesic deviation in einstein s theory .
we explicitly determined the specific motion of free test particles for all possible algebraically special spacetimes , including the corresponding subtypes , and demonstrated that the invariant quantities determining these ( sub)types are measurable by detectors via characteristic relative accelerations .
for example , the dimension of the spacetime can be measured directly by newtonian - type tidal deformations of the algebraic subtype d(b ) .
the purely transverse type n effects represent exact gravitational waves with @xmath2 polarizations , which exhibit new and peculiar observable effects in higher dimensions @xmath16 .
we gave an example of such geometric and physical interpretation of the kundt family by analyzing the class of type n or ii gravitational waves propagating on backgrounds of type o or d.
this work was supported by the grant gar p203/12/0118 .
10 bik j and podolsk j 1999 gravitational waves in vacuum spacetimes with cosmological constant . i. classification and geometrical properties of non - twisting type n solutions _ j. math .
_ * 40 * 4495505 bik j and podolsk j 1999 gravitational waves in vacuum spacetimes with cosmological constant .
ii . deviation of geodesics and interpretation of non - twisting type n solutions _ j. math .
_ * 40 * 450617 synge j l 1934 on the deviation of geodesics and null - geodesics , particularly in relation to the properties of spaces of constant curvature and indefinite line - element _ ann .
_ * 35 * 70513 ; reprinted in _ gen .
* 41 * ( 2009 ) 120514 | spacetimes kundt class defined by geometric property admit geodesic null congruence non - expanding non - twisting shear - free four - dimensional relativity vacuum radiation spacetimes type n iii o introduced studied 50 years ago by wolfgang kundt kundt class wider admits cosmological constant electromagnetic field matter fields supersymmetry solutions algebraic types extended to @xmath0 dimensions kundt spacetimes written as @xmath4 coordinate @xmath5 affine parameter along null congruence @xmath6 vanishing expansion @xmath8 are @xmath9 spatial coordinates in transverse riemannian space spatial part @xmath10 metric independent of @xmath5 other metric components @xmath11 @xmath12 functions of coordinates @xmath13 kundt class spacetimes important exact solutions in einstein s general relativity theory see chapter 31 chapter 18 contains subclasses four dimensions with interesting mathematical physical properties best known pp - waves admit covariantly constant null vector field vsi csi spacetimes @xcite polynomial scalar invariants from riemann tensor derivatives vanish constant relativistic gyratons @xcite spinning sources speed light members kundt class vacuum flat pure radiation kundt spacetimes exceptional invariant classification solutions @xcite type d pure radiation solutions known @xcite vacuum kundt solutions d found classified @xcite generalized to electrovacuum cosmological constant @xcite subfamily direct - product spacetimes bertotti robinson anti-)nariai plebaski hacyan spacetimes type o d chapter 7 @xcite extremal limits near - horizon geometries minkowski anti-)de sitter spaces backgrounds for non - expanding gravitational waves types n ii @xcite general class kundt spacetimes arbitrary dimension @xmath15 results relativity setting @xmath14 spacetime dimension parameter @xmath0 extension kundt family to @xmath16 different features unexpected properties paper contribution to research einstein s gravity extended higher dimensions explicit kundt solutions illustrate physical properties mathematical features theories investigate complete @xmath15 kundt class solutions geodesic deviation effects on free test particlessection geodevkundt summarize equation geodesic deviation introduce invariant amplitudes gravitational field discuss general kundt family geometries section specfrmes derive expressions amplitudes claasifkundt evaluate special kundt spacetimes optically privileged congruence generated multiple wand main results section subclasses discuss structure relative motion test particles algebraic types subtypes kundt geometries subsections final section example present example class type ii n non - expanding gravitational waves d o backgrounds relative motion nearby free test particles without described equation geodesic deviation @xcite important tool studies four - dimensional general relativity analyze fields gravitational waves black hole spacetimes see @xcite detailsrecent work demonstrated equation geodesic deviation-dimensional spacetime expressed invariant form einstein s field equations @xmath17{invariant form geodesic deviationl{-13.8mm}{(i\frac{2\lambda}{(d-2)(d-1)}{(i)}-\psi_{2t^{(ij(j\frac{1}{\sqrt{2}}\psi_{1t^i}-\psi_{3t^i}{(1)\hspace{+55.8 mm }\frac{1}{2}\psi_{0^{ij}}+\psi_{4^{ij}}{(j)}\hspace{-2.5 mm }\frac{8\pi}{d-2}{(i)(1){(1)}+t_{(i)(j)}(j)}-(t_{(0)(0)}{2{d-1}^{(i)}{invariant form eq geodesic deviationt{aligned @xmath19 @xmath20 spatial components separation vector two test particles orthonormal frame @xmath23 velocity vector test particle relative physical accelerationscoefficients @xmath28 denote frame components energy - momentum tensor @xmath29 trace scalars @xmath30 indices @xmath31 components weyl tensor null frame @xmath32 associated with @xmath23 via relations @xmath33 @xmath35 figure evolution separation vector @xmath36 connects particles geodesics @xmath37 @xmath38 given by equation of geodesic deviation components expressed in orthonormal frame @xmath23 with @xmath25 associated null frame @xmath32 indicated=61.0% weyl tensor components defpsicoef ] listed by boost weight generalize standard newman penrose complex scalars @xmath39 from @xmath14 case equations geodesic deviationl electric part weyl tensor represented by scalars left column [ defpsicoef ] occurs scalars respect standard symmetries weyl tensor example @xmath40\label{symmetries} relations between left right columns of defpsicoef @xmath41}}=\textstyle\frac{1}{2}}\psi_{2^{ij } \psi_{2t^{(ij)}}={\textstyle\frac{1}{2}}\psi_{2^{ikj}}{^{_k}\label{constraints}\end{aligned} notation symbols @xmath42 related to notations @xcite , @xcite @xcite . identifications components in invariant form equation of geodesic deviation summarized in table [ notationcomp ] details @xcite equivalent notations for weyl scalars in equation of geodesic deviation invariant deviationl relative motion of free test particles spacetimes described by equations [ kundt geodesic deviationa kundt geodesic deviationb ] scalars psinatn take form @xmath43\big\\nonumber\label{psiexample}\end{aligned} subtype ii(ab)@xmath44ii(abd ) simplifies to @xmath45 @xmath46 becomes type ii(abcd)@xmath44n@xmath47 spacetimes type d or o @xmath48 belong direct - product spacetimes see section 11 @xcite first term metric wave on bert @xmath49-dimensional riemannian space @xmath50 second part 2-dimensional lorentzian spacetime constant gaussian curvature @xmath51 @xmath50 constant curvature subtype d(a ) @xmath51 related to constant ricci scalar @xmath52 transverse @xmath53-dimensional space metrics represent higher - dimensional generalizations anti-)nariai plebaski hacyan bertotti robinson minkowski spacetimes types d or o non - trivial @xmath54 spacetimes bert type ii or n class exact kundt gravitational waves profile @xmath55 propagating direct - product background universes algebraic types d or o table 6 @xcite class metrics wave on bert contains pp - waves without gyratonic sources propagating flat space @xmath56 type n if @xmath57 belong class vsi spacetimes @xcite classic @xmath14 case scalars read @xmath58 kundt geometries bert type ii@xmath44ii(bcd )type n@xmath44ii(abcd @xmath59 non - vanishing weyl scalar @xmath60 subfamily @xmath61 spacetimes @xcite sections 18.618.7 @xmath62 @xmath63 @xmath64 interpreted exact kundt gravitational waves type ii d backgrounds type n o backgrounds background universes constant - curvature 2-spaces involve standard minkowski bertotti robinson anti-)nariai plebaski hacyan spacetimes analyzed kundt geometries arbitrary dimension @xmath15 geodesic deviation einstein s theory determined motion free test particles algebraically special spacetimes subtypes invariant quantities measurable relative accelerations dimension spacetime measured newtonian - type tidal deformations subtype d(b ) transverse type n effects gravitational waves @xmath2 polarizations effects higher dimensions @xmath16 geometric physical interpretation kundt family analyzing type n ii gravitational waves backgrounds type o d supported grant gar p203/12/0118 bik j podolsk j 1999 gravitational waves vacuum spacetimes cosmological constant classification geometrical properties non - twisting type n solutions40 * 4495505 bik podolsk 1999 gravitational waves vacuum cosmological constant deviation geodesics non - twisting solutions math 40 * 450617 synge 1934 deviation geodesics null - geodesics properties spaces constant curvature indefinite line element 35 * reprinted 41 2009 120514 | 0.561463 | 3,322 | 1,757 | , Saving $0.1 in GPT-4. |
177,804 | effective nuclear densities probed by kaon- and anti - kaon - nucleus systems are studied theoretically both for bound and low energy scattering states . as for the anti - kaon bound states
, we investigate kaonic atoms .
we find that the effective density depends on the atomic states significantly and we have the possibility to obtain the anti - kaon properties at various nuclear densities by observing the several kaonic atom states .
we also find the energy dependence of the probed density by kaon and anti - kaon scattering states .
we find that the study of the effective nuclear density will help to find the proper systems to investigate the meson properties at various nuclear densities . | meson - nuclear systems are very important and useful objects to extract the meson properties at finite density , which may have close connections to the chiral symmetry breaking and its partial restoration in the nucleus @xcite .
the effective nuclear densities probed by pionic atoms are known to be around @xmath0 for almost every pionic state @xcite and the change of the chiral order parameter @xmath1 in nuclei is only determined at this specific density @xcite .
thus , we are very interested in the density probed by various meson - nucleus systems to find out the proper mesonic systems to explore the meson properties at various nuclear densities , which will be important to investigate the aspects of qcd symmetries at various nuclear densities beyond the linear density approximation @xcite .
we calculate the effective density of bound states for anti - kaon ( kaonic atoms ) , and scattering states of anti - kaon and kaon with nucleus . as for the kaonic atoms , we already do the similar analysis as reported in ref .
@xcite , where we define the overlapping probability ( @xmath2 ) of the kaonic atom density ( @xmath3 ) with the nucleon density distribution ( @xmath4 ) to obtain the effective density , as @xmath5 since the integration of the @xmath2 function gives the shift of anti - kaon binding energy as @xmath6 in the first order perturbation theory for the change of the potential term depending on the density linearly , we define the nucleon density at the peak position of @xmath2 as the effective nuclear density for bound states , which is expected to be most sensitive part of @xmath7 to the meson binding energy . for scattering states ,
we newly determine the different overlapping probability ( @xmath8 ) as , @xmath9 in this case , we define the @xmath8 function so as to have the peak at the density which is most sensitive to a differential cross section , namely the change of the differential cross section is related to the @xmath8 as , @xmath10 instead of the binding energy for the bound states . here
, we consider the scattering with @xmath11c at @xmath12 . the meson wavefunction @xmath13 in eqs .
( 1 ) and ( 2 ) are obatined by solving the klein - gordon equation with meson - nucleus optical potential . in this article , we use two kinds of the optical potentials for anti - kaon , which are the chiral unitary potential @xcite and the phenomenological potential @xcite . for @xmath14 scattering
, we use the potential given in ref .
the all potentials used in this article are shown in fig .
[ fig1 ] .
c obtained by ( a ) chiral unitary approach @xcite and ( b ) phenomenological fit @xcite are shown as a function of the radial coordinate @xmath15 . the kaon-@xmath11c potential ( c ) obtained in ref .
@xcite is also shown .
the upper and lower panels show the real and imaginary part of the potential , respectively .
the kaon - nucleus potential in ref .
@xcite does nt have imaginary part .
some of the potential have the energy dependence as shown in the figures . ]
as for the kaonic atoms , we show the calculated anti - kaon densities ( @xmath3 ) and the nucelon density distributions ( @xmath4 ) , and the overlapping probabilities ( @xmath2 ) in fig . [ fig2 ] @xcite
. the nuclear density at the peak position of @xmath2 shown in the lower panels corresponds to the effective nuclear density .
we can see that the effective densities probed by atomic kaons depend on the atomic orbits significantly . comparing the results of two potentials , we find the differences of the structure of the overlapping probabilities @xmath2 .
these differences are related to the properties of the deeply bound kaonic nuclear state for both potential cases , since the kaonic nuclear states exist for both potentials . in the phenomenological potential ,
the kaonic nuclear states are deeper than those in the chiral unitary potential .
the wave functions of the atomic and nuclear states should be orthogonal .
owing to the orthogonality the wave functions of atomic states have nodes in the region of the nuclear size @xcite .
the effective densities obtained from the overlapping probabilities will have information on the difference of the potentials , which might be extracted from experimental data .
c calculated with the ( left ) chiral unitary and ( right ) phenomenological optical potentials . ] for scattering states , we show the calculated results of the energy dependence of the effective density in figs .
[ fig3 ] ( anti - kaon ) and [ fig4 ] ( kaon ) . from these results ,
we find the effective density for anti - kaon dicreases for higher anti - kaon energy .
it would be natural that the effective density for higher energy could become smaller than that for lower energy since the important partial waves become larger for higher energy in general , and the wave function with larger angular momentum will observe more outer part of the nucleus due to the centrifugal barrier .
this is the case for the anti - kaon scattering . for the anti - kaon , the increase of the effective density for each partial wave is moderate , because the anti - kaon exists inside the nucleus with small probability due to the large absorption .
in contrast , the effective density for kaon increases for higher kaon energy . this is because the potential of kaon in ref .
@xcite is weak repulsive and does not have imaginary part .
it means that the kaon with higher energy can penetrate into the nucleus without any absorption .
this makes the effective density of each partial wave increase rapidly .
this is the reason that the kaon system has different effective densities from those of the anti - kaon system .
c with chiral unitary approach ( left ) and phenomenological fit ( right ) as plotted as the functions of the anti - kaon kinetic energy [ mev ] .
the contributions from a first few partial waves in eq .
( 2 ) are also shown in the figure . ] except for the kaon scattering case . ]
we have calculated the effective nuclear density probed by anti - kaon and kaon . the effective nuclear density is defined from the overlapping probability of the meson density and the nucleon density distribution . as for the bound states , we consider the kaonic atoms .
we find the effective density probed by anti - kaon in atomic orbit depends on the quantum numbers of the atomic states .
for the scattering states , we have investigated the energy dependence of the effective density . from this study
, we find the effective density of anti - kaon decreases for higher anti - kaon energy due to the centrifugal barrier and the strong absorption inside the nucleus .
we also find the kaon for higher energy could probe higher nuclear density , which is the opposite tendency to the anti - kaon system .
r. seki and k. masutani , phys .
c * 27 * ( 1983 ) 2799 .
n. ikeno , r. kimura , j. yamagata - sekihara , h. nagahiro , d. jido , k. itahashi , l. -s .
geng and s. hirenzaki , prog .
* 126 * ( 2011 ) 483 [ arxiv:1107.5918 [ nucl - th ] ] .
d. jido and s. goda , arxiv:1108.6144 [ nucl - th ] .
j. yamagata and s. hirenzaki , eur .
j. a * 31 * , 255 ( 2007 ) [ nucl - th/0612036 ] .
a. ramos and e. oset , nucl .
phys . a * 671 * , 481 ( 2000 ) [ nucl - th/9906016 ] .
c. j. batty , e. friedman and a. gal , phys .
rept . * 287 * , 385 ( 1997 ) .
j. mares , e. friedman and a. gal , phys .
b * 606 * , 295 ( 2005 ) , nucl . phys .
a * 770 * , 84 ( 2006 ) .
e. oset and a. ramos , nucl .
a * 679 * , 616 ( 2001 ) [ nucl - th/0005046 ] .
j. yamagata , h. nagahiro , y. okumura and s. hirenzaki , prog .
phys . * 114 * , 301 ( 2005 ) [ erratum - ibid . * 114 * , 905 ( 2005 ) ] [ nucl - th/0503039 ] . | meson - nuclear systems important meson properties at finite density chiral symmetry breaking restoration in nucleus effective nuclear densities by pionic atoms around @xmath0 for every pionic state change chiral order parameter @xmath1 in nuclei determined at this specific density interested in density meson - nucleus systems mesonic systems explore meson properties at various nuclear densities investigate qcd symmetries densities beyond linear density approximation calculate effective density of bound states for anti - kaon kaonic atoms scattering states - kaon kaon with nucleus kaonic atoms similar analysis define overlapping probability kaonic atom density with nucleon density distribution @xmath4 effective density integration function shift anti - kaon binding energy perturbation theory potential density define nucleon density at peak position of @xmath2 as effective nuclear density for bound states most sensitive part @xmath7 to meson binding energyscattering states determine overlapping probability @xmath8 as @xmath9 define @xmath8 function peak at density sensitive to differential cross section change differential cross section related to @xmath8 @xmath10 instead binding energy bound states consider scattering with @xmath11c at @xmath12 meson wavefunction @xmath13 obatined klein - gordon equation with meson - nucleus optical potential use optical potentials anti - kaon chiral unitary potential @xcite phenomenological potential @xcite @xmath14 scattering use potential ref . potentials shown in fig . chiral unitary approach @xcite phenomenological fit @xcite function of radial coordinate @xmath15 kaon-@xmath11c potential c @xcite shown upper lower panels show real imaginary part potential kaon - nucleus potential @xcite imaginary part energy dependence figures kaonic atoms show calculated anti - kaon densities @xmath3 nucelon density distributions @xmath4 overlapping probabilities @xmath2 ) in fig . nuclear density at peak position @xmath2 corresponds effective nuclear densityeffective densities by atomic kaons depend on atomic orbits comparing results of two potentials differences of structure overlapping probabilities differences related to properties deeply bound kaonic nuclear state for both in phenomenological kaonic nuclear states deeper than chiral unitary potential wave functions of atomic and nuclear states should be orthogonal have nodes in nuclear size @xcite effective densities from overlapping probabilities information on difference potentials extracted from experimental data calculated with chiral unitary and phenomenological optical potentials for scattering states calculated results of energy dependence of effective density in figs effective density for anti - kaon dicreases for higher anti - kaon energy effective density for higher energy could smaller partial waves larger for wave function with larger momentum more outer part of nucleus due to centrifugal barrier for anti - kaon scattering - increase effective density for each partial wave moderate exists inside nucleus small probability large absorption effective density for kaon increases for higher energy potential of kaon in @xcite is weak repulsive imaginary partkaon higher energy nucleus without absorption effective density partial wave kaon system different effective densities from anti - kaon system chiral unitary approach left phenomenological fit right functions anti - kaon kinetic energy contributions from partial waves shown in figure kaon scattering case calculated effective nuclear density by anti - kaon kaon defined from overlapping probability meson density nucleon density distribution bound states kaonic atoms effective density anti - kaon depends on quantum numbers atomic states scattering states investigated energy dependence effective density effective density anti - kaon decreases for higher - kaon energy due to centrifugal barrier strong absorption inside nucleus kaon higher energy probe higher nuclear density opposite tendency to anti - kaon system r. seki k. masutani . 27 ( 1983 ) 2799 ikeno kimura yamagata - sekihara nagahiro jido itahashi geng hirenzaki prog . 126 * ( 2011 ) 483 arxiv:1107.5918 nucl - jido goda arxiv:1108.6144 - yamagata hirenzaki31 255 ( 2007 ) th/0612036 a. ramos e. oset phys 671 481 ( 2000 ) th/9906016 j. batty e. friedman a. gal 287 385 ( 1997 ) j. mares. friedman a. gal 606 295 ( 2005 770 84 ( 2006 ) e. oset a. ramos 679 616 ( 2001 ) th/0005046 j. yamagata h. nagahiro y. okumura s. hirenzaki . 114 301 ( 2005 ) 114 905 ( 2005 ) nucl th/0503039 | 0.31982 | 1,952 | 965 | , Saving $0.1 in GPT-4. |
177,811 | the present note refers to a result proposed in @xcite , and shows that the theorem therein is not correct .
we explain that a proof of that theorem can not be given , as the statement is not correct , and we underline a mistake occurring in their proof .
since this note is supplementary to @xcite , the reader should consult this paper for further explanations of the matter and the symbols used .
* keywords * : homoclinic orbit , heteroclinic orbit , chaos . *
chinese library classification * * 2010 mathematics subject classification * | in this note , we show that the main result proposed in @xcite , i.e. a sufficient condition for excluding the presence of homoclinic and heteroclinic orbits , is not correct .
hence this can not lead to their conjecture of a fourth kind of chaos in 3d polynomial ode systems characterized by the non - existence of homoclinic and heteroclinic orbits .
moreover , we remark that the conjecture can not be correct , as explained below in detail .
the main result of @xcite is stated in their theorem 1 , which leads to exclude the existence of a bounded trajectory for @xmath0 in any dynamical system characterized by a vector field with at least one lower bounded component , and from their proof it follows that this occurs independently of the existence of one or more equilibria in the system .
we notice that the proof they give also implies the _ non existence of any closed orbit , i.e. limit cycle_. moreover , @xcite gives an example , in eq .
( 3 ) , satisfying the assumption of the theorem 1 and showing a chaotic attractor illustrated in fig .
1 . from this finding , they conjecture the existence of a new type of chaos . however , given a system with a chaotic attractor ( and let us assume that this is the case shown in their example , indeed form fig .
[ fig1 ] of this note it can be assumed that the first return map on a suitable two - dimensional surface as qualitatively shown by the red line , leads to a two - dimensional map in chaotic regime ) then we have , by any definition of chaotic system ( see , e.g. , @xcite , @xcite , @xcite , @xcite and @xcite ) , _ the existence of infinitely many unstable limit cycles which densely cover the observed chaotic set . moreover , infinitely many homoclinic and heteroclinic orbits exist connecting these unstable limit cycles , which also are dense in the chaotic attractor_. it follows that it is not possible to identify a chaotic system characterized by the non - existence of homoclinic and heteroclinic orbits .
-plane of attractor obtained from system ( 3 ) in @xcite for @xmath1 , @xmath2 and @xmath3 .
the red line indicates a suitable plane for the return map . ]
the paper @xcite suggests that the presence of chaos is subordinated to the existence of either homoclinic orbits _ of equilibria _ or heteroclinic orbits _ of equilibria_. which is not correct .
indeed , the target of identifying a chaotic system characterized by the non - existence of homoclinic and heteroclinic orbits _ to one or more equilibria _ may be correct .
however , in our opinion this is not interesting , as it is well known that chaotic attractors may exist also when neither homoclinic nor heteroclinic orbits _ to equilibria _ are present , as it is the case in the classical lorenz system , see , e.g. @xcite .
we should emphasize that the existence of a homoclinic orbit of an equilibrium ( or heteroclinic connections between two equilibria ) , under other suitable assumptions , is relevant from a theoretical point of view , as it allows to rigorously prove the existence of chaos and also its persistence under perturbations when the homoclinic ( heteroclinic ) orbit no longer exists . in the following
, we underline the presence of an inaccuracy in the proof of theorem 1 in @xcite .
regarding the proof of theorem 1 in @xcite , an incorrect part is related to the unboundedness of the trajectories .
consider a vector field @xmath4 belonging to class @xmath5 , @xmath6 the state variable of the system , and @xmath7 the time . assuming the existence of an @xmath8 , such that for at least one @xmath9 we have that @xmath10 , @xmath11 , then ( as pointed out by the authors in eq .
( 2 ) of their paper ) we have to consider the inequality@xmath12*for * @xmath13 which implies that , given a homoclinic or heteroclinic orbit @xmath14 , we have @xmath15 which is compatible with the existence of the orbits itself .
then the authors state that the same result in holds also * for * @xmath16 getting divergence for any orbit for @xmath17 .
however this is not correct , as for @xmath0 we have ] by a simple integration from @xmath18 to @xmath19 , we have @xmath20 assuming @xmath21 @xmath22 , by basic properties of definite integrals we have that for @xmath23 @xmath24 from which inequality follows , and for @xmath0 @xmath25 from which we obtain inequality .]@xmath26 in place of .
inequality implies that @xmath27 , which does not exclude the existence of homoclinic ( or heteroclinic ) orbits .
in this note we provide some arguments showing that the statement of theorem 1 in @xcite is not correct .
moreover , we show the presence of a mistake related to the backward integration in the proof of the same theorem .
the authors are grateful to g. i. bischi and l. cerboni baiardi for helpful comments on this note .
the usual caveats apply . | note show main result in @xcite sufficient condition for excluding homoclinic and heteroclinic orbits not correct not lead to conjecture of fourth kind of chaos in 3d polynomial systems non - existence homoclinic heteroclinic orbits conjecture not be correct explained main result of @xcite in theorem 1 bounded trajectory for @xmath0 in any dynamical system vector field with one lower bounded component proof occurs independently of equilibria system proof implies non existence of closed orbit limit cycle @xcite gives example in eq . ( 3 ) satisfying assumption theorem 1 showing chaotic attractor illustrated in fig . 1 they conjecture existence new type of chaos given system with chaotic attractor first return map on two - dimensional surface leads to two - dimensional map in chaotic regime definition of chaotic system @xcite existence of infinitely many unstable limit cycles cover observed chaotic set infinitely many homoclinic and heteroclinic orbits exist connecting these unstable limit cycles dense in chaotic attractornot possible to identify chaotic system non - existence homoclinic heteroclinic orbits -plane attractor obtained from system ( 3 ) in @xcite for @xmath1 @xmath2 @xmath3 red line indicates suitable plane for return map paper @xcite suggests presence chaos subordinated to homoclinic or heteroclinic orbits not correct target identifying chaotic system non - existence homoclinic heteroclinic orbits to equilibria may be correct not chaotic attractors may exist when neither homoclinic nor heteroclinic orbits present in classical lorenz system existence of homoclinic orbit of equilibrium or heteroclinic connections between two equilibria relevant theoretical allows prove existence of chaos persistence under perturbations when homoclinic orbit no longer exists inaccuracy in proof of theorem 1 in @xcite incorrect part related to unboundedness of trajectories consider vector field @xmath4 to class @xmath5 , @xmath6 state variable system @xmath7 time assuming existence of @xmath8 for @xmath9 @xmath10 @xmath11 2 paper consider inequality@xmath12 @xmath13 implies homoclinic or heteroclinic orbit @xmath14 @xmath15 compatible with existence orbits authors state same result holds @xmath16 divergence orbit @xmath17 not correct @xmath0 simple integration from @xmath18 to @xmath19 @xmath20 assuming @xmath21 @xmath22 basic properties definite integrals @xmath23 @xmath24 inequality follows @xmath0 @xmath25 obtain inequality@xmath26 inequality implies @xmath27 exclude homoclinic or heteroclinic orbits provide arguments statement theorem 1 @xcite not correct show mistake backward integration proof same theorem authors grateful to g. i. bischi l. cerboni baiardi comments usual caveats apply | 0.190417 | 1,167 | 613 | , Saving $0.0 in GPT-4. |
177,812 | we examine the extraction of moments of parton distributions from lattice data , focusing in particular on the chiral extrapolation as a function of the quark mass .
inclusion of the correct chiral behavior of the spin - averaged isovector distribution resolves a long - standing discrepancy between the lattice moments and experiment .
we extract the @xmath0-dependence of the valence @xmath1 distribution from the lowest few lattice moments , and discuss the implications for the quark mass dependence of meson masses lying on the @xmath2 regge trajectory .
the role of chiral symmetry in spin - dependent distributions , and in particular the lattice axial vector charge , @xmath3 , is also highlighted . | parton distribution functions ( pdfs ) parameterize fundamental information on the nonperturbative structure of the nucleon , and the workings of qcd at low energy . over the past two decades considerable
experience has been gained with studies of pdfs within low energy models of the nucleon .
ultimately , however , one would like a more rigorous connection of pdfs with qcd , and presently this can only be provided through the lattice formulation of qcd . because pdfs are defined as light cone correlation functions , it is not possible to calculate them directly on the lattice in euclidean space . instead , one calculates matrix elements of local twist - two operators , which are related through the operator product expansion to moments of the pdfs . for the spin - averaged quark distributions , @xmath4 ,
the moments are defined as : @xmath5 while moments of the helicity distributions , @xmath6 , are given by : @xmath7 a number of lattice calculations of pdf moments have been performed over the last decade , initially in the quenched approximation , and more recently with dynamical quarks .
the results indicate that at the relatively large quark masses at which the calculations were made ( between @xmath8 and 190 mev ) , the unquenched results are indistinguishable from the quenched within the current errors . despite the impressive progress of the lattice calculations , for many years the moments @xmath9 have yielded results which were typically @xmath10 larger than the experimental values , when linearly extrapolated to the physical quark masses .
this discrepancy was recently resolved with the observation@xcite that a linear extrapolation in quark mass omits crucial physics associated with the nucleon s pion cloud , and that inclusion of the nonanalytic dependence on the quark mass is essential if one is to reconcile the lattice data with experiment . in this paper
we report recent progress made in connecting moments of parton distributions calculated on the lattice with experiment , focusing on the extraction of the @xmath1 moments ( which are insensitive to the poorly known disconnected contributions@xcite ) .
we discuss the extraction of the @xmath0 dependence from moments , and the implications for masses of mesons lying on the @xmath2 regge trajectory , and outline the role of chiral symmetry in helicity distributions and the axial vector charge , @xmath3 .
the spontaneous breaking of the chiral su(2)@xmath11su(2)@xmath12 symmetry of qcd generates the nearly massless goldstone bosons ( pions ) , whose importance in hadron structure is well documented . at small pion masses , pdf moments can be systematically expanded in a series in @xmath13 , with the expansion coefficients generally free parameters .
one of the unique consequences of pion loops , however , is the appearance of terms nonanalytic in the quark mass , @xmath14 , which arise from the infrared behavior of the chiral loops , and are therefore model independent .
the leading order ( in @xmath13 ) nonanalytic term in the expansion of the moments of pdfs was shown by thomas et al.@xcite to have the generic behavior @xmath15 .
on the other hand , in the heavy quark limit , in which the valence quark distributions become @xmath16-functions centered at @xmath17 , the moments @xmath18 approach @xmath19 . an extrapolation formula which explicitly satisfies both the heavy quark and chiral limits
can be written:@xcite @xmath20 in which the coefficient of the leading nonanalytic ( lna ) term , @xmath21 , is calculated from chiral perturbation theory.@xcite the mass @xmath22 reflects the scale at which the compton wavelength of the pion becomes comparable to the size of the hadron ( without its pion cloud ) .
previous fits@xcite to the lattice moments of @xmath1 suggest a value @xmath23 mev , which is consistent with values found@xcite in analyses of nucleon static properties .
the @xmath24 term in eq .
( [ xtrap ] ) , @xmath25 , is included in order to provide a linear dependence on @xmath26 , and the mass scale @xmath27 is set to be 5 gev for all @xmath28 . in fig . 1 we show the best @xmath29 fit to the lattice data@xcite for the @xmath30 moment of @xmath1 as a function of @xmath26
clearly , an extrapolation based on eq .
( [ xtrap ] ) provides a much better fit to the lattice data and experiment than a linear fit .
similar results are found@xcite for the @xmath31 and @xmath32 moments .
unfortunately , because all of the lattice data are in a region where the moments show little variation with @xmath26 , it is not possible to determine @xmath22 from the current data , and within the errors both the lattice data and the experimental values can be fitted with @xmath22 ranging from @xmath33 to 700 mev .
data at smaller quark masses are therefore crucial to constrain this parameter and guide an accurate extrapolation .
having established the appropriate way to extrapolate lattice data on pdf moments , one can now ask much information can the existing data provide on the @xmath0 dependence of the pdfs .
the reconstruction of the complete @xmath0 dependence in principle requires infinitely many moments , however , on the lattice , because of operator mixing for operators with spin @xmath34 , all calculations have so far been restricted to @xmath35 .
nevertheless , as shown by detmold et al.,@xcite considerable information on the shape of the valence @xmath36 distribution can already be inferred from just the lowest four moments .
the most efficient way to reconstruct the pdf from a limited number of moments is to determine the parameters of the pdf parameterization by directly fitting to the moments . for the isovector valence distribution accurate pdfs
can be reconstructed from the lowest four moments , using the standard parameterization , @xmath37 where @xmath38 can be related through regge theory to the intercept of the @xmath2 regge trajectory , and @xmath16 is given by perturbative qcd counting rules .
note that some care must be taken when attempting to extract information on valence quark distributions from both the even and odd moments .
while the even @xmath28 moments of @xmath1 correspond to the valence distribution , @xmath36 , the odd @xmath28 moments correspond to the combination @xmath39 .
the difference between these represents the famous violation of the gottfried sum rule .
given sufficiently many moments of @xmath1 , one can reconstruct both the valence @xmath36 and @xmath40 distributions by fitting the even and odd moments separately .
however , at present there exists only a single data point for the even moments , @xmath31 ( the @xmath41 point corresponds to normalization ) , which makes it difficult to obtain accurate information . to minimize the error associated with reconstruction of the @xmath40 distribution
, we subtract the values of the phenomenological moments of @xmath42 from the calculated odd moments .
the correction to the @xmath30 moment is @xmath43 , while for @xmath32 it is a fraction of a percent .
the resulting fits to the moments of the valence @xmath36 distribution are displayed in fig . 2 for both the linear and improved extrapolation , eq .
( [ xtrap ] ) , with the shaded region around the latter corresponding to a 1 @xmath44 variation of the fit parameters from their optimal values .
for clarity we plot @xmath45 times the moments , so that the horizontal line at unity represents the heavy quark limit .
the corresponding distributions @xmath46 are displayed in fig .
once again , the lightly shaded region represents a 1 @xmath44 deviation from the central values , while the darker band illustrates the spread between global pdf fits .
a comparison of the distribution reconstructed using the improved chiral extrapolation with the phenomenological distributions shows reasonably good agreement . on the other hand ,
the linear extrapolation gives a distribution ( scaled by a factor @xmath47 in the figure ) which has a significantly higher peak , centered at @xmath48 , reminiscent of a heavy , constituent quark like distribution .
according to regge theory , the exponent @xmath38 which governs the small-@xmath0 behavior of the distribution @xmath36 is related to the intercept ( @xmath49 ) of the isovector , @xmath50-odd , @xmath2 regge trajectory , and indeed the best fit@xcite to parameterizations of global data gives @xmath51 .
the dependence of @xmath38 on the quark mass obtained from the fits to moments in sec . 3 therefore allows one to predict the @xmath52 dependence of the regge intercept .
in addition to the intercept , one also needs to determine the slope of the trajectory as a function of @xmath52 . in the infinite mass limit ,
orbital excitations of mesons become energetically degenerate with the @xmath53 state . within regge theory ,
this is possible only if @xmath54 as @xmath55 , which is consistent with the valence distribution approaching a @xmath16-function .
one expects , therefore , that the slope should increase as @xmath52 increases from its physical value . using the lattice data for the @xmath2 meson and the values of @xmath38 generated from the fits in fig . 2
, one can then make predictions for the behavior of the masses of the orbital excitations as a function of @xmath52 . in fig .
4 we show the predicted @xmath2 regge trajectory at @xmath56 gev , corresponding to the strange quark mass , compared with the trajectory at the physical light quark mass .
a fit through the central values of @xmath38 and the @xmath2 mass@xcite at @xmath56 gev yields a slope which is larger than that of the trajectory at the physical quark mass , consistent with the expected trend towards the heavy quark limit .
although lattice data for the masses of orbital excitations are scarce , there have been some pioneering calculations of the @xmath57 and @xmath58 meson masses by the ukqcd collaboration,@xcite indicated by the filled boxes in fig . 4 ( we use the fact that the @xmath57 trajectory lies on top of the @xmath2 trajectory ) . comparing with the predictions from the pdf analysis ,
the calculated @xmath58 meson mass lies within the predicted band , albeit within large errors , while the @xmath57 mass lies on the edge of the predicted range .
needless to say , further exploration of the masses of excited mesons within lattice qcd would be very helpful in testing these predictions .
lattice simulations of the spin - averaged quark distributions provide one of the benchmark calculations of hadron structure in lattice qcd . having established confidence in the reliability of the lattice calculations through the correct treatment of the chiral behavior of the moments , we can now turn to other pdfs of the nucleon . of particular importance
are the helicity distributions , @xmath59 , which describe the distribution of the nucleon spin amongst its quark constituents .
while a complete determination of the helicity distributions requires calculation of the singlet distribution , @xmath60 , for which there have only been exploratory lattice calculations,@xcite a more basic challenge remains to understand the axial vector charge , @xmath3 , given by the @xmath41 moment of the isovector @xmath1 distribution .
the results of several lattice calculations@xcite of @xmath3 are compiled in fig .
not included are results from recent simulations using domain wall fermions,@xcite which appear to have strong finite volume dependence
see below . )
when extrapolated linearly in @xmath52 to the physical quark mass , the results are @xmath6115% lower than the experimental value .
simulations with dynamical fermions@xcite ( sesam ) yield results consistent with the earlier unquenched calculations@xcite . on the other hand , the behavior of the moments @xmath62 is known in the chiral@xcite and heavy quark limits , and can be used to constrain the extrapolation form , as in eq .
( [ xtrap ] ) : @xmath63 where @xmath64 , and @xmath65 .
using the same values of @xmath22 and @xmath27 as in fig . 1 , the results of the chiral extrapolation using eq . ( [ xtrap_ga ] )
are shown in fig .
the result for @xmath3 is a downturn with decreasing @xmath52 , making the discrepancy with the empirical value larger .
the reasons for this could be severalfold .
firstly , the chiral behavior in eq .
( [ xtrap_ga ] ) arises only from fluctuations @xmath66 , whereas it is well known that the @xmath67 plays an important role in @xmath3 through the fluctuation @xmath68 . on the other hand , in the chiral limit
the @xmath67 has been shown@xcite to give vanishing chiral contributions at order @xmath15 , suggesting that higher order effects in the chiral expansion are likely to be important .
the inclusion of these effects is currently under investigation .
more importantly , perhaps , to fully incorporate the pion cloud in a lattice simulation a sufficiently large volume must be used . indeed
, there are some indications@xcite of large finite volume effects for @xmath3 , which tend to _ increase _ @xmath3 in comparison with the results from smaller volumes .
clearly , in light of the results on the chiral extrapolation , it is imperative to perform simulations on larger lattices to understand the source of the discrepancy .
in this report we have highlighted the importance of model independent constraints from the chiral and heavy quark limits of qcd in the extrapolation of lattice data on parton distribution moments .
inclusion of the nonanalytic structure associated with the infrared behavior of goldstone boson loops leads to a resolution of a long - standing discrepancy between lattice data on low moments of the spin - averaged @xmath1 distribution and experiment .
the importance of ensuring the correct chiral behavior is further illustrated by comparing the @xmath0 distributions obtained by extrapolating the lattice data using a linear and a chirally symmetric fit . while the latter gives an @xmath0 distribution which is in quite good agreement with the phenomenological fits , the linearly extrapolated data give distributions with the wrong small-@xmath0 behavior , which translates into a much more pronounced peak at @xmath48 , reminiscent of a heavy , constituent quark like distribution .
our analysis suggests an intriguing connection between the small-@xmath0 behavior of the valence distributions and the @xmath52 dependence of meson masses on regge trajectories , which should be tested more thoroughly in future simulations of the excited hadron spectrum .
finally , we have highlighted the need for further study of moments of the helicity distributions , and the axial vector charge of the nucleon in particular , which appears to be underestimated in lattice simulations .
the inclusion of the pion cloud of the nucleon leads to a larger discrepancy at the physical quark mass , indicating that lattice artifacts such as finite volume effects may not yet be under control .
further data on larger lattices , and at smaller quark masses , will be necessary to resolve this issue .
we are grateful to d.b .
leinweber , j.w . negele and d.b .
renner for helpful discussions .
this work was supported by the australian research council , and the u.s .
department of energy contract , under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility ( jefferson lab ) . | parton distribution functions pdfs parameterize information structure nucleon workings qcd at low energy two decades experience gained with studies pdfs low energy models rigorous connection pdfs with qcd through lattice formulation qcd pdfs light cone correlation functions calculate directly on lattice in euclidean space calculates matrix elements of local twist - two operators related to moments pdfs spin - averaged quark distributions @xmath4 moments defined as @xmath5 helicity distributions @xmath6 by : @xmath7 lattice calculations of pdf moments performed decade quenched approximation dynamical quarks results at large quark masses between @xmath8 and 190 mev unquenched results indistinguishable from quenched current errors lattice moments @xmath9 yielded results @xmath10 larger than experimental values linearly extrapolated to physical quark masses discrepancy resolved with linear extrapolation quark mass omits physics nucleon s pion cloud inclusion nonanalytic dependence on quark mass essential lattice data with experimentpaper progress connecting moments parton distributions lattice experiment focusing extraction @xmath1 moments insensitive disconnected contributions discuss extraction @xmath0 dependence moments implications masses mesons @xmath2 regge trajectory outline role chiral symmetry in helicity distributions axial vector charge breaking chiral@xmath12 symmetry generates massless goldstone bosons pions importance hadron structure small pion masses pdf moments expanded @xmath13 expansion coefficients free parameters unique pion loops appearance terms nonanalytic quark mass @xmath14 from infrared behavior chiral loops model independent leading @xmath13 nonanalytic term expansion moments pdfs shown thomas et al. generic behavior @xmath15 heavy quark limit valence quark distributions @xmath16-functions centered @xmath17 moments @xmath18 approach @xmath19 extrapolation formula satisfies heavy quark chiral limits written @xmath20 coefficient leading nonanalytic term @xmath21 calculated from chiral perturbation theory mass @xmath22 reflects scale compton wavelength pion comparable to size hadron without pion cloudfits to lattice moments @xmath1 suggest value @xmath23 mev consistent with values analyses nucleon static properties @xmath24 term in eq @xmath25 included linear dependence on @xmath26 mass scale @xmath27 5 gev for @xmath28 fig . 1 best @xmath29 fit to lattice data for @xmath30 moment @xmath1 function of @xmath26 extrapolation on eq provides better fit data linear similar results for @xmath31 and @xmath32 moments lattice data in little variation with @xmath26 possible to determine @xmath22 from current data lattice data experimental values be fitted with @xmath22 from @xmath33 to 700 mev data at smaller quark masses crucial to constrain parameter accurate extrapolation established lattice data on pdf moments information existing data on @xmath0 dependence reconstruction complete @xmath0 dependence requires many moments operator mixing calculations restricted to @xmath35 information on shape valence @xmath36 distribution inferred from lowest four moments efficient to reconstruct pdf from limited moments determine parameters pdf parameterization by directly fitting to momentsisovector valence distribution reconstructed from lowest four moments standard @xmath37 @xmath38 related to @xmath2 regge trajectory @xmath16 perturbative qcd counting rules care information valence quark distributions from even odd moments even @xmath28 moments @xmath1 correspond valence distribution odd @xmath28 moments @xmath39 difference violation gottfried sum rule moments @xmath1 reconstruct valence @xmath36 @xmath40 distributions fitting even odd moments separately single data point for even moments @xmath31 @xmath41 point corresponds normalization difficult obtain accurate information minimize error subtract values phenomenological moments @xmath42 from calculated odd moments correction to @xmath30 moment @xmath43 @xmath32 fraction of percent resulting fits to moments valence @xmath36 distribution displayed in fig . 2 shaded region 1 @xmath44 variation fit parameters from optimal values plot @xmath45 times moments horizontal line unity represents heavy quark limit corresponding distributions @xmath46 displayed in fig . shaded region 1 @xmath44 deviation from central values darker band spread between global pdf fitscomparison distribution reconstructed improved chiral extrapolation with phenomenological distributions shows agreement linear extrapolation gives distribution scaled by factor @xmath47 higher peak at @xmath48 reminiscent heavy quark distribution regge theory exponent @xmath38 governs small-@xmath0 behavior distribution @xmath36 related to intercept @xmath49 ) isovector @xmath50-odd @xmath2 regge trajectory best fit gives @xmath51 dependence of @xmath38 on quark mass fits moments allows predict @xmath52 dependence regge intercept determine slope of trajectory function of @xmath52 in infinite mass limit orbital excitations degenerate with @xmath53 state regge theory possible if @xmath54 as @xmath55 consistent with valence distribution @xmath16-function slope increase as @xmath52 increases using lattice data for @xmath2 meson values @xmath38 fits predictions for behavior orbital excitations function @xmath52 fig . 4 predicted @xmath2 regge trajectory at @xmath56 gev compared with physical light quark massfit values @xmath38 @xmath2 mass @xmath56 gev yields slope larger than trajectory physical quark mass consistent with trend towards heavy quark limit lattice data orbital excitations scarce pioneering calculations @xmath57 @xmath58 meson masses by ukqcd collaboration filled boxes fig . 4 @xmath57 trajectory on top @xmath2 trajectory calculated @xmath58 meson mass within predicted band large errors @xmath57 mass edge predicted range further exploration masses mesons within lattice qcd helpful predictions lattice simulations spin - averaged quark distributions benchmark hadron structure in lattice qcd confidence reliability calculations turn to other pdfs nucleon importance helicity distributions @xmath59 describe distribution nucleon spin amongst quark constituents complete determination requires calculation singlet distribution @xmath60 exploratory lattice calculations challenge understand axial vector charge @xmath3 given by @xmath41 moment isovector @xmath1 distribution results lattice calculations @xmath3 compiled in fig . not included results recent simulations domain wall fermions strong finite volume dependenceextrapolated linearly in @xmath52 to physical quark mass results @xmath6115% lower than experimental value simulations with dynamical fermions yield results consistent with earlier unquenched calculations behavior of moments @xmath62 known in chiral@xcite heavy quark limits to constrain extrapolation form in eq . @xmath63 @xmath64 @xmath65 same values of @xmath22 @xmath27 results of chiral extrapolation using_ga shown in fig . result for @xmath3 downturn with decreasing @xmath52 discrepancy with empirical value larger chiral behavior in arises from fluctuations @xmath66 @xmath67 important role in @xmath3 chiral limit @xmath67 give vanishing chiral contributions at order @xmath15 higher order effects in chiral expansion likely important inclusion of effects under investigation to incorporate pion cloud in lattice simulation large volume must used indications of large finite volume effects for @xmath3 increase @xmath3 smaller volumes imperative to perform simulations on larger lattices to understand source discrepancyreport highlighted importance model constraints chiral heavy quark limits qcd extrapolation lattice data parton distribution moments inclusion nonanalytic structure infrared behavior goldstone boson loops discrepancy between lattice data low moments spin - averaged @xmath1 distribution experiment importance correct chiral behavior comparing @xmath0 distributions lattice linear chirally symmetric fit latter distribution fits linearly extrapolated data wrong small-@xmath0 behavior pronounced peak at @xmath48 heavy quark distribution analysis suggests connection between small-@xmath0 behavior valence distributions @xmath52 dependence meson masses on regge trajectories future simulations hadron spectrum need for further study helicity distributions axial vector charge nucleon underestimated in lattice simulations inclusion pion cloud nucleon leads larger discrepancy at physical quark mass lattice artifacts finite volume effects under control further data on larger lattices smaller quark masses necessary resolve issue grateful to d.b. leinweber j.w negele d renner discussions work supported by australian research council u.s .department energy contract southeastern universities research association operates thomas jefferson accelerator facility | 0.570868 | 3,422 | 1,755 | , Saving $0.1 in GPT-4. |
177,817 | optical stellar interferometers have demonstrated milli - arcsecond resolution with few apertures spaced hundreds of meters apart . to obtain rich direct images , many apertures will be needed , for a better sampling of the incoming wavefront .
the coherent imaging thus achievable improves the sensitivity with respect to the incoherent combination of successive fringed exposures .
efficient use of highly diluted apertures for coherent imaging can be done with pupil densification , a technique also called `` hypertelescope imaging '' .
although best done with adaptive phasing , concentrating most energy in a dominant interference peak for a rich direct image of a complex source , such imaging is also possible with random phase errors such as caused by turbulent `` seeing '' , using methods such as speckle imaging which uses several short - exposure images to reconstruct the true image .
we have simulated such observations using an aperture which changes through the night , as naturally happens on earth with fixed grounded mirror elements , and find that reconstructed images of star clusters and extended objects are of high quality . as part of the study we also estimated the required photon levels for achieving a good signal to noise ratio using such a technique .
[ firstpage ] instrumentation : high angular resolution , interferometers techniques : interferometric . | in high - resolution optical astronomy with interferometric arrays , producing better images will require more apertures for a denser sampling of the optical wave . as few as three apertures can suffice in principle to reconstruct images through `` optical aperture synthesis''@xcite , using earth rotation or baseline changes to sample the needed fourier components of the object .
but a better sensitivity can be reached with systems using more apertures simultaneously , even if these are smaller for a conserved collecting area .
the gain arises from the coherent combination of light vibrations achievable with many apertures working simultaneously , as opposed to the `` optical aperture synthesis '' approach where interference fringes are recorded with fewer apertures , repeatedly with different baseline settings , and then combined in the computer , i.e. incoherently .
n number of phased beams combined coherently with a simple fizeau arrangement indeed produces a highly constructive interference , in the form of a peak which is n times more intense than the average side - lobes .
instead , successive exposures with subsets of the sub - apertures , even if they are enlarged to conserve the photon flux and moved to improve the spatial frequency coverage , reduce the peak intensity and thus the dynamic range in the convolved image of a complex source @xcite .
fizeau combination however becomes inefficient with highly diluted apertures , since the narrow interference peak appearing in the image of a point source , at the center of the much broader envelope diffracted by the subapertures , contains a small proportion of the energy .
a way of retrieving most energy in the peak , for efficiently observing faint sources with many - aperture interferometers capable of rich direct imaging , has appeared in the form of the hypertelescope or `` densified aperture '' scheme @xcite .
practical designs for large hypertelescopes , with a spherical geometry inspired from the arecibo radio telescope , are under testing for terrestrial versions @xcite and also studied for space versions @xcite .
these future large direct imaging interferometers using many apertures will greatly benefit from adaptive optics systems for ultimate performance on faint sources , providing usable imaging with large exposure time .
but even in the absence of adaptive phasing , high resolution imaging can be done in speckle mode .
speckle mode techniques like speckle interferometry @xcite and speckle masking @xcite , heretofore successfully used with large monolithic telescopes , can also produce useful results with such interferometers .
we describe the numerical simulations done to understand the scope and explore the performance of speckle imaging techniques with this type of diluted aperture interferometers .
we have built a numerical simulation code in matlab that simulates cophased and speckle mode imaging with diluted apertures in different configurations .
the imaging performance of interferometers is much affected by their beam combination scheme , as analyzed in detail by @xcite . in our simulation
we have adopted the pupil densification scheme @xcite , also adopted for the ubaye hypertelescope project currently undergoing preliminary testing in the southern alps @xcite .
pupil densification concentrates the diffracted energy in the central part of the point spread function(psf ) without degrading the resolution .
a detailed analysis of the psf and imaging properties of these diluted aperture interferometers has been done by @xcite and @xcite . in the present article , our simulations extend these results to the case of randomly phased apertures exploited with speckle imaging , and also in the use of aperture rotation which enhances the quality of reconstructed images .
we have also tried to find out the limiting magnitude for the technique by comparing signal to noise ratio ( snr ) of reconstructions at different photon levels .
according to the usual definition of hypertelescope imaging we consider a multi - aperture fizeau interferometer equipped with a non - distorting pupil densifier i.e , one which does not distort the pattern of sub - aperture centers .
this makes the interference function field - invariant , while the spread function depends on the star s position , through its diffraction factor .
this variable spread function , hereafter called `` pseudo spread function '' , results in a pseudo - convolution process which describes the image formation on extended sources . for @xmath0 identical sub - apertures with position vectors @xmath1 ) , the cophased pseudo spread function @xcite is expressed as @xmath2 where @xmath3 is the diffraction function and the term on the right side , @xmath4 is the interference function given by @xmath5 the interference function depends only on the array configuration while the diffraction function @xmath3 depends on the beam combination scheme used @xcite . for our analysis
we have considered @xmath3 to be an airy function corresponding to the densified subaperture diameter @xmath6 . in our simulation
we are considering highly diluted arrays , approximated to array of @xmath7-functions , in which pupil densification is so strong that the shift of the airy envelope is negligible . in such a case the image formation of the hypertelescope
can be defined by a pseudo convolution equation @xmath8 where @xmath9 is the object intensity distribution .
the pseudo convolution creates the direct imaging field of the interferometer which is the diffraction function envelope corresponding to the densified subaperture .
this modeling of hypertelescope imaging is similar to the study by @xcite . to simulate the seeing conditions we generate kolmogorov phase screens ( fig .
[ fig1 ] ) with different values of fried parameter and also move these screens with different wind velocities to simulate seeing changes . in our simulations
we assume that the wavefront is coherent across an individual subaperture .
this will be the case if you use small mirrors ( @xmath10 ) in a site with good seeing conditions , and hence only piston errors associated with each subaperture was considered .
the random atmospheric piston error @xmath11 from the corresponding position of the sub - aperture in the phase screen is introduced into the interference function .
@xmath12 this new speckle interference function is used in the pseudo - convolution equation ( [ pseudoconv ] ) to simulate speckle observations .
some of the speckle images corresponding to sample objects used in the study are shown in figure [ fig3 ] .
if the phase across each subaperture is not coherent and there are phase pertubations across the individual mirrors , the diffraction envelope will break in to a speckle pattern .
we have not done an estimate of this effect in the current study .
though there are several aperture configurations under consideration for the big imaging interferometers , the one we studied was of randomly distributed mirrors with non - redundant baselines . the number of diluted apertures , @xmath0 of 20 ( fig [ fig2 ] ) and 50 were considered for this paper . the details of aperture configuration used in the simulations are in table [ tab1 ] .
details of the pupil densification beam combination scheme used in the simulations are discussed in detail by @xcite and @xcite .
densification is constrained by the smallest baseline @xmath13 in the output pupil , since any further densification will cause subapertures to overlap .
thus in maximum densification , length of the smallest baseline in output pupil will be equal to the densified subaperture diameter , i.e. @xmath14 . in our simulation for all the array configurations we have taken densified diameter of each diluted subaperture @xmath15 to be @xmath16 of the largest baseline in the output pupil .
output pupil which is partially filled creates a speckle halo surrounding the central peak in cophased case .
the dominance of this halo is dependent on the output pupil filling rate ( @xmath17 ) .
.parameters associated with aperture configuration . [ cols="<,^,^",options="header " , ] we wanted to compute the limiting magnitude of objects retrievable through such a technique . for this
we have simulated imaging of two objects , a binary star and a resolved star with spots .
both the objects were imaged assuming different magnitudes progressively and the signal to noise ratio was computed for each recontruction .
since the photon levels of speckle images used for reconstructions to compute limiting magnitude were very low , we have used another measure to compute the signal to noise ratio instead of the correlation coefficient .
the snr for the binary star was computed as the contrast of the fainter star in the background halo of noise .
@xmath18 where @xmath19 is the mean flux level from the pixel positions corresponding to the star and @xmath20 the standard deviation of the average background noise .
the average noise was computed in the encircled area in the figure ( [ fig7 ] ) . for the star spots
the contrast level of one of the spots was used as the measure of snr . to calculate the snr ,
20 reconstructions were computed at each magnitude photon level .
the resulting snr plotted against the photon levels are shown in figure [ fig9 ] for both binary star and the resolved spotted star . and [ tab2].,width=302 ]
the simulations have successfully shown that even in the absence of adaptive phasing , hypertelescope systems employing pupil densification can be used for direct imaging using speckle imaging techniques .
it is shown that with utilizing the aperture rotation of the diluted array through night , reconstruction quality could be increased substantially .
other ways of changing the aperture pattern during observation may also be considered , and should similarly be expected to improve the imaging performance . though as with cophased imaging , speckle imaging also is constrained by the field of view limitations in a hypertelescope . the use of speckle imaging for future earth - based hypertelescopes @xcite can be of interest for faint sources which can not be phased in the absence of adaptive optics , or because of the absence of a sufficiently bright guide star . with the parameters of hypertelescope used in the simulation we have obtained a good snr for the speckle technique at a magnitude of 8 - 9 for a simple binary star and 6 - 7 for a resolved star with spots .
the improvement in limiting magnitude of the technique with different amounts of pupil densification need to be studied further .
also the technique could give better results with deconvolution algorithms and pupil re - dilution as studied by @xcite .
while modified forms of adaptive optics have been proposed specifically for hypertelescopes @xcite , it remains unclear at this stage whether suitable forms of a laser guide star can also be operated for such instruments @xcite .
if not , earth - based hypertelescopes will have to use speckle - imaging on faint sources , which should provide useful results , according to our simulation results , although with lesser performance than in phased conditions . for space hypertelescopes ,
phasing is expected to be much easier @xcite .
although the huge sizes considered , 100 km for an exo - earth imager and 100,000 km for a neutron star imager , can make them sensitive to new forms of seeing such as gravitational seeing " .
99 aime c. , 2008 , a&a , 483 , 361 aime , c. ; lantri , h. ; diet , m. ; carlotti , a. , 2012 , a&a , 543 , a42 baldwin , j. e. ; haniff , c. a. ; mackay , c. d. ; warner , p. j. , 1986 , nature , 320 , 595 borkowski , v. ; labeyrie , a. ; martinache , f. ; peterson , d. , 2005 , a&a , 429 , 747 - 753 labeyrie a. , 1970 , a&a , 6 , 85 + labeyrie a. , 1996 , a&a , 118 , 517 labeyrie a. , 2007 , comptes rendus physique,8 , 426437 labeyrie a. et al .
, 2009 , exp.astron.,23 , 463 - 490 , labeyrie a. et al . , 2009 , in proc .
`` pathways toward habitable planets '' , barcelona conf .
( asp conference series , vol . 430 ) labeyrie , a. , et al . , 2012 , optical and infrared interferometry iii , proceedings of the spie , 8445 , 844511 labeyrie , a. , et al . , 2012 , optical and infrared interferometry iii , proceedings of the spie , 8445 , 844512 lardire , o. ; martinache , f. ; patru , f. , 2007 , mnras , 375 , 977 - 988 le coroller , h. ; dejonghe , j. ; arpesella , c. ; vernet , d. ; labeyrie , a. , 2004 , a&a , 426 , 721 - 728 le coroller h. et al . , 2012 , a&a , 539 , a59 lohmann , a. w. ; weigelt , g. ; wirnitzer , b. , 1983 , appl . opt .
22,4028 - 4037 martinache f. , 2004 , journal of optics a : pure and applied optics , 6 , 216 - 220 owner - petersen et al . , 2011 ,
kiruna symp .
integrated modeling of complex optomechanical systems " patru , f. ; tarmoul , n. ; mourard , d. ; lardire , o. , 2009 , mnras , 395 , 2363 - 2372 nuez , paul d. ; labeyrie , antoine ; riaud , pierre , 2014 , mnras reinheimer , t. ; hofmann , k .- h . ;
weigelt , g. , 1993 , a&a , 279 , 322 - 334 reinheimer , t. ; hofmann , k .- h . ; scholler , m. ; weigelt , g. , 1997 , a&a , 121 , 191 - 199 rondi a. , 2011 , proc .
kiruna symp .
integrated modeling of complex optomechanical systems " surya a. & saha s.k , 2014 , journal of optics , springer sridharan r. , 2000 , phd thesis , iiap bangalore weigelt g. , 1977 , optics communications , 21 , 55 - 59 | high - resolution optical astronomy with interferometric arrays better images more apertures for denser sampling optical wave few three apertures suffice reconstruct images through optical aperture synthesis'' using earth rotation baseline changes to sample fourier components object better sensitivity with systems using more apertures even smaller for conserved collecting area gain from coherent combination light vibrations with many apertures opposed to optical aperture synthesis approach where interference fringes recorded with fewer apertures different baseline settings combined computer incoherently phased beams combined with simple fizeau arrangement produces constructive interference peak n times more intense than average side - lobes successive exposures with sub - apertures enlarged reduce peak intensity dynamic range in image complex source fizeau combination inefficient with diluted apertures narrow interference peak contains small proportion energy retrieving energy in peak for observing faint sources with many - aperture interferometers hypertelescope densified aperture '' schemedesigns for large hypertelescopes spherical geometry inspired from arecibo radio telescope testing for terrestrial versions @xcite space versions future large imaging interferometers apertures benefit from adaptive optics for performance on faint sources imaging large exposure time adaptive phasing high resolution imaging in speckle mode techniques like interferometry masking used with large monolithic telescopes produce results with interferometers numerical simulations speckle imaging techniques with diluted aperture interferometers built numerical simulation code in matlab cophased speckle mode imaging with diluted apertures imaging performance affected by beam combination scheme analyzed by @xcite adopted pupil densification scheme @xcite for ubaye hypertelescope project concentrates diffracted energy in point function without degrading resolution detailed analysis of psf imaging properties of diluted aperture interferometers by @xcite simulations extend results randomly phased apertures speckle imaging aperture rotation quality images tried limiting magnitude for technique comparing signal to noise ratio of reconstructions at different photon levelsdefinition hypertelescope imaging multi - aperture fizeau interferometer non - distorting pupil densifier sub - aperture centers interference function field - invariant spread function depends on star s position diffraction factor variable spread function pseudo spread function results in pseudo - convolution process image formation on extended sources @xmath0 identical sub - apertures position vectors cophased pseudo spread function expressed as @xmath2 @xmath3 diffraction function @xmath4 interference function @xmath5 depends on array configuration diffraction function @xmath3 depends beam combination scheme @xmath3 airy function densified subaperture diameter @xmath6 highly diluted arrays @xmath7-functions pupil densification strong shift airy envelope negligible image formation defined by pseudo convolution equation @xmath8 @xmath9 object intensity distribution pseudo convolution creates direct imaging field interferometer diffraction function envelope densified subaperture modeling hypertelescope imaging similar to study @xcite simulate conditions generate kolmogorov phase screens different values fried parameter different wind velocities changes wavefront coherent across subaperturesmall mirrors @xmath10 good seeing conditions piston errors each subaperture considered random atmospheric piston error @xmath11 from sub - aperture introduced into interference function @xmath12 new speckle interference function used in pseudo - convolution equation speckle observations speckle images shown in figure fig3 if phase subaperture coherent pertubations diffraction envelope speckle pattern estimate effect in current study several aperture configurations for big imaging interferometers studied randomly distributed mirrors with non - redundant baselines diluted apertures @xmath0 20 50 considered details aperture configuration in table tab1 pupil densification beam combination scheme discussed @xcite densification constrained by smallest baseline @xmath13 in output pupil further densification subapertures overlap maximum densification length smallest baseline output pupil equal to densified subaperture diameter @xmath14 simulation densified diameter of each diluted subaperture @xmath15 to @xmath16 of largest baseline in output pupil output pupil partially filled creates speckle halo surrounding central peak in cophased casedominance halo on output pupil filling rate ( @xmath17 ) aperture configuration compute limiting magnitude of objects retrievable technique simulated imaging two objects binary star resolved star with spots different magnitudes signal to noise ratio computed for each recontruction photon levels speckle images low used another measure signal to noise ratio instead correlation snr for binary star computed as contrast of fainter star in background halo noise @xmath18 @xmath19 mean flux level from pixel positions star @xmath20 standard deviation of average background noise average noise computed in encircled area figure fig7 star spots contrast level used as measure of snr snr 20 reconstructions computed at each magnitude photon level resulting snr plotted against photon levels shown in figure fig9 ] for binary star resolved spotted star simulations shown adaptive phasing hypertelescope systems pupil densification used for direct imaging speckle imaging techniques aperture rotation diluted array night reconstruction quality changing aperture pattern observation improve imaging performancecophased imaging speckle imaging constrained by field of view limitations in hypertelescope use speckle imaging for future earth - based hypertelescopes @xcite for faint sources phased adaptive optics bright guide star obtained good snr for speckle technique at 8 - 9 for simple binary star 6 - 7 for resolved star with spots improvement in limiting magnitude with pupil densification technique could better results with deconvolution algorithms pupil re - dilution studied @xcite modified adaptive optics proposed for hypertelescopes @xcite unclear whether laser guide star operated for if not earth - based hypertelescopes use speckle - imaging on faint sources useful results lesser performance than phased conditions for space hypertelescopes phasing expected easier @xcite huge sizes 100 km for exo - earth 100,000 km for neutron star imager make sensitive to new gravitational seeing aime 2008 a&a 483 361 lantri diet carlotti 2012 543 a42 baldwin haniff mackay warner 1986 320 595 borkowski labeyriemartinache f peterson d. 2005 a&a 429 , 747 - 753 1970 a&a 6 , 85 + 1996 118 , 517 2007 comptes rendus 2009 exp.astron.,23 , 463 - 490 2009 pathways toward habitable planets barcelona conf asp conference series vol 430 2012 optical infrared interferometry iii 8445 , 844511 844512 lardire o martinache f patru 2007 mnras , 375 , 977 - 988 coroller h dejonghe j arpesella vernet d. labeyrie 2004 a&a 426 , 721 - 728 2012 a&a , 539 , a59 lohmann weigelt g wirnitzer b 1983 appl . opt . 22,4028 - 4037 martinache 2004 journal optics 6 , 216 - 220 petersen 2011 kiruna symp integrated modeling complex optomechanical systems patru f. tarmoul n. mourard d.lardire 2009 mnras 395 2363 - 2372 nuez paul labeyrie antoine riaud pierre 2014 reinheimer hofmann weigelt 1993 a&a 279 322 - 334 reinheimer hofmann scholler weigelt 1997 a&a 121 191 - 199 rondi 2011 modeling complex optomechanical systems surya saha s 2014 journal optics springer sridharan 2000 phd iiap bangalore weigelt 1977 optics communications 21 55 - 59 | 0.551795 | 3,119 | 1,595 | , Saving $0.1 in GPT-4. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.