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 |
|---|---|---|---|---|---|---|---|
54,859 | this third part of the lecture series deals with the question : who will pay for your retirement ? for western
europe the answer may be `` nobody '' , but for algeria the demography looks more promising .
* sociophysics simulations iii : retirement demography * lotfi zekri * and dietrich stauffer institute for theoretical physics , cologne university , d-50923 kln , euroland now back at u.s.t.o . , dpartement de physique , l.e.p.m .
, b.p.1505 el mnaouar , oran , algeria . | during the last two centuries in peaceful rich countries , people lived on average longer and longer , while during the last few decades the number of chidren born per women during her lifetime has sunken below the replacement rate of slightly above 2 . also in many poorer countries the number of births has fallen and the life expectancy increased .
thus the fear of overpopulation of our planet earth has to be modified by fear of old - age poverty : in the year 2030 only those goods and services can be consumed by retired people which have been produced by working - age people .
a million dollars of old - age savings can be halved by a ten - percent inflation rate over seven years , if not enough young people help me to live .
this econosociobiophysics problem is one of demography , not of money .
we present in an appendix details of the assumptions for our extrapolations into the future . in the next section we deal with conditions as are typical for western europe , to be followed by a section on the different problems of algeria .
more literature on ageing models , including one applied to our demography @xcite , is given in @xcite .
around 1970 , the contraceptive pill reduced in the then two german states the average number of babies born by a women during her lifetime below the replacement level of two , to about 1.4 .
spain and italy followed later but levelled at a lower plateau , while in france the number is higher , about 1.7 .
life expectancy rises further though slower than during the first half of the 20th century . thus if people retire at an age of about 62 years , and if around 2030 the strongest age cohort in germany are the 70-year olds , problems lie ahead . only in recent years
were they discussed in general newspapers . as in science in general
, we need open publications of extrapolation methods and results .
only if many different simulations are compared can we see to what extent they agree and thus may be relied upon .
the top curve in fig.1 shows what happens if nothing is done : the average retirement age is 62 years , and immigration and emigration cancel each other .
then @xcite the number of old people to be supported by working - age people will increase drastically , while the total population will decrease .
we added here the number of children ( up to age 20 ) to the pensioneers since both groups are not fully working in the usual sense .
for the middle curve we assumed a net immigration of 0.38 percent per year , starting now , and an increase of the retirement age by about half of the increase of the life expectancy .
thus for every year which medical progress gives us , about six month are given like a tax to the labour market , while the other six months are leasure time after retirement .
now the ratio and the population are more stable .
if we do not count in the latter simulation the children ( bottom curve ) , then the ratio of pensioneers only to working age people is lower @xcite .
however , the reduction of the expenses for children is mainly an effect of the past , not of the future .
during the first half of the previous century , the fertility was very large in north africa compared to europe .
it reached the value 8.1 during the seventies in algeria because of a low average age of marriage in this country .
thirty years later , the average number of births per women ( during her lifetime ) becomes close to 2 , whereas in france the fertility needed two centuries to pass from 6 in the middle of the 17th century to 2 in the 1930 s .
algerian people are thus young .
figure 2 shows that the number of children ( up to age @xmath0 ) added to the pensioneers ( the retirement age in algeria is @xmath1 years ) obliged workers to support about two times their number until the year @xmath2 .
sixty years afterwards the population will be older but the fractions remain constant ( no fear of increasing ) .
we assumed in fig.2 the gompertz slope @xmath3 ( see appendix ) to increase with time from @xmath4 in 1901 to @xmath5 in 1971 and to remain constant thereafter .
only fertility data from 1950 on is available in algeria .
the fertility is constant with a mean value of @xmath6 from @xmath7 to @xmath8 and then decreases abruptly til 2004 to reach a value @xmath9 ; it is assumed to stay constant at this value thereafter .
the sixty years period necessary to reach the steady state , corresponds to the age of retirement . in figure 3
, we show that the ratio of the number of babies dying in their first year to the total number of births decreased by about @xmath0 percent from 1901 to 1980 .
thus , we made a correction on the fertility data ( in fig.2 ) by reducing them by the number of children dying before they reach maturity .
we noticed also that the greatest emigration rate of algerian people was between 1950 and 1970 but remains weak compared to the rate of births and does not influence the population evolution . in our simulation
we then neglected the emigration in such calculations .
however , this simulation did not account the rate of unemployeds which was very small during the period of socialism but reaches now 17 percent of population . however , the main prediction of fig .
2 is an increase of the social load for old age by 400 percent starting from 2020 , while that for children and old age combined will stabilize at the level around the year 2000 .
with rising life expectencies and falling births , the demographic problems of rich countries can be alleviated by controlled immigration and a moderate increase of the average retirement age .
that policy requires that first the unemployment is reduced appreciably . for algeria , on the other hand , emigration could not affect sensitively the evolution of pensioneers , but their rate should be multiplied by a factor four after 15 years from now on which would create a real economic problem were it not offset by a reduction of the number of children .
lz thanks the daad for supporting a one - year part of his thesis work in cologne .
we thank w.j .
paul for suggesting to add the children to the pensioneers .
according to the azbel lectures at this seminar , in all different countries and centuries , the probability of humans to survive up to a fixed age is a universal function of the life expectancy ; we do not have to apply this universality to yeast cells for the purpose of human demography .
thus we use germany as typical western european country , without taking into account the effects of world war ii .
the mortality function @xmath10 , where @xmath11 is the number of survivors from birth to age @xmath12 , is assumed to follow a gompertz law for adults : @xmath13 $ ] since the deviations at young age occur at such low mortalities that they are not relevant if we want to be accurate within a few percent .
the deviations at old age @xcite are not yet reliably established and may also be negligible as long as the fraction of centenarians among pensioneers is very small .
the gompertz slope @xmath3 was assumed to increase linearly with time from 0.07 in 1821 to 0.093 in 1971 and to stay constant thereafter , in contrast to bomsdorf @xcite and azbel @xcite but in agreement with yashin et al @xcite ; see also wilmoth et al @xcite .
instead , the characteristic age @xmath14 was constant at 103 years until 1971 and then increased each year by 0.15 years to give a rising life expectancy .
also these deviations from universality are not yet established reliably .
( therefore we ignored the effect for algeria , keeping @xmath15 constant there . )
babies are born by mothers of age 21 to 40 with age - independent probability . the average number of children born per women over her lifetime and reaching adult age is assumed to be @xmath16 $ ] recently .
immigrants are assumed to be 6 to 40 years old with equal probability , and their number per year equals a fraction @xmath17 of the population , adjusted to give a constant total population .
after the year 2010 , the retirement age is increased by 60 percent of the increase of life expectancy at birth to 73 in 2100 at a life expectancy then of 99 years ; for the problem year 2030 these ages are 64 and 84 years
. 99 a. aszkiewicz , sz .
szymczak , s. cebrat , int . j. mod
c 14 , 1355 ( 2003 ) ; s. cebrat and a. aszkiewicz . j. insur .
medicine 37 , 3 ( 2005 ) .
d. stauffer , p. 131
in : thinking in patterns , ed . by m.m .
novak , world scientific , singapore 2004 .
for an older but much more detailed review see s. moss de oliveira , p.m.c .
de oliveira and d. stauffer , _ evolution , money , war and computers _ , teubner , stuttgart and leipzig 1999 .
e. bomsdorf , exp .
gerontology 39 , 159 ( 2004 ) .
d. stauffer , exp .
gerontology , 37 , 1131 ( 2002 ) .
s martins , d. stauffer , ingenierias ( univ .
nuevo leon , mexico ) 7 , no.22 , p.35 ( jan - mar 2004 ) .
robine , j.w .
vaupel , exp .
gerontology 36 , 915 ( 2001 ) ; k. suematsu , j. theor .
201 , 231 ( 1999 ) .
azbel , proc .
b 263 , 1449 ( 1996 ) .
yashin , a.s .
begun , s.i .
boiko , s.v.ukraintseva , j. oeppen , exp .
gerontology 37 , 157 ( 2001 ) .
wilmoth , l.j .
deegan , h.lundstrm , s. horiuchi : science 289 , 2366 ( 2000 ) . | last two centuries in rich countries people lived longer last decades chidren born per women sunken below replacement rate above 2 in poorer countries births fallen life expectancy increased fear of overpopulation modified by fear old - age poverty 2030 goods services consumed by retired produced by working - age people million dollars old - age savings halved by ten - percent inflation over seven years if not enough young people live problem of demography not money present in appendix assumptions for extrapolations future next section conditions typical for western europe followed by section problems algeria literature on ageing models @xcite around 1970 contraceptive pill reduced two german states average number babies born by women below replacement level two to 1.4 spain italy followed levelled france higher 1.7 life expectancy rises slower than first half 20th century if people retire at 62 years 2030 strongest age cohort 70-year olds problems lie ahead recent years discussed in newspapers need open publications of extrapolation methods results if simulations compared top curve in fig.1 shows if nothing done average retirement age 62 years immigration and emigration cancel other number old people supported by working - age people increase total population added children up to age 20 ) to pensioneers groups not fully working assumed net immigration 0.38 percent per year increase retirement age by half of increase life expectancy for year medical progress six month like tax to labour market other six months leasure time after retirement now ratio population stable if count children ratio of pensioneers to working age lower reduction of expenses for children effect of past not future century fertility large in north africa reached 8.1 during seventies in algeria low average age of marriage thirty years later average births per women close to 2 in france fertility needed two centuries from 6 to 2 1930 algerian people young figure 2 shows number of children up to age @xmath0 ) added to pensioneers retirement age @xmath1 years workers to support two times number until year @xmath2 sixty years afterwards population older fractions remain constant assumed gompertz slope to increase with time from 1901 to in 1971 remain constant thereafterfertility data from 1950 available in algeria fertility constant mean value @xmath6 from to @xmath8 decreases til 2004 @xmath9 assumed to stay constant sixty years steady state corresponds to age retirement figure 3 ratio babies dying in first year to total births decreased by @xmath0 percent from 1901 to 1980 correction fertility data reducing children dying before maturity greatest emigration rate algerian between 1950 and 1970 weak compared to births influence population evolution simulation neglected emigration account rate unemployeds small during socialism now 17 percent of population main prediction increase social load for old age by 400 percent from 2020 children old age stabilize around 2000 rising life expectencies falling births demographic problems rich countries alleviated by controlled immigration moderate increase average retirement age policy requires unemployment algeria emigration affect evolution pensioneers rate multiplied by factor four after 15 years economic problem not offset by reduction number children thanks daad for supporting one - year part thesis work in cologne thank w.j . paul for suggesting to add children to pensioneersazbel lectures countries probability humans to survive fixed age universal function life expectancy apply universality to yeast cells for human demography use germany as typical western european country without effects world war ii mortality function @xmath10 number survivors from birth to age @xmath12 follow gompertz law for adults deviations at young age low mortalities not relevant deviations at old age not established negligible fraction centenarians among pensioneers small gompertz slope @xmath3 increase linearly time from 0.07 1821 to 0.093 in 1971 constant contrast to bomsdorf azbel agreement with yashin characteristic age @xmath14 constant at 103 years until 1971 increased by 0.15 years rising life expectancy deviations from universality not established ignored effect for algeria keeping @xmath15 constant babies born by mothers age 21 to 40 age - independent probability average number of children born per women adult age @xmath16 immigrants 6 to 40 years old equal probability number per year equals fraction @xmath17 of population constant total population2010 retirement age increased 60 percent life expectancy 73 2100 99 years 2030 ages 64 84 years aszkiewicz szymczak cebrat j. 14 , 1355 2003 ) insur medicine 37 , 3 2005 ) d stauffer p. 131 thinking patterns m novak world scientific singapore 2004 review moss oliveira de oliveira d. stauffer evolution money war computers teubner stuttgart leipzig 1999 bomsdorf gerontology 39 , 159 2004 ) stauffer gerontology 37 , 1131 2002 ) martins stauffer ingenierias univ. nuevo leon mexico 7 no.22 p.35 jan mar 2004 robine vaupel gerontology 36 , 915 2001 ) suematsu theor. 201 , 231 1999 ) azbel 263 , 1449 1996 ) yashin begun boiko.ukraintseva oeppen gerontology 37 , 157 2001 ) wilmoth deegan h.lundstrm horiuchi science 289 , 2366 2000 ) | 0.32467 | 2,250 | 1,139 | , Saving $0.1 in GPT-4. |
54,869 | our recent studies of very short gamma ray bursts ( vsb ) of duration @xmath0 ms have indicated a significant angular asymmetry and a uniform @xmath1 distribution from the batse data . here
we update these studies , and we extend our research to events observed with konus satellite which gives a new insight into the spectra not possible with the batse data .
konus has observed 18 events with @xmath0 ms duration .
these events display considerable numbers of photons with energies above 1 mev and have photons above 5 mev in all cases .
these appear to be some of the most energetic photons observed from any classes of grb to date . | the long bursts ( @xmath2 s ) are known to originate at cosmological distances with many identified counterparts and they are widely believed to be related to the collapse of massive stars @xcite .
considerable information about the short bursts can be found @xcite including the very interesting possibility that the long and short bursts come from the same sources @xcite .
swift results seem to indicate the merger scenario @xcite but further study of short bursts is necessary . in this letter
we discuss the properties of the very short bursts ( @xmath0 ms ; hereafter vsb ) .
we have found some unusual aspects of these events including : \a ) very hard photon spectra compared to longer duration grbs .
\b ) a significant angular asymmetry on the galactic plane .
\c ) a @xmath3 value consistent with 1/2 . in this paper
we update the results of @xcite including new events .
we also analyze the vsb observed with the konus detector that have been recently reported @xcite .
sensitivity of konus detector above 1 mev allows to draw firm conclusions on the hard part of the vsb spectra .
data analysis is given in section 2 ( batse results ) and section 3 ( konus results ) , and the results are discussed in section 4 .
we use the data from the batse current gamma - ray burst catalog @xcite .
it contains the following data base : basic ( 2702 events - last update sep .
9 , 2000 ) , duration ( 2041 events - last update aug.17 , 2000 ) , cmax / cmin ( 1318 events - last update aug . 1 , 2003 ) .
the subsample of short bursts ( hereafter sb ) , with @xmath4 s @xmath5 s , consists of 448 bursts .
there are 51 bursts that have @xmath6 s , and they form our subsample of vsb .
we note that there has been evidence for an anisotropy in the sb from a previous study @xcite . the angular distribution of vsb and sb in galactic coordinates is shown in fig .
the sky was divided into 8 equal regions . in the case of isotropic distribution
the number of bursts in each region should be described by the poisson distribution .
the histograms of such values are shown in fig .
1 . the probability of any number of bursts in a single zone multiplied by the number of zones ( 8) is shown with full circles . for 448 sb ( @xmath4 s @xmath7 s )
we see conformity with isotropic distribution , but for vsb ( @xmath8 s ) the observed distribution is strongly improbable for isotropic distribution .
the number of bursts in one of the regions ( roughly in the direction of the galactic anticenter ) is 20 , which is much higher than the expected average of 51/8 .
the probability to find twenty or more events ( from the total number of fifty one ) in the region of 1/8 area is 0.00007 .
this result argues for other explanation than the statistical fluctuation .
background in the direction of the galactic center is @xmath9 cts s@xmath10 , while the mean level of the background outside the region is @xmath11 cts s@xmath10 , and the total number of sb in this regions is slightly lower than the expected average ( but within the expected error ) .
therefore , background anisotropy can not be responsible for the observed distribution of the vsb across the sky .
we support the choice of @xmath12 = 0.1 s to discriminate vsb from sb with the following consideration . in fig.2
we show the distribution of the excess number of the grbs as a function of @xmath12 .
the excess from the isotropic distribution is calculated as a difference between the grbs number in the chosen region and the sum of all grbs number in other regions divided by seven .
this study shows that the incompatibility with isotropic distribution is seen only for grbs with @xmath8 s. we observe the total excess of @xmath13 bursts in this region .
we have also reanalyzed the overall radial distribution of the vsb and the sb using the standard @xmath3 test @xcite .
we use the cmax / cmin table from batse catalog as an input for @xmath1 calculations : @xmath1 = @xmath14 and we follow the standard algorithm given in @xcite .
eight events in the table cmax / cmin have all values lower than the accepted trigger .
we can either reject these events , or attribute them the value close to the detection threshold ( @xmath15 @xmath16 0.9 ) .
the choice does not affect significantly the measured value of @xmath1 ( see table [ tab0 ] ) .
the result of @xmath1 distributions and mean values are presented in fig .
3 . sb appear to come from cosmological distances given the @xmath1 distribution we see since this is similar to the distribution for the long grbs .
no cosmological effects are seen in the distribution of vsb .
in table [ tab1 ] are listed the candidates for vsb from konus - 18 events ! this is approximately what would be expected from the relative konus and batse exposure and confirms that the batse events are real . in the case , when the same event ( the same time of registration ) was observed in both konus and batse experiments both @xmath12 values are shown in table [ tab1 ] @xcite .
independent detections by satellites other than konus are also marked , including the trigger number ( b - batse , u - ulysses , n - near ) .
we accept as vsb events where no other significant maximum ( above background ) appears within 30 seconds after the trigger .
the konus detector has a significantly larger energy acceptance out to tens of mev .
the batse detector has a smaller nai absorber and is not sensitive to 1 - 10 mev photons .
therefore the konus detector offers the first chance to measure higher energy photons from vsb . however , there are no corresponding coordinates for the konus events as in the case of batse .
we therefore use the konus data mainly for spectral information .
all vsb show appreciable number of photons above 1 mev energy .
this follows the trend observed in the batse data that the vsb events have a hard energy spectrum .
all events also show gamma rays above 5 mev . in fig.4
we compare the mean energy of sb and vsb for konus events .
we observe that in the mev region the spectra of vsb are significantly harder than the spectra of sb .
the spectra start to be flatter above 3 mev and the effect in the case of vsb is again stronger .
this extended analysis of the batse data and the results of the konus detector continue to suggest that the vsb events form a separate class of events from the sb .
strong anisotropy and @xmath17 consistent with euclidean space strongly argue for the local origin of this phenomenon .
spectra of vsb are very hard , supporting the claim that a different mechanism is here at work . detected counterparts of long bursts at other wavelengths established the relation of those events to the massive star collapse ; no such support was available for sbs @xcite .
therefore , several possibilities for the origin of this class of bursts were proposed , e.g. @xcite .
this includes neutron star mergers @xcite and neutron star collapse to strange - quark stars @xcite .
there were also suggestions that the nature of long and short bursts is the same @xcite although sb are on average harder @xcite , have @xmath18 closer to 1 and show traces of anisotropy @xcite .
vsbs are extreme examples being still harder and showing even more spectacular anisotropy , clearly indicating towards the local origin of this class of events .
primordial black hole evaporation , as an explanation of the existence of the vsb class , is also possible ( for details , see @xcite ) .
the energy expected from such phenomenon is of order of @xmath19 , and the currently evaporating black holes have masses @xmath20 g @xcite
. such energy , if converted with an efficiency @xmath21 into the form of @xmath22-rays collimated in a solid angle @xmath23 , would give the observed fluence @xmath24}$ ] .
the weakest vsb ( see table [ tab1 ] ) can come from distances of @xmath25 pc if uncollimated , and from further out if some collimation is involved .
however , less standard approach to primordial black hole mass evaporation gives much broader range of numbers @xcite .
we also note that there is some confusion between the short grb and the sgr super flares @xcite
. however , the very hard spectrum of the vsb would seem to separate those types of events .
plausible detection of x - ray afterglows @xcite , and recent claims of the identification of a host galaxy for two sb swift bursts @xcite might point toward merger scenario .
one of the localized bursts ( grb 050724 ) is a typical sb ( @xmath26 s ) .
the other one ( grb 050509b ) is intermediate between a vsb and a sb .
its galactic coordinate is not in the excess region of fig 1a and the limited energy acceptance of swift does not allow a test of the presence of mev photons to compare with other vsb studied here .
more swift data may clear the issue .
in this paper we have provided a new analysis of the most recent vsb events from batse .
we have also studied the vsb from konus for the first time .
we show that the grbs with time duration below 100 ms appear to form a separate class of grbs .
the konus data show that these events have a very hard photon spectrum up to 10 mev - this effect was already hinted in the batse hardness data .
the events appear to originate nearby within the galaxy as the asymmetry plot ( fig .
1.a ) and the @xmath1 distribution ( fig .
2.a ) indicate .
this effect is statistically significant . in previous studies of this subject ,
we have suggested that such events could come from primordial black hole evaporation @xcite , with supplementary details in @xcite .
the other local sources such as neutron stars should also be candidates .
cline , d.b . ,
matthey , c. , & otwinowski , s. 2001 , in gamma - ray bursts in the afterglow era proceedings of the international workshop held in rome cnr headquarters 17 - 20 october 2000 , eds .
enrico costa filippo frontera and jens hjorth ( berlin heidelberg : springer ) , 56 rccc @xmath27 @xmath28 1 & vsb & sb & @xmath12 @xmath29 2 s + rejected events & 0.48 @xmath30 0.05 & 0.37 @xmath30 0.02 & 0.30 @xmath30 0.01 + accepted events & 0.52 @xmath30 0.05 & 0.36 @xmath30 0.02 & 0.30 @xmath30 0.01 + nn & burst & @xmath12 [ s ] & peak flux & fluence & energy & location & @xmath12 [ s ] + & name & ( konus ) & erg @xmath31 @xmath32 & erg @xmath31 & interval kev & & ( batse ) + 110 & 010420a & 0.01 & 1.2e005 & 2.3e007 & 15 - 1000 & - & - + 32 & 970625a & 0.03 & 3.8e005 & 9.4e007 & 15 - 3000 & - & - + 18 & 960803 & 0.05 & 2.1e005 & 8.9e007 & 15 - 5000 & b(tr.5561 ) & 0.10 + 57 & 980904 & 0.05 & 3.5e005 & 1.3e006 & 15 - 2000 & b(tr.7063 ) & 0.13 + 27 & 970427 & 0.06 & 1.0e005 & 3.9e007 & 15 - 1000 & b(tr.6211 ) & 0.07 + 37 & 970921 & 0.06 & 1.0e004 & 4.5e006 & 15 - 8000 & - & - + 79 & 970427 & 0.06 & 1.0e005 & 3.9e007 & 15 - 1000 & - & - + 6 & 950610b & 0.07 & 1.6e005 & 7.1e007 & 15 - 1000 & - & - + 75 & 990504 & 0.08 & 1.5e005 & 1.2e006 & 15 - 2000 & - & - + 83 & 990831 & 0.08 & 5.9e006 & 3.8e007 & 15 - 1000 & - & - + 40 & 971118 & 0.08 & 2.7e006 & 1.7e007 & 15 - 1000 & b(tr.6486 ) & 0.07 + 86 & 991226b & 0.08 & 7.7e006 & 3.9e007 & 15 - 1000 & - & - + 117 & 020116 & 0.08 & - & - & - & - & - + 34 & 970704 & 0.09 & 1.5e003 & 4.2e005 & 15 - 10000 & b(tr.6293 ) & 0.19 + 95 & 000607 & 0.09 & 1.2e004 & 5.4e006 & 15 - 3000 & u , n , gcn693 & - + 119 & 020218a & 0.09 & - & - & - & - & - + 48 & 980330a & 0.10 & 5.2e005 & 3.0e006 & 15 - 5000 & b(tr.6668 ) & 0.12 + 112 & 010616 & 0.10 & 1.8e005 & 1.5e006 & 15 - 1000 & - & - + | long bursts ( @xmath2 s ) originate at cosmological distances related to collapse massive stars information about short bursts @xcite from same sources results indicate merger scenario further study necessary discuss properties short bursts ( @xmath0 ms ; vsb ) unusual aspects hard photon spectra longer duration significant angular asymmetry on galactic plane @xmath3 value consistent with 1/2 update results @xcite including new events analyze vsb observed with konus detector reported sensitivity konus detector above 1 mev conclusions on hard part vsb spectra data analysis in section 2 ( batse 3 ( konus results ) results discussed in section 4 use data from batse current gamma - ray burst catalog @xcite data base basic 2702 events duration ( 2041 events cmax / cmin ( 1318 events subsample of short bursts sb ) with @xmath4 s @xmath5 s 448 bursts 51 bursts have @xmath6 s form subsample of vsb evidence for anisotropy in sb from previous study @xciteangular distribution of vsb sb in galactic coordinates in fig . sky divided into 8 regions isotropic distribution bursts region described by poisson distribution histograms in fig . 1 probability of bursts zone multiplied by zones 8) shown with full circles for 448 sb ( @xmath4 s @xmath7 s conformity with isotropic distribution vsb ( @xmath8 s improbable for isotropic bursts in 20 higher than expected average 51/8 probability twenty or more events in 1/8 area 0.00007 argues other explanation statistical fluctuation background galactic center @xmath9 cts s@xmath10 mean level background outside @xmath11 cts s@xmath10 total sb slightly lower than expected average within error background anisotropy responsible for distribution vsb choice of @xmath12 = 0.1 s to discriminate vsb from sb fig.2 distribution excess grbs function of @xmath12 excess from isotropic distribution calculated as difference between grbs chosen region and divided by seven incompatibility with isotropic distribution for grbs with @xmath8 s excess of @xmath13 bursts in regionreanalyzed radial distribution vsb sb using standard @xmath3 test @xcite use cmax / cmin table from batse catalog input for @xmath1 calculations = follow standard algorithm @xcite eight events values lower than accepted trigger reject events or attribute value close to detection threshold 0.9 choice affect measured value @xmath1 table result @xmath1 distributions mean values presented in fig . 3 sb from cosmological distances distribution similar to long grbs no cosmological effects in distribution vsb table tab1 candidates for vsb from konus - 18 events confirms batse events real same event observed in both konus batse experiments @xmath12 values shown in table tab1 @xcite independent detections by satellites other than konus marked trigger number accept as vsb events where no significant maximum within 30 seconds after trigger konus detector larger energy acceptance tens of mev batse detector smaller not sensitive to 1 - 10 mev photons konus detector first chance to measure higher energy photons from vsb no corresponding coordinates for konus events batse use konus data for spectral informationvsb show photons above 1 mev energy follows trend batse data hard energy spectrum show gamma rays above 5 mev fig.4 mean energy sb vsb for konus events in mev region spectra vsb harder than sb spectra flatter above 3 mev effect vsb stronger extended analysis batse data results konus detector suggest vsb events separate class from sb strong anisotropy @xmath17 euclidean space argue for local origin spectra vsb hard different mechanism long bursts at other wavelengths established relation to massive star collapse no support for sbs @xcite possibilities for origin bursts proposed includes neutron star mergers neutron star collapse to strange - quark stars suggestions nature of long and short bursts same @xcite sb harder @xcite have @xmath18 closer to 1 show traces of anisotropy @xcite vsbs harder more spectacular anisotropy indicating local origin primordial black hole evaporation explanation vsb class possible energy expected @xmath19 evaporating black holes have masses @xmath20 g @xcite energy converted @xmath21 into @xmath22-rays collimated solid angle @xmath23 observed fluence @xmath24}$ weakest vsb table from distances @xmath25 pc if uncollimated further if collimation less standard approach primordial black hole mass evaporation gives broader range numbers @xcite confusion between short grb sgr super flares @xcite hard spectrum vsb events detection x - ray afterglows @xcite host galaxy for two sb swift bursts might point merger scenario one localized ( grb 050724 ) typical sb other grb 050509b ) intermediate between vsb and sb galactic coordinate not in excess region fig 1a limited energy acceptance of swift allow test presence mev photons vsb more swift data may clear issue new analysis of recent vsb events from batse studied vsb from konus first grbs time duration below 100 ms separate class konus data events hard photon spectrum up to 10 mev events originate nearby galaxy asymmetry plot fig . 1.a @xmath1 distribution fig. 2.a effect statistically significantprevious studies suggested events primordial black hole evaporation @xcite supplementary details local sources neutron stars candidates cline matthey otwinowski 2001 gamma - ray bursts afterglow era international workshop rome cnr headquarters 17 - 20 october 2000 enrico costa filippo frontera jens hjorth berlin heidelberg springer 56 rccc @xmath27 @xmath28 @xmath12 events 0.48 @xmath30 0.05 0.37 0.02 0.30 accepted events 0.52 0.05 0.36 0.30 burst @xmath12 peak flux fluence energy location konus @xmath31 @xmath32 interval kev 110 010420a 0.01 1.2e005 2.3e007 15 - 1000 32 970625a 0.03 3.8e005 9.4e007 15 - 3000 18 960803 0.05 2.1e005 8.9e007 15 - 5000.5561 0.10 57 980904 0.05 3.5e005 1.3e006 15 - 2000.7063 0.13 27 970427 0.06 1.0e005 3.9e007 15 - 10006211 0.07 37 970921 0.06 1.0e004 4.5e006 15 - 8000 79 970427 0.06 1.0e005 3.9e007 15 - 1000 950610b 0.07 1.6e005 7.1e007 15 - 1000 75 990504 0.08 1.5e005 1.2e006 15 - 2000 83 990831 0.08 15 - 1000 40 971118 0.08 2.7e006 1.7e007 15 - 1000 0.07 86 991226b 0.08 3.9e007 15 - 1000 117 020116 0.08 34 970704 0.09 1.5e003 4.2e005 15 - 10000.6293 0.19 95 000607 0.09 1.2e004 5.4e006 15 - 3000 gcn693 119 020218a 0.09 48 980330a 0.10 5.2e005 3.0e006 15 - 5000.6668 0.12 112 010616 0.10 1.8e005 1.5e006 15 - 1000 | 0.437107 | 3,248 | 1,758 | , Saving $0.1 in GPT-4. |
54,880 | i emphasize in this contribution how prompt photons can be used to probe nuclear parton densities as well as medium - modified fragmentation functions in heavy ion collisions .
various predictions in @xmath0a and a a collisions at lhc energies are given .
cern - ph - th/2007 - 119 ; lapth-1197/07 prompt photon production in hadronic collisions has been extensively studied , both experimentally and theoretically , over the past 25 years ( see @xcite and references therein ) . as indicated in ref .
@xcite , it is remarkable that almost all existing data from fixed - target to collider energies can be very well understood within perturbative qcd at nlo . in these proceedings , i briefly discuss how prompt photons in nuclear collisions ( @xmath0a and a a ) may allow for a better understanding of interesting aspects discussed in heavy - ion collisions , namely the physics of nuclear parton distribution functions and medium - modified fragmentation functions .
parton distribution functions in nuclei are so far poorly constrained , especially in contrast with the high degree of accuracy currently reached in the proton channel , over a wide @xmath1 and @xmath2 domain . in particular , only high-@xmath1 ( @xmath3 ) and low @xmath2 ( @xmath4 gev@xmath5 ) have been probed in fixed - target experiments . in order to predict hard processes in nuclear collisions at the lhc ,
a more accurate knowledge on a wider kinematic range is necessary .
as stressed in @xcite , the nuclear production ratio of isolated photons in @xmath0-a collisions , @xmath6 can be related to a good accuracy ( say , less than 5% ) to the parton density ratios @xmath7 with @xmath8 . to illustrate this , the ratio @xmath9 is computed for isolated photons produced at mid - rapidity in @xmath0pb collisions at @xmath10 tev in fig .
[ fig : isoy0lhc ] ( solid line ) , assuming the de florian and sassot ( ndsg ) nuclear parton distributions @xcite .
the above analytic approximation @xmath11 ( dotted line ) demonstrates how well this observable is connected to the nuclear modifications of the gluon density and structure function ; see also the agreement @xmath12 as a dash - dotted line in fig .
[ fig : isoy0lhc ] . in nucleus - nucleus scattering ,
the energy loss of hard quarks and gluons in the dense medium presumably produced at lhc may lead to the suppression of prompt photons coming from the collinear fragmentation process . in fig .
[ fig : incy00lhc ] , the expected photon quenching in pb pb collisions at @xmath13 tev is plotted .
a significant suppression due to energy loss ( taking @xmath14 gev , see @xcite for details ) is observed , unlike what is expected when only nuclear effects in the parton densities are assumed in the calculation ( dash - dotted line ) . of @xmath15 inclusive photons in pb
pb collisions at @xmath13 tev.,height=260 ] of @xmath15 inclusive photons in pb
pb collisions at @xmath13 tev.,height=260 ] finally performing momentum correlations between a prompt photon and a leading hadron in @xmath0@xmath0 and a a collisions , yet experimentally challenging , appears to be an interesting probe of vacuum and medium - modified fragmentation function , as discussed in detail in refs .
@xcite .
we refer in particular the interested reader to fig .
10 of @xcite for the predictions of @xmath16@xmath17 momentum - imbalance distributions at the lhc . | part of this work has been done in collaboration with t. gousset @xcite and p. aurenche , z. belghobsi , and j .-
guillet @xcite . | work with t. gousset p. aurenche z. belghobsi j guillet @xcite | 0.066432 | 48 | 24 | , Saving $0.0 in GPT-4. |
54,881 | we find that the ambiguity term of approximate killing vector field is responsible for the entropy production term . without the ambiguity term , pure einstein theory and
@xmath0 satisfy the relation of thermodynamic equilibrium .
considering such an ambiguity term of approximate killing vector field , we can get the entropy production term and the entropy in @xmath0 with a form defined by jacobson . in pure einstein theory ,
the shear term is the only geometric contribution of entropy production term , while in @xmath0 it can also contribute .
we believe our approach and conclusion can be generalized to other gravity theory . | the discovery of black hole entropy and hawking radiation in 1970s bekenstein:1973ur implies a profound connection between gravitation and thermodynamics .
almost twenty years ago , jacobson @xcite proposed to interpret this connection by reversing the logic and deriving the einstein s equations from a thermodynamic equation of state with two assumptions : proportionality of entropy and area for all local rindler horizons , and the clausius relation .
the first assumption is nothing but the expression of the holographic principle @xcite@xcite . in 1998 , maldacena gave a gorgeous conjecture , namely the ads / cft correspondence @xcite , which is a direct manifestation of the holographic principle .
it was suggested that gravitation is induced by a quantum field theory in lower dimensions which to a large extent support the idea of gravitation on the macroscopic scale as a manifestation of the thermodynamics of the vacuum state .
furthermore , more confidence about gravity being emergent rather than fundamental was recently inspired in verlinde:2010hp@xcite@xcite@xcite@xcite . in @xcite ,
jacobson treated @xmath0 theory as non - equilibrium thermodynamics of spacetime , therefore , an entropy production term is required to keep the entropy balance relation .
the term is an outcome of the non - vanishing expansion at @xmath1 which is related to a local boost dependence quantity . in 2008 ,
elizalde and silva @xcite proposed an alternative approach , with local thermodynamic equilibrium maintained , using the idea of local - boost - invarianceintroduced in @xcite .
in other words , local boost dependence quantity is erased under the boost - invariant truncation , so expansion at @xmath1 vanishes , without any entropy production term when we repeat jacobson s derivation .
brustein and hadad @xcite showed that the equations of motion of generalized theories of gravity are equivalent to the thermodynamics relation @xmath2 .
their proof relies on extending previous arguments by using a more general definition , namely noether charge entropy . in this paper , we focus on the origin and feature of the entropy production . we show that it is a consequence of the @xmath3 ambiguity in an approximate boost killing vector field .
if we suspend this ambiguity first , the entropy production will vanish , and vice versa . in order to research the thermodynamics of different gravity models
, we assume the validity of those gravity field equations , and then use them to derive the form of entropy . in @xmath0 gravity , if the boost variant(a lie derivative of @xmath0 along killing vector at a point @xmath1 in the spacetime ) is nonzero , the entropy production is a function of the boost variantand non - vanishing expansion at @xmath1 .
in pure general relativity ( gr ) , we can not adopt such approach . the shear term from raychauduri equation
can include the entropy production .
@xmath0 theory can also accept the shear term , instead of non - vanishing expansion term , as the entropy production term .
at last , we may find that the shear term being the entropy production term is more general .
conditions of pure gr and @xmath0 theory analyzed , we believe our method and results can also be generalized to other gravity theories .
in this section , we review jacobson s work briefly and set the background of our work .
first , we can describe spacetime in a vicinity of a free - falling local observer @xmath1 as flat through the equivalence principle .
then , we choose a local 2-surface element @xmath4 including @xmath1 and perpendicular to the worldline of @xmath1 .
the boundary of the past of @xmath4 is defined as `` local rindler horizon '' [ horizon entropy ] , whose generators are a congruence of null geodesics with vanishing expansion and shear
. therefore , the local rindler horizon reaches equilibrium at @xmath1 .
considering the local rindler horizon is actually a causal horizon , we can introduce an entropy @xmath5 , measuring the degrees of freedom beyond it . according to the holographic principle , @xmath5 is proportionate to the area elements of the horizon .
the definition of the heat flux and temperature is related to an approximate boost killing vector field @xmath6 .
@xmath6 generating boosts orthogonal to @xmath4 the causal horizon .
it vanishes at @xmath1 , its flow invariant at the tangent plane @xmath7 .
its covariant derivative @xmath8 is a timelike antisymmetric tensor orthogonal to @xmath7 .
we normalize @xmath6 by @xmath9 . in a common curved spacetime ,
no killing vectors exist .
we can only solve the killing equation @xmath10 with this `` initial data '' out to some order in the neighborhood of @xmath1 .
the equation @xmath11is equal to the killing equation . in riemann
normal coordinates @xmath12 based at @xmath1 , the zeroth and the first order parts of @xmath6 are resolved by initial conditions , while the second order part vanishes according to the killing s equation ( compare both sides of eq.[eq1 ] to get this conclusion ) .
generally , the equation can not be satisfied at third order , so we still have a @xmath3 ambiguity in killing vector @xmath6 , which would influence the integrability of equation [ eq1](refer to weinberg s text book @xcite for detail ) .
we choose the direction of @xmath6 to be future pointing on the causal horizon .
in flat spacetime , @xmath13 can be defined as @xmath14 , where @xmath15 is the horizon tangent vector and @xmath16 is a negative affine parameter that is increasing along the horizon and vanishes at @xmath1 . here
, we set light - cone coordinates as @xmath17 which satisfy @xmath18with the third order ambiguity in curved spacetime , @xmath6 should be @xmath19and therefore eq.[eq1 ] becomes @xmath20 after defining the local rindler horizon and approximate killing vectors , it is natural to define the heat .
the heat is the mean flux of the boost energy current across the horizon measured by a uniformly accelerated observer hovering inside the horizon : @xmath21@xmath22 is the expectation value of the matter stress tensor , and @xmath23 is the area of each cross section element of @xmath24 .
the integration is over a short pencil of horizon generators of @xmath24 , and @xmath16 contained in @xmath6 represents the evolution of those generators .
since it is an equilibrium state at @xmath1 , the metric and , further , its conjugate @xmath22 are also approximately stable .
it means the expectation value of the matter stress tensor do not change over a sufficiently small @xmath16 or , equivalently , quantum transitions terminate . without loss of generality , we choose @xmath25 . according to the unruh effect and rindler coordinates
, we know acceleration tends to diverge as observer approaches the horizon .
since both of the temperature and heat flux are proportional to the acceleration , they diverge the same rate .
thus , we can choose the acceleration to be unit and the temperature measured by the observer to be @xmath26 .
in jacobson s paper , he reversed the logic , using two hypotheses , universal entropy density of the horizon and the clausius relation in vicinity of @xmath4 , to derive the einstein equation .
some gravity theories , like @xmath0 gravity , may need an entropy production rate to keep the entropy balance relation .
however , we wonder that the entropy may contain un - physical quantities which will cause the entropy production . in @xcitebrustein:2009hy ,
wald entropy was introduced immediately , so the un - physical quantities were truncated . in this section ,
we postulate the @xmath0 gravity field equation first , and then ascertain the form of entropy affecting the dynamics . considering the killing equation and approximate killing vector field , we will find the entropy production rate arising from the ambiguity term .
the equation of motion of @xmath0 is @xmath27 put it into the heat , it follows that @xmath28 for simplicity , we leave the ambiguity term @xmath3 of eq.[eq1 ] alone and pick it up afterward .
terms with @xmath29 vanish since @xmath29 contracted with @xmath30 leads to @xmath31 in light - cone coordinates .
we then evaluate the equation at the leading terms in @xmath16 , @xmath32the killing equation shows @xmath33 is antisymmetric , so we can use stoke s theorem and obtain @xmath34here comes a contradiction that the integrand contain @xmath16 at beginning , while the first term at the last step do not .
in other words , @xmath35 has a term of less order @xmath36 it seems that @xmath37 should be zero by comparing the two sides s order of eq.[eq9 ] , therefore @xmath38 in another way . in jacobson s paper ,
the entropy change is @xmath39 , where @xmath40 .
it can be extracted as @xmath41 , @xmath42 is the boost variantwhich mentioned in the introduction .
the raychaudhuri equation is @xmath43the shear term is required to be zero at @xmath1 .
comparing these equations , we will obtain the entropy production term @xmath44 however , the entropy production term is supposed to vanish at @xmath1 , the equilibrium point , so the rate should be of order @xmath16 .
it requires @xmath45 and @xmath46 @xmath47 this equation is directly related to the equation of motion of @xmath48 now , if we use @xmath49 in eq.[eq7 ] , then the integrand will have @xmath50 @xmath51 which has the same order with other terms .
@xmath52 consist of two parts : one is from eq.[4 ] , the other is from using the stokes theorem .
the ambiguity term @xmath50 act as @xmath53 which also vanishes at @xmath1 and is irrelevant to the equation of motion . substituting @xmath54 parallel with the entropy production density derived by jacobson .
one thing need attention is we only use @xmath55 in this part , not each of these two terms to be zero .
it means that they are not related to the the equation of motion .
latexmath:[$\dot{f}% depandant on the choice of coordinates and so is @xmath37 which is restricted by @xmath57 hence , we find the entropy production is related to boost .
in this section , we discuss the situation of pure gr under a similar procedure . replace the equation of motion of @xmath0 with einstein equation @xmath58 we immediately get @xmath59here @xmath60 is demanded for the reason that both sides of the equation should have the same order . without @xmath61 we can get the state equation of equilibrium positively . however , if we consider ambiguity term , we can not formulate the entropy production term since expansion at @xmath1 vanishies .
however , in raychaudhuri equation ( eq.[eq11 ] ) there is a shear term @xmath62 .
if the shear term is not required to be zero , it can be used to describe the entropy production term such as @xmath63 .
this is coincident with jacobson s remarks .
actually , entropy production term comprising latexmath:[$\theta limited by latexmath:[$\dot{f}% nonzero shear like in pure gr , we will also keep @xmath60 .
it may , from another aspect , give the reason why the thermodynamic relation @xmath2 hold in generalized theories of gravity .
these two situations analyzed above indicate that the entropy production term comes from the ambiguity term of the approximate killing vector . we do need it for the entropy balance relation rather than the field equation . from this point of view , we find the entropy production term is not related to the equation of motion
. this ambiguity term can also be constructed of nonvanishing @xmath65 or other quantities .
it can be seen that our method is universal , therefore we make the conclusion that entropy production term is caused by the ambiguity term of the approximate killing vector field .
furthermore , we believe that the entropy production term can also be described by nonvanishing @xmath65 , since raychaudhuri equation is used generally on null geodesics distortion caused by gravity .
we thank miao li and xiao - dong li for valuable discussions and suggestions .
l. susskind , j. math .
phys . * 36 * , 6377 ( 1995 ) .
j. m. maldacena , adv .
theor . math .
* 2 * , 231 ( 1998 ) [ int .
j. theor .
phys.*38 * , 1113 ( 1999 ) ] .
e. p. verlinde , arxiv:1001.0785 [ hep - th ] .
w. gu , m. li and r. x. miao , arxiv:1011.3419 [ hep - th ] .
t. padmanabhan , rept .
phys . *
73 * , 046901 ( 2010 ) [ arxiv:0911.5004 [ gr - qc ] ] . c. eling , r. guedens and t. jacobson , phys .
lett . * 96 * , 121301 ( 2006 ) .
e. elizalde and p. j. silva , phys .
d * 78 * , 061501 ( 2008 ) r. m. wald , phys .
d * 48 * , 3427 ( 1993 ) .
r. brustein and m. hadad , phys . rev .
lett . * 103 * , 101301 ( 2009 ) [ erratum - ibid . *
105 * , 239902 ( 2010 ) ] .
s. weinberg , _ gravitation and cosmology : principles and applications of the general relativity theory _ , john willey & sons , ( new york , 1972 ) . | discovery black hole entropy hawking radiation 1970s implies connection gravitation thermodynamics twenty years ago jacobson proposed logic einstein s equations from thermodynamic equation two assumptions proportionality entropy area local rindler horizons clausius relation first assumption holographic principle 1998 maldacena conjecture ads / cft correspondence manifestation holographic principle gravitation induced by quantum field theory lower dimensions gravitation macroscopic scale manifestation thermodynamics vacuum state confidence about gravity emergent inspired verlinde:2010hp jacobson treated theory non - equilibrium thermodynamics spacetime entropy production term required balance outcome non - vanishing expansion at @xmath1 related local boost dependence quantity 2008 elizalde silva proposed alternative approach local thermodynamic equilibrium idea local - boost - local boost dependence quantity erased under boost - invariant truncation expansion at @xmath1 vanishes entropy production term jacobson s derivation brustein hadad showed equations of motion generalized theories gravity equivalent to thermodynamics relation @xmath2proof relies arguments general definition noether charge entropy focus on origin feature entropy production consequence of @xmath3 ambiguity in boost killing vector field suspend ambiguity entropy production research thermodynamics gravity models assume validity gravity field equations derive form entropy in @xmath0 gravity boost variant nonzero entropy production function of boost non - vanishing expansion at @xmath1 in pure general relativity adopt approach shear term from raychauduri equation include entropy production @xmath0 theory accept shear term entropy production term shear term entropy production term more general conditions pure gr @xmath0 theory analyzed method results can be generalized to other gravity theories review jacobson s work set background work describe spacetime vicinity free - falling local observer @xmath1 as flat through equivalence principle choose local 2-surface element @xmath4 including @xmath1 perpendicular to worldline @xmath1 boundary of past @xmath4 defined as local rindler horizon '' entropy generators congruence of null geodesics with vanishing expansion shear local rindler horizon equilibrium at @xmath1 causal entropy @xmath5 measuring freedom holographic principle @xmath5 proportionate to area elements horizon heat flux temperature to boost killing vector field @xmath6 boosts orthogonal to @xmath4 vanishes at @xmath1 invariant at tangent plane @xmath7 covariant derivative @xmath8 timelike antisymmetric tensor orthogonal to @xmath7 normalize @xmath6 by @xmath9 curved spacetime no killing vectors solve killing equation @xmath10 with initial data @xmath1 equation @xmath11is equal to killing equation riemann normal coordinates @xmath12 at @xmath1 zeroth first order parts of @xmath6 resolved by initial conditions second part vanishes equation at third order @xmath3 ambiguity in killing vector @xmath6 integrability equation choose direction of @xmath6 future pointing on causal horizon flat spacetime @xmath13 defined as @xmath14 @xmath15 horizon tangent vector @xmath16 negative affine parameter increasing along horizon vanishes at @xmath1set light - cone coordinates @xmath17 satisfy @xmath18with third order ambiguity curved spacetime @xmath6 @xmath19and eq.[eq1 becomes @xmath20 defining local rindler horizon killing vectors define heat heat mean flux boost energy current across horizon measured uniformly accelerated observer inside horizon @xmath21 expectation value matter stress tensor @xmath23 area cross section element @xmath24 integration over horizon generators @xmath24 @xmath16 @xmath6 evolution generators equilibrium state @xmath1 metric conjugate @xmath22 stable expectation value matter stress tensor change over small @xmath16 quantum transitions terminate choose @xmath25 unruh effect rindler coordinates acceleration observer approaches horizon temperature heat flux proportional to acceleration diverge same choose acceleration unit temperature measured observer @xmath26 jacobson s paper reversed logic hypotheses universal entropy density horizon clausius relation vicinity @xmath4 derive einstein equation gravity theories @xmath0 gravity need entropy production rate balance entropy contain un - physical quantities entropy production@xcitebrustein wald entropy introduced un - physical quantities truncated postulate @xmath0 gravity field equation ascertain entropy affecting dynamics killing equation killing vector field entropy production rate from ambiguity term equation motion @xmath0 @xmath27 heat follows @xmath28 leave ambiguity term @xmath3 eq.[eq1 pick up terms @xmath29 vanish @xmath30 leads @xmath31 light - cone coordinates evaluate equation leading terms @xmath16 killing equation shows @xmath33 antisymmetric use stoke s theorem obtain @xmath34here contradiction integrand contain @xmath16 beginning first term last step @xmath35 term less order @xmath36 @xmath37 zero sides s order eq.[eq9 @xmath38 jacobson s paper entropy change @xmath39 @xmath40 extracted @xmath41 @xmath42 boost raychaudhuri equation @xmath43the shear term zero at @xmath1 comparing equations obtain entropy production term @xmath44 at @xmath1 equilibrium point rate order @xmath16 requires @xmath45 related to motion @xmath48 use @xmath49 in eq.[eq7 ] integrand @xmath50 @xmath51 same order other terms @xmath52 two parts one from eq.[4 ] stokes theorem ambiguity term @xmath50 as @xmath53 vanishes at @xmath1 irrelevant to equation motion substituting @xmath54 entropy production density jacobson use @xmath55 not terms zero not related to equation motion latexmath depandant choice coordinates @xmath37 restricted by @xmath57 entropy production related to boost discuss situation pure gr similar procedure replace equation motion @xmath0 with einstein equation @xmath58 get @xmath59here @xmath60 demanded both sides same order without @xmath61 state equation of equilibrium ambiguity term can formulate entropy production term expansion at @xmath1 vanishies raychaudhuri equation eq.[eq11 ] shear term @xmath62 term not required zero describe entropy production term @xmath63 coincident with jacobson s remarksentropy production term latexmath limited latexmath nonzero shear gr keep @xmath60 thermodynamic relation @xmath2 generalized theories gravity entropy production term ambiguity term approximate killing vector need entropy balance relation field equation not related to equation motion ambiguity term constructed nonvanishing @xmath65 quantities method universal entropy production term caused by ambiguity term approximate killing vector field entropy described nonvanishing @xmath65 raychaudhuri equation null geodesics distortion gravity thank miao li xiao - dong li discussions suggestions l. susskind math . phys . 36 6377 ( 1995 ) m. maldacena theor math 231 ( 1998 ) 1113 ( 1999 ) p. verlinde arxiv:1001.0785 [ hep - th w. gu m. li x. miao arxiv:1011.3419 [ hep - th t. padmanabhan phys . 73 046901 ( 2010 ) arxiv:0911.5004 [ gr - qc c. eling r. guedens t. jacobson phys .96 121301 2006 ) elizalde silva phys. 78 061501 ( 2008 ) wald 48 3427 ( 1993 ) brustein m. hadad . 103 101301 ( 2009 ) erratum 105 239902 ( 2010 ) weinberg gravitation cosmology principles applications general relativity theory john willey sons new york 1972 ) | 0.501063 | 3,134 | 1,612 | , Saving $0.1 in GPT-4. |
54,885 | photoelectric , photographic , and ccd @xmath0 photometry , spectroscopic observations , and star counts are presented for the open cluster berkeley 58 to examine a possible association with the 4@xmath1.37 cepheid cg cas .
the cluster is difficult to separate from the early - type stars belonging to the perseus spiral arm , in which it is located , but has reasonably well - defined parameters : an evolutionary age of @xmath2 years , a mean reddening of e@xmath3 s.e . , and a distance of @xmath4 kpc ( @xmath5 s.d . ) .
cg cas is a likely cluster coronal member on the basis of radial velocity , and its period increase of @xmath6 s yr@xmath7 and large light amplitude describe a cepheid in the third crossing of the instability strip lying slightly blueward of strip centre . its inferred reddening and luminosity are e@xmath8 s.e . and @xmath9 .
a possible k supergiant may also be a cluster member .
[ firstpage ] stars : variables : cepheids stars : evolution galaxy : open clusters and associations : individual : berkeley 58 . | after the rediscovery in the early 1950s of spatial coincidences between cepheids and open clusters by @xcite , eggen ( see * ? ? ?
* ) , and @xcite , a number of searches for additional coincidences were made by @xcite , @xcite , and @xcite , among others .
tifft s search resulted in the discovery of a near spatial coincidence between the 4@xmath1.37 cepheid cg cassiopeiae and an anonymous open cluster , subsequently catalogued as berkeley 58 @xcite , which lies less than one cluster diameter to the west .
the field is coincident with a portion of the perseus spiral arm that is relatively rich in open clusters , and the cluster ngc 7790 with its three cepheid members lies in close proximity .
the possibility that cg cas might be an outlying member of ngc 7790 was raised at one time by @xcite , and found some support in a star count analysis by @xcite .
more detailed star counts in the field @xcite indicate otherwise , as do the available proper motion data @xcite .
the cepheid does lie in the corona of berkeley 58 @xcite , although frolov has argued that it is not a probable cluster member . given a probable distance of 3 kpc to both cg cas and
berkeley 58 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , it is not clear that existing proper motion data are precise enough to provide conclusive evidence pertaining to the cluster membership of cg cas .
the present study was therefore initiated in order to examine the case in more detail . as demonstrated here
, there is strong evidence that cg cas is a likely member of berkeley 58 and that it can serve as a calibrator for the cepheid period - luminosity ( pl ) relation .
a variety of observations were obtained for the present investigation .
table 1 presents photoelectric _ ubv _
photometry for bright members of berkeley 58 , obtained during observing runs at kitt peak national observatory in 1981 september , 1982 august , and 1984 august . the data , acquired using 1p21 photomultipliers and standard _ ubv _
filter sets used in conjunction with pulse - counting photometers on the no .
40.4-m , no .
20.9-m , and 1.3-m telescopes at kitt peak , have associated uncertainties typical of our previous investigations of cepheid clusters @xcite , namely standard internal errors for a single observation of @xmath10 in @xmath11 and _ b v _ , and @xmath12 in _ u b _ , for stars brighter than @xmath13 .
the estimated external errors for all but the faintest stars are similar in magnitude .
the stars are identified by their numbering in fig .
1 , as well as by their 2000 co - ordinates in the 2mass survey @xcite ; the number of individual observations for each star is given in column 7 of table 1 . and is centred at 2000 co - ordinates : ra = @xmath14 , dec = + 60@xmath15 56@xmath16 07@xmath17 .
the top image depicts the location of cg cas relative to the cluster core , the lower image identifies photoelectrically observed stars .
[ the national geographic society - palomar observatory sky atlas ( poss - i ) was made by the california institute of technology with grants from the national geographic society.],width=226 ] | rediscovery early 1950s of coincidences between cepheids open clusters by @xcite eggen @xcite searches for additional coincidences @xcite tifft s search near spatial coincidence between 4@xmath1.37 cepheid cg cassiopeiae anonymous open cluster berkeley 58 @xcite less than one cluster diameter west field coincident with perseus spiral arm rich in open clusters cluster ngc 7790 three cepheid members close proximity possibility cg cas might be member of ngc 7790 raised by @xcite support star count analysis @xcite detailed star counts indicate otherwise proper motion data @xcite cepheid in corona of berkeley 58 @xcite frolov not probable cluster member probable distance 3 kpc to cg cas berkeley 58 existing proper motion data precise conclusive evidence cluster membership cg cas present study initiated examine case detail strong evidence cg cas likely member of berkeley 58 calibrator for cepheid period - luminosity relation variety observations obtained for investigation table 1 presents photoelectric photometry for bright members berkeley 58 obtained kitt peak national observatory 1981 1982 august 1984 augustdata acquired using 1p21 photomultipliers ubv filter sets pulse - counting photometers 40.4-m 20.9-m 1.3-m telescopes at kitt peak uncertainties cepheid standard internal errors @xmath10 stars brighter than estimated external errors stars similar magnitude stars identified by numbering fig . 1 2000 co - ordinates in 2mass survey number observations each star in column 7 table 1 centred at 2000 co - ordinates ra = @xmath14 dec = + 60@xmath15@xmath16@xmath17 top image location cg cas cluster core lower image photoelectrically observed stars national geographic society - palomar observatory sky atlas made california institute of technology with grants national geographic society | 0.136344 | 769 | 398 | , Saving $0.0 in GPT-4. |
54,895 | we report the detection in the x rays of two weak seyfert 2 s ( ngc3393 and ngc4941 ) with the italian dutch satellite bepposax .
these are among the first sources observed in a sample of 12 seyfert 2 s which are being studied within the bepposax core program , in an effort to probe the putative torus at high x - ray energies , calibrate an isotropic luminosity indicator for absorbed nuclei , and determine the distribution of torus thicknesses , @xmath0 .
both a compton thick spectrum , with a reflected power law and a large equivalent width iron line , and a compton thin spectrum , with the intrinsic power law transmitted through a large column density absorber , provide acceptable fits to both sources , with some preference for the latter model in the case of ngc4941 . the high initial detection rate in our program points to a large final x - ray sample . | a variety of observations have provided evidence that seyfert 2 galaxies host a nuclear active source which is similar to that observed in seyfert 1 s , but is obscured along our line of sight by a pc - scale gaseous torus ( see antonucci 1993 for a review ) .
the most elementary version of such a `` unified model '' assumes seyfert 2 s and seyfert 1 s to be identical physical objects , while the orientation of the line of sight with respect to the obscuring torus accounts for all the observed differences between the two classes .
x - ray data have played a major role among the observations supporting this model .
exosat first , and ginga later clearly proved that the x - ray emission of several sy2 s is characterized by a power law spectrum similar to that observed in sy1 s ( photon index @xmath1 1.7 ) , with a cutoff at low energies due to photoelectric absorption by gas of column density between 10@xmath2 and 10@xmath3 @xmath4 ( e.g. awaki et al .
1991 ; koyama 1992 ; nandra & pounds 1994 ) .
such a dense absorbing column has been identified with the torus .
hard x - ray spectra were recognized as a powerful potential tool to test and probe various issues on active nuclei .
the direct component of the sy2 x - ray emission ( i.e. the emission at energies above the absorption cutoff ) provides direct information on the intrinsic luminosity of the active nucleus ; more specifically , it can be regarded as the only isotropic indicator of the luminosity of sy2 nuclei .
the knowledge of the intrinsic luminosity should allow to test whether sy2 and sy1 nuclei indeed have the same average power ( `` strong '' version of the unified model ) , or there is some systematic luminosity difference ( barcons et al .
1995 ; falcke et al .
the distribution of absorbing column densities @xmath0 should provide information on the geometry and physics of the obscuring material .
eventually , by assessing the distribution of luminosities , spectral indices , and column densities among sy2 s , the origin of the x - ray background should be tightly constrained without assuming any specific model for the spectra of sy2 s ( setti & woltjer 1989 ; comastri et al .
1995 ; zdziarski et al .
1995 ; barcons et al . 1995 ) .
however , such goals have not been fully addressed yet .
the available hard x - ray spectra of sy2 s are affected by selection effects towards x - ray bright sources , or sources known to be bright in either uv or ir .
ginga surveys on sy2 s , such as awaki et al .
( 1991 ) , koyama ( 1992 ) and mulchaey et al .
( 1992 ) , contain a large fraction of markarian galaxies and/or sources bright in the heao1 survey .
therefore , such studies are not representative of the full sy2 population , but they rather sample the high luminosity tail and/or the sy2 s absorbed by low column densities .
more recently , asca observations have improved the sy2 census by providing hard x - ray spectra of a larger number of sources . as a result , sy2 s with lower luminosities and larger absorbing column densities
have been detected ( e.g. ueno et al .
1996 ; awaki et al .
1996 ; makishima et al .
1994 ) . yet
, bias effects are still significant and the statistics very limited .
also the paucity of highly absorbed ( @xmath5 ) , compton thick sy2 s ( only four candidates were known at the beginning of our program , see e.g. matt 1997 and references therein ) probably does not reflect the real abundance of such objects , but is rather a consequence of a selection effect , as the x - ray spectrum of these sources is seen only in reflection and , therefore , it is orders of magnitude fainter than in other seyferts .
we have undertaken a program of observations with bepposax , the italian dutch x - ray satellite , aimed at properly assessing the distribution of luminosities and absorbing column densities in a sample of sy2 s as little biased as possible with respect to orientation effects .
objects suitable for such a survey were drawn from the seyfert sample of maiolino & rieke ( 1995 ) .
this sample is extracted from the revised shapley ames catalogue of galaxies , limited in total blue magnitude ( b@xmath6 13.2 ) , and seyfert nuclei are identified according to the properties of their ( narrow ) emission line spectrum .
although , as discussed in maiolino & rieke , the narrow line spectrum is not a completely isotropic indicator of the nuclear activity , this seyfert sample is much less biased than others , both in terms of luminosity of the nuclear source and of properties of the host galaxy .
there are 54 objects classified as seyfert 2 s in the maiolino & rieke sample ; intermediate seyferts could arise from a peculiar geometry ( see below ) , and we have not considered them . out of the 54 ,
22 have already been observed by asca , and for almost all of them hard x - ray information has been published ( see polletta et al .
1996 and references therein ; see also the heasarc online database ) .
we have prioritized the remaining targets in terms of their [ oiii ] flux , under the assumption that the [ oiii ] emission is a sufficient approximation to an isotropic measure of the nuclear luminosity .
if the [ oiii ] emission were not completely isotropic , our initial sample would be biased in favour of objects with lower optical absorption and , perhaps , lower values of @xmath0 .
the initial sample includes 12 objects , to be observed during the first year of bepposax ; we aim at the completion of the sample within the satellite lifetime ; we are also collecting optical data both from archives and from new observations . according to the `` two tori '' picture of maiolino & rieke ( 1995 ) ,
the seyfert nuclei are surrounded by a pc - scale torus , oriented at random with respect to the host galaxy , and responsible for the obscuration of the x rays and of the broad lines , and by a larger and less opaque screen ( @xmath7 @xmath4 ) , coplanar with the galaxy , and responsible for the intermediate seyfert types . within this scenario
our initial sample might be somewhat biased in favour of face - on seyfert 2 s , but would be free from biases with respect to the inner , thicker torus . in any case , our data would allow us to detect a possible bias by looking for an inverse correlation between @xmath0 and [ oiii ] at constant nuclear luminosity , i.e. , at constant intrinsic @xmath8 .
such a correlation would be important _ per se _ , since it would give information on the height of the torus ; and could be corrected for in order to restore the usefulness of our results for the goals of the program .
the x - ray astronomy satellite bepposax is a joint project of the italian space agency ( asi ) and the netherlands agency for aerospace programs ( nivr ) developed by a consortium of institutes in italy and the netherlands , including the space science department of esa ( ssd ) .
the scientific payload comprises four narrow field instruments [ nfi : low energy concentrator spectrometer ( lecs ) , medium energy concentrator spectrometer ( mecs ) , high pressure gas scintillation proportional counter ( hpgspc ) , and phoswich detector system ( pds ) ] , all pointing in the same direction , and two wide field cameras ( wfc ) , pointing in diametrically opposed directions perpendicular to the nfi common axis .
a detailed description of the entire bepposax mission can be found in butler & scarsi ( 1990 ) and boella et al .
( 1997a ) .
the mecs ( see below ) consists of three equal units , with a field of view of 28 arcmin radius , working range 1.310 kev , energy resolution @xmath18% and angular resolution @xmath10.7 arcmin ( fwhm ) at 6 kev .
the effective area at 6 kev is 155 @xmath9 ( boella et al .
1997b ) .
our targets were observed with the nfi for @xmath115 ksec starting on january 8.92 ut , 1997 ( ngc3393 ) , and for @xmath129 ksec starting on january 22.18 ut , 1997 ( ngc4941 ) . in both cases , only the mecs provided useful data , since the lecs exposure was much shorter due to the switching off of the instrument over the illuminated earth , and the hpgspc and pds were not sensitive enough for such low fluxes . for each source ,
the three mecs were aligned , equalized , and summed ; the extraction radius was chosen equal to the suggested value of 4 arcmin ; the background was taken from much longer exposures of the empty sky , and again extracted in a 4 arcmin radius centered approximately on the same detector coordinates as the source .
the net source counts are 104@xmath1016 ( 6.7 @xmath11 ) and 307@xmath1024 ( 13 @xmath11 ) for ngc3393 and ngc4941 , respectively . after grouping the total counts in energy bins with a minimum of 20 counts each , and discarding any event outside the 210 kev interval because of poor calibrations
, we carried out the spectral analysis with xspec , complemented with the release 97.1 of the bepposax response matrices .
we have tried three spectral models , all including a gaussian line plus , respectively : a power law , a fraction of which is transmitted through cold matter ( transmission model ) , a power law with zero absorption ( scattering model ) , and a reflected spectrum due to the reprocessing of a power law by cold matter ( reflection model ) .
the torus surrounding the nucleus is assumed to be compton thin in the transmission scenario , compton thick in the remaining cases ; in these latter cases a mirror is required , which can be either grey ( scattering ) or energy dependent ( reflection ) . in the case of ngc3393 the small number of bins does not allow an elaborate modeling . in order to cope with the low statistics we have frozen to standard values as many parameters as possible ; in particular , we have kept the intrinsic spectral index at 1.7 , which is the `` universal '' value outside a few schwarzschild radii .
the scattering model gives an acceptable @xmath12 ( 6.1 with 8 d.o.f . ) , but requires a hot , ionized plasma for the mirror ; if we assume that all iron is h like or he like , we find that the equivalent width of the line , ew , is larger than 4.5 kev at the 90% level ; such a value is incompatible with cosmic abundances for all but the lowest column densities ( matt et al .
the best fit parameters of the remaining models ( @xmath12 of 2.1 and 3.8 with 6 and 7 d.o.f .
, respectively ) are shown in table 1 . 0.5truecm .spectral fit parameters for ngc3393 , for transmission and reflection models , respectively .
frozen parameters are : photon indices ( 1.7 ) , line energy ( 6.4 kev ) , line width ( 0.1 kev ) .
normalizations are in photons/@xmath9/s / kev at 1 kev for power laws , and in photons/@xmath9/s for lines ; line ew are in kev , and are calculated with respect to the observed continuum ; column densities are in units of @xmath13 atoms/@xmath9 ; model fluxes are in units of @xmath14 erg/@xmath9/s in the 210 kev range ; luminosities in 10@xmath15 erg / s in the 210 kev range , after correction for absorption and assuming @xmath16=50 km / s / mpc ; the confidence intervals are 90% with one interesting parameter .
[ cols= " < , < , < " , ] the observed spectra of the two galaxies with the best fitting models superposed on them are given in figs . 1 and 2 for ngc3393 ( reflection scenario ) and ngc4941 ( transmission scenario ) , respectively .
while a firmer basis for discussion will be provided only by the completion of the program , it is tempting to put forward a few comments .
first of all , for the first targets we have a detection rate of 100% ( at the time of writing five further observations were performed , and they too resulted in detections ) : this is at least a suggestion that at the level of a few @xmath17 erg/@xmath9/s nearby seyfert 2 s are frequently visible at medium hard x - ray energies .
more troublesome appear the prospects of measuring the column density @xmath0 with a reasonable precision ; while we are certain that ngc3393 and ngc4941 both have @xmath0 larger than several times @xmath18 @xmath4 , with the x - ray data alone and with the available statistics we can not discriminate between a compton thin or a compton thick torus .
this uncertainty entails a large uncertainty on the intrinsic x - ray luminosity .
previous studies have pointed to the far infrared luminosity ( fir ) or to the [ oiii ] luminosity as isotropic indicators of the `` true '' luminosity of the nucleus ( e.g. mulchaey et al . 1994 ; awaki 1997 ) . in our case , assuming that sy1 s and sy2 s are intrinsically identical , assuming that @xmath19 and @xmath20}=100 $ ] for sy1 s ( mulchaey et al .
1994 ) , and assuming that both galaxies have compton thin tori , we find that ngc4941 is underluminous in x rays by a factor of @xmath21 , and ngc3393 by @xmath22 . on the other hand , assuming that both galaxies have compton thick tori , and we see only 1% of the intrinsic flux as reflected flux , we find reasonable agreement with the sy1 normalization . with many more data points of similar quality we will be able to check the self
consistency of the unified model . for an independent test of it , however , we would have to measure @xmath0 and the intrinsic @xmath8 without a priori assumptions ; this would require longer exposures , in order to acquire better statistics in the 210 kev range with the mecs , and to obtain detections or significant upper limits at energies @xmath23 20 kev with the pds , where in the reflection scenario a compton hump would be expected .
we are grateful to all the people who at all levels of responsibility have worked over the years at the bepposax project .
we thank in particular fabrizio fiore , for his help with the detector responses .
this research has made use of saxdas linearized and cleaned event files ( rev0 ) produced at the bepposax science data center .
two of us ( m.s . and r.m . ) acknowledge the support of the asi grant asi95rs120 . | observations seyfert 2 galaxies host nuclear active source similar to seyfert 1 s obscured by pc - scale gaseous torus antonucci 1993 elementary unified model assumes seyfert 2 s and 1 s identical objects orientation line of sight obscuring torus accounts for differences x - ray data major model exosat ginga proved x - ray emission of sy2 s power law spectrum similar to sy1 s ( photon index @xmath1 1.7 ) cutoff at low energies due to photoelectric absorption by gas column density between 10@xmath2 and 10@xmath3 @xmath4 dense absorbing column identified with torus hard x - ray spectra powerful tool to test active nuclei direct sy2 x - ray emission at energies above absorption cutoff provides information on intrinsic luminosity active nucleus only isotropic indicator of luminosity of sy2 nuclei knowledge intrinsic luminosity test sy2 and sy1 nuclei same average power or systematic luminosity difference distribution of absorbing column densities @xmath0 information on geometry physics of obscuring materialassessing distribution of luminosities spectral indices column densities among sy2 s origin x - ray background without specific model for spectra sy2 s ( setti & woltjer 1989 comastri 1995 zdziarski barcons 1995 ) goals not fully addressed hard x - ray spectra of sy2 s affected by selection effects towards x - bright sources surveys on sy2 s awaki 1991 koyama 1992 mulchaey contain large markarian galaxies sources bright in heao1 survey studies not representative of full sy2 population sample high luminosity tail sy2 s absorbed by low column densities asca observations improved sy2 census hard x - ray spectra of larger number sources sy2 s with lower luminosities larger absorbing column densities detected ( ueno 1996 awaki 1996 makishima 1994 bias effects significant statistics limited paucity of highly absorbed compton thick sy2 s four candidates known reflect real abundance objects consequence of selection effect x - ray spectrum sources seen only in reflection fainter than in other seyfertsundertaken program observations with bepposax italian dutch x - ray satellite assessing distribution luminosities absorbing column densities in sample sy2 s orientation effects objects drawn from seyfert sample maiolino & rieke ( 1995 ) sample extracted from revised shapley ames catalogue galaxies limited in total blue magnitude ( b@xmath6 13.2 ) seyfert nuclei identified narrow emission line spectrum spectrum not isotropic indicator nuclear activity seyfert sample less biased luminosity nuclear source properties host galaxy 54 objects seyfert 2 s in intermediate seyferts could arise from peculiar geometry not considered 22 observed by asca all hard x - ray information published polletta et al . 1996 heasarc online database prioritized remaining targets in [ oiii ] flux emission isotropic measure nuclear luminosity if emission not isotropic initial sample would biased objects with lower optical absorption lower values of @xmath0 initial sample includes 12 objects observed during first year of bepposax aim completion within satellite lifetime collecting optical data from archives new observations`` two tori '' picture of maiolino & rieke 1995 ) seyfert nuclei surrounded by pc - scale torus oriented random host galaxy responsible for obscuration x rays broad lines larger less opaque screen ( @xmath7 @xmath4 ) coplanar with galaxy responsible for intermediate seyfert types initial sample might be biased face - on seyfert 2 s free from biases inner thicker torus data detect possible bias inverse correlation between @xmath0 and [ oiii ] at constant nuclear luminosity constant intrinsic @xmath8 correlation important information on height torus could corrected restore usefulness results for goals program x - ray astronomy satellite bepposax joint project italian space agency netherlands agency for aerospace programs consortium institutes italy netherlands space science department of esa scientific payload comprises four narrow field instruments [ pointing same direction two wide field cameras opposed directions perpendicular to nfi common axis detailed description bepposax mission in butler & scarsi ( 1990 ) boella et al . ( 1997a )mecs three units field view 28 arcmin radius working range 1.310 kev energy resolution @xmath18% angular resolution @xmath10.7 arcmin at 6 kev effective area at 6 kev 155 @xmath9 boella et al . 1997b targets observed for @xmath115 ksec january 8.92 ut 1997 ngc3393 @xmath129 ksec january 22.18 ut 1997 ngc4941 mecs provided data lecs exposure shorter switching hpgspc pds not sensitive for low fluxes source three mecs aligned equalized summed extraction radius equal to 4 arcmin background taken from longer exposures empty sky extracted in 4 arcmin radius same detector coordinates source net source counts 104@xmath1016 ( 6.7 @xmath11 ) 307@xmath1024 ( 13 @xmath11 ) for ngc3393 ngc4941 counts in energy bins minimum 20 counts each discarding event outside 210 kev interval poor calibrations spectral analysis with xspec release 97.1 bepposax response matrices tried three spectral models gaussian line plus power law law with zero absorption reflected spectrumtorus surrounding nucleus assumed thin transmission scenario thick remaining cases mirror required grey scattering or energy dependent reflection ngc3393 small number bins elaborate modeling low statistics frozen to standard values parameters kept intrinsic spectral index 1.7 universal value outside radii scattering model acceptable @xmath12 6.1 8 d.o.f requires hot ionized plasma for mirror all iron h like or he like equivalent width line larger than 4.5 kev at 90% level incompatible with cosmic abundances lowest column densities best fit parameters remaining models @xmath12 2.1 3.8 with 6 7 d.o.f shown in table 1 fit parameters for ngc3393 transmission reflection models frozen parameters photon indices ( 1.7 line energy ( 6.4 kev line width ( 0.1 kev )normalizations in photons/@xmath9/s / kev at 1 kev for power laws photons/@xmath9/s for lines line ew in kev calculated observed continuum column densities units @xmath13 atoms/@xmath9 model fluxes units @xmath14 erg/@xmath9/s 210 kev range luminosities 10@xmath15 erg / s 210 kev range correction absorption @xmath16=50 km / s / mpc confidence intervals 90% one parameter observed spectra of two galaxies fitting models in figs . 1 and 2 for ngc3393 ngc4941 completion program first targets detection rate of 100% five observations resulted in detections @xmath17 erg/@xmath9/s nearby seyfert 2 s visible at medium hard x - ray energies troublesome measuring column density @xmath0 precision ngc3393 ngc4941 have @xmath0 larger than @xmath18 @xmath4 x - ray data discriminate between compton thin or compton thick torus uncertainty large uncertainty on intrinsic x - ray luminositystudies pointed far infrared luminosity or [ oiii ] luminosity as isotropic indicators of true luminosity nucleus mulchaey 1994 ; awaki 1997 ) assuming sy1 s and sy2 s identical @xmath19 and @xmath20}=100 $ ] for sy1 s both galaxies have compton thin tori ngc4941 underluminous in x rays by factor @xmath21 ngc3393 by @xmath22 both galaxies have compton thick tori 1% intrinsic flux as reflected flux agreement with sy1 normalization more data points similar check self consistency of unified model independent test measure @xmath0 and intrinsic @xmath8 without assumptions longer exposures better statistics in 210 kev range with detections upper limits at @xmath23 20 kev reflection scenario compton hump expected grateful to at bepposax project thank fabrizio fiore for help with detector responses research saxdas linearized cleaned event files bepposax science data center support of asi grant asi95rs120 | 0.508685 | 3,408 | 1,761 | , Saving $0.1 in GPT-4. |
54,904 | it is shown , that retardation in the @xmath0-quenching in the parker s dynamo model leads to parametric resonance .
this result is observed in the numerical simulations and can be reproduced in the simple analytic model .
the other interesting effect in the model with retardation is a appearance of the long - term processes with the period much larger than the typical time of retardation .
0.5 cm * keywords : * dynamo , instabilities , magnetic fields , turbulence . * the parametric resonance in the @xmath1-dynamo models * maxim reshetnyak _ institute of the physics of the earth ras + b.gruzinskaya 10 , moscow , 123995 , russia + _ email _ : m.reshetnyak@gmail.com_ | the main idea of the dynamo theory , which is believed can explain existence of the magnetic fields observed in cosmos , is that kinetic energy of the conductive motions is transformed into the energy of the magnetic field .
magnetic field generation is the threshold phenomenon : it starts when magnetic reynolds number @xmath2 reaches its critical value @xmath3 .
after that magnetic field grows exponentially up to the moment , when it already can feed back on the flow .
description of transition from the linear regime when influence of the magnetic field onto the flow is negligible to the nonlinear regime , when magnetic energy can exceed the kinetic energy orders of magnitude ( like it is in the planetary cores ) is a subject of the modern researches @xcite . as a result , even after quenching the saturated velocity field is still large enough , so that @xmath4 .
moreover , velocity field taken from the nonlinear problem ( when the exponential growth of the magnetic field stopped ) can still generate exponentially growing magnetic field providing that the feed back of the magnetic field on the flow is omitted ( kinematic dynamo regime ) @xcite .
it appears , that problem of stability of the full dynamo equations including induction equation , the navier - stokes equation with the lorentz force differs from the stability problem of the single induction equation with the given saturated velocity field taken from the full dynamo solution : stability of the first problem does not provide stability of the second one .
moreover , some regimes close to the case 1 from the geodynamo benchmark @xcite are stable in contrast to the solutions with the periodical boundary conditions in space and influence of the boundary conditions can be important @xcite .
one of the simple explanations of this phenomenon ( at least for some regimes ) was offered in @xcite .
using parker s dynamo model for the thin disk it was shown , that the @xmath0-effect , taken from the nonlinear oscillating saturated problem , can still generate exponentially growing magnetic field .
the origin of this effect is closely related to the parametric resonance . here
we develop these ideas and show how the phase shift in the @xmath0-quenching can effect on the behaviour of the generated magnetic field .
one of the simplest dynamo models used in the solar and galactic applications is the parker s one - dimensional model @xcite ( see its development for the galactic dynamo in @xcite ) : @xmath5 where @xmath6 and @xmath7 are azimuthal components of the vector potential and magnetic field , @xmath8 is a kinetic helicity , @xmath9 is a dynamo number , which is a product of the amplitudes of the @xmath0- and @xmath10-effects and primes denote derivatives with respect to a coordinate . for the galaxy
the only one left coordinate is cylindrical polar coordinate @xmath11 . for the thin shells , which we will have in mind in this paper ,
is a latitude @xmath12 .
equation ( [ sys11 ] ) is solved in the interval @xmath13 with the boundary conditions @xmath14 and @xmath15 at @xmath16 . system ( [ sys11 ] ) has growing solution , when @xmath17 .
putting nonlinearity of the form @xmath18 in ( [ sys11 ] ) , where @xmath19 is a magnetic energy , gives quasi - stationary solutions for the positive @xmath20 , see about various forms of nonlinearities in @xcite .
the property of the nonlinear solution is mostly predetermined by the form of its first eigenfunction .
the main result of @xcite was , that simultaneous solution of equations for @xmath21 ( [ sys11 ] , [ non ] ) and similar equations for the new magnetic field @xmath22 : @xmath23 with the same @xmath0 can lead to the exponentially growing @xmath22 , when the field @xmath24 is already saturated .
it happens when @xmath24 oscillates , and initial conditions for @xmath24 and @xmath22 are slightly different .
this effect is very similar to what was observed in the more sophisticated models @xcite , @xcite , @xcite .
below we consider this effect in more details .
here we return to the system ( [ sys11 ] ) .
so as @xmath0 is a function of the magnetic field which oscillates , in the general case we are in a position to expect appearance of the parametric resonance for @xmath24 , as well .
now we consider the more general form of the nonlinearity with retardation @xmath25 : @xmath26 and find how @xmath25 effects on the magnetic field generation .
results for @xmath27 and @xmath28 shown in fig .
[ fig1 ] are quite unexpected : in spite of the fact , that @xmath0 depends on the squared magnetic field , the mean magnetic energy is not symmetric on @xmath29 ( @xmath25 is in units of process s half period @xmath30 ) .
after some decrease of the magnetic field amplitude for @xmath31 magnetic field starts to increase being periodical , see fig .
[ fig2 ] . the sharp decrease at @xmath32 ( see fig .
[ fig2]c ) takes place up to the values at @xmath33 , accompanied by the long - term modulation with the period 8 times larger than the period of the original oscillation of @xmath34 @xmath35 at @xmath36 .
the amplitude of this oscillation increases with increase of @xmath25 , see fig .
[ fig2]d - e .
the amplitude of the new oscillation starts to change at @xmath37 , resembling beating , see fig [ fig2]e . on retardation @xmath25 in @xmath0-quenching .
@xmath38 is normalized in such a way , that @xmath39 .
@xmath25 is measured in units of @xmath35 .
, width=264 ] increase of @xmath25 up to 0.1 leads to the change of the form of the radial @xmath40 and azimuthal @xmath7 components from sinusoidal to the saw - shaped form , see fig .
[ fig3 ] , accompanied with delay of @xmath41 from @xmath7 . for the larger @xmath25 the phase shift @xmath42 decreases .
increase of @xmath25 to 0.33 leads to the growth of the magnetic field and appearance of the pike - like extremums .
this region of @xmath25 corresponds to the parametric resonance . -.4 cm -0.5 cm -2.5 cm -0.5 cm -2.5 cm -0.75 cm 0.0 cm the further increase of @xmath25 leads to the sharp decrease of the magnetic field amplitude , see figs .
[ fig1 ] , [ fig2 ] , [ fig3 ] . the form of the curve is close to that one for @xmath43 with one exception : the new large period appears .
note , that in yoshimura ( 1978a , 1978b ) such long - term modulation was riched at @xmath44 , see review of some other sources of the long - term variations in the solar dynamo in @xcite .
the different behaviour of @xmath45 for @xmath46 and @xmath47 may be explained by the accumulation of the time shift between the magnetic field and @xmath0 .
this suggestion is supported by the fact that for @xmath48 behaviour of @xmath45 is very close to that one for @xmath49 .
-0 cm -1 cm -0.1 cm up to the moment we did not consider how the magnetic field depends on @xmath12 . the dipole solution is a wave propagating from the poles to the equator with maximum at the middle latitudes for @xmath36 , see fig .
increase of @xmath25 leads to the stripe - like solution , i.e. more contrast changes of the sign of the fields and increase of the magnetic field at the poles fig .
the long term variations are well resolved in fig .
to consider how parametric resonance appears we follow @xcite and present solution for ( a , b ) in the form of the waves : @xmath50 , @xmath51 , @xmath52 , and @xmath53 we get how generation depends on @xmath25 . then putting it in ( [ sys11 ] ) we get two equations .
equation for @xmath54 does not include @xmath25 , so we consider only production of @xmath55
. then @xmath56 . if @xmath57 , where @xmath58 , then @xmath59 is unstable .
the integral for @xmath60 gives : @xmath61 then @xmath62 and for @xmath63 ( what corresponds to the phase shift between the components @xmath64 ) and @xmath65 @xmath66 grows , and parametric resonance appears , see fig . [ fig5 ] .
note , that this analysis explains small decrease of @xmath67 at small @xmath25 ( see fig .
[ fig1 ] ) which corresponds to the positive values of @xmath42 in fig .
[ fig5 ] . -0
here we considered the only one form of the nonlinearity for the fixed value of the dynamo number @xmath20 as a function of the time lag @xmath25 in the @xmath0-quenching .
however even this simple model demonstrates variety of effects : change of the form of the poloidal and toroidal fields , regions of the weaker and stronger fields , appearance of the long - term variations .
the quite natural choice of @xmath25 can lead to the sharp increase of the magnetic field amplitude concerned with the parametric resonance .
this explanation does not contradict to the simple linear analysis presented above .
it is very tempting to correspond appeared in simulations the long - term periodicities with that ones of the solar activity larger than the main 22 years period .
the difficulty is to justify the choice of the particular @xmath25 which should be obtained from the solution of the more sophisticated nonlinear model . | main idea dynamo theory magnetic fields in cosmos kinetic energy of conductive motions transformed into magnetic field magnetic field generation is threshold phenomenon starts when magnetic reynolds number @xmath2 reaches critical value @xmath3 after magnetic field grows exponentially back on flow transition from linear regime to nonlinear regime magnetic energy kinetic energy planetary cores subject of modern researches @xcite after quenching saturated velocity field still large enough @xmath4 velocity field from nonlinear problem growth stopped can generate growing magnetic field feed back on flow omitted ( kinematic dynamo regime ) problem of stability of full dynamo equations induction navier - stokes lorentz force differs from single induction equation with saturated velocity field from full dynamo solution stability of first problem provide stability second some regimes close to case 1 from geodynamo benchmark @xcite stable to solutions with periodical boundary conditions offered in @xcite using parker s dynamo model for thin disk @xmath0-effect from nonlinear oscillating saturated problem can generate exponentially growing magnetic fieldorigin effect related to parametric resonance develop ideas show phase shift in @xmath0-quenching behaviour magnetic field simplest dynamo models parker s dimensional model @xcite galactic dynamo @xmath5 azimuthal components vector potential magnetic field kinetic helicity dynamo number product of amplitudes @xmath0- @xmath10-effects primes denote derivatives coordinate galaxy left coordinate is cylindrical polar coordinate @xmath11 thin shells latitude @xmath12 equation ( sys11 solved in interval @xmath13 with boundary conditions @xmath14 @xmath15 at @xmath16 system sys11 growing solution @xmath17 nonlinearity of form @xmath18 in sys11 ] @xmath19 magnetic energy gives quasi - stationary solutions for positive @xmath20 nonlinearities property nonlinear solution predetermined by form first eigenfunction result simultaneous solution of equations for @xmath21 for new magnetic field @xmath22 @xmath23 same @xmath0 to exponentially growing @xmath22 when field @xmath24 saturated happens when @xmath24 oscillates initial conditions for differenteffect similar to sophisticated models @xcite consider effect details return to system sys11 @xmath0 function of magnetic field oscillates expect parametric resonance for @xmath24 consider nonlinearity with retardation @xmath25 : @xmath26 @xmath25 effects magnetic field generation results for @xmath27 and @xmath28 in fig unexpected @xmath0 depends on squared magnetic field mean magnetic energy not symmetric on @xmath29 @xmath25 in units of process s half period @xmath30 ) after decrease magnetic field amplitude @xmath31 magnetic field periodical sharp decrease at @xmath32 up to values at @xmath33 long - term modulation period 8 times larger than original oscillation of @xmath34 @xmath35 @xmath36 amplitude oscillation increases with @xmath25 amplitude new oscillation at @xmath37 beating retardation @xmath25 in @xmath0-quenching. @xmath38 normalized @xmath25 measured in units of @xmath35width=264 increase @xmath25 up to 0.1 form radial @xmath40 azimuthal @xmath7 components from sinusoidal to saw - shaped form fig delay @xmath41 from @xmath7 larger @xmath25 phase shift @xmath42 decreases increase @xmath25 to 0.33 growth magnetic field pike - like extremums region corresponds parametric resonance -.4 cm -0.5 cm -2.5 cm -0.75 cm 0.0 cm increase @xmath25 decrease magnetic field amplitude figs fig1 fig3 form curve close @xmath43 exception new large period appears yoshimura long - term modulation riched at @xmath44 solar different behaviour @xmath45 @xmath46 @xmath47 explained by time shift between magnetic field @xmath0 @xmath48 behaviour @xmath45 close to @xmath49 -0 cm -1 cm -0.1 cm magnetic field @xmath12 dipole solution wave propagating from poles to equator maximum middle latitudes for @xmath36 increase @xmath25 leads stripe - like solution contrast changes increase magnetic field at poles long term variations resolved in fig .consider parametric resonance follow @xcite present solution for a b ) waves @xmath50 generation depends on @xmath25 sys11 ] two equations equation @xmath54 include @xmath25 consider production of @xmath55 @xmath56 @xmath57 @xmath58 @xmath59 unstable integral for @xmath60 gives @xmath61 @xmath62 @xmath63 phase shift components @xmath64 @xmath65 @xmath66 grows parametric resonance appears fig . fig5 ] analysis explains small decrease of @xmath67 at small @xmath25 corresponds positive values @xmath42 fig fig5 ] considered nonlinearity for fixed value dynamo number @xmath20 function of time lag @xmath25 in @xmath0-quenching model demonstrates effects change form poloidal toroidal fields regions weaker stronger fields long - term variations natural choice of @xmath25 increase magnetic field amplitude parametric resonance explanation contradict linear analysis tempting correspond long - term periodicities with solar activity larger than main 22 years period difficulty justify choice @xmath25 solution sophisticated nonlinear model | 0.318745 | 2,186 | 1,130 | , Saving $0.1 in GPT-4. |
54,908 | zauner s conjecture asserts that @xmath0 equiangular lines exist in all @xmath1 complex dimensions . in quantum theory ,
the @xmath0 lines are dubbed a sic , as they define a favoured standard informationally complete quantum measurement called a sic - povm .
this note supplements a. j. scott and m. grassl [ j. math .
phys .
* 51 * ( 2010 ) , 042203 ] by extending the list of published numerical solutions .
we provide a putative complete list of weyl - heisenberg covariant sics with the known symmetries in dimensions @xmath2 , a single solution with zauner s symmetry for every @xmath3 and solutions with higher symmetry for @xmath4 and @xmath5 .
a computer study of zauner s conjecture @xcite ( sic - povm conjecture @xcite ) was reported by the author with m. grassl in ref .
@xcite . in that article we presented a putative complete list of weyl - heisenberg covariant sics in dimensions @xmath6 .
there are now compelling reasons @xcite to extend this list as far as possible , and this note sets out to do this .
background material is laid out in appendix [ sec : sic ] , including all notation and definitions used in the following .
our computer search has so far discovered sic fiducial vectors in all dimensions @xmath3 and we are confident that the list is complete for @xmath6 , where the search reported in ref .
@xcite exhausted all of @xmath7 . for @xmath8 , we continue the search by narrowing its scope to eigenspaces of the unitary matrices corresponding to @xmath9 ( zauner s matrix ) and @xmath10 ( defined below ) .
every known weyl - heisenberg covariant sic in dimensions @xmath11 can be generated by a fiducial vector from one of these eigenspaces ( up to extended clifford transformations ) and , within the eigenspaces , it is unlikely that solutions have been missed for @xmath2 ( the dimensions searched so far ) .
our list is therefore believed complete in these dimensions for weyl - heisenberg covariant sics with the known principal symmetries , @xmath9 and @xmath10 .
moreover , the lack of evidence from ref .
@xcite for other symmetries , besides those arising in conjunction with @xmath9 and @xmath10 , indicates that it could be complete for all weyl - heisenberg covariant sics . in higher dimensions ,
searching an eigenspace of zauner s matrix discovers a solution for all @xmath3 by c. a. fuchs , m. c. hoang and b. c. stacey using our code on the chimera supercomputer . ] ; searching eigenspaces ( or coneigenvectors ) of higher order symmetry matrices discovers solutions in dimensions @xmath4 and @xmath5 .
the solutions in dimensions @xmath12 are available with 150 digits of precision in the article source files ( ref .
@xcite contains solutions for @xmath6 ) where labelling generally follows the order that they are found .
appendix [ sec : num ] outlines the numerical approach to our computer search .
symmetry structures in the new solutions can be observed .
each fiducial vector solution @xmath13 that generates a weyl - heisenberg sic also generates an entire orbit of related fiducial vectors under the action of the extended clifford group : @xmath14 ( see appendix [ sec : sic ] or ref .
@xcite ) . in the table
we list the known unique fiducial vector orbits in each dimension .
note that the length of each orbit will be @xmath15 unless there is a symmetry present in @xmath16 , described by its stabiliser @xmath17\in\pec(d):|\bra{\phi}u\ket{\phi}|=1\}\:,\end{aligned}\ ] ] where @xmath18\coloneq \{e^{i\xi}u\}_{\xi\in\r}$ ] .
we then have @xmath19 .
the stabiliser of each orbit is also given .
the numerical initial fiducial vector @xmath16 and its stabiliser @xmath20 could always be chosen in a way that the stabiliser elements take the form @xmath21 ( where @xmath22 $ ] as in eq .
[ eq : eclassdefn ] ) and , therefore , only the matrices @xmath23 are quoted .
note that when @xmath1 is even , however , the subgroup of @xmath24 that defines @xmath25 will generally have an order that is a multiple of @xmath26 . in dimensions @xmath11
each stabiliser is a cyclic group of order a multiple of 3 and the vast majority , up to group conjugacy , have the symmetry described by zauner s order-3 unitary @xmath27 , where @xmath28 exceptions occur in dimensions @xmath29 , @xmath30 , @xmath31 , @xmath32@xmath33 , @xmath34 , @xmath35 , @xmath36@xmath37 , @xmath38@xmath39 , @xmath40@xmath33 , @xmath41 , in which case solutions stabilised by the order-3 unitary @xmath42 exist , as indicated in parentheses , where @xmath43 we can find additional symmetries present in the following dimensions : 1 .
@xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 : order-2 unitary @xmath57 , where @xmath58 2 .
@xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 , @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 : order-2 anti - unitary @xmath71 , where @xmath72 3 .
@xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 : order-9 unitary @xmath79 , where @xmath80 it is found that @xmath81 for @xmath82 but @xmath83 for @xmath84 , where @xmath85 denotes group conjugacy .
4 . @xmath86 , @xmath87 , @xmath88 , @xmath89 , @xmath90 , @xmath91 : order-6 anti - unitary @xmath92 , where @xmath93 note that @xmath94 since @xmath95 for @xmath96 .
we can include @xmath97 in this series and generalise to all dimensions @xmath98 , @xmath60 , @xmath99 , @xmath61 , @xmath62 , @xmath100 , by instead setting @xmath101 \begin{pmatrix } 0 & 1 \\ 1 & d+l \end{pmatrix } , & \text{otherwise } \end{cases}\end{aligned}\ ] ] then @xmath102 when @xmath103 and @xmath104 otherwise , since @xmath105 for @xmath106 . in odd dimensions , i.e. when @xmath107 , we find that @xmath108 since @xmath109 for @xmath110 \begin{pmatrix } m(2m+1)+2 & -m(2m+1)-1 \\ -1 & 1 \end{pmatrix } , & \text{otherwise } \end{cases}\end{aligned}\ ] ] therefore real fiducial vectors are expected to exist for all dimensions @xmath111 .
gaps in the current classification are marked by ?
s , which either indicate an unexplored dimension ( to be filled in at later dates ) or note the likely presence of an unknown general symmetry . [ cols="^,^,^,^,^,^ " , ] we thank marcus appleby , ingemar bengtsson , john coates , steven flammia , christopher fuchs , markus grassl , gary mcconnell , shayne waldron and jon yard for their encouragement and insights .
9 a. j. scott and m. grassl , sic - povms : a new computer study , arxiv:0910.5784 ; j. math .
phys .
* 51 * ( 2010 ) 042203
. g. zauner , quantendesigns
grundzge einer nichtkommutativen designtheorie ( in german ) , ph.d .
thesis ( university of vienna , 1999 ) ; english translation : quantum designs : foundations of a noncommutative design theory , int .
j. quantum inform . * 09 * ( 2011 ) 445 ; available online : http://www.gerhardzauner.at/qdmye.html .
j. m. renes , r. blume - kohout , a. j. scott and c. m. caves , symmetric informationally complete quantum measurements , arxiv : quant - ph/0310075 ; j. math .
phys .
* 45 * ( 2004 ) 2171 .
d. m. appleby , h. yadsan - appleby and g. zauner , galois automorphisms of a symmetric measurement , arxiv:1209.1813 ; quantum inf .
comput .
* 13 * ( 2013 ) 672 .
m. appleby , s. flammia , g. mcconnell and j. yard , generating ray class fields of real quadratic fields via complex equiangular lines , arxiv:1604.06098 .
m. appleby , s. flammia , g. mcconnell and j. yard , sics and algebraic number theory , arxiv:1701.05200 .
i. bengtsson , the number behind the simplest sic - povm , arxiv:1611.09087 .
d. m. appleby , sic - povms and the extended clifford group , arxiv : quant - ph/0412001 ; j. math .
phys .
* 46 * ( 2005 ) 052107 . | in simple terms , a sic - povm ( or sic ) may be identified as a set of @xmath0 unit vectors in @xmath1 complex dimensions , written @xmath112 , that are equiangular latexmath:[\ ] ] see ref .
@xcite for more .
another characterisation of sics comes from design theory .
a finite set @xmath216 is called a _
complex projective @xmath217-design _ if @xmath218 where @xmath219 is the haar measure . in these terms , sics are precisely equivalent to _ tight _ complex projective 2-designs , which are 2-designs that meet the absolute bound on their size : @xmath220 . all 2-designs with @xmath221 are necessarily sets of equiangular lines ( sics ) and these are the only 2-designs with this structure .
our numerical approach makes use of the _ welch bound _ : for any finite @xmath222 , and any positive integer @xmath217 , @xmath223 with equality if and only if @xmath224 is a @xmath217-design . setting @xmath225 , @xmath226 and @xmath227
, we can translate eq .
( [ eq : welchbound ] ) to our case : _ for any @xmath13 , @xmath228 with equality if and only if @xmath143 is a fiducial vector for a weyl - heisenberg covariant sic .
_ we search for sic fiducial vectors by minimising the lhs of the inequality in eq .
( [ eq : sicbound ] ) , parametrised as a function of the real and imaginary parts of the @xmath1 complex numbers @xmath229 , until the bound on the rhs is met .
in practice , this is done by repeating a local search from different initial trial vectors until a solution is found .
the local search uses a c++ implementation of l - bfgs . generally , local minima are found quickly but
an enormous number of trials are required to locate the global minima , i.e. the sics . in order to exhaust the search space to arrive at a putative complete list of unique sics , we choose initial points uniformly at random under the unitarily invariant haar measure on @xmath230 ( using the hurwitz parametrisation ) .
once enough solutions are found , and their precision refined ( using gmp / mpfr multi - precision arithmetic ) , we identify unique extended - clifford orbits by comparing each solution @xmath143 to every other translated solution @xmath231 with @xmath232 .
this brute force approach to the classification , while stubborn , gives the symmetries ( i.e. the stabiliser @xmath20 ) essentially for free , if @xmath143 is also compared to @xmath233 .
any symmetry conjectured to exist in general dimensions allows us to lower the dimension of the search space without much effort . for suppose @xmath234 = @xmath235 for some unitary @xmath236 of order @xmath237 , @xmath238 .
then the projector onto the eigenspace with eigenvalue @xmath239 , where we will have @xmath240 integer , is @xmath241 and we can search this subspace by simply replacing @xmath143 with @xmath242 in eq .
( [ eq : sicbound ] ) .
there is a similar approach to anti - unitary symmetries .
suppose instead that @xmath243 = @xmath235 for some unitary @xmath236 with @xmath244 , where @xmath245 denotes complex conjugation of matrix components in the standard basis .
we must have @xmath246 and may in fact assume @xmath247 for coneigenvalues .
now given that the projector onto the eigenspace of the unitary @xmath248 with eigenvalue 1 is @xmath249 it is easy to check that the ( unnormalised ) @xmath250 solves our coneigenvalue problem .
we can therefore replace @xmath143 with @xmath251 in eq .
( [ eq : sicbound ] ) to search the set of coneigenvetors .
contents of the folder ` .. /solutions ` + in 51, ... ,350 in a, ... ,z | sic - povm @xmath0 unit vectors in @xmath1 complex dimensions @xmath112 equiangular latexmath @xcite characterisation sics design theory finite set @xmath216 complex projective @xmath217-design @xmath218 @xmath219 haar measure sics equivalent to complex projective 2-designs meet absolute bound on size : @xmath220 2-designs with @xmath221 equiangular lines sics only structure numerical approach welch bound for finite @xmath222 positive integer @xmath217 @xmath223 if @xmath224 is @xmath217-design setting @xmath225 @xmath226 @xmath227 translate eq . welchbound @xmath13 @xmath228 if @xmath143 is fiducial vector for weyl - heisenberg covariant sic search for sic fiducial vectors by minimising lhs of inequality in eq . sicbound @xmath1 complex numbers @xmath229 until bound on rhs met local search from initial trial vectors until solution found local search uses c++ implementation l - bfgs local minima found quickly trials required to locate global minima sicsexhaust search space unique sics choose initial points random under invariant haar measure on @xmath230 hurwitz parametrisation solutions found precision refined arithmetic identify unique extended - clifford orbits comparing solution @xmath143 to @xmath231 with @xmath232 brute force approach classification gives symmetries stabiliser @xmath20 ) free if @xmath143 compared to @xmath233 symmetry lower dimension search space without effort @xmath234 = @xmath235 unitary @xmath236 @xmath237 projector onto eigenspace with eigenvalue @xmath239 @xmath240 is @xmath241 search subspace replacing @xmath143 with @xmath242 similar approach to anti - unitary symmetries @xmath243 = @xmath235 for unitary @xmath236 @xmath244 @xmath245 complex conjugation matrix components have @xmath246 assume @xmath247 for coneigenvalues projector onto eigenspace of unitary @xmath248 eigenvalue 1 is @xmath249 check unnormalised @xmath250 solves coneigenvalue problem replace @xmath143 with @xmath251 search coneigenvetorscontents folder /solutions 51 ... ,350 a ,z | 0.180256 | 920 | 506 | , Saving $0.0 in GPT-4. |
54,925 | the detection of fast quasi - periodic variability from accreting black holes and neutron stars has been used to constrain their masses , radii , and spins .
if the observed oscillations are linear modes in the accretion disks , then bounds can be placed on the properties of the central objects by assuming that these modes are locally sub - keplerian .
if , on the other hand , the observed oscillations correspond to non - linear resonances between disk modes , then the properties of the central objects can be measured by assuming that the resonant modes are excited at the same radial annulus in the disk . in this paper
, we use numerical simulations of vertically integrated , axisymmetric hydrodynamic accretion disks to provide examples of situations in which the assumptions implicit in both methods are not satisfied .
we then discuss our results for the robustness of the mass and spin measurements of compact objects from variability studies . | x - ray variability studies of accreting neutron stars and black holes offer a tool for constraining the masses and spins of the compact objects themselves . in the least model - dependent argument ,
the radial extent of the region responsible for variability observed with a given characteristic frequency is assumed to be at most equal to the size of a keplerian orbit with the same orbital frequency .
additionally , in order for this region to be able to sustain long - lived oscillatory modes , it must be larger than the radius of the innermost stable circular orbit around the compact object .
this requirement then places an upper bound on the mass of the object and a lower bound on its spin ( see , e.g. , kluzniak , michelson , & wagoner 1990 ; mushotzky , done , & pounds 1993 ; miller , lamb , & psaltis 1998 ; strohmayer 2001 ; for a discussion see psaltis 2005 ) .
the assumption implicit in the above argument is that the keplerian frequency corresponds to the fastest dynamical timescale at any radius in the accretion disk .
this is usually justified by the fact that gravity is the strongest force in the system ( see , however , alpar & psaltis 2008 ) .
indeed , the study of normal modes in a viscous accretion disk shows that the frequency of the lowest order , linear , hydrodynamic mode excited at any radius has to be smaller than the local keplerian frequency ( kato 2001 ) . observations of pairs of quasi - periodic oscillations from black holes with frequencies in small - integer ratios , however , have challenged the notion that these oscillations correspond to linear hydrodynamic modes ( strohmayer 2001 ; abramowicz & kluzniak 2001 ) .
such frequency ratios are reminiscent of non - linear interactions , which can preferentially amplify only those _ high - order _ harmonics of the normal modes that are in resonance . if this is the prevailing mechanism for compact - object variability , then meaningful bounds on or measurement of their masses and spins can only be obtained within the context of a detailed model for the oscillations . indeed , assuming that the two modes in resonance are dynamical hydrodynamic modes in the disk and using an independent dynamical measurement of the mass of the black hole may lead to an estimate of the black - hole spin ( abramowicz & kluzniak 2001 ; abramowicz et al .
the assumption implicit in this last argument is the fact that both modes are excited and trapped in the same radial annulus in the accretion disk . in this paper
, we use numerical simulations of axisymmetric viscous accretion disks in order to assess the validity of the two main assumptions discussed above : that the highest frequency of oscillation of a radial annulus in an accretion disk is the local keplerian frequency and that interactions may occur only between modes that are locally excited .
we show that neither of these two assumptions are formally justified .
we find that oscillatory modes that are excited close to the innermost stable circular orbit generate traveling sound waves that propagate radially to tens of schwarzschild radii .
the presence of such modes has been predicted in analytical studies of diskoseismic modes ( kato 2001 ) .
our study shows , however , that they can grow to such amplitudes that they can dominate the power spectrum of variability of the accretion disks over extended regions . as viewed from any one radial annulus outside their excitation region , these oscillations are highly super - keplerian . moreover , non - linear couplings may lead to resonance between the traveling modes and those that are excited locally .
both results have important implications for the phenomenological constraints imposed on the masses and spins of compact objects by the observation of high - frequency quasi - periodic oscillations , which we will discuss in detail below .
we used the numerical algorithm described in milsom & taam ( 1996 ) in order to solve the hydrodynamic equations that describe the time evolution of one - dimensional viscous accretion disks . for completeness , we summarize here briefly the relevant equations and assumptions .
the conservation of mass is described by the continuity equation @xmath0 where @xmath1 is the column density through the entire thickness of the accretion disk at radius @xmath2 and @xmath3 is the radial velocity .
the conservation of linear and angular momentum are described , respectively , by the equations @xmath4 and @xmath5 where @xmath6 is the sum of the vertically integrated gas and radiation pressures , @xmath7 is the angular velocity in the flow , @xmath8 is the local azimuthal ( keplerian ) dynamical frequency , and @xmath9 is the kinematic viscosity . finally , the conservation of energy is described by the equation @xmath10=u_r\frac{\partial p}{\partial r}+\nu\sigma \left(r\frac{\partial \omega}{\partial r}\right)^2 -\frac{4a c t^4}{3\kappa_0\sigma}\;,\ ] ] where @xmath11 is the internal energy per unit mass , @xmath12 is the mid - plane temperature in the disk , @xmath13 is the radiation constant , @xmath14 is the speed of light , and @xmath15 is the sum of the electron - scattering and free - free opacities of the flow .
this set of differential equations is augmented by a set of algebraic equations that determine the gravitational potential of the central star , the equation of state , and the kinematic viscosity in the flow .
we specify a pseudo - newtonian gravitational potential through @xmath16 where @xmath17 is the schwarzchild radius that corresponds to the mass of the star , @xmath18 .
we use an equation of state that corresponds to a perfect gas in local thermodynamic equilibrium with radiation such that @xmath19 and @xmath20 where @xmath21 is the gas constant , @xmath22 is the mean molecular weight in the flow , @xmath13 is the radiation constant , and @xmath23 is the vertical scale height of a disk in hydrostatic equilibrium .
finally , we use the @xmath24-prescription @xmath25 for the kinematic viscosity , which ensures thermal and viscous stability . in all numerical simulations we set the mass of the central object to 10 @xmath26 and specified the solution by the mass accretion rate @xmath27 in units of the eddington accretion rate @xmath28 and the viscosity parameter @xmath24 .
the domain of solution , initial conditions , boundary conditions , and method of solution are described in detail in milsom & taam ( 1996 ) .
the global timing properties of our numerical simulations have been described in detail by milsom & taam ( 1996 ) . in summary , for a given value of the viscosity parameter @xmath24 , simulations with low ( typically @xmath29 ) accretion rates result in lightcurves with noisy power spectra , simulations with intermediate accretion rates exhibit large amplitude global modes at frequencies comparable to the maximum radial epicyclic frequency in the flow and its harmonics , and simulations at even larger accretion rates ( typically @xmath30 ) exhibit no significant variability ( although epicyclic oscillations were always visible in subsequent 2d hydrodynamic simulations ; see milsom & taam 1997 ) .
the main goal of our study is to investigate the radial propagation of the traveling modes excited in the various simulations , in order to quantify the extent of the domain in which they dominate the variability .
for this reason , we performed fast fourier transforms ( ffts ) of a number of physical quantities such as the radiation flux , local mass accretion rate , vertically integrated column density , etc . , evaluated at different radii in the flow . in order to resolve the fastest possible oscillations we used a time resolution of 2 ms , and in order to resolve the power - spectral peaks of potential oscillations of high coherence we used segments of 4 s length .
finally , following standard procedure , we averaged the ffts of four sequential segments in every simulation , in order to improve the statistics , and binned the power density spectra in frequency space .
hereafter , we will be discussing the power density spectra of the vertically integrated column density in the disk .
figure [ fig : pds ] shows a typical example of the power spectrum of the vertically integrated density @xmath1 evaluated at three different radii in the accretion disk for the simulation with an accretion rate of @xmath31 and a viscosity parameter of @xmath32 .
we can clearly see that the accretion disk is variable over a wide range of timescales and , similar to the observed systems , exhibits both broad - band noise and quasi - periodic oscillations .
indeed , at @xmath33 a large number of oscillation peaks can be seen , most of which are overtones to a fundamental frequency . however , contrary to all expectations , most of the quasi - periodic oscillations that appear in the power spectra occur at frequencies larger than the local keplerian frequencies , indicated in the figure with the vertical dashed lines .
figure [ fig : freq ] shows the frequency of the dominant quasi - periodic oscillation of the vertically integrated density as a function of radius , for the simulation with @xmath34 and a viscosity parameter @xmath32 .
the radius dependence of the oscillation frequency depicted in this figure shows three distinct regions that appear in varying degrees in all simulations .
first , there is a wide region in the inner accretion disk where the dominant oscillation frequency is constant and nearly equal to the maximum of the radial epicyclic frequency .
this is surrounded by a second region in which the oscillation frequency is radius - dependent , rapidly decreasing , but super - keplerian .
finally , in a third , outer region , the oscillation frequency is radius - dependent and nearly equal to the radial epicyclic ( or local keplerian ) frequency .
the radial extent of the regions in the accretion disk in which the constant - frequency and variable - frequency modes live depends strongly on the mass accretion rate and rather weakly on the viscosity parameter @xmath24 , as shown in figure [ fig : rad ] . at accretion rates smaller than a few hundredths of the eddington critical value ,
the oscillation frequencies at all radii are nearly equal to the local epicyclic frequencies . as the accretion rate increases beyond @xmath35 , the radial extent of the constant - frequency mode also increases , reaching radii @xmath36 for accretion rates comparable to @xmath31 ; beyond that radius , only the variable frequency mode exists .
the constant frequency modes that dominate the variability in the inner accretion disk at accretion rates @xmath37 are neither excited nor trapped in that region .
they are in fact traveling inertial - acoustic modes that are generated close to the innermost stable circular orbit and propagate outwards ( see also chen & taam 1995 ) .
this is illustrated in figure [ fig : radiustime ] , which shows the radius - time plot of the vertically integrated column density for the simulation with an accretion rate of 0.1@xmath38 and a viscosity parameter @xmath32 .
we investigated the presence of oscillatory radial modes in numerical simulations of axisymmetric , viscous accretion disks . in agreement with earlier efforts ( see chen & taam 1995 )
, we find that traveling modes are generated close to the innermost stable circular orbit of the flow and propagate outwards to several tens of schwarzschild radii , without getting attenuated .
as viewed from any given annulus in the accretion disk , these waves appear as density oscillations with frequencies that are locally super - keplerian .
the presence of such super - keplerian oscillations in the simulations implies that we can not use the frequency of an observed oscillation to set a bound on the extent of the region in the accretion flow that is responsible for the modulated emission .
for example , if we were to assume , as is normally done , that the oscillation with a frequency @xmath39 hz shown in figure [ fig : freq ] is limited by the local keplerian frequency , i.e. , that @xmath40 , then we would infer that the region responsible for this oscillation is smaller than @xmath41 however , this limit would seriously underestimate the radial extent of the region where the @xmath42 hz mode resides , which is in fact four times larger ( see fig . [
fig : freq ] ) .
this is a direct consequence of the fact that the traveling mode is locally super - keplerian in the entire inner region of the disk .
it is important to emphasize here that , although the constant - frequency modes are super - keplerian in the region of the accretion disk where they are traveling , they are sub - keplerian near the innermost stable circular orbit , where they are excited .
this is true for all modes in all simulations that we explored . as a result , even though we can not use the frequency of the mode to set an upper bound on the radial extent of the region in which it resides , we can still constrain the mass and the spin of the compact object by requiring that the mode is sub - keplerian in the region where it is excited .
our simulations can only recover the excitation and propagation of axisymmetric radial oscillations .
because of this , we can not simulate the possible non - linear coupling of these to other , azimuthal or vertical , modes .
however , the presence of large amplitude traveling modes in the simulated disks suggests that the non - linear coupling and resonances may not necessarily occur between modes that are excited in the same region in the accretion disk .
indeed , figure [ fig : freq ] shows that the @xmath42 hz mode is in 2:1 resonance with an azimuthal keplerian mode at @xmath43 , in 3:2 resonance with an azimuthal mode at @xmath44 , etc . as a result , it is not justifiable to use the frequencies of observed resonant oscillations and infer the properties of the compact objects assuming that they correspond to modes in the same region in the accretion flow ( as in , e.g. , abramowicz & kluzniak 2001 ) .
finally , it is worth pointing out that our results are valid for geometrically thin , hydrodynamic accretion disks with an alpha viscosity .
traveling inertial - acoustic modes such as those we studied here are not visible in time - dependent simulations of magnetohydrodynamic accretion disks ( see , e.g. , hawley & krolik 2001 ; armitage & reynolds 2003 ; but see also arras , blaes , & turner 2006 and chan et al .
this is probably due to the combined effects of the magnetorotational instability , which damps such radial epicyclic modes , and the presence of fully developed turbulence that quickly destroys the coherence of such traveling modes .
it is not clear , at this point , what physics needs to be included ( or perhaps what assumptions need to be relaxed ) for the numerical simulations of magnetohydrodynamic disks to show large - amplitude oscillations that are ubiquitous in nature .
independent of whether the hydrodynamic inertial - acoustic modes survive in realistic simulations , our results still serve as a proof of principle that high amplitude , super - keplerian traveling modes may , in principle , exist in the inner regions of accretion flows . | x - ray variability studies neutron stars black holes tool constraining masses spins compact objects - radial extent region responsible for variability frequency equal to keplerian orbit same region sustain long - oscillatory modes must be larger than radius innermost stable circular orbit around object requirement places upper bound on mass lower bound on spin kluzniak wagoner 1990 mushotzky pounds 1993 miller 1998 strohmayer 2001 psaltis 2005 assumption keplerian frequency corresponds to fastest dynamical timescale at any radius in accretion disk justified by gravity strongest force system alpar psaltis 2008 study of normal modes in viscous accretion disk shows frequency lowest order linear hydrodynamic mode at radius smaller than local keplerian frequency ( kato 2001 ) observations quasi - periodic oscillations from black holes with small - integer ratios challenged notion to linear hydrodynamic modes strohmayer 2001 2001 frequency ratios reminiscent of non - linear interactions amplify high - order harmonics of normal modes resonanceprevailing mechanism for compact - object variability bounds on measurement masses spins obtained detailed model for oscillations assuming two modes in resonance dynamical hydrodynamic modes disk independent dynamical measurement of mass black hole estimate - hole spin ( abramowicz kluzniak 2001 assumption both modes excited trapped in same radial annulus in accretion disk numerical simulations of axisymmetric viscous accretion disks assess validity two assumptions highest frequency oscillation is local keplerian frequency interactions between modes locally excited neither assumptions justified oscillatory modes excited close to circular orbit generate traveling sound waves propagate radially to tens of schwarzschild radii presence modes predicted in studies diskoseismic modes ( kato 2001 ) study can grow dominate power spectrum variability accretion disks regions oscillations highly super - keplerian non - linear couplings lead to resonance between traveling modes excited locally results implications for phenomenological constraints on masses spins of compact objects by high - frequency quasi - periodic oscillationsused numerical algorithm milsom & taam 1996 ) solve hydrodynamic equations time evolution of viscous accretion disks summarize relevant equations assumptions conservation of mass by continuity equation @xmath0 @xmath1 column density accretion disk @xmath2 @xmath3 radial velocity conservation of linear angular momentum by equations @xmath4 and @xmath5 @xmath6 sum gas radiation pressures @xmath7 angular velocity @xmath8 local azimuthal dynamical frequency @xmath9 kinematic viscosity conservation of energy by equation @xmath10=u @xmath11 internal energy per unit mass @xmath12 mid - plane temperature in disk @xmath13 radiation constant speed of light @xmath15 sum electron - scattering free - free opacities of flow differential equations augmented by algebraic equations gravitational potential of central star equation of state kinematic viscosity in flowspecify pseudo - newtonian gravitational potential @xmath16 @xmath17 schwarzchild radius mass star @xmath18 use equation of state perfect gas local thermodynamic equilibrium radiation @xmath19 @xmath20 @xmath21 gas constant @xmath22 mean molecular weight radiation constant @xmath23 vertical scale height disk hydrostatic equilibrium use @xmath24-prescription @xmath25 kinematic viscosity ensures thermal viscous stability set mass central object to 10 @xmath26 solution mass accretion rate @xmath27 eddington accretion rate @xmath28 viscosity parameter @xmath24 domain of solution initial conditions boundary conditions method solution described milsom & taam ( 1996 ) global timing properties simulations described milsom & taam ( 1996 ) viscosity parameter @xmath24 simulations low @xmath29 accretion rates result lightcurves noisy power spectra intermediate rates large amplitude global modes maximum radial epicyclic frequency larger accretion rates @xmath30 no significant variability epicyclic oscillations visible 1997 main goal study investigate radial propagation traveling modes simulations quantify domain variabilityperformed fast fourier transforms of physical quantities radiation flux local mass accretion rate vertically integrated column density evaluated at different radii resolve fastest oscillations used time resolution 2 ms power - spectral peaks used segments 4 s length averaged ffts of four segments simulation improve statistics binned power density spectra in frequency space discussing power density spectra of vertically integrated column density in disk figure shows power spectrum of vertically integrated density @xmath1 evaluated at three radii in accretion disk simulation with accretion rate @xmath31 viscosity parameter @xmath32 accretion disk variable over timescales exhibits broad - band noise quasi - periodic oscillations at @xmath33 oscillation peaks overtones to fundamental frequency quasi - periodic oscillations occur at frequencies larger than local keplerian frequencies figure : freq ] shows frequency of dominant quasi - periodic oscillation of vertically integrated density function of radius simulation with @xmath34 viscosity parameter @xmath32radius dependence oscillation frequency shows three regions varying degrees simulations wide region inner accretion disk dominant frequency constant nearly equal to radial epicyclic frequency surrounded by second region radius - dependent super - keplerian third outer region radius - dependent nearly equal to radial epicyclic frequency radial extent regions constant - variable - frequency modes depends on mass accretion rate viscosity parameter @xmath24 figure at accretion rates smaller than hundredths eddington critical value oscillation frequencies nearly equal to local epicyclic frequencies accretion rate increases beyond @xmath35 radial extent constant - frequency mode increases reaching radii @xmath36 for rates comparable to @xmath31 beyond variable frequency mode exists constant frequency modes variability inner accretion disk at accretion rates @xmath37 excited trapped traveling inertial - acoustic modes circular orbit propagate outwards 1995 illustrated in figure radius - time plot vertically integrated column density simulation accretion rate 0.1@xmath38 viscosity parameter @xmath32 investigated oscillatory radial modes in simulations of axisymmetric viscous accretion disks agreement with earlier efforts chen & taam 1995 traveling modes generated close to innermost stable circular orbit propagate to tens of schwarzschild radii without attenuated these waves appear as density oscillations with frequencies locally super - keplerian presence super - keplerian oscillations implies can not use frequency oscillation to set bound on extent region flow responsible for modulated emission assume oscillation with frequency @xmath39 hz limited by local keplerian frequency @xmath40 infer region responsible for oscillation smaller than @xmath41 limit underestimate radial extent of region where @xmath42 hz mode resides four times larger fig direct consequence of traveling mode locally super - keplerian in entire inner region disk constant - frequency modes super - keplerian in region sub - keplerian near innermost stable circular orbit true for all modes in all simulationscan not use frequency mode to bound radial extent region can constrain mass spin of compact object by requiring mode sub - keplerian in region excited simulations recover excitation propagation of axisymmetric radial oscillations can not simulate possible non - linear coupling to other azimuthal or vertical modes presence large amplitude traveling modes in simulated disks suggests non - linear coupling resonances may not occur between modes excited in same region accretion disk figure shows @xmath42 hz mode in 2:1 resonance with azimuthal keplerian mode at @xmath43 3:2 with azimuthal mode at @xmath44 not justifiable to use frequencies of observed resonant oscillations infer properties compact objects assuming correspond to modes in same region accretion flow results valid for geometrically thin hydrodynamic accretion disks with alpha viscosity traveling inertial - acoustic modes not visible in time - dependent simulations of magnetohydrodynamic accretion disks due to magnetorotational instability radial epicyclic modes fully developed turbulence destroys coherence of traveling modesnot clear physics assumptions relaxed for numerical simulations of magnetohydrodynamic disks large - amplitude oscillations ubiquitous in nature independent hydrodynamic inertial - acoustic modes simulations results proof high amplitude super - keplerian traveling modes exist in inner regions accretion flows | 0.567292 | 3,329 | 1,669 | , Saving $0.1 in GPT-4. |
54,926 | we report on turbulent dynamo simulations in a spherical wedge with an outer coronal layer .
we apply a two - layer model where the lower layer represents the convection zone and the upper layer the solar corona .
this setup is used to study the coronal influence on the dynamo action beneath the surface . increasing the radial coronal extent gradually to three times the solar radius and changing the magnetic reynolds number
, we find that dynamo action benefits from the additional coronal extent in terms of higher magnetic energy in the saturated stage . the flux of magnetic helicity can play an important role in this context . | the solar magnetic field is produced by a dynamo operating beneath the solar surface . in the convection zone ,
the turbulent motions driven by convection and shear from the differential rotation are able to amplify and organize the magnetic field .
these fields manifest themselves at the solar surface in form of sunspots , in which the field is so strong that the heat transported by convection is suppressed , leading to dark spots on the solar disk .
one important feature of these sunspots is their latitudinally dependent occurrence . averaging over longitude
, one finds the typical behavior of equatorward migration of the underlying mean magnetic field .
this behavior gives clear evidence for the existence of a dynamo mechanism in the sun .
in dynamo theory the @xmath0-effect plays an important role , because this effect describes the amplification of large - scale magnetic field in the absent of shear . in the sun , it is believed that the @xmath0-effect produces new poloidal field from the toroidal field .
how strong its contribution for the production of toroidal field is , is currently under debate ( see e.g. * ? ? ?
. numerical simulations of turbulent dynamos have shown that the @xmath0-effect can be catastrophically quenched at high magnetic reynolds numbers ( see * ? ? ? * for a detailed discussion ) .
one possible loophole to alleviate the quenching is to allow for magnetic helicity fluxes @xcite . in this context
, it is very important to choose a realistic boundary condition for the dynamo , which allows for magnetic helicity fluxes .
preventing a transport of helicity out of the simulation domain may influence the dynamo solution and the strength of the amplified magnetic field . besides the magnetic helicity fluxes ,
commonly used boundary conditions for the magnetic field such as vertical field or perfect conductor restrict the dynamo and the magnetic field to certain solutions .
this led to the development of the so - called `` two - layer model '' , where we combine the lower layer , in which the magnetic field is generated by dynamo action , representing the solar convection zone with a upper , force - free layer , representing the solar corona .
our first application of the two - layer model led to the formation of structures reminiscent of plasmoid- and cme - like ejections , driven by a forced turbulent dynamo @xcite and , subsequently , by a self - consistently driven convective dynamo @xcite in the lower layer .
this indicates that the dynamo can be directly responsible for producing coronal ejections and form structures in the solar corona .
but in the recent work of @xcite , the authors find that differential rotation and the migration of the mean magnetic field can be also influenced by the presence of a coronal layer . in this paper
, we investigate how the corona influences the dynamo action .
we use spherical polar coordinates , @xmath1 .
the setup is the same as that of @xcite , where we use a spherical wedge with @xmath2 , @xmath3 , corresponding to @xmath4 latitude , and @xmath5 , corresponding to a longitudinal extent of @xmath6 .
@xmath7 is the radius of the sun and @xmath8 is the outer radius of the coronal layer . at @xmath9
the domain is divided at into two parts .
the lower layer mimics the convection zone , where a magnetic field gets generated by turbulent dynamo action .
the upper layer is a nearly force - free part , which mimics the solar corona .
we solve the following equations of compressible magnetohydrodynamics , @xmath10 where @xmath11 is the magnetic vector potential and the magnetic field is defined by @xmath12 , which makes equation ( [ eq : ind ] ) obey @xmath13 at all times .
@xmath14 and @xmath15 are the magnetic diffusivity and the kinematic viscosity , respectively .
@xmath16 is the advective derivative , @xmath17 is the gravitational acceleration , @xmath18 is newton s gravitational constant , and @xmath19 is the mass of the sun . we choose @xmath20 .
@xmath21 is the lorentz force and @xmath22 is the current density , where @xmath23 is the vacuum permeability . the traceless rate - of - strain tensor is defined as @xmath24 , where the semi - colons denote covariant differentiation , @xmath25 is the specific pseudo - enthalpy , with @xmath26 is the isothermal sound speed . as in the work of @xcite ,
the forcing function is only present in the lower layer of the domain .
this means that the forcing function goes smoothly to zero in the upper layer of the domain ( @xmath27 ) .
the function @xmath28 consists of random plane helical transverse waves with relative helicity @xmath29 and wavenumbers that lie in a band around an average forcing wavenumber of @xmath30 .
the forcing function also has a dependence on the helicity which is here chosen to be @xmath31 such that the kinetic helicity of the turbulence is negative in the northern hemisphere and positive in the southern .
more detailed descriptions can be found in @xcite and @xcite .
the magnetic field is expressed in units of the equipartition value , @xmath32 , where @xmath33 , @xmath34 , and @xmath35 denotes an average over @xmath36 , @xmath37 and @xmath38 , i.e. , over the whole dynamo in region .
the fluid and the magnetic reynolds numbers are defined as , @xmath39 their ratio is expressed by the magnetic prandtl number @xmath40 . as an initial condition we use gaussian noise as seed magnetic field in the dynamo region .
our domain is periodic in the azimuthal direction . for the velocity field
we use a stress - free boundary condition on all other boundaries . for the magnetic field
we apply a perfect conductor conditions in both @xmath36 boundaries and the lower radial boundary ( @xmath41 ) . on the outer radial boundary ( @xmath42 ) , we employ vertical field conditions .
we use the pencil code with sixth - order centered finite differences in space and a third - order accurate runge - kutta scheme in time ; see @xcite for the extension of the pencil code to spherical coordinates .
we perform 27 runs where we change @xmath8 and @xmath43 , but keep @xmath44 constant .
the letters for different sets indicate the coronal extents : @xmath45 , 1.5 , 2 , 3 , 1.2 , 1.1 , and 2.5 for sets a f . in the first four sets , we vary @xmath43 from @xmath46 to @xmath47 , for the last three sets we use @xmath48 . for
all runs the turbulent motion in the lower layer of the domain drives dynamo action , which amplifies the magnetic field .
after exponential growth , the field saturates and shows cycles .
the field shows an equatorward migration of the all three magnetic field components , as described in @xcite .
this is caused by an @xmath49 dynamo , where @xmath0 changes sign over the equator @xcite . in figure
[ pcorona ] , we show for all the 27 runs the normalized magnetic field energy @xmath50 as function of magnetic reynolds number @xmath43 . the value for @xmath50 is obtained by averaging in space over the lower layer of the domain @xmath51 and in time over many hundred turnover times in the saturated stage .
the error bars in figure [ pcorona ] reflect the quality of the temporal averaging . with coronal radial extent @xmath8 and magnetic reynolds number @xmath43 .
the solid black line indicates the dynamo region without corona . ] from figure [ pcorona ] , we can deduct two important results .
first , for runs with a corona the magnetic energy peaks at @xmath52 .
this seems to be not the case for runs without a corona . on the other hand , the magnetic energy declines for larger @xmath43 , as was also found by @xcite , which could be related to a change in the onset conditions for the different cases .
second , the magnetic energies for all runs with a coronal extent are larger by a factor of @xmath53 .
it seems that the actual radial size of the coronal extension is not that important as long as there exists a coronal layer .
the run of set f has just a coronal extent of @xmath54 , but the magnetic energy is closer to runs with larger coronal extent than to the one without corona .
magnetic helicity fluxes might be a key to solving this riddle .
however , the outer radial boundary condition in the runs without a corona also allow for magnetic helicity fluxes .
we recall that the simulation with a corona generates large ejections of magnetic helicity @xcite . without possessing a coronal extent
the dynamo might be not able to produce ejection of magnetic helicity and therefore has a much lower magnetic helicity flux through the boundary .
studies on the nature of helicity fluxes in these runs are already on the way ( warnecke et .
al . 2013c , in preparation )
however , magnetic helicity fluxes might be important only much larger magnetic reynolds numbers @xcite .
we have shown that a coronal layer on the top of a dynamo region can support dynamo action .
this is visible through an increase in magnetic energy by adding a corona at the top of the domain and leaving all other parameters the same .
however , it will be necessary to study magnetic helicity fluxes through the surface of the lower layer for these cases to derive any further conclusions .
the two - layer model has been used before to show the impact of a dynamo on coronal properties and generating cme - like ejections @xcite . with this model
is was also possible to generate spoke - like differential rotation and equatorward migration in global convective dynamo simulations @xcite , whereas models without a corona have not been able to reproduce these features in the same parameter regime @xcite .
besides dynamo models , this two - layer approach is successful in combination with stratified turbulence in producing a bipolar magnetic region @xcite as a possible mechanism of sunspot formation . | solar magnetic field produced by dynamo beneath solar surface in convection zone turbulent motions driven by convection shear differential rotation amplify organize magnetic field fields manifest at solar surface in sunspots field strong heat transported by convection suppressed leading to dark spots on solar disk sunspots latitudinally dependent occurrence averaging over longitude typical behavior of equatorward migration of mean magnetic field for dynamo mechanism in sun in dynamo theory @xmath0-effect describes amplification of large - scale magnetic field in absent shear produces new poloidal field from toroidal field contribution for under debate simulations of @xmath0-effect can be quenched at high magnetic reynolds numbers possible loophole to quenching allow for magnetic helicity fluxes important to choose realistic boundary condition for dynamo for magnetic helicity fluxes preventing transport helicity may influence dynamo solution strength of amplified magnetic field boundary conditions for vertical field perfect conductor restrict dynamo magnetic field to certain solutionsled to development two - layer model lower layer magnetic field generated by dynamo action solar convection zone with upper force - free layer solar corona first application led to structures plasmoid- cme - like ejections driven by forced turbulent dynamo @xcite self - consistently driven convective dynamo @xcite lower layer dynamo responsible for producing coronal ejections structures in solar corona recent work differential rotation migration mean magnetic field influenced by coronal layer paper investigate corona influences dynamo action use spherical polar coordinates @xmath1 setup same as spherical wedge with @xmath2 @xmath3 latitude @xmath5 longitudinal extent @xmath7 radius sun @xmath8 outer radius coronal layer @xmath9 domain divided into two parts lower layer mimics convection zone magnetic field generated by turbulent dynamo upper layer nearly force - free part mimics solar corona solve equations of compressible magnetohydrodynamics @xmath10 magnetic vector potential magnetic field defined by @xmath12 @xmath14 @xmath15 are magnetic diffusivity kinematic viscosity@xmath16 advective derivative @xmath17 gravitational acceleration @xmath18 newton s gravitational constant @xmath19 mass sun choose @xmath20 @xmath21 lorentz force @xmath22 current density @xmath23 vacuum permeability traceless rate - of - strain tensor @xmath24 semi - colons covariant differentiation @xmath25 specific pseudo - enthalpy @xmath26 isothermal sound speed forcing function present lower layer domain zero upper layer @xmath27 function @xmath28 random plane helical transverse waves relative helicity @xmath29 wavenumbers average forcing wavenumber @xmath30 forcing function helicity @xmath31 kinetic helicity turbulence negative northern hemisphere positive southern detailed descriptions @xcite magnetic field expressed units equipartition value @xmath32 @xmath33 average over @xmath36 @xmath37 @xmath38 whole dynamo region fluid magnetic reynolds numbers defined @xmath39 ratio expressed magnetic prandtl number @xmath40 gaussian noise seed magnetic field dynamo region domain periodic azimuthal direction velocity field stress - free boundary condition boundariesmagnetic field apply perfect conductor conditions @xmath36 boundaries lower radial boundary outer radial boundary @xmath42 employ vertical field conditions use pencil code sixth - order centered finite differences space third - order accurate runge - kutta scheme time extension pencil code spherical coordinates perform 27 runs change @xmath8 @xmath43 keep @xmath44 constant letters indicate coronal extents @xmath45 1.5 2 3 1.2 1.1 2.5 first four sets vary @xmath43 to last three sets use @xmath48 turbulent motion lower layer drives dynamo action amplifies magnetic field growth field saturates shows cycles equatorward migration three magnetic field components caused @xmath49 dynamo @xmath0 changes equator 27 runs normalized magnetic field energy @xmath50 function magnetic reynolds number @xmath43 value obtained averaging space lower layer @xmath51 time turnover times saturated stage error bars reflect temporal averaging coronal radial extent @xmath8 magnetic reynolds number @xmath43 solid black line indicates dynamo region without corona runs corona magnetic energy peaks @xmath52not case for runs without corona magnetic energy declines for larger @xmath43 related to change onset conditions magnetic energies for runs with coronal extent larger by @xmath53 radial size coronal extension not important coronal layer run of set f has coronal extent @xmath54 magnetic energy closer to runs with larger coronal extent than without corona magnetic helicity fluxes might key to riddle outer radial boundary in runs without corona magnetic helicity fluxes simulation with corona generates large ejections magnetic helicity without coronal extent dynamo helicity lower magnetic helicity flux through boundary studies on helicity fluxes on way ( warnecke et . 2013c magnetic helicity fluxes might be important only larger magnetic reynolds numbers @xcite coronal layer on top dynamo region support dynamo action visible through increase in magnetic energy by adding corona at top parameters necessary to study magnetic helicity fluxes lower layer two - layer model used impact dynamo on coronal properties cme - like ejections generate spoke - like differential rotation equatorward migration in global convective dynamo simulations models without corona not reproduce featuresdynamo models two - layer approach successful stratified turbulence bipolar magnetic region sunspot formation | 0.350997 | 2,244 | 1,134 | , Saving $0.1 in GPT-4. |
54,945 | this paper describes an innovative way to optimize a multivariate classifier , in particular a support vector machine algorithm , on a problem characterized by a biased training sample .
this is possible thanks to the feedback of a signal - background template fit performed on a validation sample and included both in the optimization process and in the input variable selection .
the procedure is applied to a real case of interest at hadron collider experiments : the reduction and the estimate of the multi - jet background in the @xmath0 plus jets data sample collected by the cdf experiment .
the training samples , partially derived from data and partially from simulation , are described in detail together with the input variables exploited for the classification . at present , the reached performance is superior to any other prescription applied to the same final state at hadron collider experiments .
lepton plus jets , multi - jet rejection , svm , multivariate analysis , cdf | a multivariate classifier is an adaptive algorithm trained to identify a signal of interest against other background events on the basis of a set of input variables .
therefore the understanding of the training samples and the input variables selection are two key elements to obtain optimal results .
in this paper we apply the previous paradigm in an innovative way to both the training and the input variable selection of a support vector machine @xcite ( svm ) algorithm .
in particular we obtain an excellent signal - background multivariate classifier when one of the training samples is biased ( i.e. it does not correctly reproduce all characteristics of the signal or of the background samples ) and statistically limited to few thousands of events .
we decided to explore the use of the svm algorithm ( described in section [ sec : svm ] ) because of several advantages with respect to other multivariate techniques .
for example , artificial neural networks , commonly used in high energy physics @xcite , require to arbitrarily set the complexity of the classifier ( i.e. the number of neurons and layers of the net ) , the training may converge to local minima and , usually , large training sets are needed to finely map the input space . on the other hand the svm algorithm , whose basic idea is the identification of the best hyper - plane separating two classes of vectors , has unique solution of the training algorithm , a small number of free parameters and good performance on low statistics training sets , as only a small number of training vectors are exploited in the final solution @xcite . other promising results using svms in high energy physics analyses
are reported in ref .
@xcite . in this work , developed in the framework of the analysis searching for the higgs boson with the collider detector at fermilab ( cdf ) experiment @xcite
, we deal with three machine - learning challenges : the reduction of the effect of a bias in the training set , a robust evaluation of the performance of the trained svm , and the optimal input variable selection .
the key point to achieve all of the above is the scan of the free parameters of the training procedure together with a cross check of the efficiency of each resulting svm ( described in section [ sec : training_set ] ) .
the efficiency cross check is performed both on the training set , with a @xmath1-fold cross validation , and on unclassified events , with a template fitting procedure over the svm output value distribution ( described in section [ sec : method ] ) .
we tested the developed methodology on a toy model ( section [ sec : toy ] ) and finally we applied it to a real physics case ( section [ sec : cdf_set ] ) . in this latest part
we exploited the described template fitting procedure to identify a reliable and optimal input variable selection .
a wide literature discusses the topic of input variable selection but we focused on the identification of the best , minimal set of inputs .
it is clear that a highly discriminative variable will improve the classification power but it has been shown in ref .
@xcite that also the performance of the svm algorithm itself improves when the variables are well chosen , especially if they have very different discrimination power @xcite .
intuitively , the introduction of too many not - significant inputs introduces a noise term to the algorithm , decreasing its performance .
furthermore , a reduced set of variables identifies the most important characteristics of the analyzed process and may decrease the number of needed validation checks . the complete procedure ( described in section [ sec : var_sel ] )
involves the automatic training and performance evaluation with several variables configurations .
the physics case under exam is the reduction and estimate of the multi - jet background contamination in a dataset enriched in leptonic @xmath2 bosons decays , selected in association with hadronic jets .
the dataset , collected at a proton - antiproton collider which was operating at @xmath3 tev ( the tevatron ) , is one of the main investigation channels at hadron collider experiments .
several interesting but rare processes ( i.e. associate @xmath4 production , diboson or top quark production , etc . )
produce a @xmath2 boson in the final state .
the @xmath2 leptonic decay is used to identify a clear event signature out of the overwhelming multi - jet background produced by generic qcd interactions between the proton and antiproton constituents .
hadronic jets faking the lepton and the neutrino identification introduce a significant multi - jet background contamination ( especially for electrons identification algorithms ) .
because of its nature , the multi - jet background is a mixture of detector effects and physics processes .
usually data - driven models obtained with a specific selection enriched in multi - jet events , are used to estimate this contamination .
these models may be statistically limited and the use of a different selection criteria often introduces unexpected biases in the simulated variables .
all this makes the application of multivariate techniques particularly challenging .
the svm is a supervised learning binary classifier whose basic function is the identification of the _ best separating hyper - plane _ between two classes of @xmath1-dimension vectors . given a training set made by two classes of vectors ( i.e. signal and background ) linearly separable , the svm algorithm produces , as a solution , a unique plane defined by the vectors at the boundary of the two classes ; those are the so called _ support vectors_. in the case of non - linear separation , the plane is found in an abstract space , defined by a transformation of the input vectors .
although the transformation can be very complex , it is not necessary to know it exactly , but we just need to know its effect on the scalar product between the vectors , named _
kernel_. finally the cases of not perfect separability of the two samples are solved by introducing a penalty parameter accounting for the contamination .
it is possible to find more details in ref .
@xcite and @xcite , but , for sake of clarity , a short overview of the algorithm is also given in the following .
for the actual , numerical , implementation of the svm algorithm we relied on the ` libsvm ` open source library @xcite .
figure [ fig : svm_linear ] shows how the linear classification problem can be formalized in the minimization of @xmath5 ( with @xmath6 = vector normal to a plane ) with the constraint : @xmath7 and the constant term @xmath8.,scaledwidth=100.0% ] the problem has a unique solution obtained by the maximization of : @xmath9 which are derived by the application of the lagrange multipliers to eq .
[ eq : constraint ] .
the solution identifies @xmath10 for some @xmath11 .
the associated vectors , i.e. the support vectors , are a subset of the training sample that defines the best hyper - plane separating the two input classes ( see figure [ fig : svm_linear ] ) . when not completely separable classes of vectors are present , a _ penalty parameter _
@xmath12 is added to account for the contamination .
the new minimization condition is : @xmath13 with the new constraint ( derived from eq .
[ eq : constraint ] ) : @xmath14 the parameter @xmath12 defines the svm implementation before the training , therefore it represents a hyper parameter of the svm . for any new vector @xmath15 considered for classification
, we evaluate its position @xmath16 with respect to the plane defined by the support vectors @xmath17 and the parameters @xmath18 : @xmath19 where @xmath8 is a constant term of the solution .
the variable @xmath20 is the final output of the svm : its sign gives the signal - background classification and , if the geometry is simple , its value is the distance of a test vector @xmath15 from the classification plane .
as we are going to see in the next paragraph , a non - linear classification is possible only thanks to a not - explicit transformation to a different vector space , where @xmath20 may loose the immediate geometrical meaning .
natively svms are used as binary classifiers but , here , we add a large degree of flexibility by exploiting the full information of the continuous variable @xmath20 .
the svm is used as a dimensionality reducer of the classification problem : the separation power and the correlations among several significant input variables are summarized into one continuous distribution .
non - linearly - separable classes of vectors can be classified by transforming them with an appropriate function , @xmath21 , that maps the elements into another space , usually with higher dimension , where the linear separation is possible . however the identification of @xmath21 is not trivial , therefore the so called _ kernel trick _ is often used . a kernel function , @xmath22 , generalizes the scalar product appearing in eq .
[ eq : lagrange_1 ] ( or eq . [ eq : lagrange_2 ] ) without the need of explicitly knowing @xmath21 .
the kernel is the composition of the mapping @xmath21 with the inner product : @xmath23 the function @xmath24 should satisfy to a general set of rules to be a kernel , but we describe only the _ gaussian _ kernel we used in this work .
it is expressed as : @xmath25 the corresponding @xmath26 is unknown and it maps the input vectors to an infinite dimension space .
the kernel is defined only by one hyper - parameter , @xmath27 , that should be set before the training of the svm .
several multivariate techniques are based on the assumption that the labelled samples used for the classifier training are drawn from the same probability distribution of the unclassified events . in our case of study , where only an approximate and statistically limited model of the background processes is available ( see section [ sec : cdf_set ] and in particular section [ sec : training_set ] for the multi - jet background description )
, we do not expect the previous assumption to hold for every portion of the phase space . to cope with this problem
, we developed an original methodology to evaluate the svm training performance .
section [ sec : svm ] shows that , for each choice of hyper - parameters and training vectors , only one optimal svm solution exists and we need to evaluate the performance of it . as a performance estimator
we use the _ confusion matrix _ of the classifier : the element @xmath28 of the matrix is the fraction of the class @xmath11 classified as member of class @xmath29 .
figure [ fig : confusion ] shows a representation of it in the two - classes case , where one class is labelled as _ background _ and the other as _ signal_. we obtain a reliable estimate of the classifier quality by filling the confusion matrix in two independent ways and combining all the available information . [ cols="^,^",options="header " , ] an example of the use of the developed svm discriminant is reported in ref .
in particular both the possibility to move the svm selection threshold and the fit of the multi - jet contamination over svm output variable @xmath20 are used .
figure [ fig : svm_phx_fit ] shows how the different physics processes contribute to the shape of the svm discriminant used in the selection of the forward electron sample . the multi - jet , or qcd ,
fraction is extracted from the fit together with the total @xmath2 plus jets component .
the svm threshold used for the final signal region identification is @xmath30 instead of the standard value of @xmath31 .
this further decreases the multi - jet contamination and reduces the impact of the systematics related to the estimate of such background .
used during the _ forward _ electron sample selection .
the multi - jet , or qcd , background fraction ( in magenta ) is extracted from the fit together with the total @xmath2 plus jets component ( in green ) .
the remaining physics processes ( in light blue ) are normalized to the expected production cross sections and acceptances .
the svm selection threshold for the final signal region identification is @xmath30 instead that the standard value of @xmath31 .
this further decreases the qcd contamination in the final analysis sample.,scaledwidth=100.0% ]
in this paper we present an innovative method for training optimization and variable selection for a svm multivariate classifier in the problematic case of biased and statistically limited training samples .
the optimization of the classifier is possible thanks to the feedback of a signal - background template fit performed on a validation sample of unclassified events over the svm output distribution @xmath20 ( defined in eq .
[ eq : distance ] ) . by construction ,
the svm output @xmath20 is a variable sensitive to the signal and background contamination .
the optimization algorithm was then applied to an actual hadron collider physics case .
the problem of multi - jet rejection in the @xmath2 plus jets channel has been analyzed in two different cases : a @xmath2 boson decaying to an electron identified in the central and in the forward region of the cdf detector and selected together with two or more hadronic jets .
the optimization algorithm allowed to scan a pool of twenty - four input variables to obtain the minimal combination of them giving the lower background contamination and the higher signal efficiency .
the resulting classifiers exploit eight and six input variables combinations , respectively for the _ central _ and _ forward _ svm classifiers , not directly related to the lepton identification algorithm .
the performance is superior to any present multi - jet rejection algorithm applied to the same final state .
the cdf ii dataset and the specific multi - jet rejection analysis have been a perfect test - bench for the svm optimization algorithm .
however the procedure can be exported to any other problem where the use of multivariate techniques is required but no accurate simulation is available or possible .
we would like to thank the university research association for the support given to f. sforza with the visiting scholars program at fnal in spring 2010 , dr .
giorgio chiarelli and dr .
sandra leone for the support , the guidance and the discussions that contributed to this paper , all the cdf higgs discovery group for the feedback and the suggestions that improved our research and , in particular , dr .
h. wolfe for the review of the manuscript .
l. ametller , l. garrido , g. stimpfl - abele , p. talavera , and p. yepes , discriminating signal from background using neural networks : application to top - quark search at the fermilab tevatron , phys . rev .
d 54 , 1233 ( 1996 ) .
t. aaltonen _
et al _ , ( cdf collaboration ) , search for the standard model higgs boson produced in association with a @xmath32 boson with 7.5 fb@xmath33 integrated luminosity at cdf , phys .
d 86 , 032011 ( 2012 ) . chen , y .- w . and lin , c .- j .
guyon , i. ; nikravesh , m. ; gunn , s. and zadeh , l. ( eds . ) , combining svms with various feature selection strategies , feature extraction , springer berlin heidelberg , 2006 , 207 , 315 .
f. sforza , v. lippi , g. chiarelli , and s. leone , rejection of multi - jet background in a hadron collider environment through a svm classifier , in : nuclear science symposium and medical imaging conference ( nss / mic ) , 2011 ieee , p1404 .
t. aaltonen _
( cdf collaboration ) , search for the standard model higgs boson decaying to a @xmath36 pair in events with one charged lepton and large missing transverse energy using the full cdf data set , phys .
( 2012 ) 111804 .
d. acosta _ et al .
_ ( cdf collaboration ) , measurement of the @xmath39 production cross section in @xmath40 collisions at @xmath41 tev using lepton + jets events with secondary vertex b - tagging , phys
d 71 , ( 2005 ) 052003 .
d. mladeni et al .
feature selection using linear classifier weights : interaction with classification models , proceedings of the 27th annual international acm sigir conference on research and development in information retrieval , ( 2004 ) 234 . | multivariate classifier is adaptive algorithm signal interest against background events input variables understanding of training samples input variables selection key optimal results in paper apply previous paradigm to training input variable selection of support vector machine @xcite ( svm ) algorithm obtain excellent signal - background multivariate classifier when one training biased not reproduce all characteristics signal background samples statistically limited to few thousands events explore svm algorithm advantages other multivariate techniques artificial neural networks require arbitrarily set complexity classifier training may converge to local minima large training sets needed to map input space svm algorithm best hyper - plane separating two classes vectors has unique solution training algorithm small free parameters good performance on low statistics training sets small number training vectors exploited in final solution promising results using svms in high energy physics analyses reported in. @xcite in this work for higgs boson fermilab experiment deal with three machine - learning challenges : reduction of bias in training set robust evaluation of performance of trained svm optimal input variable selectionkey point to is scan of free parameters training procedure cross check of efficiency of each resulting svm described in section training_set efficiency cross check performed on training set @xmath1-fold cross validation unclassified events with template fitting procedure over svm output value distribution method tested methodology on toy model applied to real physics case_set exploited template fitting procedure to identify reliable optimal input variable selection literature discusses input variable selection focused on identification best minimal set of inputs highly discriminative variable classification power performance svm algorithm improves when variables well chosen especially different discrimination power introduction of too many not - significant inputs introduces noise term algorithm performance reduced set of variables identifies important characteristics process may decrease needed validation checks complete procedure involves automatic training performance evaluation with several variables configurations physics case under exam is reduction and estimate of multi - jet background contamination in dataset enriched in leptonic bosons decays with hadronic jets dataset collected at proton - antiproton collider at main investigation at hadron collider experiments rare processesassociate @xmath4 production diboson top quark production produce @xmath2 boson final state @xmath2 leptonic decay clear event signature multi - jet background generic qcd interactions proton antiproton constituents hadronic jets faking lepton neutrino identification introduce significant multi - jet background contamination electrons identification algorithms multi - jet background mixture detector effects physics processes data - driven models specific selection multi - jet events estimate contamination models statistically limited different selection criteria introduces unexpected biases simulated variables makes application multivariate techniques challenging svm supervised learning binary classifier function identification best separating hyper - plane between two classes @xmath1-dimension vectors training set two classes vectors separable svm algorithm produces unique plane defined by vectors boundary _ support vectors_ non - linear separation plane found in abstract space defined by transformation input vectors transformation complex need effect on scalar product between vectors _ kernel_ cases not perfect separability solved introducing penalty parameter accounting for contamination more details in ref . @xcite and @xcite short overview algorithm followingimplementation svm algorithm relied on libsvm open source library @xcite figure svm_linear shows linear classification problem minimization of @xmath5 @xmath6 = vector normal to plane constraint @xmath7 constant term @xmath8.,scaledwidth=100.0% problem unique solution maximization of @xmath9 derived application lagrange multipliers to eq constraint solution identifies @xmath10 @xmath11 associated vectors support vectors subset training sample defines best hyper - plane separating input classes figure svm_linear not separable classes vectors penalty parameter @xmath12 added contamination new minimization condition @xmath13 new constraint @xmath14 parameter @xmath12 defines svm implementation training represents hyper parameter svm new vector @xmath15 classification evaluate position @xmath16 plane defined support vectors @xmath17 parameters @xmath18 : @xmath19 @xmath8 constant term solution variable @xmath20 final output svm sign gives signal - background classification value distance of test vector @xmath15 from classification plane non - linear classification possible not - explicit transformation to different vector space @xmath20 may loose geometrical meaningsvms as binary classifiers add flexibility full information continuous variable @xmath20 svm used as dimensionality reducer classification problem separation power correlations among input variables summarized into continuous distribution non - linearly - separable classes vectors classified by transforming with function @xmath21 maps elements into another space higher dimension linear separation possible identification of @xmath21 not trivial _ kernel trick _ often used kernel function @xmath22 generalizes scalar product in lagrange_1 or_2 without knowing @xmath21 kernel is composition of mapping @xmath21 with inner product : @xmath23 function @xmath24 should satisfy rules be kernel describe only gaussian _ kernel used work expressed as @xmath25 @xmath26 unknown maps input vectors to infinite dimension space kernel defined by one hyper - parameter @xmath27 set before training svm multivariate techniques labelled samples for classifier training from same probability distribution of unclassified events our case approximate limited model of background processes available expect previous assumption to hold for every phase space developed original methodology to evaluate svm training performancesection svm shows hyper - parameters training vectors one optimal svm solution exists evaluate performance performance estimator use confusion matrix classifier element @xmath28 is fraction class @xmath11 member class @xmath29 figure confusion shows representation two - classes case one class labelled background other signal_ obtain reliable estimate classifier quality by filling confusion matrix combining information example use developed svm discriminant reported in ref possibility move svm selection threshold fit multi - jet contamination over svm output variable @xmath20 used figure svm_phx_fit shows physics processes contribute to shape svm discriminant selection forward electron sample multi - jet fraction extracted from fit total @xmath2 plus jets component svm threshold for final signal region identification is @xmath30 instead @xmath31 decreases multi - jet contamination reduces impact systematics estimate sample selection multi - jet background fraction extracted from fit with total @xmath2 plus jets component remaining physics processes normalized to expected production cross sections acceptances svm selection threshold for final signal region identification is @xmath30 instead @xmath31 decreases qcd contamination in final analysis sample,scaledwidth=100.0% paper innovative method for training optimization variable selection for svm multivariate classifier in biased statistically limited training samples optimization thanks to feedback signal - background template fit validation sample unclassified events over svm output distribution @xmath20 defined eq distance svm output @xmath20 sensitive to signal background contamination optimization algorithm applied to hadron collider physics case problem multi - jet rejection in @xmath2 plus jets channel analyzed in two cases @xmath2 boson decaying electron central forward region cdf detector with two hadronic jets optimization algorithm twenty - four input variables minimal combination lower background contamination higher signal efficiency resulting classifiers exploit eight six input variables combinations for central forward svm classifiers not related to lepton identification algorithm performance superior to present multi - jet rejection algorithm final state cdf ii dataset multi - jet rejection analysis test - bench for svm optimization algorithm procedure can exported to other problem multivariate techniques no accurate simulation university research association dr . giorgio chiarelli dr . sandra leone cdf higgs discovery group feedback suggestions dr . h. wolfe for review ametller garrido stimpfl - abele talavera yepes discriminating signal neural networks top - quark search fermilab tevatron phys . rev 54 , 1233 1996 aaltonen collaboration standard model higgs boson @xmath32 boson 7.5 fb@xmath33 luminosity cdf . d 86 032011 2012 chen lin guyon nikravesh gunn zadeh combining svms feature selection strategies extraction springer berlin heidelberg 2006 , 207 , 315 sforza lippi chiarelli leone multi - jet background hadron collider svm classifier nuclear science symposium medical imaging conference 2011 p1404 aaltonen standard model higgs boson @xmath36 pair one charged lepton missing transverse energy full cdf data set . 2012 ) 111804 acosta collaboration measurement @xmath39 production cross section @xmath40 collisions @xmath41 lepton + jets events secondary vertex b - tagging phys d 71 2005 ) 052003 mladenifeature selection linear classifier weights interaction classification models proceedings 27th international conference information retrieval 2004 234 | 0.570573 | 3,452 | 1,708 | , Saving $0.1 in GPT-4. |
54,967 | we address scaling in inhomogeneous and anisotropic turbulent flows by decomposing structure functions into their irreducible representation of the so(3 ) symmetry group which are designated by @xmath0 indices . employing simulations of channel flows with re@xmath1
we demonstrate that different components characterized by different @xmath2 display different scaling exponents , but for a given @xmath2 these remain the same at different distances from the wall .
the @xmath3 exponent agrees extremely well with high re measurements of the scaling exponents , demonstrating the vitality of the so(3 ) decomposition .
.#1 version of 0.5 cm most of the available data analysis and theoretical thinking about the universal statistics of the small scale structure of turbulence assume the existence of an idealized model of homogeneous and isotropic flow .
in fact most realistic flows are neither homogeneous nor isotropic .
accordingly , one can analyze the data pertaining to such flows in two ways .
the traditional one has been to disregard the inhomogeneity and anisotropy , and proceed with the data analysis assuming that the results pertain to the homogeneous and isotropic flow .
the second , which is advocated in this letter , is to take the anisotropy explicitly into account , to carefully decompose the relevant statistical objects into their isotropic and anisotropic contributions , and assess the degree of universality of each component separately .
we analyze here direct numerical simulations of a channel flow with re@xmath4 @xcite .
the main conclusion of this letter is that this procedure is unavoidable ; in particular it highlights the universality of the scaling exponents of the isotropic sector which are presumably those governing the universal small scale statistics at very high reynolds numbers . in agreement with recent studies of the this subject @xcitewe report that different irreducible representations of the symmetry group ( characterized by indices @xmath0 ) exhibit scalar functions that scale with apparently universal exponents that differ for different @xmath2 .
the exponents found at low values of the reynolds number for the @xmath3 ( isotropic ) sector are in excellent agreement with high re results ; these exponents are invariant to the position in the inhomogeneous flow , leading to reinterpretation of recent findings of position dependence as resulting from the intervention of the anisotropic sectors .
the latter have nonuniversal weights that depend on the position in the flow .
we consider here channel flow simulations on a grid of @xmath5 points in the stream - wise direction @xmath6 , and @xmath7 in the other two directions , @xmath8 .
we denote by @xmath9 the direction perpendicular to the walls and by @xmath10 the span - wise direction in planes parallel to the walls .
we employ periodic boundary conditions in the span - wise and stream - wise directions and no - slip boundary conditions on the walls .
the reynolds number based on the taylor scale is @xmath11 in the center of the channel @xmath12 .
the simulation is fully symmetric with respect to the central plane .
the flow correctly develops a mean profile in the stream - wise direction which depends only on the distance from the wall , @xmath13 .
the mean profile shows the three typical regimes : a laminar linear mean profile inside the viscous sublayers , a logarithmic profile for intermediate distances and finally a parabolic mean profile in the core of the channel . for more details on the averaged quantities and on the numerical code the reader
is referred to @xcite .
previous analysis of the same data - base @xcite as well as of other dns @xcite and experimental data @xcite in anisotropic flows found that the scaling properties of energy spectra , energy co - spectra and of longitudinal structure functions exhibit strong dependence on the local degree of anisotropy .
for example , in @xcite the authors studied the longitudinal structure functions at fixed distances from the walls : @xmath14 where @xmath15 denotes a spatial average on a plane at a fixed height @xmath16 , @xmath17 .
for this set of observables they found that : ( i ) these structure functions did not exhibit clear scaling behavior as a function of the distance @xmath18 .
consequently , one needed to resort to extended - self - similarity ( ess ) @xcite in order to extract a set of relative scaling exponents @xmath19 ; ( ii ) the relative exponents , @xmath20 depended strongly on the height @xmath16 . moreover , only at the center of the channel and very close to the walls the error bars on the relative scaling exponents extracted by using ess were small enough to claim the very existence of scaling behavior in any sense . similarly ,
an experimental analysis of a turbulent flow behind a cylinder @xcite showed a strong dependence of the relative scaling exponents on the position behind the cylinder for not too big distances from the obstacle , i.e. where anisotropic effects may still be relevant in a wide range of scales . in the following
we present an interpretation of the variations in the scaling exponents observed in non - isotropic and non - homogeneous flows upon changing the position in which the analysis is performed .
in particular , we will show that decomposing the statistical objects into their different @xmath21 sectors rationalizes the findings , i.e. scaling exponents in given @xmath21 sector appear quite independent of the spatial location ; only the _ amplitudes _ of the so(3 ) decomposition depend strongly on the spatial location .
these findings , if confirmed by other independent measurements , would suggest that the apparent dependence of scaling exponents for longitudinal structure functions on the location in a non - homogeneous flow results from of a superposition of power laws each of which is characterized by its own _
universal _ scaling exponent .
the amplitudes of the various contributions may depend on the local degree of anisotropy and non - homogeneity .
our method of analysis is quite simple @xcite .
we start by a direct measurement of the longitudinal structure functions @xmath22^p
\right > \label{spdef}\ ] ] note that the two velocity fields are measured at the extremes of the diameter of a sphere of radius @xmath18 centered at @xmath23 . due to the inhomogeneity
this function depends explicitly on @xmath23 .
due to the anisotropy the function depends on the orientation of the separation vector @xmath24 as well as on its magnitude .
the average must be taken over different time frames .
typically we have used 160 time frames for such an average .
the time frames are separated by about one eddy turn over time . in each time frame
we also improved the statistics by averaging over one fourth of the total number of spatial points in the plane at fixed @xmath16 , invoking the homogeneity in the span - wise and stream - wise directions , @xmath25 .
thus we have finally about @xmath26 contributions to each average .
having computed @xmath27 we decompose it into the irreducible representations of the so(3 ) symmetry group according to : @xmath28 we expect that when scaling behavior sets in ( presumably at high enough re ) we should find : @xmath29 in other words , we expect @xcite the scaling exponent @xmath30 to be independent of @xmath31 .
the first result that we want to display is that by applying the so(3 ) decomposition we seem to improve significantly the very existence of scaling behavior . in fig.1
we show ( i ) the log - log plot of the raw structure function ( [ spdef ] ) with @xmath32 measured on the central plane with the vector @xmath33 in the streamwise direction , @xmath34 , and ( ii ) the fully isotropic sector @xmath35 with the average in ( [ spdef ] ) taken on the sphere centered on the central plane @xmath36 .
= 8truecm = 6 cm it appears that already at this fairly low reynolds number the @xmath3 sector shows decent scaling behavior as a function of @xmath18 .
this is in marked contrast with the raw structure function for which no scaling behavior is detectable ( @xmath37 symbols in fig.1 ) . for the raw quantity
the method of extended - self - similarity " ( ess ) @xcite is unavoidable if one wants to extract any kind of apparent scaling exponent . in our analysis
we found similar results also for higher order structure functions .
the scaling behavior is improved dramatically for the components and it can be seen even without ess .
nevertheless we will use ess below for _ quantitative _ reading of the exponents within every sector .
the second point we would like to stress is the apparent _ invariance _ of the scaling exponents belonging to the same @xmath21 sector with respect to changing the spatial location in the flow . to study this issue quantitatively
we resort to ess , and examine the relative scaling of , say , structure functions of order @xmath38 with respect to the structure function of order @xmath39 for @xmath40 .
the ess method is applied in each @xmath21 sector separately .
in fig.2 we show two typical ess plots for longitudinal structure functions of order 4 vs longitudinal structure functions of order 2 both at the center @xmath41 and at @xmath42 in the sector @xmath3 .
also , in the inset the quality of the scaling can be appreciated by looking at the _ logarithmic local slopes _ of @xmath43 vs @xmath44 as a function of @xmath18 for the same two different central position of the sphere : at the center of the channel ( @xmath36 ) and at one quarter of the total channel height ( @xmath45 ) .
the two curves give the same global relative scaling exponent .
the best fits for the relative scaling exponents in the sector @xmath46 give @xmath47 at the center and @xmath48 at @xmath45 .
this result is remarkable and together with the experimental result of ref.@xcite it provides strong evidence for the universality of scaling exponent as defined in distinct @xmath21 sectors .
we recall that the accepted value of this relative exponent in high - re experiments is @xmath49 @xcite .
= 7truecm = 6 cm similarly , but affected from larger error bars , one recovers the same invariance with respect to higher order moments .
for instance we measure @xmath50 . as for relative scaling exponents of higher @xmath2 sectors ,
the scaling is less clean and therefore we may only quote qualitative estimates . as an example , for relative scaling exponents of the @xmath51 and @xmath52 sectors we have @xmath53 , @xmath54 , @xmath55 , @xmath56 . to underline the quantitative improvement resulting from the application of the so(3 ) decomposition
we show in fig 3 the _ logarithmic local slopes _ of the raw structure functions @xmath57 vs. @xmath58 at @xmath59 and at @xmath60 .
also the _ logarithmic local slopes _ of the projection on the @xmath3 sector at the same two distances from the walls are presented .
as is evident , the raw structure function at the center of the channel and the two @xmath3 projections are is in good agreement with the high - reynolds numbers estimate @xmath61 while a clearly spurious departure is seen for the raw structure functions at @xmath62 .
= 8truecm = 6 cm notice that due to the invariance of eq.([spdef ] ) under the inversion @xmath63 all the amplitudes @xmath64 belonging to sectors with odd @xmath2 vanish .
similarly , at the center of the channel the symmetry with respect to the center @xmath65 forces all the amplitudes of the components with @xmath66 odd to vanish as well . as a consequence ,
the sector @xmath67 is relevant only when the center of mass is not in the central plane .
when @xmath45 we recover indeed good scaling behavior also for this sector but with a relative scaling exponent @xmath68 slightly larger then the relative scaling exponents observed for the other @xmath69 sectors .
this fact , which seems to violate the supposed foliation in the @xmath2 index asserted in eq.([scale ] ) is not well understood at the moment and it may be correlated with the presence of large scale coherent structures ( hairpin ) oriented at @xmath70 with respect to the walls observed in all channel flow simulations @xcite . finally , we discuss briefly the determination of the scaling exponents associated with higher @xmath2 sectors .
the scaling exponent @xmath71 was estimated by a number of authors on the basis of dimensional analysis @xcite , and the result is @xmath72 .
there is no theoretical knowledge of the actual value of this exponent with intermittency corrections .
our direct analysis for @xmath69 seems to confirm the dimensional expectation , in agreement with the previous experimental @xcite finding .
= 8truecm = 6 cm in figure 4 we show the log - log plot of @xmath73 vs @xmath74 for @xmath51 at the center of the channel , and for @xmath51 and @xmath52 at @xmath45 , superimposed with the straight line with slope @xmath75 .
the agreement is quite good . considering the relatively low reynolds numbers and the fact that the projections on the different sectors depend on the non - universal prefactors @xmath64 in the decomposition ( [ scale ] ) , we think that together with the experimental result reported in @xcite the present finding gives strong support to the view that the scaling exponents in the @xmath69 sector are universal .
we are not able yet to offer the similar support to the possibility that all the scaling exponents in the higher @xmath2 sectors are universal .
such a conclusion calls for additional careful analysis of the scaling of higher order structure functions and higher @xmath2 sectors .
it is outside the scope of this letter , but it is currently under active study . in summary , we presented three important results that follow from the so(3 ) decomposition of the longitudinal structure functions measured in channel flow simulations @xcite : these are ( i ) the scaling behavior is better defined in separated @xmath21 sectors .
this is in contradistinction with the raw longitudinal structure function which fails to exhibit any scaling at all .
( ii ) the isotropic @xmath76 component of the structure functions exhibits a universal scaling exponent which is invariant to the spatial location in the flow and the distance from the walls .
( iii ) the @xmath69 component exhibits a scaling exponent which is compatible with the theoretical expectation and is in excellent agreement with the experimental measurement @xcite , indicating universality .
the picture that emerges is that the higher order sectors are characterized by scaling exponents that are larger than the fundamental exponent in the isotropic sector .
if this is so , it may explain the decay of anisotropy at small scales for high re flows .
in the limit re@xmath77 we expect scaling behavior at very small values of @xmath78 with @xmath79 being the outer scale . at such small scales
only the smallest exponent survives , and this is how the alleged universality of the small scales is achieved . | we are strongly indebted to f. toschi for helping in the set - up of the data - analysis and for a continuous and fruitful collaboration on the subject .
lb is partially supported by infm ( pra - turbo ) .
ip acknowledges the partial support of the german - israeli foundation , the israel science foundation , the european commission under the training and mobility of researchers program and the naftali and anna backenroth - bronicki fund for research in chaos and complexity .
g. amati , r. benzi and s. succi , phys .
e * 55 * 6985 ( 1997 ) .
f. toschi , g. amati , s. succi .
r. benzi and r. piva , `` intermittency of structure functions on channel flow turbulence '' , phys .
lett . submitted ( 1998 ) .
g. amati , s. succi and r. piva , int .
c * 8 * 869 ( 1997 ) . | indebted f. toschi data - analysis collaboration supported by infm support german - israeli foundation israel science foundation european commission naftali anna backenroth - bronicki fund research chaos complexity g. amati r. benzi s. succi phys . e * 55 * 6985 1997 ) toschi amati s succi benzi piva intermittency structure functions channel flow turbulence . lett. submitted 1998 ) amati succi piva int . c * 8 * 869 ( 1997 ) | 0.070376 | 220 | 119 | , Saving $0.0 in GPT-4. |
54,972 | we apply our deformation theory of periodic bar - and - joint frameworks to tetrahedral crystal structures .
the deformation space is investigated in detail for frameworks modelled on quartz , cristobalite and tridymite . * keywords : * periodic frameworks , deformations , flexibility , silica polymorphs . | in this paper we present specific applications of our general deformation theory of periodic frameworks @xcite .
considerations related to _ framework flexibility _ appear already in the early structural investigations based on @xmath0-ray crystallography @xcite . for framework materials , envisaged as corner sharing polyhedra , an intuitive notion of a ` coordinated tilting ' of the polyhedra is used in classifying similar structures @xcite or in studies of thermal and pressure effects .
a most important area of theoretical and experimental studies where geometric models of deforming frameworks have been implicated is that concerned with _ displacive phase transitions _
@xcite .
regarding the use of geometrical facts , it should be observed that , for most framework structures , only a confined sample of geometrical possibilities has been explored in the literature , typically one - parameter families which are intuitively ` accessible ' .
the deformation theory developed in our paper @xcite shows that one may expect , in general , a rich and diverse geometry .
the present undertaking describes the deformation spaces for tetrahedral periodic frameworks modeled on _ quartz , cristobalite and tridymite_.
the ideal structure considered here is made of congruent regular tetrahedra .
the oxygen atoms would correspond with the vertices , each oxygen being shared by two tetrahedra .
the silicon atoms should be imagined at the centers of the tetrahedra .
we shall examine all the geometric deformations of the periodic framework described in figure [ figq ] , without concern for self - collision or any prohibition of a physical nature .
equivalence under euclidean motions is eliminated by assuming the tetrahedron marked @xmath1 as fixed .
since all edges maintain their length , the positions of the two tetrahedra which share the vertices @xmath2 and @xmath3 are completely described by two orthogonal transformations @xmath4 , respectively @xmath5 as follows : @xmath4 fixes @xmath2 and takes @xmath6 to @xmath7 , @xmath8 , while @xmath5 fixes @xmath3 and takes @xmath9 to @xmath10 , @xmath11 .
the figure , by depicting only the ` visible ' edges , implies that both @xmath4 and @xmath5 are orientation reversing , that is , as orthogonal matrices @xmath12 .
if we denote the edge vectors @xmath13 by @xmath14 , we have : @xmath15 @xmath16 @xmath17 @xmath18 it follows that the dependency condition of a zero sum for these four generators of the periodicity lattice takes the form @xmath19 under our regularity assumptions , the three vectors @xmath20 , @xmath21 and @xmath22 have the same length and form an equilateral triangle .
this restricts @xmath21 to the circle on the sphere of radius @xmath23 ( which corresponde with an angle of @xmath24 with @xmath25 ) .
thus , @xmath26 is constrained to a surface , which is differentiably a two - torus @xmath27 . for each choice of @xmath28 on this torus , @xmath20 is determined by ( [ 0sum ] ) , hence @xmath29 is restricted to a circle @xmath30 in @xmath31 .
it follows that [ quartz ] the deformation space of the ideal quartz framework is given by a three dimensional torus @xmath32 minus the degenerate cases when the span of the four vectors is less than three dimensional .
the case of the ` ideal @xmath33 cristobalite ' structure illustrated in figure [ figc ] is already covered in @xcite .
the periodicity group of the framework is give by all the translational symmetries of the ideal crystal framework .
as a result , there are @xmath34 orbits of vertices and @xmath35 orbits of edges . adopting the notations of figure [ figcc ]
, we may assume the tetrahedron @xmath36 as fixed and parametrize the possible positions of the other tetrahedon by a rotation around the origin @xmath37 .
[ cristobalite ] the deformation space of the ideal high cristobalite framework is naturally parametrized by the open set in @xmath31 where the depicted generators remain linearly independent .
the tetrahedral framework @xmath38 of tridymite is depicted in figure [ figt ] .
we consider the ideal case made of regular tetrahedra .
the quotient graph has @xmath39 and @xmath40 .
all deformations can be described by three orthogonal transformations ( matrices ) @xmath41 acting with centers at @xmath42 and respectively @xmath43 . with @xmath37 as the origin and the tetrahedron @xmath44 assumed fixed , we put : we may interpret this system as a problem about a _
spherical four - bar mechanism _ in the following way .
all the vectors implicated in ( [ tsystem ] ) are unit vectors and can be depicted as points on the unit sphere @xmath65 . for a given @xmath66 , we mark by @xmath67 the midpoint of the spherical geodesic segment @xmath68 $ ] and trace the circle with center @xmath67 and diameter @xmath68 $ ] .
this is illustrated in figure [ figs ] .
it is an elementary observation that any solution @xmath69 and @xmath70 determines diameters of the corresponding circles for @xmath71 , with the two geodesic arcs @xmath72 $ ] , like @xmath73 $ ] and @xmath74 $ ] , of length @xmath75 .
thus , the two spherical quadrilaterals with vertices at @xmath76 and respectively @xmath77 , @xmath78 , @xmath79 , @xmath80 are two configurations of the same four - bar mechanism and moreover , the distance between the midpoints of the opposite edges represented by diameters is the same .
it follows from the theory of the spherical four - bar mechanism that , for a generic @xmath66 near the identity of @xmath31 , the abstract configuration space is made of two loops which correspond by reflecting the corresponding realizations .
each loop component has two configurations with the prescibed distance @xmath81 $ ] .
thus , there are four configurations with the prescribed distance .
we observe that if we replace @xmath82 by @xmath83 and @xmath83 by @xmath82 in the labeling of the vertices of a realization , the orientation is reversed , hence the configuration belongs to the other component .
thus , the two obvious solutions of ( [ tsystem ] ) , namely correspond with configurations on the two different loop components , as do the remaining two , which are also paired by relabeling .
this discussion shows that all four solutions are obtained from the quadrilateral @xmath76 and its reflection in the geodesic @xmath85 $ ] , by the two relabelings with @xmath82 and @xmath83 possible in each case . in figure [ figsr ]
we have depicted the quadrilateral @xmath76 as @xmath86 , with reflection in @xmath81 $ ] marked as @xmath87 .
then , , the solutions @xmath88 of the system ( [ tsystem ] ) are the following four solutions : @xmath89 , @xmath90 , @xmath91 and @xmath92 .
the deformation space of the tridymite framework is singular in a neighbourhood of the aristotype and can be represented as a ramified covering with four sheets of a three - dimensional domain .
there is a natural @xmath93 action on this covering which fixes the aristotype framework . indeed , the two involutions , inverting the labeling and reflecting in @xmath85 $ ] , commute ang give a @xmath94 action on the covering .
the dimension of the tangent space at the aristotype framework is computed from the linear version of ( [ tsystem ] ) and is six . | paper present applications deformation theory of periodic frameworks @xcite considerations framework flexibility appear early structural investigations @xmath0-ray crystallography framework materials corner sharing polyhedra notion coordinated tilting polyhedra used classifying structures studies thermal pressure effects important area studies geometric models deforming frameworks implicated displacive phase transitions use geometrical facts framework structures confined sample geometrical possibilities explored typically one - parameter families deformation theory paper shows rich diverse geometry present undertaking describes deformation spaces for tetrahedral periodic frameworks on quartz cristobalite tridymite ideal structure congruent regular tetrahedra oxygen atoms correspond with vertices shared by two tetrahedra silicon atoms at centers of tetrahedra examine geometric deformations of periodic framework described figure without concern self - collision prohibition physical nature equivalence under euclidean motions eliminated assuming tetrahedron marked as fixededges maintain length positions two tetrahedra vertices @xmath2 described by orthogonal transformations @xmath4 @xmath4 fixes @xmath2 takes to @xmath5 fixes to figure edges implies @xmath4 and @xmath5 orientation reversing orthogonal matrices @xmath12 denote edge vectors @xmath13 by @xmath14 @xmath15 dependency condition of zero sum for four generators periodicity lattice takes form @xmath19 three vectors @xmath20 same length form equilateral triangle restricts @xmath21 to circle on sphere radius @xmath23 angle of @xmath24 @xmath25 @xmath26 constrained to surface two - torus @xmath27 each choice of @xmath28 @xmath20 determined by [ 0sum ] ) @xmath29 restricted to circle @xmath30 @xmath31 deformation space of ideal quartz framework given by three dimensional torus @xmath32 minus cases span four vectors less than three dimensional ideal @xmath33 cristobalite structure in figure covered in @xcite periodicity group framework translational symmetries ideal crystal framework @xmath34 orbits vertices @xmath35 orbits edges assume tetrahedron @xmath36 fixed parametrize positions other tetrahedon rotation around origin @xmath37 deformation space ideal high cristobalite framework parametrized by open set @xmath31 generators linearly independent tetrahedral framework @xmath38 tridymite depicted in figure [ figt ideal case regular tetrahedra quotient graph has @xmath39 @xmath40 deformations described by three orthogonal transformations @xmath41 centers @xmath42 @xmath43 @xmath37 origin tetrahedron @xmath44 fixed interpret system problem spherical four - bar mechanism vectors unit vectors depicted as points on unit sphere @xmath65 @xmath66 mark @xmath67 midpoint spherical geodesic segment @xmath68 trace circle with center @xmath67 diameter @xmath68 $ illustrated in figure [ figs solution @xmath69 @xmath70 determines diameters circles @xmath71 two geodesic arcs @xmath72 length @xmath75 two spherical quadrilaterals with vertices at @xmath76 @xmath77 are configurations of same four - bar mechanism distance between midpoints opposite edges same for generic @xmath66 @xmath31 abstract configuration space two loops realizations each loop component has two configurations with prescibed distance @xmath81 four configurations with prescribed distance replace @xmath82 by @xmath83 orientation reversed configuration belongs to other component two solutions of tsystem ] correspond with configurations on two loop components remaining two paired by relabeling all four solutions from quadrilateral @xmath76 reflection in geodesic @xmath85 by two relabelings with @xmath82 and @xmath83 figure quadrilateral @xmath76 as @xmath86 reflection @xmath81 as @xmath87 solutions @xmath88 of system are four solutions : @xmath89 @xmath90 @xmath91 @xmath92 deformation space of tridymite framework singular in neighbourhood aristotype represented as ramified covering with four sheets of three - dimensional domain natural @xmath93 action on covering fixes aristotype frameworktwo involutions inverting labeling reflecting @xmath85 commute give @xmath94 action covering dimension tangent space aristotype framework computed from linear version tsystem six . | 0.309027 | 1,729 | 887 | , Saving $0.1 in GPT-4. |
54,981 | we present some new results on searches for new physics at the tevatron run 1 ( 1992 1996 ) .
the topics covered are searches for r - parity violating and conserving msugra , large extra dimensions in di - photon and monojet channels , leptoquark in jets + channel , and two model independent searches .
all results were finalized during the past year .
= 14.5pt | tevatron run i has been a great success for high energy physics . for the period between october 1992 and february 1996 ,
about @xmath0 of data were collected by the two competing experiments and collaborations : cdf and d .
both collaborations have made many important measurements and discoveries with their powerful and multi - purpose detectors@xcite , culminated by the discovery of the top quark in 1995 .
they are also engaged in search for new physics beyond the standard model ( sm ) . though no convincing evidence of new physics was found ,
the searches have extended our understanding of the fundamentals of the universe and have led our quest for ultimate understanding in a concerted direction . this paper reports nine results on searches for new physics conducted recently at the tevatron by cdf and d .
we cover the topics of susy , large extra dimension , leptoquark , and model independent searches .
minimal supergravity or msugra@xcite is a model which provides a framework for the spontaneous breaking of the supersymmetry@xcite . in this model ,
susy is broken in the hidden sector of the theory and this breaking is communicated to the physical sector of the theory through gravitational interactions .
there are five parameters to completely determine the susy sector of the theory : * @xmath1 : common scalar particle mass at the susy breaking scale @xmath2 is usually the gut scale ( @xmath3 ) or the planck scale ( @xmath4 ) . ] ; * @xmath5 : common gaugino mass at the @xmath2 scale ; * @xmath6 : common trilinear coupling at the @xmath2 scale ; * @xmath7 : ratio of the vacuum expectation values of the two higgs doublets ; * @xmath8 : @xmath9 is the higgsino mass parameter . an additional parameter called r - parity is introduced and is defined as : @xmath10@xcite , where @xmath11 and @xmath12 are baryon and lepton numbers , respectively , and @xmath13 refers to spin . a superpotential for mssm , the minimal supersymmetric extension of the standard model@xcite , can be written as the following : @xmath14 where the first line describes @xmath15-conserving couplings and the second line describes @xmath15-violating couplings ; @xmath16 , @xmath17 , and @xmath18 are @xmath19 yukawa coupling matrices ; @xmath20 and @xmath12 are left - handed quark and lepton supermultiplets , respectively ; @xmath21 , @xmath22 , @xmath23 are the right - handed singlets of the up and down type ( s)quarks and ( s)leptons , respectively ; @xmath24 and @xmath25 are the two higgs doublets ; @xmath9 is the higgsino mass parameter ; @xmath26 , @xmath27 , and @xmath28 are the coupling strengths for lepton number violating interactions and @xmath29 is the coupling strength for baryon number violating interactions .
we describe in this paper four searches for msugra under various additional constraints . in this analysis
, we assume that stop pair are produced through @xmath15-conserving processes and then decay through @xmath15-violating process : @xmath30 , where @xmath31 and @xmath32 are leptonically and hadronically decayed @xmath33 , respectively .
we also assume that @xmath34 in eq .
( [ eqn : superpotential ] ) dominates the couplings .
the key to this analysis is the identification of @xmath32 .
the following criteria are used to select @xmath32 : * @xmath32 candidates are clusters with @xmath35 and @xmath36 ; * number of tracks and @xmath37 s in a narrow cone around a @xmath33 cluster candidate are consistent with those coming from a @xmath33 ; * @xmath38 and isolation energy of tracks and reconstructed @xmath39 mass are consistent with those of a @xmath33 .
figure [ fig : rpv_stop_echan_tau_ntracks ] shows the number of tracks in a @xmath33 cone .
the 1-prong and 3-prong structures of the @xmath33 candidates are prominent .
a total of @xmath40 of data are used in this analysis .
the major sm backgrounds come from @xmath41 , diboson , @xmath42 , @xmath43 , and multijet events .
the first two are physics backgrounds which have the same final states as the signal while the rest are @xmath33 fakes of by jets .
leptonically decayed @xmath31 is identified with a tagging electron or muon .
we require @xmath44 , @xmath45 for the electron channel or @xmath46 , @xmath47 for the muon channel . in order to increase the signal significance ,
the following additional selection cuts are applied : * transverse mass of the lepton and the : @xmath48 ; * the scalar sum of @xmath49 of the lepton , @xmath32 , and : @xmath50 ; * @xmath51 with @xmath52 .
the distribution of these variables are shown in figure [ fig : rpv_stop_echan ] .
[ h!tb ] the results after the cuts are shown in table [ tbl : rpv_stop_result ] . since no signal is found in our data .
we set the signal limit in terms of @xmath53 , the production cross section of @xmath54 as a function of @xmath55 .
the limits are shown in figure [ fig : rpv_stop_combined_limit ] . from the figure
, we set the 95% c.l .
limit on stop mass : @xmath56 .
the previous limit from aleph collaboration is @xmath57@xcite .
[ h!tb ] 0.1 in ._number of expected background events @xmath58 , observed events @xmath59 in data and the expected total signal efficiency ( for @xmath60 ) after all the cuts in search for rpv msugra in decays of stop pair are applied .
_ [ cols="^,^,^,^",options="header " , ] [ tbl : gamma_met_led_limit ] quaero , which means `` to seek '' in latin , uses d data which are publically available on the internet and automatically optimizes searches for any signal provided by the user .
the data sets are categorized according to their final states .
the backgrounds and their respective fractions have been calculated and are available to the user .
it is understood that the data are well explained by the expected background and that the goal of the analysis is to set @xmath61 , the 95% c.l .
upper limit , on the cross section of the model . at the quaero web page , http://quaero.fnal.gov
, users can use pythia@xcite to generate their model events which results in one of the available final states .
they can then define a variable set @xmath62 to be used to optimize the search .
the optimization algorithm has the following steps : * kernel density estimation@xcite is used to obtain a signal and background probability distribution @xmath63 and @xmath64 , respectively ; * a discriminant function is defined as@xcite : @xmath65 * the sensitivity @xmath13 is defined as the reciprocal of @xmath66 , which is the 95% c.l . limit on model cross section as a function of @xmath67 on @xmath68 .
an optimal @xmath67 on @xmath68 is chosen to minimize @xmath66 , thus maximize @xmath13 ; * the region of variable space having @xmath69 is used to determine the actual 95% c.l .
cross section upper limit @xmath61 . based on run 1 data ,
quaero has set @xmath61 for various models , including sm higgs production : @xmath70 , @xmath71 , @xmath72 , and @xmath73 ; @xmath74 and @xmath75 production : @xmath76 , @xmath77 ; and leptoquark production : @xmath78 .
we have presented in this paper nine analyses which were finalized in year 2001 . even though we did not observe any signature of new physics , we are able to set stronger limits on them . new tools and techniques
have been developed to equip us for more challenging searches . with more powerful detectors for both cdf and d , and an order of magnitude of increase in luminosity for expected for run 2
, we are looking forward to more exciting searches and possibly discoveries .
the author is thankful to both cdf and d collaborations for the opportunity to present these new results at la thuile 2002. in particular the author would like to thank the following persons who actually carried out the analyses and provided the detail information : abdelouahab abdesselam , sudeshna banerjee , and hai zheng of the dcollaboration , and david gerdes , chris hays , teruki kamon , minjeong kim , bruce knuteson , yoshiyuki miyazaki , simona murgia , peter onyisi , and masashi tanaka of the cdf collaboration . | tevatron run i success for high energy physics october 1992 and february 1996 @xmath0 data collected by two competing experiments collaborations cdf and d collaborations made important measurements discoveries with detectors@xcite discovery top quark in 1995 for new physics beyond standard model no convincing evidence new found searches extended understanding of fundamentals universe led quest for ultimate understanding paper reports nine results on searches for new physics at tevatron by cdf and d cover susy large extra dimension leptoquark model independent searches minimal supergravity msugra@xcite model framework for spontaneous breaking of supersymmetry@xcite susy broken in hidden sector theory breaking communicated to physical sector through gravitational interactions five parameters determine susy sector : @xmath1 common scalar particle mass at scale gaugino mass trilinear coupling ratio of vacuum expectation values of two higgs doublets higgsino mass parameter additional parameter r - parity introduced defined as @xmath10@xcite @xmath11 are baryon and lepton numbers @xmath13 refers to spinsuperpotential for mssm minimal supersymmetric extension of standard model@xcite written @xmath14 first line-conserving second line-violating couplings @xmath19 yukawa coupling matrices @xmath20 left - handed quark and lepton supermultiplets right - handed singlets of up down)quarks)leptons @xmath24 are two higgs doublets @xmath9 higgsino mass parameter @xmath26 coupling strengths for lepton number violating interactions strength baryon number violating interactions four searches for msugra under additional constraints stop pair produced through @xmath15-conserving processes decay through @xmath15-violating process @xmath30 @xmath31 @xmath32 leptonically hadronically decayed @xmath33 @xmath34 dominates couplings key identification of @xmath32 criteria select @xmath32 candidates clusters with @xmath35 @xmath36 number of tracks @xmath37 s in around @xmath33 cluster candidate consistent with @xmath33 @xmath38 isolation energy of tracks reconstructed @xmath39 mass consistent with @xmath33figure rpv_stop_echan_tau_ntracks shows tracks in @xmath33 cone 1-prong 3-prong structures candidates prominent @xmath40 data used major sm backgrounds from @xmath41 diboson multijet events first two physics backgrounds same final states as signal rest @xmath33 fakes jets leptonically decayed @xmath31 identified with tagging electron or muon require @xmath44 @xmath45 for electron @xmath46 @xmath47 muon channel increase signal significance additional selection cuts applied transverse mass of lepton @xmath48 scalar sum of @xmath49 @xmath51 with @xmath52 distribution variables shown in figure results after cuts shown in table rpv_stop_result no signal found set signal limit @xmath53 production cross section @xmath54 function @xmath55 limits shown in figure_stop_combined_limit set 95% c.l . limit on stop mass : @xmath56 previous limit is @xmath57@xcite expected background events @xmath58 observed events @xmath59 expected total signal efficiency for @xmath60 ) after cuts rpv decays stop paircols="^,^^",options="header " tbl : gamma_met_led_limit quaero latin uses data available optimizes searches for signal user data sets categorized final states backgrounds fractions calculated available data explained by expected background goal analysis to set @xmath61 95% c.l . upper limit on cross section model quaero web page http://quaero.fnal.gov users use pythia@xcite generate model events final states define variable set @xmath62 optimize search optimization algorithm kernel density estimation@xcite signal background probability distribution @xmath63 @xmath64 discriminant function defined as@xcite : @xmath65 @xmath13 reciprocal of @xmath66 95% c.l . limit on model cross section function @xmath67 @xmath68 optimal @xmath67 on @xmath68 minimize @xmath66 maximize @xmath13 region of variable space @xmath69 95% c.l . cross section upper limit @xmath61 quaero set @xmath61 for models including sm higgs production @xmath70 @xmath74 @xmath75 @xmath76 @xmath77 leptoquark production @xmath78presented paper nine analyses finalized 2001 observe new physics set stronger limits new tools techniques developed challenging searches powerful detectors cdf d increase luminosity run 2 looking forward exciting searches discoveries author thankful cdf d collaborations opportunity present new results la thuile 2002. persons carried analyses provided detail information abdelouahab abdesselam sudeshna banerjee hai zheng dcollaboration david gerdes chris hays teruki kamon minjeong kim bruce knuteson yoshiyuki miyazaki simona murgia peter onyisi masashi tanaka cdf collaboration | 0.320996 | 2,046 | 1,061 | , Saving $0.1 in GPT-4. |
54,982 | we analyzed continuous maxi / gsc data of the x - ray binary pulsar 4u 162667 from 2009 october to 2013 september , and determined the pulse period and the pulse - period derivative for every 60-d interval by the epoch folding method .
the obtained periods are consistent with those provided by the fermi / gbm pulsar project . in all the 60-d intervals ,
the pulsar was observed to spin up , with the spin - up rate positively correlated with the 220 kev flux .
we applied the accretion torque model proposed by ghosh & lamb ( 1979 , apj , 234 , 296 ) to the maxi / gsc data , as well as the past data including both spin - up and spin - down phases .
the ghosh & lamb relation was confirmed to successfully explain the observed relation between the spin - up / down rate and the flux . by comparing the model - predicted luminosity with the observed flux ,
the source distance was constrained as 513 kpc , which is consistent with that by chakrabarty ( 1998 , apj , 492 , 342 ) .
conversely , if the source distance is assumed , the data can constrain the mass and radius of the neutron star , because the ghosh & lamb model depends on these parameters .
we attempted this idea , and found that an assumed distance of , e.g. , 10 kpc gives a mass in the range of 1.811.90 solar mass , and a radius of 11.411.5 km , although these results are still subject to considerable systematic uncertainties other than that in the distance . | an x - ray binary pulsar is a system consisting of a magnetized neutron star and a stellar companion .
the x - ray emission is powered by gravitational energy of accreting matter . in a system with a low - mass companion star ,
the accretion flow is considered to take place via roche - lobe overflow , and to form an accretion disk in the vicinity of the pulsar .
the angular momentum of the accreting gas is transferred to the pulsar at the inner edge of the accretion disk , and accelerates the pulsar rotation until it finally reaches an equilibrium determined by the accretion rate and the magnetic moment of the pulsar .
so far , numbers of theoretical studies have been performed on the interaction between the pulsar magnetosphere and the accretion flows . along this scenario ,
@xcite , @xcite ( gl79 hereafter ) and @xcite ( lrb95 hereafter ) proposed their accretion models and presented equations describing the pulse - period derivative @xmath0 as a function of the luminosity @xmath1 , the pulse period @xmath2 , the mass @xmath3 , the radius @xmath4 and the surface magnetic field @xmath5 of the neutron star .
these models have been examined against observational data ( e.g. ; ; ; @xcite ; ; @xcite ) .
the results show that the observed @xmath0@xmath1 relations are grossly consistent with the model predictions . however , the validity of the models has not yet been fully confirmed , because it requires long - term monitoring of some suitable objects with known @xmath5 , covering significant pulse - period and luminosity changes with a sufficient sampling rate .
when these models describing the accretion torque are better calibrated , the observed @xmath0@xmath1 relations can give us observational constraints on @xmath3 and @xmath4 , which are very important because they are directly connected to the equation of state ( eos ) of nuclear matter .
so far , various observational attempts to measure @xmath3 and @xmath4 have been carried out , and generally yielded @xmath6 and @xmath7 km ( e.g. reviews by @xcite and @xcite )
. however , the measurements are not yet accurate enough to constrain the eos .
furthermore , the values of @xmath3 ( mainly from x - ray binary pulsars ) and @xmath4 ( mainly from weakly - magnetized neutron stars ) have been derived from different populations of neutron stars .
therefore , further studies of the @xmath0@xmath1 relations are expected to be valuable .
4u 162667 is a low - mass x - ray binary pulsar first detected with the uhuru satellite @xcite , and its 7.6-s coherent pulsation was discovered by @xcite . because no period modulation due to orbital motion has been detected beyond an upper limit of @xmath8 lt - ms ( @xmath9 is the orbital semi - major axis of the pulsar and @xmath10 is the orbital inclination ) , the mass of the companion star is estimated to be very low ( @xmath11 for @xmath12 ; @xcite ) .
it is hence classified as an ultra compact x - ray binary .
the bepposax observation revealed a cyclotron resonance scattering feature at @xmath13 , indicating the surface magnetic field of @xmath14 , where @xmath15 is the gravitational redshift , @xmath16 represented by the gravitational constant @xmath17 and the velocity of light @xmath18 @xcite .
the feature was confirmed by the suzaku observation @xcite .
the source distance was estimated to be 513 kpc from the optical flux by assuming that the effective x - ray albedo of the accretion disk is high ( @xmath190.9 ) @xcite . since the discovery of the 7.6-s pulsation in 1977 , the period of 4u 162667 has been repeatedly measured with various x - ray satellites ( e.g. references in @xcite ; @xcite ) .
table [ tab : past_fx_p_pdot ] summarizes the x - ray fluxes , periods and period derivatives observed from 1978 to 2008 , and figure [ fig : all_fx_p_pdot ] visualizes long - term behavior of these quantities .
it clearly shows that the source made transitions twice between the spin - up and the spin - down phases , at mjd @xmath20 48000 ( 1990 june ) and mjd @xmath20 54000 ( 2008 february ) , separated by @xmath20 18 years . in each phase @xmath0
were almost constant , and its absolute values were very similar as @xmath21 .
these period - change behavior suggests that 4u 162667 is close to an equilibrium state in which the net torque transfer from the accreting matter to the pulsar is approximately zero . at the last transition in 2008 when the source turned from the spin - down into the spin - up phase , the flux increased by a factor of @xmath202.5 @xcite .
these properties , together with the accurate knowledge of @xmath5 , make this object ideal for our study .
monitor of all - sky x - ray image ( maxi ; @xcite ) is an x - ray all - sky monitor on the international space station .
since the in - orbit operation started in 2009 august , its main instrument , the gsc ( gas slit camera ; @xcite ; @xcite ) , has been scanning the whole sky every 92 min in the 220 kev band .
the gsc field of view typically scans a celestial point source for about 60 s in each transit , which is long enough to study the 7.6-s pulsation from 4u 162667 .
thus , the maxi / gsc data are useful to study the long - term variation of the flux , @xmath2 , and @xmath0 . in this paper
, we analyze the maxi / gsc data of 4u 162667 from mjd 55110 ( 2009 october 6 ) to mjd 56550 ( 2013 september 15 ) and determine the flux , @xmath2 , and @xmath0 .
we then apply the spin - up / down models proposed by gl79 and lrb95 to the previous and the maxi / gsc data , to examine whether either model can explain the observed behavior of 4u 162667 , and to evaluate how these data can constrain the source distance , as well as the mass and radius of the neutron star .
akmal , a. , pandharipande , v. r. , & ravenhall , d. g. 1998 , , 58 , 1804 angelini , l. , white , n. e. , nagase , f. , et al .
1995 , , 449 , l41 basko , m. m. , & sunyaev , r. a. 1975 , , 42 , 311 bhattacharyya , s. 2010 , advances in space research , 45 , 949 bildsten , l. , chakrabarty , d. , chiu , j. , et al .
1997 , , 113 , 367 camero - arranz , a. , finger , m. h. , ikhsanov , n. r. , wilson - hodge , c. a. , & beklen , e. 2010 , , 708 , 1500 camero - arranz , a. , pottschmidt , k. , finger , m. h. , et al .
2012 , , 546 , aa40 chakrabarty , d. 1998 , , 492 , 342 chakrabarty , d. , bildsten , l. , grunsfeld , j. m. , et al .
1997 , , 474 , 414 douchin , f. , & haensel , p. 2001
, , 380 , 151 elsner , r. f. , darbro , w. , leahy , d. , et al .
1983 , , 266 , 769 finger , m. h. , wilson , r. b. , & chakrabarty , d. 1996 , , 120 , 209 ghosh , p. , & lamb , f. k. 1979 , , 234 , 296 giacconi , r. , murray , s. , gursky , h. , et al .
1972 , , 178 , 281 iwakiri , w. b. , terada , y. , mihara , t. , et al .
2012 , , 751 , 35 jain , c. , paul , b. , & dutta , a. 2010 , , 403 , 920 joss , p. c. , & rappaport , s. a. 1984 , , 22 , 537 kii , t. , hayakawa , s. , nagase , f. , ikegami , t. , & kawai , n. 1986 , , 38 , 751 kirsch , m. g. , briel , u. g. , burrows , d. , et al .
2005 , , 5898 , 22 klochkov , d. , staubert , r. , postnov , k. , shakura , n. , & santangelo , a. 2009 , , 506 , 1261 klus , h. , ho , w. c. g. , coe , m. j. , corbet , r. h. d. , & townsend , l. j. 2014 , , 437 , 3863 krauss , m. i. , schulz , n. s. , chakrabarty , d. , juett , a. m. , & cottam , j. 2007 , , 660 , 605 lattimer , j. m. , & schutz , b. f. 2005 , , 629 , 979 leahy , d. a. 1987 , , 180 , 275 levine , a. , ma , c. p. , mcclintock , j. , et al .
1988 , , 327 , 732 lovelace , r. v. e. , romanova , m. m. , & bisnovatyi - kogan , g. s. 1995 , , 275 , 244 matsuoka , m. , kawasaki , k. , ueno , s. , et al .
2009 , , 61 , 999 mavromatakis , f. 1994 , , 285 , 503 mihara , t. 1995 , ph.d .
thesis , dept . of physics ,
univ . of tokyo mihara , t. , nakajima , m. , sugizaki , m. , et al .
2011 , , 63 , 623 orlandini , m. , dal fiume , d. , frontera , f. , et al .
1998 , , 500 , l163 zel , f. 2013 , reports on progress in physics , 76 , 016901 pravdo , s. h. , white , n. e. , boldt , e. a. , et al .
1979 , , 231 , 912 rappaport , s. , & joss , p. c. 1977 , , 266 , 683 rappaport , s. , markert , t. , li , f. k. , et al .
1977 , , 217 , l29 reynolds , a. p. , parmar , a. n. , stollberg , m. t. , et al .
1996 , , 312 , 872 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 shen , h. , toki , h. , oyamatsu , k. , & sumiyoshi , k. 1998a , nuclear physics a , 637 , 435 shen , h. , toki , h. , oyamatsu , k. , & sumiyoshi , k. 1998b , progress of theoretical physics , 100 , 1013 shinoda , k. , kii , t. , mitsuda , k. , et al . 1990 , , 42 , l27 sugizaki , m. , mihara , t. , serino , m. , et al . 2011 , , 63 , 635 sugizaki , m. , yamamoto , t. , mihara , t. , nakajima , m. , & makishima , k. 2015 , , 210 van haaften , l. m. , voss , r. , & nelemans , g. 2012 , , 543 , a121 vaughan , b. a. , & kitamoto , s. 1997 , arxiv : astro - ph/9707105 yagi , k. , & yunes , n. 2013 , , 88 , 023009 | x - ray binary pulsar system magnetized neutron star stellar companion x - ray emission powered by gravitational energy of accreting matter in system with low - mass companion star accretion flow via roche - lobe overflow accretion disk in vicinity pulsar angular momentum of accreting gas transferred to pulsar accelerates pulsar rotation until equilibrium determined by accretion rate magnetic moment theoretical studies performed on interaction between pulsar magnetosphere accretion flows @xcite gl79 lrb95 proposed accretion models equations pulse - period derivative @xmath0 function of luminosity pulse period mass radius surface magnetic field of neutron star models examined against observational data results show observed @xmath0@xmath1 relations consistent with model predictions validity not confirmed requires long - term monitoring of objects with @xmath5 pulse - period luminosity changes rate when models accretion calibrated observed @xmath0 relations can give observational constraints on @xmath3 @xmath4 connected to equation of state of nuclear matter observational attempts to measure @xmath3 @xmath4 yielded @xmath6 @xmath7 km reviews by @xcite measurements not accurate constrain eos values of @xmath3 x - ray pulsars @xmath4 from weakly - magnetized neutron stars derived from different populations further studies of @xmath0@xmath1 relations valuable 4u 162667 low - mass x - ray binary pulsar detected with uhuru satellite @xcite 7.6-s pulsation discovered by @xcite no period modulation detected beyond upper limit @xmath8 lt - ms mass of companion star estimated low classified as ultra compact x - ray binary bepposax observation revealed cyclotron resonance scattering feature at @xmath13 surface magnetic field of gravitational redshift gravitational constant velocity of light feature confirmed by suzaku observation @xcite source distance estimated 513 kpc from optical flux effective x - ray albedo high since discovery 7.6-s pulsation 1977 period of 4u 162667 repeatedly measured with x - ray satellites@xcite table past_fx_p summarizes x - ray fluxes periods derivatives 1978 to 2008 figure_fx visualizes long - term behavior source made transitions twice between spin - up down phases at @xmath20 48000 1990 june 54000 2008 february separated 18 years @xmath0 almost constant absolute values similar as @xmath21 period - change behavior suggests 4u 162667 close to equilibrium state net torque transfer from matter to pulsar approximately zero last transition 2008 up flux increased by factor @xmath202.5 @xcite properties accurate knowledge of @xmath5 make object ideal for study monitor all - sky x - ray image @xcite ) international space station 2009 august gsc scanning sky every 92 min 220 kev band gsc scans celestial point source 60 s each transit 7.6-s pulsation from 4u 162667 maxi / gsc data useful study long - term variation of flux @xmath2 @xmath0 analyze maxi / gsc data of 4u 162667 from 55110 2009 to mjd 56550 2013 september 15 determine flux @xmath2 @xmath0 apply spin - up models gl79 lrb95 previous maxi / gsc data examine behavior 4u 162667 evaluate source distance mass radius neutron star akmal pandharipande ravenhall d. 1998 , 58 , 1804 angelini white nagase. 1995 449 , l41 basko sunyaev 1975 , 42 , 311 bhattacharyya . 2010 advances space research 45 , 949 bildsten . chakrabarty d. chiu j.. 1997 113 , 367 camero - arranz finger ikhsanov wilson - hodge beklen 2010 708 , 1500 camero - arranz pottschmidt finger m.. 2012 546 , aa40 chakrabarty 1998 492 , 342 bildsten grunsfeld . 1997 474 , 414 douchin haensel 2001 380 , 151 elsner darbro leahy .. 1983 266 , 769 finger wilson chakrabarty d. 1996 120 , 209 ghosh lamb f. 1979 234 , 296 giacconi murray gursky h. 1972 178 , 281 iwakiri terada mihara t. 2012 751 , 35 jain paul dutta 2010 403 , 920 joss rappaport s. 1984 22 , 537 kii hayakawa nagase ikegami kawai n. 1986 38 , 751 kirsch briel burrows d. . 2005 5898 , 22 klochkov staubert postnov shakura santangelo 2009 506 , 1261 klus h. ho coe corbet townsend . 2014 437 , 3863 krauss schulz chakrabarty d juett cottam j. 2007 660 , 605 lattimer schutz . 2005 629 , 979 leahy d.. 1987 180 , 275 levine ma p mcclintock j. 1988 327 , 732 lovelace romanova bisnovatyi - kogan 1995 275 , 244 matsuoka kawasaki ueno 2009 61 , 999 mavromatakis 1994 285 , 503 mihara 1995 ph.d thesis dept . univ . tokyo mihara nakajima sugizaki. 2011 63 , 623 orlandini dal fiume frontera f. 1998 500 , l163 zel 2013 reports progress physics 76 , 016901 pravdo white boldt 1979 231 , 912 rappaport joss 1977 266 , 683 rappaport markert li . 1977 217 , l29 reynolds parmar stollberg . 1996 312 , 872 shakura sunyaev 1973 24 , 337 shen h toki oyamatsu sumiyoshi 1998a nuclear physics 637 , 435 shen toki oyamatsu sumiyoshi 1998b progress theoretical physics 100 , 1013 shinoda kii mitsuda 1990 42 , l27 sugizaki mihara serino 2011 63 , 635 sugizaki yamamoto mihara nakajima makishima 2015 210 van haaften voss nelemans . 2012 543 , a121 vaughan kitamoto s. 1997 arxiv astro - ph/9707105 yagi yunes n. 2013 88 , 023009 | 0.440862 | 3,001 | 1,572 | , Saving $0.1 in GPT-4. |
54,984 | we searched for spectral signatures from the exosphere of hd209458 in the uv and the near - ir with uves and isaac at the vlt .
we looked in particular for the helium absorption feature at 10 830 predicted to be among the strongest ones .
the upper limit of the he i line derived is 0.5% at 3 @xmath0 for a 3 bandwith .
planet - induced chromospheric activity search on hd209458 was also performed .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | the he i feature at 10 830 is predicted to be one of the deepest absorption lines of the diffuse and extended exosphere in the optical and near - ir transmission spectrum the goal of the present study is to search for exospheric signatures in the spectrum of hd209458 , during transits of the planet .
shkolnik et al .
( 2003 ) suggested that a chomospheric activity can be induced by the planet orbit .
we searched for such a signature in stellar chromospheric lines .
we performed observations of hd209458 in the uv with vlt / uves in the 0.33 - 0.67 @xmath1 m region at high spectral resolution .
1 transit was covered with r=70000 in december 1999 in two spectral range : blue ( 0.328 - 0.4562 @xmath1 m ) and red ( 0.4583 - 0.6686 @xmath1 m ) . 6 transits were covered with r=100000 : 1 in june 2000 in blue and red and june to september 2002 in red ( 0.475 - 0.680 @xmath1 m ) .
we searched for ions and neutral molecules originating in the planet s exosphere and located in the evaporated material around the planet .
observations of 1 transit in the infrared was also performed with vlt / isaac at medium resolution ( r=10000 ) around the he i line at1.083 @xmath1 m ( 1.06 - 1.1 @xmath1 m ) .
a sample of these spectra centered around this feature is shown is fig .
1a . from observations during one full transit with isaac in june 2001 of the spectral region around 1.083 @xmath1 m
, no detection of he was obtained due to strong instrumental fringing affecting the data .
an 3 @xmath0 upper limit of 0.5% deep absorption feature was derived .
following the suggestion of shkolnik et al.(2003 ) , we searched signatures of a chromospheric activity induced by the ( magnetic more probably than tidal ) heating of the star s outer atmosphere by the planet .
we have recovered high - resolution uves spectra in the blue arm during one transit ( and its off - transit ) period in 2000 .
the ca ii k at 3933 , ca ii k at 3968 and al i at 3944lines can be related to chromospheric activity . a sample of these spectra centered around the ca ii k feature is shown is fig .
our spectra show emission enhancement observed in the ca ii h band but we found nothing significant in the two otherbands .
the variations could be indicative of the stellar chromospheric activity with the planet s magnetic field .
uves data from the 5 transits in june - september 2002 are currently being analysed to look for exospheric planetary signature of other constituents .
further and improved observations are required .
we have time allocation this year .
moutou , c. , coustenis , a. , schneider , j. , queloz , d. & mayor , m. , 2003 , in , 405 , 341 and references therein + shkolnik , e. , walker , g. a. h , & bohlender , d. a. , 2003 , submitted to + | he i feature at 10 830 deepest absorption exosphere optical near - transmission spectrum goal present study exospheric signatures spectrum hd209458 during transits planet shkolnik et al . 2003 suggested chomospheric activity induced planet orbit searched signature stellar chromospheric lines observations hd209458 uv vlt / uves 0.33 - 0.67 @xmath1 m region high spectral resolution 1 transit covered r=70000 december 1999 blue 0.328 - 0 red - 0.6686 6 transits covered r=100000 june 2000 june september 2002 red 0.475 - 0.680 searched for ions neutral molecules planet s exosphere evaporated material 1 transit infrared vlt / isaac medium resolution r=10000 ) he i line at1.083 @xmath1 m ( - 1.1 sample spectra fig . 1a full transit isaac june 2001 1.083 @xmath1 m no detection strong instrumental fringing 3 @xmath0 upper limit 0.5% deep absorption feature derived shkolnik searched signatures chromospheric activity induced by heating star s outer atmosphere planetrecovered high - resolution uves spectra blue arm transit 2000 ca ii k at 3933 3968 al i at 3944lines related chromospheric activity sample spectra ca ii k feature shown fig . spectra show emission enhancement ca ii h band nothing significant two otherbands variations indicative stellar chromospheric activity planet s magnetic field uves data 5 transits june - september 2002 analysed exospheric planetary signature further observations required time allocation this year moutou coustenis schneider j queloz mayor m. 2003 , 405 , 341 shkolnik walker bohlender 2003 submitted | 0.134629 | 735 | 387 | , Saving $0.0 in GPT-4. |
54,986 | jet fragmentation in @xmath0+@xmath0 and au+au collisions is studied via back - to - back correlations of direct photons and charged hadrons .
the direct photon correlations are obtained by statical subtraction of the background from decay photons .
results on the nuclear modification to the associated charged hadron yields are reviewed .
further studies of jet fragmentation in @xmath0+@xmath0 using isolated direct photons are also presented .
a -smeared lo pqcd calculation is used to interpret the data .
the sensitivity of the data to the underlying fragmentation function is tested and the results are found to be compatible with expectations of a sample dominated by quark jet fragmentation . | the suppression of high hadrons and the modification of back - to - back di - hadron correlations in nuclear collisions are generally understood to be manifestations of in - medium parton energy loss @xcite .
the quantitive interpretation of such measurements is , however , complicated by the fact that the initial parton energy is unknown .
the dominant mechanism for direct photon production at lo in pqcd is the quark - gluon compton scattering process , @xmath1 . compared to partons
, photons have a small cross section to interact with the medium .
modulo higher order effects , the direct photon can hence be used to determine the initial of the recoil parton before energy loss , since they exactly balance . using the photon as a proxy for the parton
, the fragmentation function of the recoil jet , which may be effectively modified by the medium , can be studied in detail @xcite .
experimentally , the challenge of direct photon measurements is to separate them from the large background of photons produced from hadron decays .
direct photon correlations with charged hadrons may be obtained by a statistical subtraction of the decay component as follows .
first the per - trigger yield ( @xmath2 ) of all photons is measured , which is the number of photon - hadron pairs divided by the number of photons , i.e. , @xmath3 .
this quantity can be decomposed into contributions from direct and decay sources : @xmath4 where direct is taken to signify any photon not from hadron decay .
we may solve for @xmath5 , @xmath6 where the direct photon excess , @xmath7 , is known from measurements of the direct photon cross section @xcite .
@xmath8 is estimated from and @xmath9-hadron correlations which compose more than 95% of the decay background .
the daughter photon correlations are determined from those of their parent mesons by a simulation which takes into account decay kinematics as well as detector resolution , acceptance and efficiency @xcite .
figure [ fig : dphi_stat ] shows examples of azimuthal correlations , @xmath10 , for direct photons in @xmath0+@xmath0 and central au+au collisions , as well as the corresponding total and decay photon correlations used in the statistical subtraction @xcite . on the near - side @xmath5 is consistent with zero , as would be expected at lo .
the disappearance of back - to - back correlations in central au+au , a phenomenon well - known from di - hadron correlations , is also evident in @xmath5 .
fig [ fig : iaa ] shows the nuclear modification factor , which is the ratio of @xmath2 in au+au to @xmath2 in @xmath0+@xmath0 , as a function of collision centrality as quantified by the number of participants .
also shown is for -hadron correlations @xcite as well as for single s @xcite which both demonstrate a level of modification consistent with the direct photon correlations within sizable uncertainties .
the parton path - length through the medium is expected to vary between the three observables such that for a medium with an extended region of partial transmission , one should observe a different level of modification amongst these observables @xcite .
in contrast , the results suggest that the medium is sufficiently absorbent that the path - length difference is beyond the sensitivity of the current measurements .
although initial measurements of the fragmentation function using direct photon correlations have been made ( not shown ) , no modification is apparent within the sizable statistical and systematic uncertainties @xcite .
nuclear modification factors for ( ) and direct photon - associated and -associated charged hadrons ( ) . ] nuclear modification factors for ( ) and direct photon - associated and -associated charged hadrons ( ) . ]
the uncertainties in the direct photon correlations are predominantly due to the statistical subtraction of the large decay background .
higher precision may be achieved by removing decay photons event - by - event .
this is accomplished by tagging decay photon pairs from and @xmath9 based on their invariant mass and by applying an isolation cut .
given a finite detector acceptance and efficiency , a residual background remains after the application of these cuts which is subtracted at the statistical level @xcite .
these methods are difficult to apply in high multiplicity nuclear collisions , but are relatively straight - forward to apply in @xmath0+@xmath0 .
figure [ fig : dphi_iso ] compares azimuthal correlations for direct photons using the tagging and isolation to those obtained from a fully statistical subtraction .
the uncertainties are notably reduced . to study the fragmentation of the away - side jet , the quantity ( @xmath11 ) is used as a proxy for the fragmentation variable @xmath12 , which is a good approximation for @xmath13 .
figure [ fig : xe ] shows the away - side per - trigger yield as a function of for isolated direct photon triggers in @xmath0+@xmath0 collisions .
charged hadron yield per isolated direct photon as a function of ( scaled for visibility ) with power - law fits @xcite . ]
charged hadron yield per isolated direct photon as a function of ( scaled for visibility ) with power - law fits @xcite . ]
the values of the parameter @xmath14 of power - law fits to the distributions ( dn / d@xmath15 ) are shown in fig .
[ fig : slope ] . to test the sensitivity of the data to the underlying fragmentation function
, a lo pqcd calculation was performed adding a phenomenological gaussian smearing whose width was determined from data @xcite .
the calculation was performed for both the compton scattering and annihilation sub - processes , which correspond to quark and gluon fragmentation , respectively .
the data lie closer to the calculation for the compton sub - process which is expected to dominate .
photons are more likely to oppose an up than a down quark due to the valence quark content of the proton and the dependence of the compton scattering amplitude on the electric charge of the quark .
hence one should observe an excess of positive charge for away - side hadrons which is confirmed by data in fig [ fig : charge ] .
also shown is the lo+smearing calculation .
charge asymmetry of partner hadrons for direct photon and triggers @xcite . ] charge asymmetry of partner hadrons for direct photon and triggers @xcite . ]
we have presented results on two - particle correlations using direct photon triggers in @xmath0+@xmath0 and au+au collisions . using a purely statistical method , we bserve the disappearance of back - to - back correlations in central au+au collisions similar to previous observations in the di - jet channel .
the level of suppression is consistent to that observed for single and di - hadrons , within the current statistical and systematic errors . in @xmath0+@xmath0 collisions ,
more precise results were obtained by the event - by - event identification of a large fraction of the decay photon background contribution .
this was achieved using invariant mass tagging of decay pairs as well as isolation cuts .
the data were interpreted in terms of a model in which a gaussian -smearing was applied to lo pqcd .
the recoil jet was found to be compatible with quark fragmentation rather gluon fragmentation , as expect for sample dominated by lo quark - gluon compton scattering .
a charge asymmetry of the away - side jet fragments was observed , further confirming the dominance of the quark - like nature of the away - side jet fragmentation .
the application of event - by - event methods points the way to future studies in which the uncertainties on the heavy - ion measurement may be reduced .
furthermore , the results demonstrate that direct photons are an effective quark jet tag , a feature which may be exploited in heavy - ion collisions to study the color factor dependence of parton energy loss .
99 adcox k _
et al . _ ( phenix ) 2002 suppression of hadrons with large transverse momentum in central au+au collisions at @xmath16 = 130 gev , _ prl _ * 88 * 022301(_preprint _ nucl - ex/0109003 ) adcox k _
_ ( phenix ) 2003 dense - medium modifications to jet - induced hadron pair distributions in au+au collisions at @xmath16 = 200-gev , _ prl _ * 97 * 052301(_preprint _ nucl - ex/0507004 ) for a review of medium modifications to fragmentation functions see arleo f 2009 ( medium - modified ) fragmentation functions , _ epj _ * c61 * 603 ( _ preprint _
arxiv:0810.1193 ) adler s s _ et al . _
( phenix ) 2005 centrality dependence of direct photon production in @xmath16 = 200 gev au + au collisions , _ prl _ * 94 * 23230 ( _
preprint _ nucl - ex/0503003 ) adler s s _ et al . _
( phenix ) 2006 measurement of direct photon production in @xmath0+@xmath0 collisions at @xmath17 = 200 gev , _ prl _ * 98 * 012002 ( _ preprint _ hep - ex/0609031 ) the statistical subtraction method and estimate of decay photon correlations are describe in detail in adare a _ et al . _
( phenix ) 2010 photon - hadron jet correlations in @xmath0+@xmath0 and au+au collisions at @xmath17 = 200 gev , _ prc _ * 80 * 024908 ( _ preprint _
arxiv:0903.3399 ) adare a _ et al . _
( phenix ) 2010 transition in yield and azimuthal shape modification in dihadron correlations in relativistic heavy ion collisions , _ prl _ * 104 * 252301 ( _ preprint _
arxiv:1002.1077 ) adare a _ et al . _
( phenix ) 2008 suppression pattern of neutral pions at high transverse momentum in au + au collisions at @xmath16 = 200 gev and constraints on medium transport coefficients , _ prl _ * 101 * 232301(_preprint _ arxiv:0801.4020 ) zhang h , owens j f , wang e and wang x - n 2007 dihadron tomography of high - energy nuclear collisions in nlo pqcd , _ prl _ * 98 * 12301(_preprint _ nucl - th/0701045 ) + zhang h , owens j f , wang e and wang x - n 2009 tomography of high - energy nuclear collisions with photon - hadron correlations , _ prl _ * 103 * 032302 ( _ preprint _
arxiv:1006.1347 ) connors m 2009 direct photon - hadron correlations measured with phenix , _ npa _ * 830 * 447c-450c ( _ preprint _
arxiv:0907.457 ) adare a _ et al . _
( phenix ) 2010 high @xmath18 direct photon and -triggered azimuthal jet correlations in @xmath19 gev @xmath0+@xmath0 collisions _ preprint _ | suppression high hadrons modification di - hadron correlations in nuclear collisions manifestations in - medium parton energy loss quantitive interpretation complicated initial parton energy unknown dominant mechanism direct photon production at lo pqcd quark - gluon compton scattering process photons small cross section interact with medium direct photon used determine initial recoil parton before energy loss using photon proxy for parton fragmentation function recoil jet modified by medium studied challenge direct photon measurements separate from large background photons hadron decays direct photon correlations with charged hadrons obtained by statistical subtraction of decay component per - trigger yield of all photons measured number of photon - hadron pairs divided by number of photons quantity decomposed into contributions from direct decay sources direct any photon not from hadron decay direct photon excess known from measurements direct photon cross section estimated from-hadron correlations more than 95% of decay background daughter photon correlations determined from parent mesons by simulation decay kinematics detector resolution acceptance efficiencyfigure dphi_stat shows azimuthal correlations @xmath10 for direct photons @xmath0+@xmath0 central au+au collisions total decay photon correlations statistical subtraction @xcite @xmath5 consistent with zero disappearance of back - to - back correlations in central au+au evident in @xmath5 fig shows nuclear modification factor ratio of @xmath2 in au+au to @xmath2 in @xmath0+@xmath0 function of collision centrality quantified by number of participants for -hadron correlations @xcite single s @xcite demonstrate modification consistent with direct photon correlations sizable uncertainties parton path - length medium vary between three observables modification results suggest medium absorbent path - length difference beyond sensitivity current measurements initial measurements of fragmentation function using direct photon correlations no modification within sizable statistical systematic uncertainties @xcite nuclear modification factors for ( ) direct photon - associated and -associated charged hadrons uncertainties in direct photon correlations due to statistical subtraction of large decay background higher precision by removing decay photonsaccomplished tagging decay photon pairs @xmath9 mass applying isolation cut finite detector acceptance efficiency residual background remains after subtracted at statistical level @xcite methods difficult in high multiplicity nuclear collisions straight - forward in @xmath0+@xmath0 figure compares azimuthal correlations for direct photons using tagging isolation statistical subtraction uncertainties reduced fragmentation away - side jet quantity @xmath11 ) proxy for fragmentation variable @xmath12 approximation for @xmath13 figure shows away - side per - trigger yield isolated direct photon triggers in @xmath0+@xmath0 collisions charged hadron yield per isolated direct photon scaled visibility fits @xcite values parameter @xmath14 power - law fits distributions dn d@xmath15 shown in fig test sensitivity fragmentation function pqcd calculation performed adding phenomenological gaussian smearing width determined from data @xcite calculation performed for compton scattering annihilation sub - processes quark gluon fragmentation data closer to calculation for compton sub - processphotons oppose up down quark due to valence quark content proton dependence compton scattering amplitude electric charge quark excess positive charge away - side hadrons confirmed by data fig charge lo+smearing calculation charge asymmetry partner hadrons direct photon triggers presented results two - particle correlations direct photon triggers in @xmath0+@xmath0 au+au collisions method disappearance back - to - back correlations in central au+au collisions di - jet channel level suppression consistent single di - hadrons current statistical systematic errors @xmath0+ collisions precise results obtained event - by - event identification decay photon background contribution achieved invariant mass tagging decay pairs isolation cuts data interpreted model gaussian -smearing applied to lo pqcd recoil jet compatible with quark fragmentation gluon fragmentation charge asymmetry away - side jet fragments observed confirming dominance quark - like nature jet fragmentation application event - by - event methods future studies uncertainties heavy - ion measurementresults demonstrate direct photons effective quark jet tag exploited heavy - ion collisions study color factor dependence parton energy loss adcox k 2002 suppression hadrons large transverse momentum central au+au collisions @xmath16 = 130 gev 88 * 022301 ex/0109003 2003 dense - medium modifications jet - induced hadron pair distributions au+au collisions @xmath16 = 200-gev 97 * 052301 ex/0507004 review medium modifications fragmentation functions arleo 2009 medium - modified fragmentation functions 603 arxiv:0810.1193 adler 2005 centrality dependence direct photon production @xmath16 = 200 gev collisions 94 * 23230 ex/0503003 adler 2006 measurement direct photon production @xmath0+ collisions @xmath17 = 200 gev 98 * 012002 ex/0609031 statistical subtraction method estimate decay photon correlations adare a 2010 photon - hadron jet correlations @xmath0+ au+au collisions @xmath17 = 200 gev 80 * 024908 arxiv:0903.3399phenix 2010 transition yield shape modification dihadron correlations heavy ion collisions prl 104 * 252301 arxiv:1002.1077 adare a phenix 2008 suppression neutral pions high transverse momentum au au collisions @xmath16 200 gev constraints medium transport coefficients 101 * 232301 arxiv:0801.4020 zhang owens wang wang 2007 dihadron tomography high - energy nuclear collisions 98 * 12301 2009 tomography high - energy nuclear collisions photon - hadron 103 * 032302 arxiv:1006.1347 connors 2009 direct photon hadron correlations 830 * 447c-450c arxiv:0907.457 adare a phenix 2010 high @xmath18 direct photon -triggered azimuthal jet correlations @xmath19 gev @xmath0 collisions | 0.358835 | 2,420 | 1,201 | , Saving $0.1 in GPT-4. |
55,004 | we present new photometric observations of eclipsing binary systems v1241 tau and gq dra .
we use the following methodology : initially , wd code is applied to the light curves , in order to determine the photometric elements of the systems .
then the residuals are analysed using fourier transformation techniques .
the results show that one frequency can be barely attributed to the residual light variation of v1241 tau , while there is no evidence of pulsation on the light curve of gq dra . | v1241 tau was first observed by henrietta leavitt ( @xcite ) .
although @xcite defended a @xmath0 sct type variation on the light curve of the system , @xcite mentioned that no trace of pulsation can be seen .
the light variation of gq dra was first determined by hipparcos ( @xcite ) .
bvri light curves of v1241 tau were obtained with 0.4-m telescope of the university of athens observatory in november 2012 .
gq dra was observed with the 1.22-m telescope of the onsekiz mart university observatory in 7 nights between march - april 2013 .
light curves of both systems were analysed using phoebe ( pra and zwitter 2005 ) software .
results show that the system v1241 tau has a semi - detached configuration where the inclination is about 81@xmath1.5 and the mass ratio is 0.44 .
the hotter and cooler components has the temperature values of 7500 k and 4906 k , respectively .
we concluded that 95 percent of the light in @xmath2 filter comes from the primary component .
our solution of semi - detached binary gq dra is the first light curve solution in the literature .
mass ratio of the system is found to be 0.25 and the orbital inclination was calculated as 75@xmath1.3 during the solution .
the temperature of the secondary derived as 5050 k where the primary s value was fixed to 8750 k. the agreement between results and the observations is drawn in fig .
we tried to represent the residuals with a periodic variation using fourier transformation techniques . in the case of v1241 tau
, our research has resulted in one frequency ( f=2.13 c / d ) which can be attributed to a change in the light curve .
however , it is worth to emphasize that this kind of variation may be assigned to other reasons as well , such as observational effects and atmospheric conditions .
the same method was applied to the residuals yielded from the binary solution of gq dra which showed no remarkable oscillation - like variation .
the main result regarding the pulsational behaviour of the systems is that there is no trace of pulsations in neither of them .
the residual light curve of the system v1241 tau can be represented by a periodic variation , however , it is not satisfactorily enough to mention any physical ( low amplitude ) oscillation and it may be assigned to other reasons which mentioned in the previous section . | v1241 tau observed by henrietta leavitt @xcite defended variation no pulsation light variation gq dra determined by hipparcos ) light curves v1241 tau obtained with 0.4-m telescope university of athens observatory november 2012 gq dra observed with 1.22-m telescope onsekiz mart university observatory 7 nights march - april 2013 light curves analysed using phoebe software v1241 tau semi - detached configuration inclination 81@xmath1.5 mass ratio 0.44 hotter cooler components temperature values 7500 4906 k 95 percent light in @xmath2 filter from primary component semi - detached binary gq dra first light curve solution literature mass ratio 0.25 orbital inclination 75@xmath1.3 temperature secondary 5050 k primary s 8750 k agreement between results observations in fig . residuals with periodic variation using fourier transformation techniques v1241 tau one frequency ( f=2.13 c / d ) to change in light curve variation other reasons observational effects atmospheric conditions method applied to residuals binary solution gq dra no oscillation - variationmain result pulsational behaviour systems no trace pulsations in neither residual light curve system v1241 tau represented by periodic variation not enough physical low amplitude oscillation may assigned to other reasons previous section | 0.131673 | 538 | 274 | , Saving $0.0 in GPT-4. |
55,014 | in central regions of non - axisymmetric galaxies high - resolution hydrodynamical simulations indicate spiral shocks , which are capable of transporting gas inwards . the efficiency of transport is lower at smaller radii ,
therefore instead of all gas dropping onto the galactic centre , a roughly uniform distribution of high - density gas develops in the gaseous nuclear spiral downstream from the shock , and the shear in gas is very low there .
these are excellent conditions for star formation .
this mechanism is likely to contribute to the process of ( pseudo- ) bulge formation . | galaxies with non - axisymmetric mass distribution induce efficient gas inflow , which can span throughout most of the galaxy , leading to the accumulation of gas in the galaxy centre , on scales comparable to the resolution limits of observations or computation . in some cases ,
high - resolution data indicate presence of nuclear rings in the innermost few hundred parsecs of a galaxy , where gas density is high and star formation occurs .
kormendy & kennicutt ( 2004 ) postulated that stars created there eventually give rise to pseudobulges .
although formation of nuclear rings can be studied in detail with grid - based hydrodynamical modelling within the eulerian scheme ( e.g. piner et al 1995 ) , evolution of galaxies is most often studied within the lagrangian scheme ( e.g. sph methods ) , which appears more flexible .
however , these studies usually do not resolve the central gas concentration resulting from inflow ( e.g. di matteo et al . 2007 ) , usually called the central blob . sph studies that resolve nuclear rings in some cases , in other cases still produce a central blob ( patsis & athanassoula 2000 ) . on the other hand , high - resolution eulerian models indicate that aside for nuclear rings , nuclear spirals can form in centres of galaxies ( englmaier & shlosman 2000 , maciejewski 2000 , maciejewski et al .
recently , most detailed studies within the lagrangian scheme ( ann & thakur 2005 ) demonstrated that the central blob can be resolved into a spiral , and is equivalent to nuclear spirals seen in the models built with the eulerian scheme . in maciejewski ( 2004b , hereafter m04b )
, i showed that nuclear ring and spiral are two modes of wave propagation in centres of galaxies . here
i focus on nuclear spirals , since nuclear rings have been already studied in detail .
generation of nuclear spirals can be well explained within the framework of gas response to a fixed rotating stellar potential ( englmaier & shlosman 2000 ) .
gravity torques generate waves in gas that for any power - law rotation curve can propagate between the centre of the galaxy and the single inner lindblad resonance present in that case ( maciejewski 2004a ) .
the waves in this region give rise to a spiral morphology in gas with the pitch angle proportional to the velocity dispersion in gas .
centres of non - axisymmetric galaxies act as resonant cavities , and generation of waves is inevitable there .
( contours ; only for div@xmath0 ) in model 8s20r of gas flow in a barred galaxy from m04b .
darker shading indicates higher density .
_ right : _ for the same model , @xmath1 ( shear ) is shown in greyscale , with the highest gas density overplotted in contours ( unlike in the left panel ) .
darker shading indicates larger shear .
the snapshot was taken at 0.5 gyr , after the flow has stabilized .
the bar is 6 kpc long and is vertical on the plots .
the box is 2 kpc long.,title="fig : " ] ( contours ; only for div@xmath0 ) in model 8s20r of gas flow in a barred galaxy from m04b .
darker shading indicates higher density .
_ right : _ for the same model , @xmath1 ( shear ) is shown in greyscale , with the highest gas density overplotted in contours ( unlike in the left panel ) .
darker shading indicates larger shear .
the snapshot was taken at 0.5 gyr , after the flow has stabilized .
the bar is 6 kpc long and is vertical on the plots .
the box is 2 kpc long.,title="fig : " ] in m04b , i studied nuclear spirals with grid - based models constructed within the eulerian scheme .
i showed that models for small departures from axial symmetry are in excellent agreement with the analytical solution , which provided an anchor for considering strongly barred models .
gas density and the location of shocks in nuclear spirals was given in m04b .
shocks were traced by large negative values of the divergence of gas velocity field .
divergence is a measure of expansion of the fluid , and its highly negative values should indicate gas compression in shocks .
here , i add one more characteristic of the flow : its shear .
the measure of shear in the plane of the galactic disc should take a form @xmath2 , but this form is not invariant under rotation of the coordinate system .
however , shear can still be characterized by a single invariant value for a two - dimensional flow in a plane .
this value can be derived from the stress tensor @xmath3 , after extracting its symmetric part , and leaving out the asymmetric one , responsible for the curl of the velocity field .
furthermore , the trace has to be separated out , since it solely contains terms corresponding to the expansion of the fluid , i.e. the divergence of the velocity field . as a result ,
one is left with the tensor @xmath4 , which describes the shear of the flow only .
this is a traceless , symmetric tensor .
the measure of its magnitude can be obtained from its eigenvalues . in two dimensions ,
the two eigenvalues of a traceless tensor have opposite signs , but the same magnitude @xmath5 , which describes the amplitude of shear , and is expressed by @xmath6 here i calculated the shear of gas velocity field in model 8s20r from m04b , where the nuclear spiral is present .
fig.1 displays the gas density , square of divergence of the velocity field , div@xmath7 , as the shock indicator , and @xmath1 as the indicator of the shear .
following the flow in fig.1 from outside in , one can notice two almost straight lanes , with density slightly above average , and strong shock and shear .
these lanes , almost vertical in fig.1 , mark the principal shocks in the bar that often manifest themselves as a pair of straight dust lanes . as expected , the highest density in gas is located downstream from the shock ( gas rotates counterclockwise in fig.1 ) , and high shear extends downstream from the shock as well . when the flow is followed inwards , the principal straight shock takes a spiral morphology , which i call the nuclear spiral ( see also englmaier & shlosman 2000 , maciejewski et al .
2002 , m04b ) .
the divergence of gas velocity field remains highly negative , which indicates that the nuclear spiral is a shock in gas .
the highest density in gas is still downstream from the shock , as expected . on the other hand ,
the distribution of shear in gas in the region of the nuclear spiral is more complicated and very interesting . as the principal shock curves inwards and turns into the nuclear spiral , at azimuthal angles somewhere half - way between the minor and the major axis of the bar , the thick dark lane marking large shear in the right - hand panel of fig.1 splits .
one branch is exactly co - spatial with the shock this indicates large shear in the nuclear spiral shock .
the other branch curves less than the spiral , splits again , and disappears around the major axis of the bar . between these two branches of high shear
, there is a pocket of the lowest shear in the region , immediately downstream from the nuclear spiral shock , and exactly co - spatial with the post - shock gas of highest density .
thus the high - density post - shock gas that forms the gaseous nuclear spiral experiences very little shear , contrary to the post - shock gas in the straight shock in the bar .
this difference becomes clear once the region presented in fig.1 is divided into elements , and for these elements the shear , quantified by @xmath1 , is plotted against overdensity , i.e. the ratio of the measured density to the initial one in the model , as in fig.2 .
the points in the plot group in two perpendicular stripes : the vertical one , with overdensity about 10 and high shear , and the horizontal one , with low shear , below 0.5 , but highest overdensities , reaching 100 .
it can be shown that the vertical stripe contains gas emerging from the straight shock in the bar , while the post - shock gas in the nuclear spiral groups in the horizontal stripe .
elements of the model from fig.1 . ]
presence of shocks and shear influences star formation : shocks trigger the collapse of molecular clouds , while shear may destroy the clouds .
thus one should expect enhanced star formation in the presence of shocks , but this may be inhibited by strong shear . in model 8s20r , analyzed above , gas orbiting around the galaxy centre passes through the nuclear spiral shock roughly every 10 myr .
immediately downstream from the shock , it is compressed in gaseous spiral arms , whose density is a few times higher than the inter - arm density , and where shear is negligible .
these are excellent conditions for star formation .
therefore in nuclear spirals star formation should occur on much higher rate than in the case when the inflowing gas gathers in an amorphous blob , which lacks shocks and post - shock high - density low - shear gas .
note however that star formation should not be confined to the gaseous spiral arms .
its timescale is of the order of crossing time between the arms ( 10 myr ) , and one should expect that only the first step of star formation , the formation of self - gravitating cores , occurs in the arms .
however , young stars should appear at any azimuthal angle . therefore , although star formation is triggered by the spiral shock , it is distributed in a disc or wide ring .
the dynamics of nuclear spirals presented here is determined only by gas response to an external gravitational potential .
this process is well understood and its workings can be studied in detail . obviously , self - gravity in gas , and feedback from star formation will make the picture presented here more complicated .
however , once these processes are included , results become uncertain , as star formation is poorly understood .
since in nuclear spirals compression of gas is induced by the spiral shock , the role of self - gravity in gas may not be dominant there .
also the feedback from star formation is likely to be weakened by the fact that most of the stars are not formed in the dense gaseous nuclear spiral .
i presented here the mechanism of central mass accumulation in nuclear spirals . when a nuclear spiral is present in the centre of a galaxy , gas passes through the nuclear spiral shock every 10 myr or so , with no shear in the post - shock compressed gas .
this reoccurring condition for star formation makes something like a cycle engine that efficiently converts gas into stars in the galaxy centre .
stars are also formed in nuclear rings , but the distribution of material accumulated in nuclear spirals is more bulge - like , since the spiral spans over a considerable radial range
. star formation rate in nuclear spiral is likely to be higher than in nuclear ring , since the nuclear spiral shock accompanies the gaseous spiral throughout its extent , while the nuclear ring damps the shock ( see m04b ) .
nuclear spirals might have played a crucial role in the early universe , since they form in gas with high velocity dispersion , likely present in newly formed galaxies .
ann ( 2005 ) also showed that nuclear spirals form preferably in less massive galaxies . like nuclear rings
, nuclear spirals may be responsible for the formation of pseudobulges , or , as it was pointed out during this meeting , disky bulges , since we still do not know how to lift gas or stars from the galactic plane . | galaxies with non - axisymmetric mass distribution induce gas inflow to accumulation gas in galaxy centre comparable to resolution limits high - resolution data indicate nuclear rings in innermost parsecs galaxy where gas density high star formation occurs kormendy & kennicutt 2004 ) stars give rise to pseudobulges formation of nuclear rings studied with grid - hydrodynamical modelling eulerian scheme evolution often studied within lagrangian scheme sph methods ) more flexible studies do not resolve central gas concentration from inflow 2007 central blob sph studies resolve nuclear rings produce central blob ( athanassoula 2000 high - resolution eulerian models indicate nuclear spirals can form in centres galaxies detailed studies lagrangian scheme 2005 demonstrated central blob can be resolved into spiral equivalent to nuclear spirals in models eulerian scheme maciejewski 2004b showed nuclear ring and spiral are two modes of wave propagation in centres galaxies focus on nuclear spirals studied generation of nuclear spirals explained within gas response to fixed rotating stellar potential ( 2000gravity torques generate waves in gas propagate between centre galaxy inner lindblad resonance maciejewski 2004a waves give spiral morphology gas pitch angle proportional to velocity dispersion centres of non - axisymmetric galaxies as resonant cavities generation waves inevitable div@xmath0 model 8s20r of gas flow barred galaxy from m04b darker shading indicates higher density @xmath1 ( in greyscale highest gas density overplotted in contours larger shear snapshot taken at 0.5 gyr bar 6 kpc long vertical on plots box 2 kpc long model 8s20r gas flow barred galaxy m04b darker shading indicates higher density @xmath1 ( in greyscale highest gas density overplotted in contours darker shading indicates larger shear snapshot taken at 0.5 gyr bar 6 kpc long vertical plots box 2 kpc long in m04b studied nuclear spirals with grid - based models eulerian scheme models for small departures from axial symmetry with analytical solution anchor for considering strongly barred modelsgas density location of shocks in nuclear spirals in m04b shocks traced by negative values divergence of gas velocity field divergence measure expansion fluid negative values indicate gas compression in shocks characteristic flow shear measure shear in plane galactic disc form @xmath2 not invariant under rotation coordinate system shear characterized by single invariant value for two - dimensional flow value derived from stress tensor @xmath3 extracting symmetric part asymmetric curl velocity field trace separated contains terms expansion fluid divergence velocity field tensor @xmath4 describes shear flow traceless symmetric tensor measure magnitude from eigenvalues two eigenvalues traceless tensor opposite signs same magnitude @xmath5 describes amplitude of shear expressed by @xmath6 calculated shear of gas velocity field in model 8s20r from m04b nuclear spiral present fig.1 displays gas density square of divergence of velocity field div@xmath7 shock indicator @xmath1 indicator shear flow fig.1 two straight lanes density above average strong shock shear lanes vertical in fig.mark principal shocks in bar as straight dust lanes highest density in gas downstream from shock gas rotates counterclockwise in fig.1 high shear extends downstream flow inwards principal straight shock takes spiral morphology nuclear spiral englmaier & shlosman 2000 maciejewski et 2002 m04b divergence of gas velocity field negative nuclear spiral is shock in gas highest density in gas downstream distribution of shear in gas in nuclear spiral complicated principal shock curves inwards turns into nuclear spiral minor major axis dark lane large shear splits one branch co - spatial with shock large shear in nuclear spiral shock other branch curves less splits disappears around major axis between pocket of lowest shear downstream from nuclear spiral shock co - spatial with post - shock gas highest density high - density post - shock gas gaseous nuclear spiral experiences little shear contrary to post - shock gas in straight shock bar difference clear once region.1 divided into elements shear quantified by @xmath1 plotted against overdensity ratio of measured density to initial modelpoints in plot group two stripes vertical overdensity 10 high shear horizontal low shear below 0.5 highest overdensities 100 vertical stripe contains gas from straight shock post - shock gas nuclear spiral groups horizontal stripe model fig.1 shocks shear influences star formation shocks trigger collapse molecular clouds shear clouds enhanced star formation shocks inhibited by strong shear model 8s20r gas galaxy centre passes through nuclear spiral shock every 10 myr compressed in gaseous spiral arms density higher inter - arm shear negligible excellent conditions for star formation nuclear spirals star formation higher rate than inflowing gas in amorphous blob shocks post - shock high - density low - shear gas star formation not confined to gaseous spiral arms timescale crossing time between arms 10 myr first step formation self - gravitating cores in arms young stars appear at any azimuthal angle star formation triggered by spiral shock distributed in disc wide ring dynamics of nuclear spirals determined by gas response to external gravitational potential process studied in detailself - gravity in gas feedback from star formation make picture complicated results uncertain star formation poorly understood in nuclear spirals compression gas induced by spiral shock self - gravity may not dominant feedback from star formation weakened by stars not formed in dense gaseous nuclear spiral mechanism of central mass accumulation in nuclear spirals nuclear spiral centre gas passes through every 10 myr no shear in post - shock compressed gas reoccurring condition for star formation cycle engine converts gas into stars stars formed in nuclear rings distribution material in nuclear spirals more bulge - like spiral spans radial range star formation rate in nuclear spiral higher than nuclear ring shock accompanies gaseous spiral nuclear ring damps shock nuclear spirals in early universe form in gas with high velocity dispersion in newly formed galaxies spirals form in less massive galaxies spirals may responsible for formation pseudobulges disky bulges know how to lift gas stars from galactic plane | 0.363197 | 2,472 | 1,202 | , Saving $0.1 in GPT-4. |
55,017 | weak interactions in two - nucleon system at low energies are explored in the framework of effective field theory .
we review our resent calculations of parity - violating observables in radiative neutron capture on a proton at threshold where both pionful and pionless theories are employed .
parity violation , effective field theory 21.30.fe | exchanges of @xmath0 , @xmath1 and @xmath2 mesons have long been a standard way to describe the weak interactions of the baryon at the hadronic level .
one of the important issues with the one - meson exchange ( ome ) weak interaction is the value of the weak coupling constants .
a bench marking calculations for the weak coupling constants was done with a quark model @xcite .
calculation of the weak coupling constants since then employed various models and theories , and the results as a whole turned out to be consistent to the best values " in ref .
however , experimental determinations are yet quite controversial , so a revealing true values of the weak coupling constants still remains as an open question . in the present situation
, one can cast a question : is the ome picture sufficient to describe the hadronic weak dynamics ?
it is quite recent that people started to consider the effective field theory ( eft ) , which already achieved good successes in the description of strong interactions and electromagnetic ( em ) transitions in the low energy two - nucleon systems , in search for an answer to the question in the hadronic weak interaction problem @xcite . with the counting rules of eft
, we can expand the two - nucleon weak potential systematically and perturbatively .
interestingly , eft gives terms that are absent in the ome potential , but are more important than the heavy - meson terms in it . in this work we consider parity - violating ( pv ) observables in radiative capture of thermal neutron on a proton . in the polarized neutron capture
, it is known that the pv asymmetry , @xmath3 , is dominated by the pion - exchange term in the weak ome potential .
we thus employ an eft that explicitly includes pions while heavier degrees of freedom are integrated out , and the short range dynamics are represented by two nucleon contact terms . in the unpolarized neutron capture , on the other hand , pv polarization , @xmath4 , is dependent on the @xmath1- and @xmath2- exchange terms in the ome weak potential .
since heavy mesons are integrated out in the eft , interactions relevant to @xmath4 are only the two - nucleon contact terms .
therefore we are naturally lead to consider @xmath4 with a theory where all the interactions are represented only with contact terms , so called the pionless theory .
in the next sections theories are briefly reviewed . in the following section
, we present the results and discussion , and in the last section we summarize the work .
order of a contribution ( equivalently a feynman diagram ) is counted in powers of @xmath5 , where @xmath6 is the momentum scale of the incoming and outgoing particles , and @xmath7 is the chiral symmetry breaking scale .
we count the order of @xmath5 of a diagram with the counting rules of the heavy baryon chiral perturbation theory @xcite .
weak pion - nucleon lagrangian is given by @xmath8 where @xmath9 is a weak pion - nucleon coupling constant .
at the leading order ( lo ) ( @xmath10 ) , only the one - pion - exchange ( ope ) diagram contributes to the weak potential .
there is no diagram at the order of @xmath11 , and the next - to - next leading order ( nnlo ) consists of the two - pion - exchange ( tpe ) contributions and a two - nucleon contact term ( ct ) which has an unknown low energy constant ( lec ) .
diagrams at lo and nnlo are shown in fig .
[ fig:1 ] . weak potentials in fig .
[ fig:1 ] in coordinate space can be written in the form @xmath12 , \label{eq : general_form}\end{aligned}\ ] ] where @xmath13 is the momentum operator conjugate to the relative coordinate @xmath14 . in the fourier transformation from momentum space to coordinate one , we encounter ultraviolet divergence for the tpe and contact terms . in order to regularize the divergence
, we introduce the monopole cutoff function @xmath15 for tpe and contact terms .
thus the regularized ope , tpe and ct terms @xmath16 in eq .
( [ eq : general_form ] ) read @xmath17 \right\ } , \label{eq : vr_tpe } \\
v^\lambda_{\rm ct}(r ) & = & - c^r_6 \lambda^2 \frac{{\rm e}^{-\lambda r}}{4 \pi r}\ , .
\label{eq : vr_ct}\end{aligned}\ ] ] analytic expressions of the functions @xmath18 and @xmath19 can be found in ref .
@xcite , and @xmath20 is the renormalized lec in the contact term . the lec @xmath20 can be determined model independently with experimental data , but error bars of the data for the hadronic weak interaction are too large to constrain the lec with reasonable uncertainty . in the theoretical calculation , @xmath20 is dependent on the regularization schemes , and several choices are possible . in this work
we consider the minimal subtraction scheme , in which the renormalized lec takes the form @xmath21,\ ] ] where @xmath22 is the bare lec , a function of the renormalization point @xmath23 @xcite , and @xmath24 .
when the parity - odd potential is added to the parity - conserving ( pc ) potential , it causes parity mixture in the wave function , and this parity admixture can contribute to em transitions that are forbidden in the pure parity states . with dibaryon fields in the pionless eft
, it has been shown in recent works @xcite that low energy observables can be calculated with good convergence and accuracy .
details for the strong and em interactions in the pionless dibaryon formalism can be found in those papers , and in this work we discuss the pv terms in the theory . at sufficiently low energies ,
pion mass can be treated as a large scale . in this case , just as heavy mesons are integrated out in the pionful theory , the contribution of pion exchange can be subsumed in the two - nucleon contact terms .
then since we do nt have mesons dynamically in the theory , all the interactions are represented by the contact terms .
effective weak lagrangian for the low energy in the pionless eft have been obtained in ref . @xcite .
pionless lagrangian with dibaryon fields can be obtained by replacing a two - nucleon field with a dibaryon one of the same quantum numbers .
we obtain lagrangian for the pv dibaryon - nucleon - nucleon ( @xmath25 ) interaction as @xmath26 where @xmath27 and @xmath28 denote weak @xmath25 coupling constants for parity mixing for the @xmath29 and @xmath30 dibaryon states , respectively .
spin - isospin operator @xmath31 in eq . (
[ eq : lwk0s ] ) projects two - nucleon system to @xmath32 state .
pv vertex given by eq .
( [ eq : lwk0s ] ) therefore mixes @xmath32 in the @xmath29 state .
similarly , @xmath33 in eq .
( [ eq : lwk0 t ] ) is projection operator to @xmath34 state , and thus the lagrangian mixes @xmath34 state in the @xmath30 state .
we note that the terms in eqs .
( [ eq : lwk0s],[eq : lwk0 t ] ) are relevant to the pv polarization in unpolarized neutron capture .
they are part of the general lagrangian that accounts for all the possible parity - mixing modes @xcite .
parity - violating asymmetry @xmath3 in @xmath35 is defined as @xmath36 where @xmath37 is an angle of the photon momentum with respect to the neutron polarization .
we employ a hybrid scheme in the calculation , where only weak potentials are obtained from eft , while strong potential is described with a phenomenological model , argonne v18 and the em operators are assumed to satisfy siegert s theorem .
weak pion - nucleon coupling constant @xmath9 is of the order of @xmath38 so it is sufficient to work at its linear order .
pv asymmetry is then proportional to @xmath9 , and thus we write the result as @xmath39 the coefficient @xmath40 depends on various inputs such as strong interaction , weak interaction , em operators , cutoff value .
the results in table [ tab : result1 ] show the contribution to @xmath40 from ope and tpe terms , and their dependence on the cutoff @xmath41 and renormalization scale @xmath23 .
lcrrrr @xmath41 ( mev ) & & 500 & 1000 & 1500 & 2000 + @xmath42 & & @xmath43 & @xmath44 & @xmath45 & @xmath46 + & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath50 + @xmath51 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 + & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 + dependence on the cutoff @xmath41 is non - negligible for both ope and total value ( ope and tpe ) , and it is more drastic for the nnlo ( total subtracted by ope ) .
difference is relatively large between @xmath62 mev and 1000 mev , i.e. about 11% increase in magnitude for lo and about 54% for nnlo .
however , the values become quite stable when @xmath41 is more than 1000 mev , and the change reduces to about 1% for lo and 18% for nnlo . dependence on the renormalization scale @xmath23
is already sizable , and the largest gap is about 15% of the value of the smallest magnitude , @xmath58 .
however , the cutoff dependence almost disappears when the renormalization scale is two times the pion mass , @xmath63 .
these values can be regarded as fixed - point results up to nnlo .
parity - violating polarization @xmath4 in @xmath64 is defined as @xmath65 where @xmath66 and @xmath67 are the total cross section for the photons with right and left helicity , respectively . while we calculate @xmath3 in @xmath68 in a hybrid way , @xmath4 is calculated self - consistently in the pionless dibaryon formalism .
both pc and pv amplitudes are calculated with the same counting rules , and the pc and pv diagrams are truncated at the same order . for the pc amplitude for @xmath64 , we employ the result in ref .
@xcite , @xmath69.\end{aligned}\ ] ] feynman diagrams that contribute to the pv amplitude at lo are shown in fig .
[ fig:2 ] .
taking the zero momentum approximation , we obtain the pv amplitude at lo as @xmath70 , \label{eq : pvz}\end{aligned}\ ] ] and the pv polarization @xmath4 at lo reads @xmath71 since the parameters other than @xmath28 and @xmath27 are well fixed , measurements of @xmath4 provides a relation that can constraint the values of @xmath28 and @xmath27 .
we considered weak interaction at the hadron level with effective field theories . with the pionful theory , we derived weak two - nucleon potential up to nnlo .
parity - violating asymmetry @xmath3 calculated with the weak potential shows that the perturbative expansion converges reasonably , and the results are non negligibly dependent on the cutoff value and renormalization scale .
however , the lo contribution is sufficiently dominant that we can determine the first significant digit of the weak pion - nucleon coupling constant @xmath9 if @xmath3 is measured precisely . with the pionless theory , we calculated the parity - violating polarization @xmath4 in @xmath64 . in the hybrid calculation adopted in the calculation of @xmath3 , orders of the strong interaction and em operators
are not countable because they are quoted from the formulations where perturbative expansion is absent .
therefore , order mismatch could be unavoidable in the hybrid calculation .
on the other hand , @xmath4 is calculated in a single theory , and the order of the diagrams is fixed to a single order .
the result is dependent on the weak coupling constants for @xmath72 mixing and @xmath73 one .
measurement of @xmath4 is expected to provide a constraint to reduce the possible space for these weak coupling constants .
this research was supported by the daegu university research grant , 2008 .
b. desplanques , j.f .
donoghue , b.r .
holstein , ann .
124 ( 1980 ) 449 .
kaplan , m.j .
savage , r.p .
springer , m.b .
wise , phys .
b 449 ( 1999 ) 1 . c.h . hyun .
park , d .-
min , phys .
b 516 ( 2001 ) 321 . m.j .
savage , nucl .
phys . a 695 ( 2001 ) 365 .
zhu , c. m. maekawa , b. r. holstein , m. j. ramsey - musolf and u. van kolck , nucl .
phys . a 748 ( 2005 ) 435 .
liu , phys .
c 75 ( 2007 ) 065501 .
hyun , s. ando , b. desplanques , phys .
b 651 ( 2007 ) 257 . j.w .
shin , s. ando and c.h .
hyun , arxiv:0907.3995v1 [ nucl - th ] .
park , d .-
min , m. rho , phys .
rep . 233 ( 1993 ) 341 .
s. ando and c.h .
hyun , phys .
c 72 ( 2005 ) 014008 .
s. ando , r.h .
cyburt , s.w .
hong , c.h .
hyun , phys .
c 74 ( 2006 ) 025809
. s. ando , j.w .
shin , c.h .
hyun , s.w .
hong , k. kubodera , phys .
b 668 ( 2008 ) 187 . d.r .
phillips , m.r .
schindler , r.p .
springer , nucl .
phys . a 822 ( 2009 ) 1 . | exchanges @xmath0 mesons standard weak interactions baryon at hadronic level important one - meson exchange weak interaction value weak coupling constants bench calculations for constants with quark model @xcite calculation employed various models theories results consistent to best values experimental determinations controversial revealing true values weak coupling constants open question is ome picture sufficient describe hadronic weak dynamics ? recent effective field theory ( eft ) successes strong interactions electromagnetic transitions low energy two - nucleon systems answer hadronic weak interaction problem counting rules of eft expand two - nucleon weak potential systematically perturbatively eft gives terms absent in ome potential more important than heavy - meson terms work consider parity - violating pv ) observables in radiative capture thermal neutron on proton polarized neutron capture pv asymmetry @xmath3 dominated by pion - exchange term in weak ome potential employ eft includes pions heavier degrees freedom integrated out short range dynamics represented by two nucleon contact termsunpolarized neutron capture pv polarization @xmath4 dependent on @xmath1- @xmath2- exchange terms weak potential heavy mesons integrated in eft interactions relevant to @xmath4 two - nucleon contact terms consider @xmath4 interactions contact terms pionless theory next theories reviewed present results discussion summarize work order contribution feynman diagram counted in powers of @xmath5 @xmath6 momentum scale @xmath7 chiral symmetry breaking scale count order @xmath5 with counting rules heavy baryon chiral perturbation theory weak pion - nucleon lagrangian given by @xmath8 @xmath9 weak pion - nucleon coupling constant leading order @xmath10 one - pion - exchange ) diagram contributes to weak potential no diagram at order @xmath11 leading order two - pion - exchange contributions two - nucleon contact term unknown low energy constant diagrams at lo nnlo shown in fig ] weak potentials coordinate space written in form @xmath12 @xmath13 momentum operator conjugate to relative coordinate @xmath14fourier transformation from momentum space to coordinate one encounter ultraviolet divergence for tpe contact terms regularize divergence introduce monopole cutoff function @xmath15 for tpe contact terms regularized ope tpe ct terms @xmath16 eq read @xmath17{eq vr_tpe\lambda ct{eq vr_ct}{aligned expressions functions @xmath18 @xmath19 in ref @xcite @xmath20 renormalized lec contact term lec @xmath20 determined independently with experimental data error bars hadronic weak interaction large constrain lec uncertainty theoretical calculation @xmath20 dependent on regularization schemes several choices possible consider minimal subtraction scheme renormalized lec form @xmath21 @xmath22 bare lec function of renormalization point @xmath23 @xcite @xmath24 parity - odd potential added to parity - conserving pc potential causes parity mixture in wave function transitions forbidden in pure parity states dibaryon fields in pionless eft low energy observables calculated with good convergence accuracystrong em interactions pionless dibaryon formalism in papers discuss pv terms theory low energies pion mass large scale heavy mesons integrated theory contribution pion exchange subsumed in two - nucleon contact terms interactions represented by contact terms weak lagrangian low energy pionless eft obtained in ref @xcite pionless lagrangian with dibaryon fields replacing two - nucleon field with dibaryon same quantum obtain lagrangian pv dibaryon - nucleon - nucleon @xmath25 interaction @xmath26 @xmath27 @xmath28 denote weak @xmath25 coupling constants parity mixing @xmath29 @xmath30 dibaryon states spin - isospin operator @xmath31 projects two - nucleon system to @xmath32 state pv vertex mixes @xmath32 in @xmath29 state @xmath33 projection operator @xmath34 state lagrangian mixes @xmath34 state @xmath30 state terms in eqs relevant to pv polarization in unpolarized neutron capture part of general lagrangian parity - mixing modes @xciteparity - violating asymmetry @xmath3 in @xmath35 defined as @xmath36 @xmath37 angle of photon momentum neutron polarization hybrid scheme calculation weak potentials from eft strong potential described with phenomenological model argonne v18 em operators satisfy siegert s theorem weak pion - nucleon coupling constant @xmath9 order of @xmath38 sufficient work at linear order pv asymmetry proportional to @xmath9 write result as @xmath39 coefficient @xmath40 depends on inputs strong interaction weak interaction em operators cutoff value results in table show contribution to @xmath40 from ope tpe terms dependence on cutoff @xmath41 renormalization scale @xmath23 @xmath41 mev 500 1000 1500 2000 + dependence on cutoff @xmath41 non - negligible for ope and total value drastic for nnlo ( difference large between @xmath62 mev and 1000 mev 11% increase in magnitude for lo 54% for nnlovalues stable when @xmath41 1000 mev change reduces to 1% lo 18% nnlo dependence on renormalization scale @xmath23 sizable largest gap 15% smallest magnitude @xmath58 dependence disappears when renormalization scale two times pion mass @xmath63 fixed - point results up to nnlo parity - violating polarization @xmath4 @xmath64 defined as @xmath65 @xmath66 total cross section photons right left helicity @xmath3 @xmath68 @xmath4 self - consistently pionless dibaryon formalism pc pv amplitudes calculated same counting rules diagrams truncated pc amplitude @xmath64 result @xcite @xmath69 feynman diagrams pv amplitude lo shown in fig pv amplitude at lo as @xmath70 pv polarization @xmath4 lo reads @xmath71 parameters @xmath28 @xmath27 fixed measurements @xmath4 values @xmath28 @xmath27 considered weak interaction hadron level with field theories derived weak two - nucleon potential up to nnloparity - violating asymmetry @xmath3 calculated weak potential shows perturbative expansion converges results dependent on cutoff value renormalization scale lo contribution dominant determine first digit weak pion - nucleon coupling constant @xmath9 if @xmath3 measured precisely pionless theory calculated parity - violating polarization @xmath4 @xmath64 hybrid calculation @xmath3 orders strong interaction em operators not countable quoted from formulations perturbative expansion absent order mismatch unavoidable @xmath4 calculated single theory order diagrams fixed single order result dependent on weak coupling constants @xmath72 mixing @xmath73 one measurement @xmath4 reduce space weak coupling constants research supported daegu university research grant 2008 desplanques donoghue holstein 124 ( 1980 ) 449 kaplan savage springer wise 449 ( 1999 ) hyun park 516 ( 2001 ) 321 savage 695 ( 2001 ) 365 zhu maekawa holstein ramsey - musolf van kolck 748 ( 2005 ) 435liu phys . 75 2007 ) 065501 hyun s ando desplanques phys b 651 ( 2007 ) 257 j shin s ando hyun arxiv:0907.3995v1 park min rho phys . rep 233 ( 1993 ) 341 s ando hyun 72 ( 2005 ) 014008 ando cyburt hyun 74 ( 2006 ) 025809 ando shin hyun hong kubodera phys . b 668 ( 2008 ) 187 phillips schindler springer . 822 ( 2009 ) 1 | 0.499893 | 3,255 | 1,647 | , Saving $0.1 in GPT-4. |
55,018 | shocks are ubiquitous in the interstellar medium of galaxies , where they contribute to the energetic balance and to the cycle of matter , and where they are thought to be the primary sites for cosmic rays acceleration .
most of the time : in jets and outflows , supernova remnants , or colliding flows , they are linked with star formation .
the study of shocks is hence a powerful tool to probe the evolution of the interstellar medium and to better understand star formation . to these aims ,
the most precise observations must be compared with the most precise models of shocks .
the sofia / great instrument represents a powerful observational tool to support our progresses , as it allows to observe numerous shock tracers in the far - infrared range . | the interstellar medium ( ism ) is constantly evolving due to unremitting injection of energy in various forms .
energetic radiation transfers energy to the ism : from the uv photons , emitted by the massive stars , to x- and @xmath0-ray ones . cosmic rays
are another source of energy .
finally , mechanical energy is injected through shocks or turbulence .
shocks are ubiquitous in the interstellar medium of galaxies . they are associated to star formation ( through jets and bipolar outflows ) , life ( via stellar winds ) , and death ( in agb stellar winds or supernovae explosion ) .
the dynamical processes leading to the formation of molecular clouds also generate shocks where flows of interstellar matter collide . because of their ubiquity
, the study of interstellar shocks is also a useful probe to the other mechanisms of energy injection in the ism .
this study must be conducted in order to understand the evolution of the interstellar medium as a whole , and to address various questions detailed in this review . to this aim , it is paramount to interpret the most precise observations with the most precise shock models . from the observational point of view , the sofia telescope , and particularly the great receiver and
the spectral resolution it offers , represents a powerful tool to better address the above questions , as it allows to observe numerous shock tracers in the far - infrared range .
[ ] co spectra ( lower panels ) showing the sofia / great observations .
the bhr71 image is an overlay of 8 @xmath1 m emission ( colours , _
spitzer_/irs ) with co ( 32 ) ( white contours , apex ) .
the cep e image from pdbi shows the jet , the terminal bowshock , and cavity in the southern lobe .
each spatial element corresponds to a spectral range centered on the vertical dashed lines of the lower panel , in the same colours.,title="fig:",scaledwidth=60.0% ] the paradigm for the formation of low - mass stars has been established over the past decades .
it involves an accretion of matter through a disk - like structure , as well as ejection through jets and outflows , generally bipolar ( e.g. fig .
[ figure1 ] ) .
however , a lot of shock - related questions remain : what are the ejection mechanisms of the molecular gas ?
what are the mechanisms leading to the peculiar chemistry observed in numerous low - mass star forming environments ?
what are the energetic impacts of jets and outflows on their environment ? can cosmic rays be accelerated in the associated shocks ? to address these questions , the observation of emission from high-@xmath2 rotational lines of co is an important probe to the temperature and density conditions .
indeed , co is an abundant molecule whose levels for which @xmath3 trace the warm ( @xmath4 k ) and typically shocked medium .
today , sofia / great is the only telescope that allows for their observation .
the chemically rich outflows l1157 ( eislffel _ et al . _
@xcite ) and bhr71 ( gusdorf _ et al . _
@xcite ) have been observed with sofia / great . in the latter case ,
their comparison with state - of - the - art shock models led to precise determinations of masses , momentum and energy transfers .
combined with h@xmath5 and sio observations , they also allowed to better constrain the silicon chemistry .
[ ] m map from cunningham_et al .
_ @xcite shows the complex outflow system , while the spectra show the oh observations by sofia / great at 117 and 163 @xmath1 m . for g5.890.39 , the large - scale ( top , hunter _
@xcite ) and zoomed - in ( middle , su _
@xcite ) maps also illustrate the complexity of the star forming region , while the spectrum shows the velocity - resolved emission / absorption structure in the [ oi ] 63 @xmath1 m line , as seen by sofia / great.,title="fig:",scaledwidth=50.0% ] when climbing the protostellar mass ladder , new questions are asked .
the scenario for massive star formation is less clear ( is it a scaled - up version from the low - mass one ?
do massive stars result from competitive accretion mechanisms , or should other mechanisms be invoked ? ) , and the role of the important protostellar radiation field must be quantified . to these aims , observing co is still paramount , as well as its photodissociation and photoionization products : c and c@xmath6 .
this is also true for a molecule widely observed by _
, h@xmath5o : oh and oi . high - frequency , velocity - resolved observations of such tracers ( c@xmath6 , oh , and oi ) can only be performed with sofia / great . in the cep e intermediate - mass jet / outflow system (
see fig .
[ figure1 ] , gomez - ruiz _ et al .
_ @xcite and lefloch _ et al . _
@xcite ) , co observations by sofia / great were used to show that the influence of the protostellar radiation field is negligible , at least in the terminal shock regions . in the cep
a and g5.890.39 ( gusdorf _ et al . _
@xcite , gusdorf _ et al . _
@xcite , leurini _ et al . _
@xcite ) massive star - forming regions , oh , oi , and c@xmath6 observations were used to perform detailed energetics evaluations , to highlight the shortcomings of current shock models and to illustrate the necessity of developing irradiated shock models . in a forthcoming steps , these observations will serve as a basis for more thorough chemical studies , specially on the chemistry of h@xmath5o .
co ( 2 - 1 ) ( carlhoff _ et al . _
@xcite , colour map ) and @xmath7co ( 11 - 10 ) ( gusdorf _ et al .
_ , in prep
. , blue contours ) .
the white contours delimitate the regions a , b , c , and d defined in louvet _
@xcite based on column density values inferred from _ herschel _ observations .
the white hexagons mark the massive dense cores identified by louvet _
@xcite based on continuum observations from the pdbi .
the red hexagon is a dense core .
_ right - hand panel : _
@xmath7co spectra over a 24 beam size , extracted from regions a and b ( gusdorf _ et al .
_ , in prep.).,scaledwidth=85.0% ] a significant legacy of _ herschel _ is the study of filaments and ridges .
these structures are the result of dynamical processes of matter accretion , themselves at the base of the formation of molecular clouds : colliding flows .
the ridges are the densest filaments , where clustered , massive star formation is usually observed ( see fig .
[ figure3 ] and e.g. nguyen - luong _ et al .
_ @xcite ) .
their observation yields numerous questions : what is the chemistry that takes place in such regions of low - velocity , dense shocks ?
what are the accretion mechanisms leading to the formation of the very massive protostars that seem to be detected ?
can the most sophisticated models of ism evolution account for the formation of such filaments with embedded massive stars ?
what is the contribution of these low - energy but extended regions to the high-@xmath2 co emission as seen by _ herschel _ on a larger , galactic scale ?
again , only sofia / great can provide the answers to such questions .
in louvet _
@xcite , we used pdbi observations to compare ongoing star formation with large simulations . in gusdorf
, we will use co and c@xmath6 observations by sofia / great to characterize the shocks , and directly evaluate their contribution to the galactic emission .
these results will support our new models of sio formation in low - velocity shock environments ( louvet _ et al .
_ , in prep . )
99 gusdorf , a. , gerin , m. , goicoechea , j. r. , 2015c , a&a , accepted leurini , s. , wyrowski , f. , wiesemeyer , h. 2015 , a&a , in press gusdorf , a. , gsten , r. , menten , k. m. , 2015b , a&a , in press lefloch , b. , gusdorf , a. , codella , c. , 2015 , a&a , 581 , 4 gusdorf , a. , riquelme , d. , anderl , s. , 2015a , a&a , 575 , 98 louvet , f. , motte , f. , hennebelle , p. , 2014
, a&a , 570 , 15 nguyen - luong , q. , motte , f. , carlhoff , p. , 2013 ,
apj , 775 , 88 carlhoff , p. , nguyen - luong , q. , schilke , p. 2013 , a&a , 560 , 24 gusdorf , a. , anderl , s. , gsten , r. , 2012 , a&a , 542 , 19 eislffel , j. , nisini , b. , gsten , r. , 2012 , a&a , 542 , 11 gmez - ruiz , a.i .
, gusdorf , a. , leurini , s. , 2012 , a&a , 542 , 9 su , y .- n . , liu , s .- y . , chen , h .-
r . , and tang , y .- w . , 2012 , apj , 744 , 26 cunningham , n. j. , moeckel , n. , and bally , j. , 2009 , apj , 692 , 943 hunter , t. r. , brogan , c. l. , indenbetouw , r. , and cyganowski , c. j. , 2008 , apj , 680 , 1271 | interstellar medium ) evolving due to injection of energy energetic radiation transfers energy from uv photons to x- and @xmath0-ray ones cosmic rays another source energy mechanical energy injected through shocks turbulence shocks ubiquitous in interstellar medium associated to star formation life via stellar winds death dynamical processes to formation molecular clouds generate shocks where interstellar matter collide study of interstellar shocks useful other mechanisms of energy injection study to understand evolution interstellar medium address questions review paramount to interpret precise observations with precise shock models sofia telescope great receiver spectral resolution powerful tool to allows shock tracers in far - infrared range co spectra lower panels showing sofia / great observations bhr71 image overlay of 8 @xmath1 m emission with co ( 32 ) white contours cep e image from pdbi shows jet terminal bowshock cavity in southern lobe each spatial element corresponds to spectral range centered on vertical dashed lines of lower panel in same coloursparadigm for formation of low - mass stars established past decades involves accretion of matter through disk - like structure ejection through jets outflows generally bipolar fig shock - related questions remain : ejection mechanisms molecular gas mechanisms peculiar chemistry in low - star forming environments energetic impacts of jets outflows environment can cosmic rays accelerated in shocks ? observation of emission from high-@xmath2 rotational lines of co important temperature density conditions co abundant molecule levels trace warm shocked medium sofia / great only telescope allows observation chemically rich outflows l1157 bhr71 observed with sofia / great comparison with shock models led to precise determinations of masses momentum energy transfers combined with observations constrain silicon chemistry map from shows complex outflow system spectra show observations by sofia / great at 117 and 163 @xmath1 m . for g5.890.large - scale top zoomed - in middle maps illustrate complexity star forming region spectrum shows velocity - resolved emission / absorption structure 63 @xmath1 m line seen by sofia / great climbing protostellar mass ladder new questions asked scenario for massive star formation less clear scaled - up version from low - mass massive stars result from competitive accretion mechanisms or other mechanisms ? role protostellar radiation field observing co paramount photodissociation photoionization products : c and c@xmath6 true for molecule observed by h@xmath5o oh oi high - frequency velocity - resolved observations of tracers with sofia / great cep e intermediate - mass jet / outflow system fig . co observations by sofia / great influence protostellar radiation field negligible in terminal shock regions cep a g5.890.39 massive star - forming regions , oh oi c@xmath6 observations used detailed energetics evaluations highlight shortcomings of current shock models illustrate necessity of developing irradiated shock modelsforthcoming observations basis for chemical studies chemistry of h@xmath5o . co ( 2 - 1 ) ( carlhoff @xcite colour map and @xmath7co ( 11 - 10 ) ( gusdorf al. blue contours ) white contours delimitate regions a b c d in louvet _ @xcite white hexagons mark dense cores red hexagon dense core right - hand panel @xmath7co spectra over 24 beam size regions a b ( gusdorf al.,scaledwidth=85.0% legacy of herschel _ study of filaments ridges structures result of matter accretion base formation of molecular clouds ridges densest filaments clustered massive star formation observed ( see fig . figure3 nguyen - luong @xcite )observation yields questions chemistry in low - velocity dense shocks ? accretion mechanisms massive protostars ? models ism evolution account for formation filaments with massive stars ? contribution low - energy extended regions to high-@xmath2 co emission ? sofia / great provide answers louvet _ @xcite used pdbi observations compare star formation with large simulations gusdorf co and c@xmath6 observations characterize shocks evaluate contribution to galactic emission results support models of sio formation in low - velocity shock environments ( louvet et al . prep ) 99 gusdorf a. gerin goicoechea 2015c leurini wyrowski wiesemeyer h 2015 gusdorf gsten menten 2015b lefloch . gusdorf codella 2015 a&a 581 , gusdorf riquelme d anderl s 2015a 575 , louvet f. motte hennebelle p. 2014 a&a , 570 nguyen - luong motte carlhoff 2013 apj 775 , 88 carlhoff nguyen - luong schilke 2013 a&a 560 , 24 gusdorf anderl gsten 2012 a&a 542 , 19 eislffel nisini gsten 2012&a 542 , 11 gmez - ruiz gusdorf leurini 2012 542 , 9 su , liu chen tang 2012 apj 744 , 26 cunningham n moeckel n bally j 2009 apj , 692 , 943 hunter brogan indenbetouw cyganowski c. j 2008 apj 680 , 1271 | 0.354481 | 2,261 | 1,115 | , Saving $0.1 in GPT-4. |
55,019 | fundamental mode classical cepheids have light curves which repeat accurately enough that we can watch them evolve ( change period ) . the new level of accuracy and quantity of data with the _ kepler _ and _ most _ satellites probes this further .
an intriguing result was found in the long time - series of kepler data for v1154 cyg the one classical cepheid ( fundamental mode , p = 4.9@xmath0 ) in the field , which has short term changes in period ( @xmath120 minutes ) , correlated for @xmath110 cycles ( period jitter ) . to follow this up
, we obtained a month long series of observations of the fundamental mode cepheid rt aur and the first overtone pulsator sz tau .
rt aur shows the traditional strict repetition of the light curve , with the fourier amplitude ratio @xmath2 remaining nearly constant .
the light curve of sz tau , on the other hand , fluctuates in amplitude ratio at the level of approximately 50% . furthermore prewhitening the rt aur data with 10 frequencies reduces the fourier spectrum to noise . for sz tau
, considerable power is left after this prewhitening in a complicated variety of frequencies . | the quality and quantity of satellite data is revealing subtle features in cepheid pulsation .
specifically , we discuss here recent findings from two long series of observations by the _ kepler _ and _ most _ satellites .
there is only one classical cepheid ( v1154 cyg ) in the _ kepler _ field , and it is a fundamental mode pulsator with a period of 4.9@xmath0 .
analysis of the _ kepler _ data by ( * ? ? ?
* ; * ? ? ?
* derekas et al . ( 2012 ) ) found small cycle to cycle variations of the period .
typically the period excursions might be 20 minutes and last up to 15 cycles before the sign of the period change ( @xmath3 ) reverses .
to look further for the small fluctuations made visible by the long continuous strings of accurate satellite data , we observed two cepheids with the _ most _ satellite ( ( * ? ? ?
* ; * ? ? ?
* walker et al .
2003 ) , ( * ? ? ?
* ; * ? ? ?
* matthews et al . 2004 ) ) .
motivation for the structure of the observation request is provided by known period changes in cepheids ( fig .
[ fig1 ] ) .
the data taken from ( * ? ? ?
* szabados ( 1983 ) ) show the well known increase in period fluctuations with period or luminosity of the cepheid .
this is consistent with period changes ( @xmath3 ) determined by evolution through the instability strip , with more luminous stars evolving more quickly .
the exception to the trend , however , is the group of cepheids pulsating in an overtone mode which apparently show unusually large period jitter rather than the smooth and continuous evolution of fundamental mode pulsators . for the _ most _ observations we wanted to contrast the behavior of a fundamental mode pulsator ( rt aur ) with a first overtone pulsator ( sz tau ) . the background information on @xmath3 is shown in fig . [ fig2 ]
. for rt aur the period change o@xmath4c ( observed minus computed )
diagram has a decrease followed by an increase , which is typically fitted with a parabola . for sz tau
, the o@xmath4c diagram shows both increasing and decreasing periods , as well as about 10000 days when it remains constant . the _ most _ observations ( evans et al .
, in preparation ) phased for pulsation period are shown in fig .
[ fig3 ] . for rt aur
, the observations were interleaved with another target , resulting in gaps in the light curve . however , the light curve repeats very precisely from cycle to cycle . for sz tau , the overtone pulsator ,
only the maximum of the phased light curve is shown .
this emphasizes the fact that there are variations in maximum brightness from cycle to cycle ( also easily visible at minimum light ) .
a small instrumental signal ( differing earthshine through the satellite orbit ) is seen in fig .
[ fig3]b , which is being removed through additional processing .
the data from the two stars was subjected to a number of comparisons . in fig .
[ fig4 ] , the fourier spectra are shown after pre - whitening for a fourier fit of 10 terms .
the instrumental signal is seen at the orbital frequency ( @xmath114 d@xmath5 ) .
the fourier series describes the fundamental mode pulsator ( rt aur ) well and little power is left at low frequencies . in sz tau ,
on the other hand , power still remains in a complicated set of low frequencies , indicating that the single periodicity does not fully describe the pulsation .
the fourier parameters themselves were compared for the two stars . as an example , fig .
[ fig5 ] shows the amplitude ratio @xmath6 changes for the six cycles observed for each of the stars . for rt aur
, the variation is estimated to be 3% . for sz
tau the variation is much larger ( 45% ) .
the variation in fig .
5 ( right ) is typical of the variation in other fourier coefficients of the overtone cepheid .
thus the _ most _ observations show that the overtone pulsator sz tau has a larger variability in the parameters we examined , or a larger instability in the pulsation than the fundamental mode cepheid .
we present here a brief summary of the period changes or possible period changes in classical cepheids and the characteristics of pulsation which they suggest .
most probably the @xmath3 which we observe results from a combination of causes . *
flickering @xmath3 ( _ kepler _ ) : it is cyclic and reasonably short term
. it could be due either to a pulsation phenomenon or star spots .
* instability in overtones ( _ most _ ) : this seems to be most probably pulsation related .
financial support was provided by cxc nasa contract nas8 - 03060 ( nre ) , estec contract no .
4000106398/12/nl / kml ( ls ) , european community s seventh framework programme ( fp7/2007 - 2013 ) under grant agreement no . 269194 ( irses / ask ) ( rs , ad ) , the hungarian otka grant k83790 , the ktia urkut 10 - 1 - 2011 - 0019 grant , jnos bolyai research scholarship of the hungarian academy of sciences ( rs , ad ) , the lendlet-2009 young researchers program of the hungarian academy of sciences ( ad ) , and iau travel grant ( rs ) . | quality satellite data features in cepheid pulsation discuss recent findings from two observations by _ kepler _ and _ most _ satellites one classical cepheid ( v1154 cyg ) in _ kepler _ field fundamental mode pulsator with period 4.9@xmath0 analysis _ kepler _ data by found small cycle variations of period period excursions 20 minutes last to 15 cycles before period change ( @xmath3 ) reverses small fluctuations observed two cepheids with _ most _ satellite 2004 motivation for observation by known period changes in cepheids ( fig . fig1 ] ) data from szabados 1983 ) show increase in period fluctuations with luminosity cepheid consistent with period changes ( @xmath3 ) determined by evolution through instability strip more luminous stars evolving quickly exception trend cepheids pulsating in overtone mode show large period jitter fundamental mode contrast behavior of fundamental mode pulsator ( rt aur ) with first overtone pulsator ( sz tau ) background information on @xmath3 shown in fig . [ fig2 ]rt aur period change o@xmath4c diagram decrease increase fitted with parabola sz tau diagram shows increasing decreasing periods 10000 days constant observations phased pulsation period shown in fig . [ fig3 ] rt aur observations interleaved with target gaps in light curve light curve repeats cycle to cycle sz tau overtone pulsator maximum phased light curve shown variations in maximum brightness cycle to cycle visible minimum light small instrumental signal differing earthshine satellite orbit in fig . [ fig3]b removed through additional processing data two stars subjected comparisons fig fig4 fourier spectra shown after pre - whitening fit 10 terms instrumental signal at orbital frequency @xmath114 d@xmath5 ) fourier series describes fundamental mode pulsator rt aur little power left at low frequencies sz tau power remains in complicated low frequencies single periodicity describe pulsation fourier parameters compared for two stars fig . fig5 ] shows amplitude ratio @xmath6 changes six cycles rt aur variation 3% sz tau variation larger 45% ) variation in fig .5 right ) typical variation fourier coefficients overtone cepheid observations show overtone pulsator sz tau larger variability parameters instability pulsation than fundamental mode cepheid summary period changes classical cepheids characteristics pulsation probably @xmath3 results combination causes flickering @xmath3 cyclic short term due to pulsation or star spots instability in overtones probably pulsation related financial support by cxc nasa contract nas8 - 03060 estec contract. 4000106398/12/nl / european community s seventh framework programme fp7/2007 - 2013 ) grant agreement 269194 hungarian otka grant k83790 ktia urkut 10 - 1 - 2011 - 0019 grant jnos bolyai research scholarship hungarian academy sciences lendlet-2009 young researchers program hungarian academy sciences iau travel grant | 0.186789 | 1,206 | 603 | , Saving $0.0 in GPT-4. |
55,027 | the lowest three moments of generalized parton distributions are calculated in full qcd and provide new insight into the behavior of nucleon electromagnetic form factors , the origin of the nucleon spin , and the transverse structure of the nucleon . | matrix elements of the light cone operator + [ .2 cm ] @xmath0 + and the tower of twist - two operators @xmath1 provide a wealth of precise information about the quark and gluon structure of the nucleon .
the diagonal nucleon matrix element @xmath2 measures the light cone momentum distribution , @xmath3 , and @xmath4 specifies the @xmath5 moment of this distribution , @xmath6 .
off - diagonal matrix elements of @xmath7 measure the generalized parton distributions@xcite @xmath8 and @xmath9 : @xmath10 where @xmath11 , @xmath12 , @xmath13 , and @xmath14 .
off - diagonal matrix elements of the twist - two operators @xmath15 yield moments of these generalized parton distributions , and this work considers the generalized form factors @xmath16 and @xmath17 .
two special cases are important for the present work .
the zeroth moments correspond to the familiar electromagnetic form factors ( weighted with appropriate quark charges ) , @xmath18 and @xmath19 .
the first moments yield the total quark angular momentum , @xmath20 $ ] .
combined with the angular momentum from the quark spin , @xmath21 $ ] , this enables decomposition of the quark contribution to the nucleon spin .
burkardt@xcite has shown that the generalized parton distribution @xmath22 is the fourier transform of the impact parameter dependent parton distribution @xmath23 where @xmath24 is the probability of finding a quark with longitudinal momentum fraction @xmath25 and transverse position ( or impact parameter ) @xmath26 , and @xmath27 is the transverse momentum transfer .
physically , we expect the transverse size of the nucleon to decrease significantly as @xmath25 increases .
as @xmath28 , a single parton carries all the momentum , the proton wave function is reduced to a single fock space component , and the transverse wave function has zero spatial extent . since @xmath29 is the fourier transform of the transverse distribution , the slope in @xmath30 at the origin measures the rms transverse radius .
the slope should decrease significantly with @xmath25 and as @xmath31 , it should approach zero . hence , the slope of the generalized form factors @xmath32 should decrease with increasing @xmath33 , and we demonstrate this effect in our lattice calculations .
. ] by calculating an overdetermined set of lattice observables to improve the statistical accuracy of generalized form factors as described in ref
. @xcite , we calculate the lowest three moments + @xmath34 \langle p ' | { \cal o}^{\lbrace \mu_1 \mu_2\rbrace } | p \rangle = \bar p^{\lbrace\mu_1}\langle \ ! \langle \gamma^{\mu_2\rbrace}\rangle \ !
\rangle a_{20}(t ) \\ \mbox{\,\,\ , } + \frac{{\mbox{i}\ } } { 2 m } \bar p^{\lbrace\mu_1 } \langle \ ! \langle \sigma^{\mu_2\rbrace\alpha}\rangle \ ! \rangle \delta_{\alpha } b_{2 0}(t ) + \frac{1}{m}\delta^{\ { \mu_1 } \delta^ { \mu_2 \ } } c_{2}(t ) , \\[.3 cm ] \langle p ' | { \cal o}^{\lbrace\mu_1 \mu_2 \mu_3\rbrace } | p \rangle = \bar p^{\lbrace\mu_1}\bar p^{\mu_2 } \langle \ ! \langle \gamma^{\mu_3\rbrace } \rangle \ ! \rangle a_{30}(t)\\ \mbox{\quad}+ \frac{{\mbox{i}\ } } { 2 m } \bar p^{\lbrace \mu_1}\bar p^{\mu_2 } \langle \ !
\langle \sigma^{\mu_3\rbrace\alpha } \rangle \ ! \rangle \delta_{\alpha } b_{3 0}(t ) \\
\mbox{\quad } + \delta^{\lbrace \mu_1}\delta^{\mu_2 } \langle \ ! \langle \gamma^{\mu_3\rbrace}\rangle \ !
\rangle a_{32}(t ) \\ \mbox{\quad } + \frac{{\mbox{i}\ } } { 2 m } \delta^{\lbrace\mu_1}\delta^{\mu_2 } \langle \ !
\langle \sigma^{\mu_3\rbrace\alpha}\rangle \ ! \rangle \delta_{\alpha } b_{3 2}(t)$ ] , + where @xmath35 .
we calculated connected diagram contributions using @xmath37 sesam@xcite full qcd configurations with wilson fermions at @xmath38 on @xmath39 lattices at each of three quark masses , @xmath40= 0.1570 , 0.1565 , and 0.1560 , corresponding to pion masses defined by @xmath41 of 744 , 831 , and 897 mev respectively .
figure 1 shows our result for the electromagnetic form factor ratio @xmath42 divided by the next to leading order light cone wave function result@xcite @xmath43 with @xmath44 = 0.3 gev .
the @xmath45 dependence is in excellent agreement with the recent jlab data@xcite plotted in ref .
@xcite , but the overall ratio is a factor of four too high in the heavy quark world in which @xmath46 700 - 900 mev .
the total quark contribution to the nucleon spin is given by the extrapolation to @xmath47 of @xmath48 and @xmath49 shown in figure 2 . since @xmath48 is calculated directly at @xmath47 and @xmath49 is well fit by a constant that is measured to be nearly zero with small errors , the connected contribution to the angular momentum is measured to within a few percent .
combined with the results of @xmath50 from reference @xcite , we obtain the connected diagram contributions to the decomposition of nucleon spin shown in table 1 .
similar results have been obtained in refs .
@xcite . to the extent that the disconnected diagrams do not change the qualitative behavior ,
we conclude that of the order of 70% of the spin of the nucleon arises from the quark spin and a negligible fraction arises from the quark orbital angular momentum in a heavy pion world where @xmath46 700 - 900 mev . in the chiral limit , the quark spin contribution must decrease to @xmath51 30% to agree with experiment . *
3c|c @xmath52 & 0.1570 & 0.1565 & 0.1560 + @xmath53 & 0.67@xmath54 .04 & 0.73 @xmath54 .03 & 0.68 @xmath54 .02 + @xmath55 & 0.06@xmath54 .05 & -0.04 @xmath54 .04 & 0.00 @xmath54
.03 + @xmath56 & 0.73@xmath54 .04 & 0.69@xmath54 .02 & 0.68 @xmath54 .03 + figure 3 shows the generalized form factors @xmath57,@xmath48 , and @xmath58 for @xmath52 = 0.1560 and 0.1570 , corresponding to the lowest three moments of @xmath59 . as explained above , the decrease in slope with increasing moment
is a clear manifestation of the decrease of the transverse size of the light cone wave function as @xmath31 .
note that the error bands are sufficiently narrow that the dramatic change of slope is clearly determined , strongly ruling out a factorized ansatz for the momentum transfer dependence of generalized form factors .
qualitatively similar behavior is obtained for the connected contributions to @xmath60 and for the spin - dependent @xmath61@xcite . * 3c normalized generalized form factors @xmath62 for n=1 ( diamonds ) , n=2 ( triangles ) and n=3 ( squares ) . ,
title="fig:",scaledwidth=47.0% ] + normalized generalized form factors @xmath62 for n=1 ( diamonds ) , n=2 ( triangles ) and n=3 ( squares ) . ,
title="fig:",scaledwidth=47.0% ] + d. mller , d. robaschik , b. geyer , f. m. dittes and j. horejsi , fortsch .
* 42 * ( 1994 ) 101 [ arxiv : hep - ph/9812448 ] , x. d. ji , phys . rev .
* 78 * ( 1997 ) 610 [ arxiv : hep - ph/9603249 ] , a. v. radyushkin , phys .
d * 56 * ( 1997 ) 5524 [ arxiv : hep - ph/9704207 ] .
m. burkardt , int .
j. mod .
a * 18 * , 173 ( 2003 ) [ arxiv : hep - ph/0207047 ] .
p. hagler , j. negele , d. b. renner , w. schroers , t. lippert and k. schilling [ lhpc collaboration ] , phys .
d * 68 * , 034505 ( 2003 ) [ arxiv : hep - lat/0304018 ] .
n. eicker et al ( sesam coll . ) , phys .
d * 59 * , 014509 ( 1999 ) [ arxiv : hep - lat/9806027 ] .
o. gayou _ et al . _
[ jefferson lab hall a collaboration ] , phys .
* 88 * , 092301 ( 2002 ) [ arxiv : nucl - ex/0111010 ] .
d. dolgov _ et al . _
[ lhpc collaboration ] , phys .
d * 66 * , 034506 ( 2002 ) [ arxiv : hep - lat/0201021 ] . | elements light cone operator @xmath0 tower twist - two operators @xmath1 provide information quark gluon structure nucleon diagonal nucleon matrix element @xmath2 measures light cone momentum distribution @xmath4 specifies @xmath5 moment off - diagonal matrix elements @xmath7 measure generalized parton distributions @xmath8 @xmath9 elements twist - two operators @xmath15 yield moments generalized parton distributions considers form factors @xmath16 @xmath17 special cases zeroth moments correspond electromagnetic form factors quark charges @xmath18 @xmath19 first moments yield total quark angular momentum combined momentum quark spin enables decomposition quark contribution nucleon spin generalized parton distribution @xmath22 fourier transform impact parameter dependent parton distribution @xmath23 @xmath24 probability quark longitudinal momentum fraction transverse position @xmath26 transverse momentum transfer transverse size nucleon @xmath25 increases @xmath28 single parton carries momentum proton wave function reduced single space component transverse wave function zero spatial extent@xmath29 fourier transform transverse distribution slope @xmath30 origin measures rms transverse radius slope decrease with @xmath25 @xmath31 approach zero slope generalized form factors @xmath32 decrease @xmath33 demonstrate effect in lattice calculations calculating overdetermined lattice observables improve statistical accuracy form factors calculate lowest three moments + @xmath34 \langle{\lbrace \mu_1 \mu_2\rbrace\rangle \bar p{\lbrace\mu_1}\langle\gamma^{\mu_2\rbrace}\rangle{20}(t \mbox{\frac{{\mbox{i} 2 \bar{\lbrace\mu_1\sigma^{\mu_2\rbrace\alpha}\rangle\delta_{\alpha }_{2 0}(t ) \frac{1}{m}\delta^{ \mu_1\mu_2_{2}(t )3 cm \langle\lbrace\mu_1_2_3\rbrace\rangle\bar\lbrace\mu_1_2\langle\gamma^{\mu_3\rbrace a_{30}(t \mbox{\quad} \frac{{\mbox{i} 2 m \bar{\lbrace \mu_1\mu_2\sigma^{\mu_3\rbrace\alpha \delta_{\alpha }{3 0}(t ) \mbox{\quad\delta{\lbrace \mu_1}\delta{\mu_2 }\gamma{\mu_3\rbrace}_{32}(t ) \mbox{\quad }\frac{{\mbox{i} 2 m \delta^{\lbrace\mu_1}{\mu_2 }\sigma^{\mu_3\rbrace\alpha}\rangle\delta_{\alpha }_{3 2}(t)$ @xmath35 calculated diagram contributions @xmath37@xcite qcd configurations wilson fermions @xmath38 @xmath39 lattices quark masses @xmath40= 0.1570 0.1565 0.1560 pion masses @xmath41 744 831 897 mev figure 1 electromagnetic form factor ratio @xmath42 divided by light cone wave function result @xmath43 @xmath44 = 0.3 gev @xmath45 dependence jlab data ratio four too high heavy quark world @xmath46 700 - 900 mev total quark contribution nucleon spin extrapolation to @xmath47 @xmath48 @xmath49 figure 2 @xmath48 calculated connected contribution angular momentum measured within few percent combined with results @xmath50 connected diagram contributions to decomposition nucleon spin table 1 similar results in @xcite 70% spin nucleon from quark spin negligible fraction quark orbital angular momentum heavy pion world @xmath46 700 - 900 mev quark spin contribution decrease to @xmath51 30% experiment @xmath52 & 0.1570 0.1565 0.1560 + @xmath53 0.67@xmath54 .04 0.73 @xmath540.68 @xmath54 .02 + 0.06 .05 -0.04 .04 0.00 .03 + 0.73 .04 0.69 .02 0.68 .03 figure 3 shows form factors @xmath57 @xmath52 = 0.1560 0.1570 lowest three moments @xmath59 decrease slope increasing moment decrease transverse size light cone wave function @xmath31 error bands narrow dramatic change slope determined ruling out factorized ansatz momentum transfer dependence similar behavior contributions @xmath60 spin - dependent @xmath61 normalized form factors @xmath62 n=1 diamonds ) n=2 triangles n=3 )% @xmath62 d. mller robaschik b. geyer f. m. dittes j. horejsi fortsch 42 * ( 1994 ) 101 arxiv hep - ph/9812448 ]. ji phys . rev. 78 1997 ) 610 ph/9603249 radyushkin phys . 56 * ( 1997 ) 5524 hep - ph/9704207 ]. burkardt int . j18 173 ( 2003 ) arxiv hep - ph/0207047 p. hagler negele renner schroers lippert schilling lhpc collaboration phys . d 68 034505 2003 ) hep - lat/0304018 n eicker sesam coll phys 59 014509 ( 1999 ) hep - lat/9806027 gayou jefferson lab hall collaboration . 88 092301 ( 2002 ) nucl - ex/0111010 dolgov collaboration d 66 * 034506 ( 2002 ) hep - lat/0201021 | 0.349443 | 2,319 | 1,247 | , Saving $0.1 in GPT-4. |
55,032 | jet - matter interaction remains a central question and a theoretical challenge in heavy - ion physics and might become important in high - multiplicity events in proton - proton collisions at lhc energies .
full jet measurement at lhc offer the proper tool to investigate energy loss process and fragmentation of hard parton in the medium .
since jet reconstruction will be constrained to small cone sizes , then study of the connection between jets and surrounding environment provides a further possibility to extend our exploration .
we study jets at @xmath0 tev and @xmath1 collisions at @xmath2 tev .
we analyze the flavor components in jet - like environments .
we introduce a definition for surrounding cones / belts and investigate flavor dependence and correlation of different hadron species produced in jets . here , we focus on proton - triggered correlations . our analysis can be extended for heavy ion collisions . | the state - of - the - art particle detectors and very - high energies reached at the large hadron collider ( lhc ) opened a new window to create and investigate high momentum particle showers in hadron - hadron collisions .
these showers originate from color partons and they are identified as _
jets_. jets were discovered in electron - positron ( @xmath3 ) collisions , where it was assumed : partons are travelling in qcd vacuum .
however , in proton - proton and heavy - ion collisions partons are propagating in a color medium .
the deconfined medium modifies the parton properties e.g. induces gluon radiation and parton energy loss @xcite .
this phenomena has been seen and investigated in details in heavy - ion collisions at rhic energies @xcite .
high - multiplicity proton - proton events may behave similarly displaying intense interaction between jets and the surrounding matter. thus , study of jet energy loss and jet - matter interaction can be accomplished with the proper determination of the jets and the remaining background , the so called _ underlying events_. generally speaking , underlying events ( ue ) contain particles originating from partons outside identified jet(s ) . the first standard definition of ue was given by the cdf collaboration at fermilab @xcite . in an earlier work we introduced _ surrounding belts _ ( sb ) around the cones of identified jets @xcite .
here , we focused on properties of near and away side jets , and their surrounding regimes , especially hadron contents and correlations .
we display our recent results on the generalized definition of the underlying event .
we analyzed mean-@xmath4 vs. multiplicity for the newly defined areas in case of @xmath0 tev proton - proton collisions .
we performed pythia simulations for @xmath2 tev @xmath1 collisions .
we used pythia 6.4 @xcite with atlas - csc tune @xcite for both cases and generated @xmath5 m minimum bias events .
furthermore , we have studied proton - triggered correlations in surrounding belts characterized by various radii and thicknesses at @xmath2 tev .
transition area between jet cones and underlying event carries information about jet - matter interaction .
the investigation of this area demands proper definitions and solid basis for quantitative analysis .
first we considered the mean transverse momenta vs. multiplicity in @xmath6 tev proton - proton collisions .
the newly defined areas based on _ right panel _ of fig .
[ ang - corr ] . here ,
the main areas are highlighted : near and away refers for the the near- and away - side jets respectively .
we marked the inner surrounding belt with sb1 and the outer sb2 for both near and away side jets .
based on this definition , the underlying event is everything outside the outer belts .
hadron multiplicities can be given in each region however , for reference we used @xmath7 , which is the generalized underlying event area
outside both outer surrounding belts .
we plotted the mean transverse ( @xmath8 ) momenta of above areas as the function of @xmath7 on the _ left panel _ of fig .
[ ang - corr ] .
the ( @xmath9 for near- and away - side jet is drawn by _
full squares _ and _ full disks _ respectively .
mean-@xmath4 for surrounding belts are plotted with _
open squares _ and _ open triangles _ for inner belts : sb1@xmath10 and sb1@xmath11. open circles and diamonds denote results on outer belts : sb2@xmath10 and sb2@xmath11 , respectively . _ full triangles _ display the @xmath9 for the total event .
finally , we indicated @xmath9 for the underlying event with _
stars_. as we plotted on fig .
[ ang - corr ] jet - like events has high-@xmath8 values , with low multiplicity , due to the produced jets in the @xmath6 tev @xmath1 collisions .
mean @xmath4s for these jet - like events are falling quickly as the @xmath7 multiplicity is increasing . in parallel
, the @xmath9 for the total event has a similar structure .
surrounding belts have almost the flat value @xmath12 gev / c , practically at every @xmath7 value . testing the underlying event , @xmath13 is increasing up to @xmath14 gev / c , indicating a mini - jet like structure at high multiplicities is starting to play the role , as we pointed out in ref .
@xcite .
_ left panel : _
mean-@xmath4 vs. ue multiplicity outside surrounding cones in @xmath1 collisions at @xmath6 tev .
_ right panel : _ illustration of inner - belt ( sb1 ) and outer - belt ( sb2 ) .
( color online.),title="fig:",width=283,height=245 ] _ left panel : _
mean-@xmath4 vs. ue multiplicity outside surrounding cones in @xmath1 collisions at @xmath6 tev .
_ right panel : _ illustration of inner - belt ( sb1 ) and outer - belt ( sb2 ) .
( color online.),title="fig:",width=283,height=245 ]
the aim of our study is to map differences in surrounding belts , connected to near - side and away - side correlation peaks .
such differences carry information about hadronization processes and/or possible jet - matter interactions .
the wanted differences can be amplified displaying the ratio of hadron yields in the different belts : sb1@xmath10/sb1@xmath11 and sb2@xmath10/sb2@xmath11 .
following the illustration of fig .
[ ang - corr ] , we define 3 different geometrical sets to be analyzed : cdf - set : : : @xmath15 ; @xmath16 ; @xmath17 .
@xmath18-set : : : @xmath19 ; @xmath20 ; @xmath17 .
@xmath21-set : : : @xmath22 ; @xmath23 ; @xmath17 .
the cdf - set is motivated by the original definition from cdf collaboration @xcite .
the @xmath18-set is motivated by a smaller , jet - cone - like size .
the @xmath21-set is determined from pythia simulations on correlations at @xmath24 tev at corresponding trigger and associated momentum regions . here
we will use @xmath25 and @xmath26 displayed on fig .
[ ang - corr ] . comparing the above selections ,
the area of the underlying event is the smallest in the case of cdf - set , and it is the largest for @xmath21-set .
the widths of the surrounding belts ( @xmath27 ) is a result of an optimalization to have proper statistics for analysis and comparison .
we have calculated the wanted ratios of sb1@xmath10/sb1@xmath11 and sb2@xmath10/sb2@xmath11 for all possible charged hadron triggers , namely choosing one of the identified hadrons ( @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 ) as trigger particle and extracting angular correlation with all other identified hadron species .
differences in the proton - triggered correlation functions , enhanced by ratios of surrounding belts , sb1@xmath10/sb1@xmath11 and sb2@xmath10/sb2@xmath11 as a function of transverse momentum of the associated hadrons , @xmath34 .
see text for details on legend .
( color online . ) , width=604,height=340 ] fig .
[ 7t - p - trigger ] displays our results extracted from pythia simulations for @xmath1 at @xmath24 tev , using proton trigger in the momentum window defined earlier .
inner belt ratios , sb1@xmath10/sb1@xmath11 are marked by _
full squares _ for cdf - set , _ full disks _ for @xmath18-set , and _ full triangles _ for @xmath21-set. outer belt ratios , sb2@xmath10/sb2@xmath11 are marked by _
open squares _ for cdf - set , _ open circles _ for @xmath18-set , and _ open triangles _ for @xmath21-set.
further results for other identified hadron triggers can be visualized similarly , but we want to focus on proton trigger .
results from cdf - set , indicated by _ full and open squares _ are close to be unity ( no effect ) in all correlations .
this means we are testing homogeneous ue region far from correlation peaks .
results from @xmath18-set , indicated by _ full and open circles _ go below unity for most of the hadrons ( except antiprotons ) , because of the expected momentum conservation for leading hadrons .
antiprotons behave differently due to a strong correlation in baryon - antibaryon production closer to the correlation peaks .
this effect is based on baryon number conservation .
a slight difference can be already seen between the ratios from inner belt and outer belt . in the case of @xmath21-set ( _ full and open triangles _ ) we are closer to the correlational peaks , but the results are very similar to the @xmath18-set , except the deviation in proton - antiprotons correlation is even more strong .
thus , we are really in the fragmentation region with very strong baryon - antibaryon correlation
. a slight difference between the inner belt and outer belt can be seen , also .
taking @xmath30 trigger we will see strong @xmath35 correlation driven by strangeness conservation .
the @xmath36proton and @xmath36antiproton correlations are just similar to e.g. @xmath37 ones .
we have investigated identified hadron correlations in @xmath1 collision at @xmath24 tev , by means of pythia event generator .
we introduced different geometrical setups .
we have found strong proton - antiproton correlation , based on baryon number conservation and strong @xmath35 correlation driven by strangeness conservation . introducing different surrounding belts into the analysis
further details were revealed .
these correlations should be extracted from real data .
the alice hmpid and the planned vhmpid detectors can play an important role in these explorations @xcite .
this work was supported by hungarian otka nk77816 , pd73596 and etvs university .
one of the authors ( ggb ) is also thanks to jnos bolyai research scholarship of the has .
99 a. a. affolder _ et al . _ [ cdf collaboration ] , phys . rev .
d * 65 * , 092002 ( 2002 ) .
p. levai and a. g. agocs , pos e * ps - hep2009 * ( 2009 ) 472 .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) | particle detectors high energies at large hadron collider opened high momentum particle showers in hadron collisions showers originate from color partons identified as jets discovered in electron - positron collisions partons in qcd vacuum in proton - proton heavy - ion collisions partons in color medium medium parton properties gluon radiation parton energy loss in heavy - ion collisions at energies high - multiplicity proton - proton events interaction between jets surrounding matter study of jet energy loss - matter interaction with determination of jets background underlying events underlying events contain particles from partons outside jet first standard definition of by cdf collaboration at fermilab @xcite introduced surrounding belts around jets focused on properties near side jets surrounding regimes hadron contents correlations recent results on generalized definition underlying event analyzed mean-@xmath4 vs. multiplicity for areas proton - proton collisions performed pythia simulations for collisions used pythia 6.4 @xcite atlas - csc tune generated @xmath5 m minimum bias eventsstudied proton - triggered correlations in belts at @xmath2 tev transition area between jet cones underlying event information jet - matter interaction investigation demands definitions quantitative analysis considered mean transverse momenta vs multiplicity in @xmath6 tev proton - proton collisions defined areas right panel fig . main areas highlighted near away jets marked inner surrounding belt with sb1 outer sb2 for underlying event outside outer belts hadron multiplicities each region used @xmath7 generalized underlying event area outside outer belts plotted mean transverse @xmath8 ) momenta function @xmath7 left panel fig . @xmath9 for near- away - side jet drawn by full squares full disks mean-@xmath4 surrounding belts plotted with open squares triangles inner belts sb1@xmath10 sb1@xmath11 open circles diamonds results outer belts full triangles display @xmath9 for total event indicated @xmath9 underlying event with stars fig jet - like events high-@xmath8 values low multiplicity due produced jets @xmath6 @xmath1 collisions mean @xmath4s falling @xmath7 multiplicity increasing @xmath9 total event similar structure surrounding belts flat value @xmath12 gev / c every @xmath7 value @xmath13 increasing to @xmath14 gev / c mini - jet like structure high multiplicities mean-@xmath4 vs. ue multiplicity outside cones @xmath1 collisions @xmath6 tev illustration inner - belt sb1 outer - belt sb2 ) mean-@xmath4 vs. ue multiplicity cones @xmath1 collisions @xmath6 tev illustration inner - belt sb1 outer - belt sb2 aim study map differences surrounding belts connected near - side away - side correlation peaks differences information hadronization processes possible jet - matter interactions differences amplified displaying ratio hadron yields different belts sb1@xmath11 illustration fig define 3 geometrical sets cdf - set @xmath15 @xmath18-set @xmath19 @xmath21-set @xmath22 cdf - set motivated by original definition cdf collaboration @xcite @xmath18-set motivated smaller jet - cone - like size @xmath21-set determined from pythia simulations correlations @xmath24 trigger momentum regions use @xmath25 and @xmath26 fig . comparing area underlying event smallest cdf - set largest for @xmath21-set widths of surrounding belts @xmath27 ) result optimalization statistics analysis comparison calculated wanted ratios of sb1@xmath10/sb1@xmath11 sb2 for charged hadron triggers choosing identified hadrons @xmath28 ) as trigger particle extracting angular correlation with hadron species differences in proton - triggered correlation functions enhanced by ratios of surrounding belts sb1@xmath10 function of transverse momentum of associated hadrons @xmath34 see text for details legend width=604,height=340 fig .7t - p - trigger displays results from pythia simulations for @xmath1 @xmath24 using proton trigger momentum window inner belt ratios sb1@xmath10/sb1@xmath11 by full squares full disks full triangles-set outer belt ratios sb2@xmath10/sb2@xmath11 open open circles open triangles-set results for other hadron triggers focus on proton trigger results cdf - set full open squares close to unity correlations testing homogeneous ue region far from correlation peaks results @xmath18-set full open circles below unity for hadrons except antiprotons momentum conservation antiprotons behave differently strong correlation in baryon - antibaryon production closer to correlation peaks baryon number conservation slight difference between ratios inner outer belt @xmath21-set full open triangles closer to correlational peaks results similar to @xmath18-set deviation in proton - antiprotons correlation strong fragmentation region with strong baryon - antibaryon correlation slight difference between inner belt outer belt@xmath30 trigger strong @xmath35 correlation strangeness conservation @xmath36proton correlations similar @xmath37 investigated hadron correlations @xmath1 collision @xmath24 tev pythia event generator introduced geometrical setups strong proton - antiproton correlation baryon number conservation @xmath35 correlation strangeness conservation introducing surrounding belts details revealed correlations extracted real data alice hmpid planned vhmpid detectors important role explorations work supported hungarian otka nk77816 pd73596 etvs university thanks jnos bolyai research scholarship affolder cdf collaboration . rev . d * 65 * 092002 2002 levai. agocs hep2009 2009 ) 472 sjostrand mrenna skands jhep 0605 026 2006 | 0.35273 | 2,416 | 1,165 | , Saving $0.1 in GPT-4. |
55,039 | electronic structure calculation were used to predict a new material for spintronic applications .
co@xmath0mn@xmath1fe@xmath1si is one example which is stable against on - site correlation and disorder effects due to the position of the fermi energy in the middle of the minority band gap .
experimentally the sample were made exhibiting @xmath2 structure and a high magnetic order . | half - metallic ferromagnets have been proposed as ideal candidates for spin injection devices because they have been predicted to exhibit 100 % spin polarization at the fermi energy ( @xmath3 ) @xcite . from the applications point of view , a high curie temperature for a half - metallic ferromagnet may be an important condition .
for this reason , heusler alloys ( @xmath2 structure ) have recently attracted great interest .
some of these alloys exhibit high curie temperatures and , according to theory , should have a high spin polarization at the fermi energy @xcite .
calculations also show that anti - site disorder will destroy the high spin polarization @xcite , implying that precise control of the atomic structure of the heusler alloys is required .
the heusler alloy co@xmath0mnsi has attracted particular interest because it is predicted to have a large minority spin band gap of 0.4 ev and , at 985 k , has one of the highest curie temperature , among the known heusler compounds @xcite .
structural and magnetic properties of co@xmath0mnsi have been reported for films and single crystals @xcite . from tunnelling magneto resistance ( tmr ) data with one electrode consisting of a co@xmath0mnsi film sakuraba _
et al _ @xcite measured a tmr ration of 159 % at 2k and @xmath4 % at 300k . if using co@xmath0fesi as one electrode inomata _
et al _ @xcite obtained tmr ratios of 60 % at 5k and 41 % at 300k .
an important quantity for the application of the half - metallic ferromagnets is the size of the gap in the minority states and the position of @xmath3 inside of the gap .
small gaps may be easily destroyed by temperature effects or quasi - particle excitations @xcite .
the half - metallicity may also be easily destroyed if @xmath3 is located close to the band edges , either of the minority valence or conduction bands .
the recent tmr results @xcite show that elctrodes made of heusler compounds result in high tmr ratios at low temperatures but the temperature dependence is still a challenge which has to be solved to use heusler electrodes in applications .
the present investigation focuses on searching for a mixed compound where the half - metallic behaviour is stable against temperature effects .
the self - consistent electronic structure calculations were carried out using the full potential linearized augmented plane wave method ( flapw ) as provided by wien2k @xcite .
the lda@xmath5 method @xcite was used to account for on - site correlation at the transition metal sites .
semi - empirical values corresponding to 7.5% of atomic values of coulomb - exchange parameter have been used as suggested in our previous publication @xcite .
the size of the minority band gap and position of @xmath3 can be seen from the calculated energies displayed in fig .
[ fig-1 ] for the heusler alloy co@xmath0mn@xmath6fe@xmath7si . in co@xmath0mnsi
, @xmath3 is close to the top of the valence band . in co@xmath0fesi ,
the situation is different and @xmath3 is near to the bottom of the conduction band .
both of the compounds are at the two corners of the gap . in co@xmath0mn@xmath1fe@xmath1si @xmath3
is directly in the middle of the minority band gap and therefore this compound should be more stable against temperature effects or quasi - particle excitations compared to pure co@xmath0mnsi and co@xmath0fesi .
co@xmath0mn@xmath1fe@xmath1si samples were prepared by arc melting of stoichiometric amounts of the constituents in an argon atmosphere at 10@xmath8 mbar .
the polycrystalline ingots were then annealed in an evacuated quartz tube at 1273 k for 21 days .
the samples exhibiting the heusler type @xmath2 structure with a lattice parameter of 5.64 ( see fig .
another proof of the high order of the samples are the results from the @xmath9fe mbauer spectra taken of the samples ( see fig .
the spectrum is dominated by an intense sextet with a line width of approximately @xmath10 mm / s .
in addition to the sextet , a much weaker line at the centre of the spectrum is visible .
its contribution to the overall intensity of the spectrum is approximately 3.5 % .
the origin of the singlet may be caused by anti - site disorder leading to a small fraction of paramagnetic fe atoms .
the hyperfine field ( hff ) at the fe sites amounts to @xmath11 a / m . for more details and other properties of the whole series of co@xmath0mn@xmath6fe@xmath7si with the fe concentration ranging from @xmath12 to 1 in steps of 0.1
see ref.@xcite .
electronic structure calculations predicted the heusler compound co@xmath0mn@xmath1fe@xmath1si due to the position of @xmath3 in the middle of the minority band gap as very stable against any kind of effects destroying the halfmetalicity and therefore well suited for spintronic applications .
polycristaline samples were prepared exhibiting @xmath2 structure and a high magnetic order . | half - metallic ferromagnets proposed ideal for spin injection devices predicted exhibit 100 % spin polarization at fermi energy @xmath3 ) @xcite high curie temperature for ferromagnet important heusler alloys @xmath2 structure attracted interest exhibit high curie temperatures high spin polarization at fermi energy @xcite calculations show anti - site disorder high spin polarization precise control of atomic structure heusler alloys required heusler alloy co@xmath0mnsi interest predicted large minority spin band gap 0.4 ev 985 k highest curie temperature structural magnetic properties of co@xmath0mnsi reported for films single crystals tunnelling resistance data one co@xmath0mnsi film measured tmr ration 159 % at 2k % at 300k co@xmath0fesi obtained tmr ratios 60 % at 5k 41 % at 300k important for application half ferromagnets size gap in minority states position of @xmath3 inside gap small gaps destroyed by temperature effects quasi - particle excitations metallicity destroyed if @xmath3 close to band edgesrecent tmr results show elctrodes heusler compounds high tmr ratios at low temperatures temperature dependence challenge present investigation mixed compound half - metallic behaviour stable against temperature effects self - consistent electronic structure calculations using full potential linearized augmented plane wave method lda@xmath5 method used for on - site correlation at transition metal sites semi - empirical values to 7.5% atomic values coulomb - exchange parameter used previous size minority band gap position of @xmath3 from calculated energies in fig . for heusler alloy co@xmath0mn@xmath7si @xmath3 close to top valence band near bottom of conduction band compounds at two corners of gap in @xmath3 middle of minority band gap should more stable against temperature effects@xmath0mnsi samples prepared by arc melting constituents in argon atmosphere at 10@xmath8 mbar polycrystalline ingots annealed in evacuated quartz tube at 1273 k for 21 days samples heusler type @xmath2 structure lattice parameter 5.64proof high order samples results @xmath9fe mbauer spectra fig spectrum dominated by intense sextet line width @xmath10 mm / s weaker line centre visible contribution to intensity 3.5 % origin singlet anti - site disorder small fraction paramagnetic fe atoms hyperfine field hff at fe sites @xmath11 a / m details co@xmath0mn@xmath6fe@xmath7si fe concentration @xmath12 to 1 0.1 see ref.@xcite electronic structure calculations heusler compound co@xmath0mn@xmath1fe@xmath1si @xmath3 band stable against effects halfmetalicity suited for spintronic applications polycristaline samples prepared @xmath2 structure high magnetic order | 0.186158 | 1,215 | 617 | , Saving $0.0 in GPT-4. |
55,043 | we present apex observations of hcn ( 9 - 8 ) and ( 4 - 3 ) lines toward a sample of hot cores .
the spectral shapes of the main transitions are asymmetric and self - absorbed , as expected for high optical depth in a possibly infalling envelope . for spherical symmetry , the large column densities of these sources would mean that the central region is obscured by dust above a certain frequency .
however , we detected the vibrationally excited satellite lines ( @xmath0 ; @xmath19 - 8 ) at 797 ghz , which originate from the inner regions .
this indicates that high - frequency alma observations of hot core centers will be feasible . | the early stages of star formation are known to occur deep inside interstellar clouds of dust and molecular gas .
this is particularly true when massive protostars are produced : the hydrogen - burning reactions in a high - mass protostar are ignited while the object is still accreting and deeply embedded in its parent cloud .
hence , due to its much higher energy output , it stirs and heats up the natal core in a much more effective way than a low - mass object , giving rise to a hot molecular core .
this stage of high - mass star formation is characterized by processes that occur in the dense and heated ( to @xmath2 100 k ) surroundings of the protostar : ice mantles around dust grains evaporate and the frozen molecules are thereby released into the gas phase . due to the high temperature and density , the molecules are easily excited , and due to the large column densities , they can readily be observed . since hot cores display a plethora of molecular lines ( see figure [ fig : spec798 ] )
, they can be studied very well using molecular lines , particularly of highly excited transitions ( for a review , see e.g. van der tak 2004 @xcite ) .
although molecular line studies of hot cores are quite popular at mm and lower submm wavelengths ( e.g. hatchell et al . 1998 @xcite ) , a systematic study of all submm atmospheric windows has , so far , only been carried out for orion - kl with the cso ( 350 and 650 ghz , schilke et al . 1997 @xcite , 2001 @xcite ; 850 ghz , comito et al . 2005
@xcite ) .
often considered a template for the study of the hot - core phase of massive star formation , orion - kl is in fact not representative at all : * the mass involved , 10 @xmath3 , is about 2 orders of magnitude lower than the mass of many other massive hot cores ( e.g. g10.47 , olmi et al .
1996 @xcite ) , and its extremely strong molecular lines are due mostly to its proximity ( i.e. , to the relatively large beam filling factors ) ; * its anomalous geometry ( cf . the `` blister '' model , genzel & stutzki 1989 @xcite ) may result in an abnormally low dust opacity along the line of sight , a condition that is not representative of `` typical '' hot - core sources .
in fact , for a typical h@xmath4 column density of @xmath5 @xmath6 , * the dust opacity @xmath7 , with @xmath8 , will become @xmath2 1 at a frequency of about 550 ghz*. thus , while orion does look similar to many other hot cores at frequencies of @xmath9 ghz , an extrapolation to shorter wavelengths is by no means obvious ( see figure [ fig : dust ] ) .
molecular observations of massive ( @xmath2 100 m@xmath10 ) cores in the 690 or 850 ghz ( 450 or 350 @xmath11 m ) band are thus likely to show not only quantitative but also qualitative differences with respect to lower - frequency observations .
at least two scenarios may be encountered : \(i ) a geometrically straightforward one , where the observed object is spherical in shape and isotropic in density structure . in this case
, the high dust opacity at 350/450 @xmath11 m will obscure the innermost , hottest and densest regions of the hot molecular core , while opening the path to absorption - line studies of the outer envelope .
\(ii ) one in which the hot - core component is actually made up by a cluster of high mass clumps . here , in spite of the high nominal h@xmath4 column densities ( hence high @xmath12 )
, the radiation from close to the central object(s ) could still be observable , since the radiation can leak out between the clumps .
the same applies to many non - spherical geometries ( e.g. face - on disks , hollow outflow cones ) .
scenario ( i ) above would be bad news for high - frequency alma observations , but it would be fatal for planned high frequency herschel / hifi surveys , which will extend to even higher frequencies , 1.9 thz , corresponding to a wavelength of 150 @xmath11 m . of course , one would not lose all spectral features , since low excitation lines would still show up in absorption and would provide valuable information ( similar to the ir absorption lines ) , but the direct view to the central engine would be blocked .
to discern between the proposed scenarios , and to help selecting suitable objects for the herschel / hifi spectral scan key project , we observed the vibrationally excited hcn(9 - 8 , @xmath0 ) line at 797 ghz .
because of its high critical density ( @xmath13 @xmath14 ) , this transition can hardly be excited by collisions , but can be pumped by 14 @xmath11 m ir radiation ( corresponding to the the energy of the @xmath0 state ) which is emitted by warm ( @xmath2 150 k ) dust ( carroll & goldsmith 1981 @xcite , hauschildt et al . 1993 @xcite ) .
it therefore comes from either unusually dense and warm clumps or from regions of a high ir radiation field , i.e. the line originates most likely from close to the central star . using the flash ( heyminck et al .
2006 @xcite ) and apex-2a ( risacher et al .
2006 @xcite ) instruments on apex ( gsten et al . 2006
@xcite ) , we observed the @xmath15=9 - 8 and 4 - 3 transitions in the @xmath16=0 and 1 vibrational states , as well as the 4 - 3 transition of h@xmath17cn and hc@xmath18n . using two fast fourier
transform spectrometers ( klein et al .
2006 @xcite ) as backends , three 1.8 ghz wide bands were covered , centered around 798/810 , 342/354 and 345/357 ghz ( see figure [ fig : spec798 ] as an example ) .
the source sample consists of 10 hot cores ( see figure [ fig : hcnlines ] plus 17233 - 3606 and w33a ) .
the hcn(4 - 3 , @xmath0 ) should not suffer significantly from high dust opacities at 850 @xmath11 m and can be used to calibrate the intrinsic strength of the lines .
we found evidence for the ( 9 - 8 , @xmath0 ) line toward all sour- ces where the ( 4 - 3 , @xmath0 ) was detected , indicating that , at least at this wavelength and for this source sample , the central core is still visible .
hence alma observations of the central engines of hot cores even at the highest frequencies appear to be feasible , although the cores may shut down at higher frequencies relevant for herschel / hifi .
the ground state lines often display self - absorption features .
their asymmetric line shapes with the blue part being the stronger one could be indicative of infall motion . to extract the information on the source structure from the line strengths and shapes , we plan to analyze the data with detailed radiative transfer modeling ( hogerheijde & van der tak 2000 @xcite ) .
preliminary results indicate that modelling the source as a spherical cloud can not reproduce all observed lines , indicating deviations from this simple structure .
investigations of effects of such deviations are under way . | early stages star formation occur inside interstellar clouds dust molecular gas particularly true when massive protostars produced hydrogen - burning reactions in high - mass protostar ignited while object embedded in parent cloud due to higher energy output stirs heats up natal core rise to hot molecular core stage high - mass star formation characterized by processes in dense heated ( to @xmath2 100 k ) surroundings protostar ice mantles around dust evaporate frozen molecules released into gas phase due to high temperature density molecules easily excited large column densities readily observed hot cores display molecular lines figure can be studied well using molecular lines particularly highly excited transitions 2004 molecular line studies of hot cores popular at mm lower submm wavelengths systematic study of all submm only carried out for orion - kl with cso considered template for study hot - core phase massive star formation orion - kl not representative mass involved , 10 @xmath3 about 2 orders of magnitude lower than mass many other massive hot cores1996 @xcite ) strong molecular lines due to proximity large beam filling factors anomalous geometry blister model low dust opacity sight not representative of typical hot - core sources typical h@xmath4 column density @xmath5 @xmath6 dust opacity @xmath7 @xmath8 @xmath2 1 at frequency 550 ghz orion similar to hot cores at frequencies @xmath9 ghz extrapolation to shorter wavelengths obvious figure molecular observations of massive @xmath2 100 m@xmath10 ) cores in 690 or 850 ghz ( 450 or 350 @xmath11 m ) band likely show quantitative qualitative differences lower - frequency observations two scenarios geometrically straightforward observed object spherical in shape isotropic in density high dust opacity at 350/450 @xmath11 m innermost hottest densest regions hot molecular core to absorption - line studies hot - core component by cluster of high mass clumps high h@xmath4 column densities high @xmath12 radiation from central object(s observable leak between clumpsapplies to non - spherical geometries face - on disks hollow outflow cones scenario ( i ) bad news for high - frequency alma observations fatal for planned high frequency herschel / hifi surveys to higher frequencies 1.9 thz wavelength of 150 @xmath11 m lose all spectral features low excitation lines show up in absorption provide valuable information direct view to central engine blocked scenarios / observed vibrationally excited hcn(9 - 8 , @xmath0 ) line at 797 ghz high critical density transition excited by collisions pumped by 14 @xmath11 m ir radiation energy @xmath0 state emitted by warm ( @xmath2 150 k ) dust comes from dense warm clumps or high ir radiation field line originates from close to central star using flash apex-2a instruments observed @xmath15=9 - 8 and 4 - 3 transitions in @xmath16=0 and 1 vibrational states 4 - 3 transition of h@xmath17cn and hc@xmath18n using two fast fourier transform spectrometers2006 @xcite backends three 1.8 ghz bands covered around 798/810 342/354 345/357 ghz figure fig spec798 source sample 10 hot cores hcnlines plus 17233 - 3606 w33a hcn(4 - 3 , @xmath0 ) high dust opacities at 850 @xmath11 m calibrate intrinsic strength lines evidence ( 9 - 8 , @xmath0 ) line sour- ces ( 4 - 3 , @xmath0 ) detected central core visible observations central engines hot cores highest frequencies feasible cores shut down higher frequencies ground state lines display self - absorption features asymmetric line shapes blue part stronger infall motion source structure line strengths plan analyze data detailed radiative transfer modeling van der tak 2000 @xcite preliminary results modelling source spherical cloud reproduce all lines deviations investigations effects deviations under way | 0.268361 | 1,657 | 827 | , Saving $0.0 in GPT-4. |
55,057 | we report preliminary results for the @xmath0 pion scattering length @xmath1 , calculated with chirally improved fermions .
after chiral extrapolation our results are in good agreement with both , experimental and theoretical predictions . | since the pion is the goldstone boson of the spontaneous breaking of chiral symmetry , observables related to the pion give insight into the chiral dynamics of qcd .
for example the pion scattering length is a parameter entering the effective chiral lagrangian of qcd and a precise determination of this observable in an ab - initio lattice calculation is desirable . on the lattice scattering data of stable particles can be calculated using lscher s formula @xcite which relates the scattering length to the energy @xmath2 of the scattering state .
the crucial technical challenge is a very precise determination of @xmath2 and advanced strategies for error reduction have to be used . for chirally sensitive quantities , such as the pion scattering length ,
it is important to be able to work with small pion masses . using fermions based on the ginsparg - wilson equation it is now possible to reach pion masses much lower than what was accessible with more traditional formulations .
this implies that the chiral extrapolation of the results to the physical pion mass becomes more reliable .
our calculation is based on the chirally improved ( ci ) dirac operator @xcite .
ci fermions are a systematic approximation of a solution of the ginsparg - wilson equation and have been tested successfully @xcite in spectroscopy down to pion masses of @xmath3 mev . [ cols="^,^,^,^,^,^",options="header " , ] table 2 . our final results for @xmath4 compared to experimental data and to 2-loop chiral perturbation theory .
we thank jimmy juge for important comments and discussions .
our computations were done on the hitachi sr8000 at lrz in munich and at the rz regensburg . c. gattringer , phys .
d 63 ( 2001 ) 114501 ; c. gattringer , i. hip and c. b. lang , nucl . phys .
b 597 ( 2001 ) 451 . c. gattringer _
[ bgr collaboration ] , nucl .
b 677 ( 2004 ) 3 ; nucl .
suppl . 119 ( 2003 ) 796 ;
c. gattringer , nucl . phys .
suppl . 119 ( 2003 ) 122 .
c. gattringer , r. hoffmann and s. schfer , phys .
d 65 ( 2002 ) 094503 .
juge [ bgr collaboration ] , nucl .
suppl . 129 ( 2004 ) 194 .
j. gasser and h. leutwyler , annals phys .
158 ( 1984 ) 142 . | pion goldstone boson chiral symmetry observables chiral dynamics qcd pion scattering length parameter chiral lagrangian qcd precise determination in lattice calculation desirable lattice scattering data stable particles calculated using lscher s formula @xcite relates scattering length to energy @xmath2 scattering state challenge precise determination @xmath2 advanced strategies error reduction for chirally sensitive quantities pion scattering length important work with small pion masses using fermions based ginsparg - wilson equation reach pion masses lower chiral extrapolation to pion mass reliable calculation based on improved ci dirac operator @xcite fermions ginsparg - wilson equation tested successfully @xcite in spectroscopy to pion masses @xmath3 mev table 2 final results for @xmath4 compared to experimental data 2-loop chiral perturbation theory thank jimmy juge for comments discussions computations done on hitachi sr8000 at lrz munich rz regensburg c. gattringer phys . d 63 ( 2001 ) 114501 lang 597 ( 2001 ) 451677 2004 ) 3 nucl 119 2003 ) 796 gattringer 119 2003 ) 122 gattringer hoffmann schfer 65 ( 2002 ) 094503 collaboration nucl 129 ( 2004 ) 194 j gasser h leutwyler 158 ( 1984 ) 142 | 0.129567 | 611 | 304 | , Saving $0.0 in GPT-4. |
55,060 | we have been searching for surviving companions of progenitors of galactic type - ia supernovae , in particular sn 1572 and sn 1006 .
these companion stars are expected to show peculiarities : ( i ) to be probably more luminous than the sun , ( ii ) to have high radial velocity and proper motion , ( iii ) to be possibly enriched in metals from the snia ejecta , and ( iv ) to be located at the distance of the snia remnant .
we have been characterizing possible candidate stars using high - resolution spectroscopic data taken at 10m - keck and 8.2m - vlt facilities .
we have identified a very promising candidate companion ( tycho g ) for sn 1572 ( see @xcite , however for a different view see @xcite ) but we have not found any candidate companion for sn 1006 , suggesting that sn event occurred in 1006 could have been the result of the merging of two white dwarfs ( see @xcite ) . adding these results to the evidence from the other direct searches , the clear minority of cases ( 20% or less ) seem to disfavour the single - degenerate channel or that preferentially the single - degenerate escenario would involve main - sequence companions less massive than the sun .
therefore , it appears to be very important to continue investigating these and other galactic type - ia sne such as the johannes kepler sn 1604 . | type ia supernovae ( sne ia ) are the best known cosmological distance indicators at high redshifts .
their use led to the discovery of the currently accelerating expansion of the universe ( see @xcite , @xcite ) .
these sne are thought to occur when a white dwarf made of carbon and oxygen accretes sufficient mass to trigger a thermonuclear explosion .
the explosion could happen via accretion from a companion star ( single degenerate ( sd ) channel ) or via merging of two white dwarfs ( double - degenerate ( dd ) channel ) .
therefore , a companion star will survive the explosion only in the sd channel ( see these reviews @xcite , for further information ) .
both channels might contribute to the production of type - ia supernovae but the relative proportions of their contributions remain unclear .
one way to investigate type - ia sne is by performing direct survey of the field of historical events @xcite .
depending on the date of the sn explosion and the distance to the sn remnant , we are able to define the region where to find the possible companion star , close to the geometrical center of the sn remnant @xcite .
there are a few known historical galactic type - ia supernova events : sn 1572 ( tycho brahe supernova ) , sn 1006 , sn 1604 ( kepler supernova ) , and three new recently identified type - ia supernova remnants : sn 185 ( rcw86 ) , g299.2 - 2.9 and g272.2 - 3.2 .
our group have been investigating these galactic historical snia , trying to search for companion stars of progenitors of historical galactic type - ia sne with the aim of clarifying the origin of these cosmological candles .
we present here the results we have got in two of them , sn 1572 and sn 1006 .
ruiz - lapuente et al .
@xcite investigated the type - ia sn 1572 and found one promising candidate ( called tycho g ) to be the surviving companion of the progenitor of sn 1572 remnant , thus the origin of tycho s supernova remnant ( snr ) might be attributed to the sd channel . on the other hand , for instance ,
the absence of any ex - companion in the supernova remnant snr 0509@xmath067.5 , in the large magellanic cloud ( lmc ) , down to very faint magnitudes , strongly suggests that the sn explosion there was produced by a dd system .
gonzlez hernndez et al .
@xcite studied the star tycho g and derived its stellar parameters and chemical abundances , using hires@10m - kecki high - resolution spectra . in fig .
[ fig1 ] , we show a re - determination of the ni abundance in tycho g compared with those of similar stars of the galactic thin and thick disk . a slightly ni enhancement is seen in tycho g ( see @xcite ) , which may suggest ni pollution from tycho sn 1572 ejecta .
we have been also improved our previous proper motion determination of tycho g together with many other stars in the tycho sn field using images from hst programmes go-9729 , go-10098 and go-12469 ( pi : ruiz - lapuente ) , which span about 8 yr . in fig .
[ fig1 ] we compare the proper motion ( pm ) of tycho g , perpendicular to the galactic plane , together with those of other stars of the sn 1572 field , in comparison with the distribution of pms predicted by the besanon model of the galaxy @xcite .
tycho g is the only peculiar star in fig .
[ fig1 ] , and thus possibly the only viable candidate to be the surviving companion of sn 1572 @xcite .
several models have been proposed to evaluate the luminosity , peculiar radial velocity and proper motion , and rotational velocity of tycho g and support this star as the possible surviving companion @xcite .
_ left panel : _ [ ni / fe ] abundance ratio of tycho - g ( filled triangle ) in comparison with the abundances of f- , g- and k - type metal - rich dwarf stars from @xcite .
thin - disc stars are depicted as filled circles , whereas transition and thick - disc stars are the empty circles .
solar analogues are shown as empty squares from @xcite .
the size of the error bars indicates the 1@xmath1 uncertainty .
the dashed - dotted lines indicate solar abundance values . _
right panel : _ the distributionof proper motions , @xmath2 , perpendicular to the galactic plane , as a function of distance , in the direction of the x - ray centroid of tycho s snr , for thin - disc and thick - disc stars together ( and [ fe / h ] @xmath3 ) , according to the besanon model of the galaxy ( see @xcite ) .
1-@xmath1 , 2-@xmath1 and 3-@xmath1 regions are indicated .
the position of tycho - g is depicted as a triangle , @xmath4[kpc ] @xmath5 and @xmath2[mas yr@xmath6 @xmath7 .
, title="fig:",width=268 ] _ left panel : _ [ ni / fe ] abundance ratio of tycho - g ( filled triangle ) in comparison with the abundances of f- , g- and k - type metal - rich dwarf stars from @xcite .
thin - disc stars are depicted as filled circles , whereas transition and thick - disc stars are the empty circles .
solar analogues are shown as empty squares from @xcite .
the size of the error bars indicates the 1@xmath1 uncertainty .
the dashed - dotted lines indicate solar abundance values . _
right panel : _ the distributionof proper motions , @xmath2 , perpendicular to the galactic plane , as a function of distance , in the direction of the x - ray centroid of tycho s snr , for thin - disc and thick - disc stars together ( and [ fe / h ] @xmath3 ) , according to the besanon model of the galaxy ( see @xcite ) . 1-@xmath1 , 2-@xmath1 and 3-@xmath1 regions are indicated .
the position of tycho - g is depicted as a triangle , @xmath4[kpc ] @xmath5 and @xmath2[mas yr@xmath6 @xmath7 .
, title="fig:",width=268 ]
the sn 1006 has been also investigated with the aim of finding the surviving companion of the sn progenitor ( see @xcite ) .
we selected all targets up to a limiting magnitude of @xmath8 , and we acquired high - resolution spectroscopy using uves@8.2m - vlt within a circle of radius 4 around the geometrical centre of the sn 1006 remnant . in fig .
[ fig2 ] we show uves@vlt observed spectra of several giant stars located at distances consistent with the sn 1006 remnant .
model simulations of the impact of snia ejecta indicate that red giant should have been stripped off most of its hydrogen envelope and therefore , these stars , which are normal giant stars , are not expected to be the surviving companion .
the stellar parameters were derived from the measurements of equivalent widths of fei - ii lines using the code stepar @xcite . in fig .
[ fig3 ] we depict the abundance ratios [ x / fe ] of several fe - peak elements of the stars in the sn 1006 field compared to those in stars belonging to the galactic disk .
none of the stars reveals any remarkable abundance peculiarity .
small portion of the high - resolution uves spectra of giant stars at the distance of the sn 1006 remnant sorted in decreasing effective temperature order from top to bottom .
, width=381 ] chemical abundances of fe - peak elements derived using the ew technique .
we performed a differential analysis on a line - by - line basis , using the solar uves spectrum of the moon as reference .
red triangles correspond to the four giant stars whose distances are marginally compatible with that of the remnant of sn 1006 .
blue squares , to the rest of the stars in the sample .
, width=381 ] using optical and near - infrared photometry together with the stellar parameters we derive the distances to the stars of the sn 1006 field ( see fig . [ fig4 ] ) .
we clearly see that only giant stars are compatible within the error bars with the distance to the sn remnant .
we inspect the 2mass photometry catalogue and use the irfm method ( see @xcite ) to derive effective temperatures of the additional candidates , with the aim of identifying those at the distance of the sn 1006 remnant .
we found no main - sequence brighter than @xmath9 , which brings the limit down to @xmath10 , corresponding to @xmath11 ( approximately equal to , or slightly less than , solar luminosity ) .
hydrodynamical simulations suggest that even if the pre - sn companion is main - sequence solar - mass star , about 1000 yr after the strong impact of the snia ejecta , the star does not have enough time to become dimmer than the sun .
however , di stefano , voss and claeys ( @xcite ) have suggested that , by the time that the exploding white dwarf has reached the critical mass for explosion ( within the single degenerate scneario ) , the companion ( donor ) might have become a wd .
this finding deserves further investigation for its implications in the surviving companion of sne ia .
distances of the target stars within the sn 1006 field as a function of their stellar parameters , metallicity and radial velocity .
giant , subgiant and dwarf stars are shown as red , green and blue filled circles , respectively .
the black dashed line gives the distance of the sn 1006 remnant .
the violet dashed - dotted line show the well - defined distance error range of the sn 1006 remnant . ,
width=381 ] several stars analyzed in @xcite are compared to model predictions , but those models gives hotter candidates than those observed ( see @xcite for further details ) .
thus , sn 1006 , the brightest event ever observed in our galaxy , should have been possibly produced by merging of two white dwarfs . adding this result to the evidence from the other direct searches
, the single - degenerate channel appears either to happen in the minority of the cases ( 20% or less ) , or preferentially it involves main - sequence companions with masses more probably below that of the sun .
j.i.g.h . acknowledges financial support from the spanish ministry projects mineco aya2011 - 29060 and mineco aya2011 - 26244 , and also from the spanish ministry of economy and competitiveness ( mineco ) under the 2011 severo ochoa program mineco sev-2011 - 0187 .
p.r.l . is grateful for the support of the spanish ministry project mineco aya201236353 . h.m.t and
d.m acknowledge financial support from the universidad complutense de madrid ( ucm ) , the spanish ministry of econ- omy and competitiveness ( mineco ) from pojects aya2011 - 30147-c03 - 02 , and the comunidad de madrid under pricit project s2009/esp-1496 ( astro- madrid ) .
h.m.t also acknowledges the financial support of the spanish ministry of economy and competitiveness ( mineco ) under grants bes-2009 - 012182 and eebb - i-12 - 04038 . | type ia supernovae are cosmological distance indicators at high redshifts led to accelerating expansion universe occur when white dwarf carbon oxygen accretes mass thermonuclear explosion explosion could happen via accretion from companion star or merging two white dwarfs companion star explosion only in sd channel both channels contribute to - ia supernovae relative proportions unclear investigate type - ia sne direct survey of historical events depending on date sn explosion distance to remnant define region possible companion star close to center remnant few known historical galactic type - ia supernova events : sn 1572 sn 1006 sn 1604 three new type - supernova remnants : sn 185 ( rcw86 ) , g299.2 - 2.9 g272.2 - 3.2 group investigating historical snia for companion stars of progenitors type - ia sne clarifying origin of cosmological candles present results in sn 1572 and sn 1006 . ruiz - lapuente et al @xcite investigated sn 1572 found promising candidate tycho g ) surviving companion of 1572 remnant origin of tycho s supernova remnant might to sd channel absence of ex - companion in supernova remnant snr 0509@xmath067.5 in large magellanic cloud suggests sn explosion produced by dd system @xcite studied star tycho g derived stellar parameters chemical abundances using hires@10m - kecki high - resolution spectra fig - determination of ni abundance in tycho g compared similar stars galactic ni enhancement seen in tycho g may suggest ni pollution from tycho sn 1572 ejecta improved proper motion determination of tycho g stars using images from hst programmes go-9729 go-10098 go-12469 proper motion of tycho g galactic plane other stars sn 1572 field distribution pms predicted by besanon model of galaxy tycho g only peculiar star in fig possibly only viable candidate surviving companion of sn 1572models proposed evaluate luminosity radial velocity motion rotational velocity of tycho g support possible surviving companion @xcite left panel abundance ratio tycho - g triangle with f- g- k - type metal - rich dwarf stars @xcite thin - disc stars filled circles transition thick - disc stars empty circles solar analogues empty squares error bars 1@xmath1 uncertainty dashed - dotted lines indicate solar abundance values right panel distributionof motions perpendicular galactic plane function distance direction x - ray centroid tycho s snr thin - disc thick - disc stars besanon model galaxy 1-@xmath1 2- 3- regions indicated position of tycho - g triangle left panel abundance ratio tycho - g filled triangle with f- g- k - type metal - rich dwarf stars from @xcite thin - disc stars filled circles transition thick - disc stars empty circles solar analogues empty squares error bars 1@xmath1 uncertainty dashed - dotted lines indicate solar abundance valuesright panel distributionof motions @xmath2 perpendicular galactic plane function distance direction x - ray centroid tycho s snr thin - thick - disc stars fe / h @xmath3 besanon model galaxy 1-@xmath1 2- 3-@xmath1 regions indicated position tycho - g triangle @xmath4[kpc @xmath5 @xmath2@xmath6 @xmath7 sn 1006 investigated finding surviving companion selected targets magnitude @xmath8 acquired high - resolution spectroscopy uves@8.2m - vlt circle radius 4 around centre sn 1006 remnant fig fig2 uves@vlt observed spectra giant stars distances consistent sn 1006 remnant simulations impact snia ejecta indicate red giant stripped hydrogen envelope normal giant stars not expected surviving companion stellar parameters derived from measurements equivalent widths fei - ii lines code stepar @xcite fig fig3 abundance ratios x / fe fe - peak elements stars sn 1006 field compared stars galactic disk none remarkable abundance peculiarityhigh - resolution uves spectra giant stars distance sn 1006 remnant sorted decreasing temperature order top to bottom chemical abundances fe - peak elements derived ew technique performed differential analysis line - by - line solar uves spectrum moon red triangles four giant stars marginally compatible with remnant sn 1006 blue squares rest stars sample optical near - infrared photometry stellar parameters distances stars sn 1006 field only giant stars compatible within error bars distance sn remnant 2mass photometry catalogue irfm method derive effective temperatures additional candidates identifying distance sn 1006 remnant no main - sequence brighter than @xmath9 limit to @xmath10 @xmath11 equal or less than solar luminosity hydrodynamical simulations suggest pre - sn companion main - sequence solar - mass star 1000 yr after impact snia ejecta star enough time become dimmer than sun voss claeys exploding white dwarf reached critical mass explosion companion might have become wd finding deserves further investigation implications surviving companion sne iadistances target stars within sn 1006 field stellar parameters metallicity radial velocity giant subgiant dwarf stars shown as red green blue circles black dashed line distance of sn 1006 remnant violet line distance error range remnant stars analyzed in @xcite compared to model predictions models hotter candidates sn 1006 brightest event galaxy produced by merging two white dwarfs single - degenerate channel in minority cases 20% or less or involves main - sequence companions with masses below sun acknowledges financial support from spanish ministry projects mineco aya2011 - 29060 - 26244 spanish ministry of economy and competitiveness 2011 program - 0187 grateful for support spanish ministry project mineco aya201236353 acknowledge financial support from universidad complutense de madrid spanish ministry of econ omy competitiveness mineco comunidad de madrid project s2009/esp-1496 acknowledges financial support spanish ministry of economy competitiveness mineco under grants bes-2009 - 012182 eebb - i-12 - 04038 | 0.359578 | 2,530 | 1,237 | , Saving $0.1 in GPT-4. |
55,067 | a complex approach phonon quantum discrete model ( pqdm ) was developed to describe dynamics , kinetics and statistics of phonons in carbon nanostructures with zero - chirality of both zig - zag and armchair geometry .
the model allows include into the pure phonon problem existing interaction with others subsystems : electrons , photons , impurities and defects .
we predict that planar c- structures are geometrically stable and may bridge interelectrode space in strong external electric field .
the exact solution of generalized thermal conductivity ( tc ) equation was obtained for nanotubes .
temperature distribution along the tube was derived analytically .
the diagonalization procedure for the case ofstrong ph - ph interaction was proposed .
it was shown the quadratic increasing of heat conductivity with the growth of the phonon mean free path ( pmfp ) .
heat capacitance and the entropy of carbon linear tubes were calculated as the function of temperature .
our theoretical approach explains the nature of good tc in carbon and carbon - like materials by existing of the soft vibration branch ( low frequency radial breathing mode phonons with high density of states at thermal energies ) accompanied by structure hardness ( high frequency @xmath0- and z - branches ) providing big pmfp .
tc coefficient for high conducting channel in surrounding medium was calculated .
the mechanism of heat conductivity temperature damping was analyzed .
two competitive tendencies produce tc maximum at intermediate temperatures ( 100 - 300)k .
it was shown the strongly non - linear increasing of effective heat conductivity with the growth nanotubes concentration .
it was shown that insertion of armchair nanotube inside a medium or its coating by polyacetilene molecule considerably changes the structure of radial breathing phonons .
e.ya.glushko ( 1 and 2 ) , v.n.evteev ( 3 ) , m.v.moiseenko ( 3 ) , n.a.slusarenko ( 3 ) and a.a.zakhidov ( 4 ) ( ( 1 ) institute for semiconductors physics nas ukraine , pr.nauki .
21 , kiev , 03047 , ukraine ; ( 2 ) slavonic university , anry barbusse str .
9 , kiev 03150 , ukraine ; ( 3 ) pedagogical university , gagarin avenue 54 , krivoy rog , 50086 , ukraine ; ( 4 ) texas university at dallas , richardson , tx 75083 - 0688 , usa ) ( e - mail : eugene.glushko@scientist.com ) i. motivation \ii .
introduction \iii .
systems under consideration \iv .
phonon quantum discrete model ( pqdm ) v. vibrational dynamics .
linear approximation .
\vi .
frequency spectrum and dos .
zig - zag geometry .
\vii .
frequency spectrum and dos .
armchair geometry .
\viii .
amplitude distribution and knot s theorem ix .
heat transfer x. phonon mean free path and size dependences xi .
thermal coefficient temperature dependence xii .
generalized phonon transport equation \xiii .
summary and discussion | phonon dynamics in open and closed carbon and carbon - like nets .
how many phonon branches are present ? whether exists a simple correlation between differed phonon types for different nt geometries ?
vibrational eigenmode and its amplitudes distribution in the net .
vibrational eigenmodes , phonons , micro - sound and heat propagation .
phonon mean free path .
sound waves boundary conditions .
the mechanism of temperature damping at high temperatures and concrete law of increasing at low temperatures .
maximal thermal conductivity for solitary carbon tube is not measured reliably now and changes in wide interval from 200 w / mk to 3000 w / mk for different authors .
the temperature of maximal thermal conductivity for carbon tubes is not surely measured also and may change in wide interval from 150 k to 300 k for different authors .
the question is whether exist differences in thermal conductivity of solitary carbon tubes depending on type ( armchair or zig - zag ) , chirality and diameter . which physical factors are essential in electron transport through the molecule ?
do phonons ( vibrons ) participate essentially in electron transport through the molecule ?
are differences in thermal conductivity of solitary carbon tubes , flat carbon structures like graphene and graphite ( in plane ) essential ?
@xmath1 @xmath2
zig - zag nt @xmath3 @xmath4 @xmath5 armchair nt @xmath6 @xmath7 @xmath8 @xmath9 .
@xmath10 .
@xmath11 .
@xmath12 @xmath13 for 3-polar contact @xmath14 and for 4-polar contact @xmath15 @xmath16 @xmath17 approximation @xmath18 @xmath19
@xmath20 @xmath21 the mean free path along the tube lph .
@xmath22 @xmath23 .
@xmath24 .
@xmath25 @xmath26 @xmath27 @xmath28 temperature distribution along the tube @xmath29z } } } \right ) \label{eq12.4}\end{aligned}\ ] ] .
the boundary conditions for microsound spreading through periodic nt complexes @xmath30 atomic vibration dynamics was considered for carbon zero - hirality nanotubes connected with solid matrix .
vibrational eigenmodes , density of states and amplitude distribution for tubes inserted into a cylindrical well inside the medium and bounded with its hard inner wall were calculated .
phonon band parameters were analyzed in linear approximation for three types of vibration : athimuthal or tangential j - branch , radial breathing mode ( r - branch ) and longitudinal z - branch .
it s shown that phonon propagation in actual nanotubes is characterized by a kind of `` compactification '' of circular freedom s degree due to the big phonon mean free path .
nanotubes of actual diameters are ideal one - dimensional phonon qnd heat conductors .
thermal fluxes and thermal conductivity were considered in pqdm .
temperature dependences were obtained .
the mechanism of heat conductivity high temperature damping is cleared .
two competitive tendencies produce thermal conductivity maximum at intermediate temperatures ( 100 - 300)k .
the explanation of the nature of good thermal conductivity in carbon and carbon - like materials by existing of the soft vibration branch ( low frequency rbm phonons with high dos at thermal energies ) accompanied by structure hardness ( high frequency j- and z - branches ) providing big mean free path for phonons ( fig.5.2 ) .
boundary problem for temperature distribution in channel - medium system was considered in cylindrical symmetry .
the method of partial solution building for cylindrical boundary problem was proposed .
temperature distribution isotherms were calculated .
concentration dependences were considered for thermal conductivity of nt - solid matrix composite .
it was shown the strongly non - linear increasing of effective heat conductivity with the growth nanotubes concentration .
phonon dynamics was considered for carbon zero - chirality nanotubes embraced by polymer molecule .
vibrational eigenmodes , density of states and amplitude distribution for a bounded tube were calculated .
phonon band parameters were obtained in linear approximation for three types of vibration . | phonon dynamics in open closed carbon nets how many phonon branches ? correlation between phonon types nt geometries ? vibrational eigenmode amplitudes distribution in net vibrational eigenmodes phonons micro - sound heat propagation phonon mean free path sound waves boundary conditions temperature damping at high temperatures law increasing at low temperatures maximal thermal conductivity for solitary carbon tube not measured changes from 200 w mk to 3000 w / mk authors temperature maximal thermal conductivity carbon tubes not measured change 150 k to 300 k differences in thermal conductivity solitary carbon tubes depending on type armchair zig - zag chirality diameter physical factors essential in electron transport molecule ? phonons vibrons participate in electron transport ? differences in thermal conductivity of solitary carbon tubes flat carbon structures like graphene graphite essential ? zig - zag armchair nt 3-polar contact 4-polar contact approximation mean free path along tube temperature distribution along tubeboundary conditions microsound through complexes atomic vibration dynamics carbon zero - hirality nanotubes connected solid matrix vibrational eigenmodes density amplitude distribution for tubes cylindrical well wall calculated phonon band parameters analyzed three types j radial longitudinal z - phonon propagation in nanotubes compactification big phonon free path nanotubes ideal phonon heat conductors thermal fluxes conductivity considered temperature dependences obtained mechanism heat conductivity high temperature damping cleared competitive tendencies produce thermal conductivity maximum at intermediate temperatures 100 - 300 good thermal conductivity in carbon materials soft vibration branch structure hardness high frequency z - branches mean free path phonons boundary problem temperature distribution channel - medium system considered cylindrical symmetry partial solution building cylindrical boundary problem proposed temperature distribution isotherms calculated concentration dependences thermal conductivity nt - solid matrix composite non - linear heat conductivity with growth nanotubes concentration phonon dynamics considered for carbon zero - chirality nanotubes embraced polymer moleculevibrational eigenmodes density amplitude bounded tube calculated phonon band parameters three types vibration | 0.180259 | 895 | 378 | , Saving $0.0 in GPT-4. |
55,101 | we note that the fine structure at the endpoint region of the beta decay spectrum is now essentially known using neutrino oscillation data , if the mass of one neutrino is specified .
this may help to identify the effects of nonzero neutrino masses in future experiments .
we also give a compact description of the entire range of allowed neutrino masses as a function of the third neutrino mass , @xmath0 .
a three neutrino assumption is being made .
an exact treatment of phase space kinematics is used , in contrast to the conventional approximate formula .
this work is independent of theoretical models ; however , additional restrictions due to the assumption of a complementary ansatz " for the neutrino mass matrix are also discussed .
the ansatz implies that the values of the three neutrino masses should approximately be able to form a triangle .
it is noted that most of the presently allowed neutrino mass sets have this triangular property . | measurement of the maximum electron energy in beta decay processes is the original approach @xcite to finding the absolute value of a possible neutrino mass .
it is still a subject of great interest @xcite and may yield the first direct measurement of this crucial quantity .
especially interesting is the process of tritium decay @xcite , @xmath1 in this paper , we discuss how results obtained from present neutrino oscillation experiments @xcite yield characteristic shapes for the beta decay endpoint spectrum which could help in signal identification @xcite .
the main point is that the present oscillation data already predict fairly reliably a bi - unique description of the shape of the endpoint spectrum if the mass of any neutrino is specified .
this fortunate situation arises since , as we will discuss , the only unknown mixing angle is strongly bounded and hence can not have much effect on the endpoint spectrum shape .
so , running over the possible values of the third neutrino mass @xmath0 , we would get a catalog " of shapes which can be compared with the experimental shape to find a best fit .
of course , there are a number of practical corrections to the observed end point spectrum other than just the phase space and neutrino mixing effects to be considered here .
these include @xcite ( i ) different final masses of @xmath2 due to different final atomic electron states , ( ii ) corrections due to binding of @xmath3 in a molecule , ( iii ) resolution function of the detector , ( iv ) final state effects .
the corrections should be made on each page " of the catalog obtained .
we start in section ii with an exact numerical treatment of the phase space , which contains a relevant correction to the approximate treatment often used . in section iii
we show that , as @xmath0 decreases from about the highest value considered to be consistent with information from cosmology , there are , in general , two solutions : type i where @xmath0 is the largest of the neutrino masses and type ii where @xmath0 is the smallest of all the neutrino masses .
below about @xmath4 ev only the type ii solutions are allowed .
a characteristic feature is that the neutrinos 1 and 2 are extremely close in mass ; their splittings range from about a ten thousandth of an ev to a hundredth of an ev . the end point spectrum shapes , showing the slope discontinuities corresponding to the vanishing of each neutrino s contribution , are plotted to illustrate these features . with a very great accuracy that might be achieved in the future , the spacings of the neutrinos could conceivably survive the needed corrections mentioned above
. however the most practical procedure would probably be to integrate each corrected predicted spectrum shape ( of definite @xmath0 and type ) over an energy interval and then look to see which one is best fit by the experimental value . in section
iv we discuss some model restrictions on the ranges of @xmath0 which lead to type i and type ii solutions .
these restrictions arise by assuming a so - called complementary ansatz " for the neutrino mass matrix .
the observation of @xmath0 in the non - allowed ranges would constitute a falsification of the ansatz . finally , section v contains a brief summary and some discussion .
taking the tritium decay example and assuming one massive neutrino for the time being , let @xmath5 be the mass of the tritium atom , @xmath6 the mass of @xmath2 , @xmath7 the electron mass and @xmath8 the neutrino mass .
often @xcite , the kinetic part of the recoil nucleus energy is neglected so one has an easy approximate formula for the maximum electron energy : @xmath9 the exact formula , corresponding to the physical situation where the neutrino and the recoil nucleus both emerge with the same velocity is @xmath10 .
\label{exactform}\ ] ] to get an idea of the accuracy of the approximate formula we use @xcite the input masses ( in mev ) : @xmath11 to obtain table i. notice that in eq .
( [ decimals ] ) we have , for the purpose of conveniently tracking " digits in this illustration , included more digits than warranted for the experimental accuracy of @xmath5 and @xmath6 .
.comparison of exact and approximate maximum electron energy for different neutrino masses .
note that @xmath12 . [ cols="^,^,^",options="header " , ] it may be amusing to see why the ansatz leads to more restrictions .
take the simplified case , treated in detail in ref .
@xcite , where @xmath13
. then the ansatz condition amounts to the lengths corresponding to the three neutrino masses forming a triangle in the complex plane .
( the two independent majorana phases are related to the internal angles of this triangle ) .
the procedure outlined here obtains , for a given assumed value of @xmath0 , the three neutrino masses .
however it is not guaranteed that they form a triangle , since the mass of one neutrino may be greater than the sum of the masses of the two others .
the case where @xmath14 in table [ typicalsolutions ] is an obvious example of this situation . in the case
when @xmath15 is non - zero , treated in detail in ref .
@xcite , there is also a triangle condition but the triangle is not made simply using the masses as sides . at the present moment ,
the non - triangular " regions of @xmath0 space constitute a small , but interesting , part of the allowed range ( considering of course the type i and type ii cases separately ) .
the allowed regions are summarized in figs .
[ restrictionsi ] and [ restrictionsii ] .
so , if in the future , the neutrino masses are found from an endpoint spectrum analysis to give ( assuming suitably updated numbers ) @xmath0 outside the range predicted by the complementary ansatz , the latter can be ruled out .
we have demonstrated that , given the value of the third neutrino mass @xmath0 , the endpoint spectrum of the electron energy in beta decay is biuniquely predicted to a reasonable approximation using the experimental results on neutrino oscillations and their analyses . for definiteness some characteristic endpoint spectrum shapes were explicitly illustrated .
this seems important not only as an indication of how much our understanding of neutrinos has advanced in the last ten years but as a convenient benchmark for helping to analyze future beta decay experiments .
of course the actual experimental results require other important corrections to this `` ideal '' theoretical picture .
some novel features of this paper which may be of interest include : \1 . while the kinematics of the endpoint spectrum has been discussed many times in the literature , the exact formula for the phase space dependence on the electron energy ( obtained by performing one integration over the dalitz diagram while holding the invariant matrix element constant ) , eq .
( [ phasespace ] ) , does not seem to have been previously used .
these results may not make big changes in other analyses but one can use them with confidence since there will then be no question of kinematical ( nucleon recoil ) corrections needing to be separately taken into account .
the method of analysis of the presently allowed neutrino masses using the squared mass differences obtained from the neutrino oscillation experiments presented in section iii involves listing the results for the masses of neutrinos one and two while choosing various values of the mass @xmath0 .
the cases where @xmath0 is heaviest ( type i ) and where it is lightest ( type ii ) are treated separately .
for example , as discussed in section iii , one sees that the currently allowed neutrino mass spectrum is characterized for a large part of its range as almost degenerate rather than hierarchical .
the main point is that for any choice of @xmath0 and type , the shape of the endpoint spectrum is already pretty well known from the neutrino oscillation experiments .
even though the value of @xmath16 is only bounded , it will not affect the shape much .
our treatment easily shows that the slope discontinuities corresponding to the mass differences of individual neutrinos occur very close to each other . that would make detailed verification of the shape difficult in the near future .
however , knowing the shape for any value of @xmath0 makes it possible to integrate the final electron distribution over an energy interval @xmath17 corresponding to a given experiment .
then the number of predicted electrons could be plotted as a function of @xmath0 and @xmath17 to get an `` ideal '' estimate of the experimental sensitivity to distinguishing different neutrino mass scenarios .
this ideal estimate should of course be corrected for effects mentioned in section i , but that is beyond the scope of the present paper .
further applications of this approach will be presented elsewhere .
the complementary ansatz for the neutrino mass matrix @xcite-@xcite enables one to predict , given @xmath0 and the `` dirac '' cp phase , the two majorana phases of the lepton mixing matrix .
it is of interest because it enables the estimation of the neutrinoless double beta decay quantity @xmath18 and , in a certain model , the baryon asymmetry by the leptogenesis mechanism without making arbitrary assumptions about these phases .
an amusing feature of the ansatz is that ( to the approximation that the effect of the dirac cp phase @xmath15 is negligible ) the magnitudes of the three neutrino masses must make up a triangle .
if they do nt , the ansatz would be falsified . in section iv ,
the regions of allowed neutrino masses which falsify the ansatz were specified ( including the effect of @xmath15 ) .
it was pointed out that a large portion of the currently allowed range of @xmath0 actually does correspond to a triangular pattern of neutrino masses .
however , the small non - triangular region near @xmath19 ev is the only place where @xmath0 is several times higher than @xmath20 .
-.5 cm we are grateful to mariam tortola for checking our equations and correcting an overall factor in eq .
( [ edistrib ] ) .
s.s.m would like to thank the physics department at syracuse university for their hospitality .
the work of s.n is supported by national science foundation grant no .
the work of j.s . is supported in part by the u. s. doe under contract no .
de - fg-02 - 85er 40231 .
a symmetrical parameterization of the lepton mixing matrix , @xmath21 can be written as : @xmath22 .
\label{writeout}\ ] ] some advantages of this choice are discussed in sections v and vi of @xcite .
as written , it holds for the case of majorana type neutrinos .
there are three cp violating phases , @xmath23 and @xmath24 .
invariant " combination @xmath25 corresponds to the dirac phase " .
if neutrinos are of dirac type , only a single phase ( say @xmath24 ) may be taken to be non - zero .
kamland collaboration , k. eguchi et al , phys .
lett . * 90 * , 021802 ( 2003 ) ; sno collaboration , q. r. ahmad et al , arxiv : nucl - ex/ 0309004 ; k2k collaboration , m. h. ahn et al , phys .
lett . * 90 * , 041801 ( 2003 ) ; for recent reviews see , for examples , s. pakvasa and j. w. f. valle , special issue on neutrino , proc .
indian natl .
acad . part a * 70 * , 189 ( 2004 ) [ arxiv : hep - ph/0301061 ] and v. barger , d. marfatia and k. whisnant , int . j. mod . phys .
e * 12 * , 569 ( 2003)[arxiv : hep - ph/0308123 ] .
relevant discussions are given by r. e. shrock , phys .
b * 96*,159 ( 1980 ) ; f. vissani , arxiv : hep - ph/0102235 ; y. farzan , o. l. g. peres and a. yu .
smirnov , nucl .
b * 612 * , 59 ( 2001 ) ; j. studnik and m. zralek , arxiv : hep - ph/0110232 ; f. feruglio , a. strumia and f. vissani , nucl .
b * 637 * , 345 ( 2002 ) ; y. farzan and a. yu .
smirnov , phys .
b * 557 * , 224 ( 2003 ) ; s. m. bilenky , c. giunti , j. a. grifols and e. masso , phys . rep . * 379 * , 69 ( 2003 ) .
these are discussed in chap . 2 of f. boehm and p. vogel , physics of massive neutrinos , cambridge univ .
press ( 1987 ) .
for more recent work , see s. gardner , v. bernard and u .- g .
meissner , phys .
b * 598 * , 188 ( 2004 ) and l. platter , h .- w . hammer and u .- g .
meissner , phys .
b * 254 * ( 2005 ) . | measurement maximum electron energy in beta decay processes original approach to finding value possible neutrino mass still subject of interest may yield first direct measurement of crucial quantity interesting process of tritium decay discuss results from present neutrino oscillation experiments yield characteristic shapes for beta decay endpoint spectrum help signal identification present oscillation data predict bi - unique description shape endpoint spectrum if mass neutrino specified unknown mixing angle strongly bounded effect on endpoint spectrum shape running over possible values third neutrino mass catalog of shapes compared with experimental shape find best fit practical corrections to observed end point spectrum other phase space mixing effects considered include different final masses of @xmath2 due to final atomic electron states corrections due to binding of @xmath3 molecule resolution function of detector final state effects corrections made on each page of catalog start in section ii with exact numerical treatment of phase space relevant correction to approximate treatment section iii as @xmath0 decreases from highest value consistent with information cosmology two solutions : type i where largest and type ii smallest@xmath4 ev type ii solutions allowed characteristic neutrinos 1 and 2 close in mass splittings range ten thousandth to hundredth ev end point spectrum shapes slope discontinuities vanishing neutrino s contribution plotted illustrate features accuracy spacings neutrinos could survive corrections practical procedure integrate each corrected spectrum shape @xmath0 type over energy interval best fit experimental value section iv model restrictions on ranges @xmath0 type i and type ii solutions restrictions arise assuming complementary ansatz for neutrino mass matrix observation @xmath0 in non - allowed ranges falsification of ansatz section v summary discussion tritium decay example one massive neutrino @xmath5 be mass tritium atom @xmath6 mass @xmath2 @xmath7 electron mass @xmath8 neutrino mass @xcite kinetic recoil nucleus energy neglected approximate formula for maximum electron energy : @xmath9 exact formula situation neutrino recoil nucleus emerge same velocity is @xmath10 use @xcite input masses in mev ) : @xmath11 obtain table i. eq .decimals ] tracking digits included more digits than for experimental accuracy of @xmath5 and @xmath6.comparison of maximum electron energy for different neutrino masses @xmath12 cols="^^",options="header ansatz leads to more restrictions simplified case ref @xcite @xmath13 ansatz condition amounts to lengths to three neutrino masses forming triangle in complex plane two independent majorana phases related to internal angles triangle procedure obtains for assumed value @xmath0 three neutrino masses not guaranteed form triangle mass of one neutrino may be greater than two others case @xmath14 example @xmath15 is non - zero @xcite triangle condition triangle not made using masses as sides non - triangular regions of @xmath0 space constitute small part of allowed range type allowed regions summarized in figs . [ restrictionsi ] and [ restrictionsii ] if neutrino masses found from endpoint spectrum analysis give @xmath0 outside range predicted by complementary ansatz latter can be ruled outdemonstrated given value third neutrino mass @xmath0 endpoint spectrum of electron energy in beta decay predicted using experimental results on neutrino oscillations analyses characteristic endpoint spectrum shapes illustrated important indication understanding of neutrinos benchmark for future beta decay experiments experimental results require corrections to ideal theoretical picture novel features paper include kinematics of endpoint spectrum discussed exact formula for phase space dependence on electron energy obtained one integration over dalitz diagram invariant matrix element constant phasespace previously used results may not make big changes other analyses use with confidence no question of kinematical nucleon recoil ) corrections method of analysis of allowed neutrino masses using squared mass differences neutrino oscillation experiments section iii involves listing results for masses of neutrinos one and two choosing values of mass @xmath0 cases where @xmath0 heaviest and lightest type ii ) treated separately currently allowed neutrino mass spectrum characterized as almost degenerate hierarchical for any choice of @xmath0 type shape of endpoint spectrum known from neutrino oscillation experimentsvalue @xmath16 bounded affect shape treatment shows slope discontinuities to mass differences neutrinos occur close detailed verification shape difficult knowing shape for value @xmath0 integrate final electron distribution over energy interval @xmath17 experiment number predicted electrons plotted as function of @xmath0 and @xmath17 ideal estimate experimental sensitivity neutrino mass scenarios ideal estimate corrected for effects section i beyond scope present paper further applications presented elsewhere complementary ansatz for neutrino mass matrix @xcite-@xcite enables predict @xmath0 cp phase two majorana phases of lepton mixing matrix enables estimation neutrinoless double beta decay quantity @xmath18 baryon asymmetry by leptogenesis mechanism without arbitrary assumptions effect dirac cp phase @xmath15 magnitudes of three neutrino masses must make triangle ansatz falsified in section iv regions of allowed neutrino masses falsify ansatz specified including effect @xmath15 large portion of range of @xmath0 to triangular pattern neutrino masses small non - triangular region near @xmath19 ev only @xmath0 times higher than @xmath20grateful to mariam tortola checking equations correcting factor physics department syracuse university hospitality work s.n supported national science foundation grant j.s supported u. s. doe contract de - fg-02 - 85er 40231 symmetrical parameterization lepton mixing matrix @xmath21 written as @xmath22 advantages sections v vi @xcite holds majorana type neutrinos three cp violating phases @xmath23 @xmath24 invariant combination @xmath25 corresponds dirac phase neutrinos dirac type single phase @xmath24 non - zero kamland collaboration k. eguchi phys . lett 90 021802 2003 sno collaboration q r. ahmad - 0309004 m. h. ahn phys . 041801 2003 recent reviews see s. pakvasa j w. f. valle special issue neutrino indian natl . acad . 70 189 ( 2004 ) ph/0301061 v. barger d. marfatia k. whisnant int . j. mod . . 12 569 ( 2003 - ph/0308123 discussions by r. e.shrock phys b 96,159 1980 ) f vissani hep - ph/0102235 farzan peres smirnov nucl b 612 59 2001 ) studnik m zralek hep - ph/0110232 feruglio strumia vissani nucl b 637 345 2002 ) farzan smirnov 557 224 2003 ) bilenky giunti grifols e. masso phys . * 69 ( 2003 ) discussed chap 2 f. boehm p vogel physics massive neutrinos cambridge univ press 1987 ) recent work s. gardner v. bernard meissner phys . b 598 * 188 ( 2004 ) l. platter hammer meissner 254 * 2005 ) | 0.439266 | 2,878 | 1,454 | , Saving $0.1 in GPT-4. |
55,103 | suppose @xmath0 is the space of all absolutely 1-summing operators between two banach spaces @xmath1 and @xmath2 . we show that if @xmath2 has a copy of @xmath3 , then @xmath0 will have a copy of @xmath3 , and under some conditions
if @xmath1 has a copy of @xmath4 then @xmath5 would have a complemented copy of @xmath4 .
= msam10 at 10pt = tibi at 13.5pt c = rmtmib at 13.5pt | many studies on copy of @xmath3 or @xmath4 in spaces of bounded operators , weakly compact operators , and compact operators have been made @xcite . here
we intend to prove similar results in the space of absolutely 1-summing operators .
absolutely 1-summing operators were introduced and studied by grothendick in his famous rsume @xcite . for two banach spaces @xmath1 and @xmath2 , the operator @xmath6 is called absolutely 1-summing operator , if given any sequence @xmath7 of @xmath1 for which @xmath8 , for each @xmath9 , we have @xmath10 is the banach space of all absolutely 1-summing operators from @xmath1 to @xmath2 , endowed with @xmath11-norm as follows : @xmath12 & \leq \left . \rho
\ { \sup}_{x^ { * } \in b^{*}_{e } } \ \left(\sum\limits_{i = 1}^{n}|x^{*}(x_{i})|\right ) ; \ x_{i}\in e , 1 \leq i \leq n\right\rbrace.\end{aligned}\ ] ] by grothendick
pietsch dominated theorem @xcite , for @xmath13 there exists a regular probability measure @xmath14 defined on @xmath15 ( with its weak@xmath16 topology ) for which @xmath17 holds for each @xmath18 .
therefore , @xmath19 .
the following lemma discusses the main results of our study .
-.3pc @xcite .
let @xmath20 and @xmath21 be , in @xmath15 and @xmath22 , the unit balls of two banach spaces @xmath23 and @xmath2 respectively
. then @xmath24 and @xmath25.@xmath26
here , we present a result showing complemented copy of @xmath4 in @xmath0 .
suppose @xmath27 is a sequence in @xmath15 such that @xmath28 and @xmath29 for each @xmath30 , where @xmath7 is a weakly unconditionally cauchy sequence in @xmath1 and @xmath31 ( for example @xmath32 .
then @xmath0 has a complemented copy of @xmath4 if @xmath2 has a copy of @xmath4 . from the assumption
there is an @xmath4-basic sequence @xmath33 in @xmath2 and so @xmath34 such that for each @xmath35 we have @xmath36 therefore , @xmath37 & \leq \sum\limits_{n = 1}^{m } |a_{n}| \vert x_{n}^ { * } \vert \cdot \left\vert y_{n } \right\vert \leq m_{2 } \sum\limits_{n = 1}^{m } |a_{n}|.\end{aligned}\ ] ] on the other hand , @xmath38 & = { \sup}_{x^ { * * } \in b_{e^ { * * } } } \left\vert \sum\limits_{n = 1}^{m } a_{n } x^ { * * } ( x_{n}^ { * } ) y_{n}\right\vert\\[.2pc ] & \geq m_{1 } \sum\limits_{n = 1}^{m } |a_{n}| \cdot |x_{0}^ { * * } x_{n}^{*}| \geq cm_{1 } \sum\limits_{n = 1}^{m } |a_{n}|.\end{aligned}\ ] ] this shows that @xmath0 contains @xmath39 as a copy of @xmath4 .
we show this copy is complemented in @xmath40 .
define @xmath41 $ ] by @xmath42 , where @xmath43 $ ] is the closed linear span of @xmath44 in @xmath0 .
since @xmath7 is @xmath45 and @xmath46 is the absolutely 1-summing operator , @xmath47 is a well - defined linear map . from the closed graph theorem
one can easily show that @xmath47 is a projection from @xmath48 onto @xmath43 $ ] which completes the proof .
@xmath49 suppose for @xmath52 in @xmath2 there is a positive null sequence @xmath53 such that @xmath54 consider @xmath55 and @xmath56 in the unit sphere of @xmath23 and @xmath1 respectively such that @xmath57 .
then @xmath58 which completes the proof .
@xmath49 suppose @xmath59 contains a sequence @xmath60 equivalent to the standard unit vector basis of @xmath3 such that for @xmath61 ( respectively @xmath62 ( respectively @xmath63 ) are basic sequences .
then @xmath64 or @xmath65 has a copy of @xmath3 .
@xmath34 such that for any @xmath75 , @xmath76 but it is easy to see that @xmath74 is weakly null . from the assumption @xmath77
, it follows from bessaga
peczynski theorem @xcite that it would be a basic sequence with coefficient functionals @xmath67 such that @xmath78 .
we would have , @xmath79 & \leq ( 1/c ) \left\vert \sum\limits_{n = 1}^{\infty } a_{m } t_{m } \right\vert\\[.3pc ] & \leq ( 1/c ) \pi_{i } \left(\sum_{n = 1}^{\infty } a_{m } t_{m } \right)\\[.3pc ] & \leq m - ( 2/c)\ \hbox{sup}_{n } |a_{n}|,\end{aligned}\ ] ] which shows that @xmath80 is a @xmath3-basic sequence in @xmath2.@xmath49 similar to the proof of theorem 2 we may assume @xmath33 is a complemented copy of @xmath3 in @xmath2 and @xmath81 the coefficient functional of it in @xmath82 $ ] the closed linear span of @xmath81 . since @xmath82 $ ] is complemented in @xmath2 , we can assume @xmath67 is a weak@xmath16 null convergence sequence in @xmath83 .
define @xmath84 $ ] by @xmath85 , where @xmath20 is an arbitrary element in the unit sphere of @xmath23 and @xmath86 $ ] is the closed linear subspace generated by @xmath87 in @xmath40 . for any @xmath88 in @xmath1 , we have @xmath89 & \leq \sum\limits_{i = 1}^{n } c_{2 } \hbox{sup}_{n}| ( y_{n}^ { * } t(x_{i}))|\\[.2pc ] & = c_{2 } \sum\limits_{i = 1}^{n } \| t(x_{i})\| \leq c_{2 } \pi_{1}(t ) \sum\limits_{i = 1}^{n } |x^{*}x_{i}|.\end{aligned}\ ] ] | studies on @xmath3 or @xmath4 in bounded weakly compact made @xcite prove similar results in 1-summing operators introduced studied by grothendick in @xcite two spaces @xmath1 and @xmath2 operator @xmath6 absolutely 1-summing operator sequence @xmath7 of @xmath1 @xmath8 @xmath9 @xmath10 is banach space of 1-summing operators from @xmath1 to @xmath2 endowed with @xmath11-norm @xmath12 \leq 1 by grothendick theorem @xcite for @xmath13 regular probability measure @xmath14 defined on @xmath15@xmath16 topology for @xmath17 holds for each @xmath18 @xmath19 lemma discusses main results study @xmath20 and @xmath21 be unit balls of two banach spaces @xmath23 and @xmath2 @xmath24 and @xmath25.@xmath26 present result showing complemented copy of @xmath4 in @xmath0 @xmath27 sequence in @xmath15 @xmath28 @xmath29 each @xmath30 @xmath7 weakly cauchy sequence in @xmath1 @xmath31 @xmath32 @xmath0 complemented copy @xmath4 @xmath2 copy @xmath4 @xmath4-basic sequence @xmath33 @xmath2 @xmath34 each @xmath35 @xmath36 @xmath37 \sum\limits_{n = 1}^{m }|a_{n} x_{n}^\vert y_{n_{2\limits_{n = 1}^{m }_{n}{aligned} @xmath38 }_{x^ { * * }\vert\sum\limits_{n = 1}^{m }_{n_{n}_{n}\vert\\ m_{1\sum\limits_{n = 1}^{m }|a_{n}_{0}^ { * *_{n}^{*}\limits_{n = 1}^{m }_{n}{aligned @xmath0 contains @xmath39 copy of @xmath4 copy complemented in @xmath40define @xmath41 by @xmath42 @xmath43 closed linear span of @xmath44 @xmath0 @xmath7 @xmath45 @xmath46 1-summing operator @xmath47 - defined linear map closed graph theorem @xmath47 projection from @xmath48 @xmath43 completes proof @xmath49 @xmath52 @xmath2 positive null sequence @xmath53 @xmath54 @xmath55 @xmath56 unit sphere of @xmath23 @xmath1 @xmath57 @xmath58 completes proof @xmath49 @xmath59 sequence @xmath60 equivalent standard unit vector basis @xmath3 @xmath61 @xmath62 @xmath63 basic sequences @xmath64 @xmath65 copy of @xmath3 @xmath34 @xmath75 @xmath76 @xmath74 weakly null assumption @xmath77 peczynski theorem basic sequence with coefficient functionals @xmath67 @xmath78 @xmath79 \leq 1/c\left\vert \sum\limits_{n = 1}^{\infty t__{n 13pc \leq 2/c)\hbox{sup}_{n}{aligned @xmath80 @xmath3-basic sequence @xmath2. proof theorem 2 @xmath33 complemented copy @xmath3 @xmath2 @xmath81 coefficient functional @xmath82 closed linear span @xmath81 @xmath82 complemented @xmath2 assume @xmath67 weak@xmath16 null convergence sequence @xmath83 define @xmath84 @xmath85 @xmath20 arbitrary element unit sphere @xmath23 @xmath86 closed linear subspace @xmath87 @xmath40 @xmath88 @xmath1 @xmath89\leq \sum\limits_{i = 1}^{n c_{2 }{sup}_{n}(x_{i} c_{2\sum\limits_{i = 1}^{n(x_{i c_{2{1}\limits{i 1}^{n{_{i}{aligned} | 0.262202 | 1,643 | 934 | , Saving $0.0 in GPT-4. |
55,111 | we present a method to confirm the planetary nature of objects in systems with multiple transiting exoplanet candidates .
this method involves a fourier - domain analysis of the deviations in the transit times from a constant period that result from dynamical interactions within the system .
the combination of observed anti - correlations in the transit times and mass constraints from dynamical stability allow us to claim the discovery of four planetary systems kepler25 , kepler26 , kepler27 , and kepler28 , containing eight planets and one additional planet candidate .
[ firstpage ] celestial mechanics ; stars : individual ( kic 4349452 , kic 9757613 , kic 5792202 , kic 6949607 ) ; methods : data analysis ; techniques : photometric | to date , nasa s _ kepler _ mission has announced the discovery of nearly a thousand candidate exoplanet systems including a significant number of systems with multiple transiting objects @xcite .
such a large sample of exoplanet candidates can provide valuable insight into the nature of the general population of planets and planetary systems , even if there were a sizeable fraction that are false - positive signals .
nevertheless , a sample of systems that are known with high confidence to be planetary enables a significant amount of additional science both for individual objects and for the planet population generally .
moreover , it yields more efficient use of resources in these endeavors by indicating those systems where supplementary observations will likely be most fruitful .
historically , planetary systems have been confirmed primarily by making dynamical measurements of the planet mass and orbital properties via radial velocity ( rv ) measurements .
more recently @xcite , this mass measurement has been accomplished by detailed modeling of transit timing variations , or ttvs the deviations from a constant period that result from gravitational perturbations among multiple planets and their host star @xcite .
another contemporary development is the validation of planets by the process of elimination excluding false positive signals by scrutinizing a wide variety of data on individual systems ( e.g. , ` blender ` @xcite ) .
yet detailed dynamical modeling of systems , the gathering of rv and other complementary data , and the synthesis of this information is time consuming for individual systems and is prohibitive for the entirety of the list of _ kepler _ objects of interest ( kois ) .
the development of simpler , and more rapid , validation techniques is crucial for the timely advancement of the exoplanetary science enabled by _
kepler_. in this work , we present a method to confirm the planetary nature of candidates in multiple transiting systems .
complementary methods with the same goal are simultaneously being developed by @xcite , @xcite , and @xcite .
these methods are specifically intended to confirm planetary systems using a small number of assumptions and easily implemented analysis .
the method presented here broadly uses conservation of energy and dynamical stability to show that alternative ( non - planetary ) explanations of the observed photometric data are excluded with high confidence and that the objects must be both dynamically interacting and planetary in mass .
specifically , the properties of the ttv signature for the multiple candidates and the maximum allowed masses for objects in the systems demonstrate that exoplanet candidates in several _ kepler _ systems are indeed planets . generally speaking , it is difficult to construct astrophysical false positives that can mimic a multi - transiting system @xcite .
yet , the analysis presented herein does not invoke this enhancement in planet likelihood , and we confirm the discovery of 8 total planets in the systems kepler25 ( koi 244 , kic 4349452 ) , kepler26 ( koi 250 , kic 9757613 ) , kepler27 ( koi 841 , kic 5792202 ) , and kepler28 ( koi 870 , kic 6949607 ) .
this paper will proceed as follows .
first , we discuss the essential components of the photometric data reduction and the stellar properties ( [ photometry ] ) . the analysis method is discussed in [ method ] .
the results of the analysis for the various systems is found in [ results ] .
we discuss the implications of these results in [ discussion ] .
finally , additional obervational information and photometric diagnostics for some of the systems are given in [ extradata ] .
the kepler mission was designed to detect terrestrial - size planets in the habitable zone of the host star , necessitating both a large sample size and sensitivity to a much larger range of orbital distances than ground - based surveys @xcite .
the instrument is a differential photometer with a wide ( @xmath0 square degrees ) field - of - view ( fov ) that continuously and simultaneously monitors the brightness of approximately 150,000 main - sequence stars .
a discussion of the characteristics and on - orbit performance of the instrument and spacecraft is presented in @xcite .
each of the systems considered here was found using the transiting planet search pipeline ( tps ) which identifies significant transit - like features , or threshold crossing events ( tce ) , in the kepler light curves @xcite .
data showing tces are then passed to the data validation ( dv ) pipeline @xcite .
the dv pipeline fits a transiting planet model to the data , removes it from the light curve , and returns the result to tps in an effort to find additional transit features .
dv also completes a suite of statistical tests that are applied to the data after all tces are identified in an effort to assess the likelihood of false - positives .
the results of these diagnostic tests are consistent with the planet interpretation and do not warn of potential pitfalls .
after pipeline data processing and the photometry extraction , the time series is detrended as described in @xcite in order to measure important stellar and planet candidate parameters . a first estimate for the stellar mass and radius ( @xmath1 and @xmath2 )
is obtained by comparing the stellar @xmath3 and @xmath4 values derived from an analysis of the stellar spectrum or from the kepler input catalog ( kic , * ? ? ?
* ) to a set of cesam ( code devolution stellaire adaptatif et modulaire , @xcite ) stellar evolution models computed in steps of @xmath5 for solar composition . with @xmath1 and
@xmath2 fixed to their initial values , a transit fit is then computed to determine the orbital inclination , planetary radius , and depth of the occultation ( passing behind the star ) assuming a circular orbit .
the transit lightcurve is modeled using the analytic expressions of @xcite using non - linear limb darkening parameters derived for the kepler bandpass @xcite .
the best fit model is found using a levenberg - marquardt minimization algorithm @xcite .
the best fit transit model is then removed from the lightcurve and the residuals are used to fit for the characteristics of next transiting candidate identified by tps .
the transit times obtained from these model fits are used in our analysis below and a brief excerpt is given in table [ tabtts ] . since these systems show visible ttvs we display their light curves in figures [ lc244 ] , [ lc250 ] , [ lc841 ] , and [ lc870 ] , where data are prepared ( with ttvs corrected ) as described in @xcite .
lcccc koi & n & @xmath6 & ttv@xmath7 & @xmath8 + & & bjd-2454900 & ( d ) & ( d ) + 244.01 & + 244.01 & 0 & 86.0821 & -0.0050 & 0.0029 + 244.01 & 2 & 111.5291 & 0.0014 & 0.0008 + 244.01 & 3 & 124.2463 & -0.0019 & 0.0011 + 244.01 & 4 & 136.9696 & 0.0011 & 0.0010 + 244.01 & 5 & 149.6898 & 0.0010 & 0.0008 + 244.01 & 6 & 162.4084 & -0.0008 & 0.0009 + 244.01 & 7 & 175.1296 & 0.0000 & 0.0008 + 244.01 & 8 & 187.8486 & -0.0013 & 0.0011 + 244.01 & 9 & 200.5706 & 0.0003 & 0.0024 + 244.01 & 10 & 213.2911 & 0.0005 & 0.0005
+ + a centroid measurement , which gives the displacement of the flux centroid during transit compared with the flux centroid in data taken out of transit , is calculated for each candidate planet .
the proper interpretation of the value derived for the centroid motion requires knowledge of the positions and brightnesses of the stars near the target .
motion can result in two different ways ; a nearby star with the transit signature or a nearby star ( or stars ) that dilute the light from the target star , which has the transit signal . an analysis of the centroid motion can give important information in identifying the host of the transiting object and can help disentangle complicated , projected systems blended within the target aperture .
one target star kepler25 is very bright ( kp = 10.7 ) and saturates the central pixels of its image on the photometer .
thus , the centroid information can only identify displacements of the size of a pixel and is of limited use .
the centroid of a second target , kepler27 , does displace significantly from the nominal location during transit .
the dimness of kepler27 ( kp = 15.9 ) presents a challenge to the centroid analysis .
a higher - resolution image of the target taken in the j band with the united kingdom infrared telescope ( ukirt ) shows a faint , neighboring star ( kp = 19.5 ) that is 2 arcseconds to the northeast of the target and that is not identified in the kic ( see figure [ ukirt841 ] ) .
the centroid motion induced by this faint star is consistent with the interpretation that the faint star has constant brightness and the transit event arises from the target .
the centroid analysis of the remainder of the eight planets shows no displacement beyond the 3@xmath9 radius of confusion for each target and therefore do not raise concern . when modeling the planet and stellar properties for each of these systems , additional light from neighboring stars has been accounted for .
the _ kepler _ follow - up observing program ( fop ) has obtained spectra of the host stars for three of the four systems ( kepler25 , kepler26 , and kepler27 ) .
the spectra were inspected for a prominent second set of spectral lines which would indicate the presence of an additional star .
for host stars with modest snr follow - up spectroscopy ( roughly 20 per resolution element from the 1.5 m tillinghast reflector at flwo ) , we fit spectra to a library of theoretical spectra using the tools of @xcite . for host stars with high - snr follow - up spectroscopy from keck ,
we also report stellar parameters derived from a spectroscopy made easy ( sme ) analysis @xcite with mass and radius estimates generated from yonsei - yale isochrones . in cases without follow - up spectroscopy
, we adopt the stellar parameters from the kic @xcite .
the adopted stellar parameters are given in table [ tabstars ] along with the estimate of contamination from nearby stars derived from the kic . | nasa s _ kepler mission announced discovery of nearly thousand candidate exoplanet systems including systems with multiple transiting objects @xcite large sample exoplanet candidates insight into nature planets planetary systems even false - positive signals sample of systems known with high confidence to be planetary enables additional science for individual objects planet population yields efficient use of resources indicating systems where supplementary observations fruitful planetary systems confirmed by dynamical measurements planet mass orbital properties via radial velocity rv ) measurements recently @xcite mass measurement accomplished by detailed modeling of transit timing variations deviations gravitational perturbations planets contemporary development validation of planets by elimination excluding false positive signals by scrutinizing variety of data on systems detailed dynamical modeling synthesis time consuming prohibitive for kepler objects interest development of simpler rapid validation techniques crucial for advancement exoplanetary science work present method to confirm planetary nature of candidates in multiple transiting systems complementary methods developed by @xcite , @xcite methods intended confirm planetary systems using small assumptions easily implemented analysismethod uses conservation energy dynamical stability show alternative non - planetary explanations photometric data objects must be dynamically interacting planetary in mass ttv signature for multiple candidates maximum allowed masses demonstrate exoplanet candidates in several kepler systems are planets difficult to construct astrophysical false positives mimic multi - transiting system @xcite analysis invoke enhancement planet likelihood discovery of 8 planets in systems kepler25 244 4349452 kepler26 250 kepler27 841 5792202 kepler28 870 6949607 paper discuss components photometric data reduction stellar properties analysis method in [ method ] results analysis in [ results ] implications in [ discussion ] additional obervational information photometric diagnostics for in [ extradata ] kepler mission designed to detect terrestrial - size planets in habitable zone host star large sample size sensitivity to larger range orbital distances than surveys @xcite instrument is differential photometer with wide field - of - view monitors brightness of 150,000 main - sequence stars discussion of characteristics on - orbit performance of instrument spacecraft in @xcite systems considered found using transiting planet search pipeline identifies significant transit - like features in kepler light curves @xcite data showing tces passed to data validation dv pipeline @xcite dv pipeline fits transiting planet model removes from light curve returns result to tps additional transit features dv completes statistical tests after tces identified false - positives results diagnostic tests consistent with planet interpretation warn potential pitfalls after pipeline data processing photometry extraction time series detrended in @xcite to measure stellar planet candidate parameters first estimate for stellar mass radius obtained by comparing stellar @xmath3 @xmath4 values to cesam stellar evolution models computed in @xmath5 for solar composition @xmath1 fixed to values transit fit computed to orbital inclination planetary radius depth of occultation circular orbit transit lightcurve modeled using analytic expressions @xcite non - linear limb darkening parameters best fit model found using levenberg - marquardt minimization algorithm best fit transit model removed from lightcurve residuals used to fit for characteristics next transiting candidate identified by tpstransit times from model fits used in analysis excerpt in table [ tabtts systems show ttvs display light curves in figures lc244 lc250 lc841 lc870 ] data prepared ttvs corrected in @xcite @xmath6@xmath7 @xmath8 bjd-2454900 + 244.01 244.01 0 86.0821 -0.0050 0.0029 + 244.01 2 111.5291 0.0014 0.0008 + 244.01 3 124.2463 -0.0019 0.0011 + 244.01 4 136.9696 + 244.01 5 149.6898 0.0008 + 244.01 6 162.4084 -0.0008 0.0009 + 244.01 & 7 175.1296 0.0000 0.0008 + 244.01 8 187.8486 -0.0013 0.0011 + 244.01 9 200.5706 0.0003 0.0024 + 244.01 & 10 213.2911 0.0005 centroid measurement displacement flux centroid during transit calculated for each candidate planet interpretation requires knowledge positions brightnesses stars near target motion can nearby star with transit signature or light from target transit signal analysis of centroid motion host transiting object disentangle complicated systems target aperturetarget star kepler25 bright kp = 10.7 ) saturates central pixels photometer centroid information displacements pixel limited use centroid second target kepler27 from location during transit dimness kepler27 kp = 15.9 ) centroid analysis higher - resolution image telescope shows faint neighboring star ( kp = 19.5 ) 2 arcseconds northeast target not identified in kic centroid motion induced by faint star consistent with interpretation constant brightness transit event arises from target centroid analysis of eight planets shows no displacement beyond 3@xmath9 radius of confusion concern planet stellar properties additional light from neighboring stars accounted for kepler follow - up observing program obtained spectra of host stars for three systems kepler25 kepler26 kepler27 spectra inspected for prominent second set spectral lines presence additional star for host stars with modest snr follow - up spectroscopy fit spectra to theoretical spectra using tools @xcite for stars high - snr follow - up spectroscopy report stellar parameters from spectroscopy analysis @xcite with mass radius estimates from yonsei - yale isochronescases without spectroscopy adopt stellar parameters from kic @xcite adopted parameters given in table estimate contamination nearby stars kic | 0.357146 | 2,306 | 1,172 | , Saving $0.1 in GPT-4. |
55,112 | we present an exact procedure that allows one to calculate the scattering length for any potential expressed as an algebraic sum of inverse powers of the inter - atomic distance .
we apply it to @xmath0 lennard - jones potentials , with different values of @xmath1 .
the procedure is suitable for a very precise determination of the intensities of the potential corresponding to zeros or poles of the scattering length . | in the study of ultracold dilute gases , the effect of atom - atom collisions is , in most cases , described by means of a regularized zero - range pseudopotential whose coupling constant is related to the scattering length @xmath2 of the two - atom system @xcite . due to this crucial role ,
several authors have been concerned to calculate the scattering length for interatomic potentials .
numerical methods are unavoidable in the case of general realistic interactions .
there are , however , particular forms of potentials whose scattering length is susceptible of being computed algebraically .
examples are the hard - core potential with a @xmath3 tail @xcite , and the potentials discussed by pade @xcite .
these , besides of being analytical at any distance , are able to reproduce the general trend of experimental data . motivated by the mentioned papers of pade , we present here an exact method of computing the scattering length applicable to any sum of inverse power potentials . in particular , to illustrate the method , we consider here a ( 12 , s ) lennard - jones type potential @xmath4 where @xmath5 is the reduced mass of the two interacting atoms , @xmath6 represents the classical turning point , and @xmath7 is a dimensionless parameter giving the intensity of the interaction . for the exponent @xmath1 of the long range part of the potential we have considered three different values : @xmath8 , as in the familiar van der waals potential , @xmath9 , as in the case of atom - ion collisions @xcite , and @xmath10 , as in the retarded atom - atom potential @xcite .
this last case is exactly solvable , as shown by pade ( * ? ? ?
* eq . ( 5 ) with @xmath11 ) .
nevertheless , as a test , we apply our method to solve it .
our procedure , quite unconventional , has been successfully applied to the solution of the schrdinger equation with a polynomial potential @xcite .
( in fact , the transformation @xmath12 converts the problems at hand into different anharmonic oscillators . )
the reduced radial schrdinger equation with the potential ( [ uno ] ) , in the case of energy and angular momentum equal to zero , reads @xmath13 this differential equation presents two singular points : a regular one at infinity , of indices @xmath14 and @xmath15 , and an irregular one at the origin .
two independent solutions can be immediately written as power series of @xmath16 , @xmath17 with @xmath18 and coefficients @xmath19 given by the recurrence relation @xmath20 those series expansions are convergent for any value of @xmath21 , except for @xmath22 . the general solution of eq .
( [ dos ] ) is but a linear combination of the two basic solutions @xmath23 .
let us suppose that the ( unnormalized ) physical solution is given by @xmath24 where @xmath25 and @xmath26 are constants , to be determined , such that @xmath27 becomes regular at the origin .
then , for large @xmath21 , @xmath28 and , consequently , the scattering length would be given by @xmath29 the determination of @xmath25 and @xmath26 requires the knowledge of the behaviour of @xmath30 and @xmath31 at the origin .
this is what is known as the _ connection problem _ for the two singular points of the differential equation ( [ dos ] ) . in what follows ,
we present our solution of that problem .
let us simplify the notation by using , instead of @xmath21 , the dimensionless variable @xmath32 in terms of the new variable , the schrdinger equation becomes @xmath33 with @xmath34 the basic solutions given in eq .
( [ tres ] ) are , in terms of @xmath35 , @xmath36 in order to represent the behaviour of these functions at the origin , it is convenient to consider the two ( formal ) thom solutions at that singular point , @xmath37 the label @xmath38 being to be replaced by @xmath39 or @xmath40 , according to the behaviour , regular or irregular , of these formal solutions at the origin , which is determined by the sign of @xmath41 . substitution of these formal expressions in eq .
( [ nueve ] ) gives the condition @xmath42 , and consequently @xmath43 the exponent @xmath44 and the recurrence relation for the coefficients @xmath45 @xmath46 in the cases of @xmath9 or @xmath47 , or @xmath48 if @xmath10 .
( notice that in this case the recurrence relation reduces to a linear difference equation of first order and , therefore , it is exactly solvable . ) then , the behaviour of the basic solutions at the origin can be expressed in the form @xmath49 the coefficients @xmath50 and @xmath51 are called _
connection factors_. their determination solves the connection problem .
it allows also one to obtain immediately the scattering length , as we are going to show .
substitution of the last expressions of @xmath52 in @xmath53 gives , for @xmath54 , @xmath55 since the physical solution has to be regular at the origin , it must be @xmath56 giving for the scattering length @xmath57 for the computation of the connection factors , we have found convenient to express them as quotients of ( constant ) wronskians of the concerned solutions . taking the wronskian of both sides of eq .
( [ dsiete ] ) with @xmath58 and isolating @xmath59 one obtains @xmath60/\mathcal{w}\left[w_{\rm{irr}},w_{\rm{reg}}\right ] , \label{vdos}\ ] ] that , substituted in eq .
( [ vuno ] ) , gives the scattering length as a quotient of wronskians , @xmath61/\mathcal{w}\left[w_2,w_{\rm{reg}}\right]\right ) \,r_{0}\ , , \label{vtres}\ ] ] a direct computation of the values of these wronskians written in the form @xmath62 = w_j(z)\,\frac{dw_{\rm{reg}}(z)}{dz } - w_{\rm{reg}}(z)\,\frac{dw_j ( z)}{dz}\ ] ] is to be discarded due to the fact that the expansion in the definition , eq .
( [ ddos ] ) , of @xmath63 is an asymptotic one and , in general , does not converge .
our indirect way of evaluating that kind of wronskians was explained in refs . @xcite and
@xcite , where potentials sum of positive powers of the distance were considered . for convenience of the reader , we apply , in the next section , the method to the present case .
let us introduce the auxiliary functions @xmath64 obviously , @xmath65 = \exp(-\beta_{\rm{reg}}\,z^{-5}/5)\ , \mathcal{w}\left[w_j , w_{\rm{reg}}\right ] \label{vseis}\ ] ] the left hand side of this equation can be written formally , by direct computation , as a doubly infinite sum of powers of @xmath35 , @xmath65 \sim \sum_{p=-\infty}^\infty \gamma_{p , j}\,z^{p+\nu_j+\mu } , \label{vsiete}\ ] ] with the notation @xmath66 the value of the @xmath67 $ ] can be immediately obtained if we are able to write an expansion of @xmath68 , in the right hand side of ( [ vseis ] ) , with the same powers of @xmath35 as the expansion ( [ vsiete ] ) of the left hand side . with this purpose , we construct , for each value of @xmath69 , five formal expansions @xmath70 of @xmath68 .
such expansions are but particular forms of the so called heaviside s exponential series @xcite @xmath71 for integer @xmath72 , it reduces to the familiar series expansion of the exponential function . for any other value of @xmath72 , it is an asymptotic expansion whenever @xmath73 .
it becomes evident that , for any set of constants @xmath74 ( @xmath75 ) satisfying the restriction @xmath76 , \label{tuno}\ ] ] one has from ( [ vseis ] ) @xmath77\sim\sum_{k=0}^{4 } c_{k , j}\,\mathcal{e}_{k , j}(z ) .
\label{tdos}\ ] ] by choosing for the @xmath78 in the expansions @xmath79 the values @xmath80 a comparison of the resulting expansion in ( [ tdos ] ) with that in ( [ vsiete ] ) can be done , term by term .
one obtains in this way @xmath81 for any positive integer @xmath82 . by substituting in ( [ tuno ] ) the values of @xmath83 obtained from ( [ tcuatro ] ) one
has finally @xmath84 = \sum_{k=0}^4\frac{\gamma(n+1+\delta_{k , j})}{(\sqrt{\lambda}/5)^{n+\delta_{k , j } } } \,\gamma_{-5n - k , j } , \label{tcinco}\ ] ] with @xmath78 and @xmath85 given , respectively , by ( [ ttres ] ) and ( [ vocho ] ) . the convergence of series of the type of that appearing in the right - hand side of eq .
( [ vocho ] ) was discussed in ref .
it is rather slow , namely , like that of the series @xmath86 .
consequently , a sufficiently precise arithmetic should be used .
the integer @xmath82 in the right hand side of eq .
( [ tcinco ] ) can be chosen arbitrarily with the only restriction @xmath87 .
in fact , a test of the procedure is the necessary coincidence of the results obtained for different choices of @xmath82 .
we show in fig .
1 the behavior of the values of @xmath88 obtained with our method , for the particular case of @xmath8 , as the intensity @xmath7 of the potential goes from 0 to 2500 . notice that we have represented , on the horizontal axis , the values of @xmath89 instead of those of @xmath7 . in his study of solvable potentials of the lennard - jones type with exponents @xmath90 , pade @xcite
has shown that the difference of the values of the square root of the intensity corresponding to two consecutive zeros or poles of @xmath2 is a constant , namely @xmath91 . in the case of figure 1 ,
such difference is not exactly , but nearly a constant .
similar graphics are obtained for other values of @xmath1 in the potential ( [ uno ] ) .
nevertheless , as far as the sign of the scattering length determines the attractive or repulsive nature of the zero - range pseudopotential and , consequently , the possibility of formation of bose - einstein condensates , the precise location of the zeros and poles of @xmath88 is crucial .
moreover , as it is well known , a new bound state arises whenever the increasing intensity of the potential coincides with one of the poles .
our procedure allows to locate zeros and poles , by requiring cancelation of the wronskians @xmath92 $ ] and @xmath93 $ ] given by eq .
( [ tcinco ] ) , with a precision limited only by the number of digits carried in the computation .
we report , in tables 1 and 2 , a list of the first ten zeros and poles of the scattering length for the potential ( [ uno ] ) with @xmath94 and @xmath95 . the approximate ( exact in the case @xmath10 )
constancy of the difference of consecutive zeros or poles is manifest .
rrr @xmath9 & @xmath8 & @xmath10 + 1.135708 & 2.944907 & 4 + 4.281230 & 10.307414 & 14 + 7.627058 & 17.758560 & 24 + 10.991652 & 25.220363 & 34 + 14.361060 & 32.685259 & 44 + 17.732554 & 40.151469 & 54 + 21.105133 & 47.618360 & 64 + 24.478348 & 55.085650 & 74 + 27.851968 & 62.553194 &
84 + 31.225862 & 70.020910 & 94 + rrr @xmath9 & @xmath8 & @xmath10 + 2.650141 & 4.728696 & 6 + 5.949138 & 12.165518 & 16 + 9.308435 & 19.622908 & 26 + 12.675992 & 27.086171 & 36 + 16.046629 & 34.551611 & 46 + 19.418744 & 42.018080 & 56 + 22.791679 & 49.485114 & 66 + 26.165118 & 56.952491 & 76 + 29.538887 & 64.420092 & 86 + 32.912885 & 71.887847 & 96 + in order to facilitate the comparison of the values of the scattering length for different exponents @xmath1 , we give in figure 2 a synoptic representation , for @xmath94 and @xmath95 , of a variable related to the scattering length , namely @xmath96 , _ vs. _
such a variable , that presents the advantage of being free of the singularities of @xmath2 , was already used by ouerdane _
@xcite to solve numerically the riccati equation obeyed by the scattering length , in the variable phase theory @xcite , when the potential is truncated at a large distance . as a consequence of the pade s
study @xcite and of ours , it seems plausible to formulate the following conjecture : given a potential @xmath97 of the lennard - jones type with exponents @xmath98 , let us denote by @xmath99 and @xmath100 ( @xmath101 ) the values of the intensity corresponding respectively to zeros and poles of the scattering length .
then , one can write the quasi - linear laws @xmath102 where @xmath103 and @xmath104 are nearly independent of @xmath82 . in the particular case @xmath105 , pade @xcite obtained @xmath106
the dependence of @xmath107 on its arguments is such that , keeping @xmath108 fixed , @xmath109 diminishes as @xmath110 increases and , with fixed @xmath110 , @xmath109 increases with @xmath108 .
the recommendations of two anonymous referees have been decisive to improve the presentation of this article .
the authors acknowledge financial support of departamento de ciencia , tecnologa y universidad del gobierno de aragn and fondo social europeo ( project e24/1 ) and of ministerio de ciencia e innovacin ( project mtm2009 - 11154 ) . | study ultracold dilute gases effect of atom - atom collisions described regularized zero - range pseudopotential coupling related to scattering length @xmath2 of two - atom system @xcite authors calculate scattering length for interatomic potentials numerical methods unavoidable realistic interactions particular forms potentials scattering length computed algebraically examples hard - core potential with @xmath3 tail @xcite potentials discussed by pade @xcite analytical reproduce general trend of experimental data motivated present exact method of computing scattering length to inverse power potentials consider lennard - jones type potential @xmath4 where @xmath5 is reduced mass of two interacting atoms @xmath6 classical turning point @xmath7 dimensionless parameter intensity of interaction for exponent @xmath1 of long range part potential considered three values : @xmath8 @xmath9 - ion @xmath10 retarded atom - atom potential @xcite last case solvable shown by pade with @xmath11 test apply our method to solve procedure applied to solution schrdinger equation with polynomial potential @xcite transformation @xmath12 converts problems into anharmonic oscillators reduced radial schrdinger equation potential ( [ uno energy angular momentum to zero reads @xmath13 presents two points regular at infinity irregular at origin solutions written as power series of @xmath16 @xmath17 @xmath18 coefficients @xmath19 recurrence relation @xmath20 series expansions convergent for value @xmath21 except @xmath22 solution of eq . [ dos linear combination of two solutions @xmath23 unnormalized physical solution by @xmath24 @xmath25 @xmath26 constants @xmath27 becomes regular at origin large @xmath21 @xmath28 scattering length by @xmath29 determination of @xmath25 @xmath26 requires knowledge behaviour of @xmath30 @xmath31 at origin connection problem for two singular points differential equation solution simplify notation @xmath21 dimensionless variable @xmath32 schrdinger equation becomes @xmath33 with @xmath34 basic solutions in eq .[ tres ] ) @xmath35 @xmath36 represent behaviour functions at origin consider two formal solutions @xmath37 @xmath38 replaced by @xmath39 or @xmath40 according behaviour regular or irregular origin determined by sign @xmath41 substitution formal expressions in eq . gives condition @xmath42 @xmath43 exponent @xmath44 recurrence relation for coefficients @xmath45 @xmath46 @xmath9 @xmath47 @xmath48 @xmath10 recurrence relation reduces to linear difference equation solvable behaviour basic solutions origin expressed form @xmath49 coefficients @xmath50 @xmath51 _ connection factors_ determination solves connection problem obtain scattering length substitution of last expressions @xmath52 in @xmath53 gives @xmath54 @xmath55 physical solution regular at origin @xmath56 scattering length @xmath57 connection factors express as quotients of constant wronskians of solutions wronskian of both sides of eq .dsiete @xmath58 isolating @xmath59 obtains @xmath60/\mathcal{w}\left[w_{\rm{irr}},w_{\rm{reg}}\right \label{vdos} substituted eq . vuno gives scattering length quotient wronskians @xmath61/\mathcal{w}\left[w_2,w_{\rm{reg}}\right]\right_{0}\label{vtres} direct computation values wronskians form @xmath62 = w_j(z)\,\frac{dw_{\rm{reg}}(z)}{dz -_{\rm{reg}}(z) discarded expansion definition @xmath63 asymptotic not converge indirect evaluating wronskians explained in refs . @xcite @xcite , potentials sum positive powers distance considered apply next section method present caseintroduce auxiliary functions @xmath64 @xmath65 = \exp(-\beta_{\rm{reg}},z^{-5}/5)\ \mathcal{w}\left[w_j w_{\rm{reg}}\right left side equation doubly infinite sum powers @xmath35 @xmath65 \sum_{p=-\infty}^\infty \gamma_{p , j},z^{p+\nu_j+\mu notation @xmath66 value @xmath67 obtained expansion @xmath68 right side same powers @xmath35 expansion left side construct each value @xmath69 five expansions @xmath70 @xmath68 expansions forms heaviside s exponential series @xmath71 integer @xmath72 reduces familiar series expansion exponential function other value @xmath72 asymptotic expansion @xmath73 constants @xmath74 ( @xmath75 ) satisfying restriction @xmath76{tuno vseis @xmath77\sim\sum_{k=0}^{4 } c_{k , j}\mathcal{e}_{k , j}(z ){tdos} choosing @xmath78 expansions @xmath79 values @xmath80 comparison expansion [ tdos with [ vsiete ] term by term obtains @xmath81 for positive integer @xmath82 substituting [ tuno values @xmath83 from [ tcuatro ] ) @xmath84 = \sum_{k=0}^4\frac{\gamma(n+1+\delta_{k , j}\sqrt{\lambda}/5)^{n+\delta_{k , j }\gamma_{-5n - k , j }{tcinco} with @xmath78 and @xmath85 given by ( [ ttres ] ) and [ vocho ] ) convergence of series right - side slow like series @xmath86 precise arithmetic used integer @xmath82 chosen arbitrarily with restriction @xmath87 test procedure coincidence of results for different choices @xmath82 fig . 1 behavior of values @xmath88 method case @xmath8 intensity @xmath7 potential goes 0 to 2500 represented horizontal axis values @xmath89 instead of @xmath7study solvable potentials lennard - jones type exponents @xmath90 @xcite shown difference values square root intensity two consecutive zeros poles @xmath2 constant @xmath91 figure 1 difference nearly constant similar other values @xmath1 potential uno scattering length zero - range pseudopotential formation bose - einstein condensates precise location zeros poles @xmath88 crucial new bound state arises increasing intensity potential coincides with procedure allows locate zeros poles cancelation wronskians @xmath92 @xmath93 precision limited by number digits computation report tables 1 2 list first ten zeros poles scattering length potential uno @xmath94 @xmath95 approximate @xmath10 constancy difference consecutive zeros poles manifest @xmath9 @xmath8 @xmath10 + 1.135708 & 2.944907 4 + 4.281230 10.307414 14 + 7.627058 17.758560 24 + 10.991652 25.220363 34 + 14.361060 32.685259 44 + 17.732554 40.151469 54 + 21.105133 47.618360 64 + 24.478348 55.085650 74 + 27.851968 62.553194 84 + 31.225862 70.020910 94 + @xmath9 + 2.650141 4.728696 6 + 5.949138 12.165518 16 + 9.308435 19.622908 26 + 12.675992 27.086171 36 + 16.046629 34.551611 46 + 19.418744 42.018080 56 + 22.791679 49.485114 66 + 26.165118 56.952491 76 + 29.538887 64.420092 86 + 32.912885 71.887847 96 comparison scattering length exponents @xmath1 figure 2 synoptic representation @xmath94 @xmath95 variable scattering length @xmath96 vs free singularities @xmath2 used ouerdane @xcite solve riccati equation scattering length variable phase theory potential truncated large distance formulate conjecture potential @xmath97 lennard - jones type exponents @xmath98 denote @xmath99 @xmath100 values intensity zeros poles scattering length quasi - linear laws @xmath102 @xmath103 @xmath104 independent @xmath82case @xmath105 obtained @xmath106 dependence @xmath107 arguments @xmath108 fixed @xmath109 diminishes @xmath110 increases @xmath109 increases @xmath108 recommendations two anonymous referees presentation authors acknowledge financial support departamento ciencia tecnologa universidad gobierno aragn fondo social europeo ministerio ciencia innovacin project - 11154 | 0.55958 | 3,348 | 1,878 | , Saving $0.1 in GPT-4. |
55,121 | this is a comment to the paper , limitations on the superposition principle : superselection rules in non - relativistic quantum mechanics by c cisneros et al 1998 eur .
j. phys .
19 237 .
doi:10.1088/0143 - 0807/19/3/005 .
+ the proof that the authors construct for the limitation on the superposition of state vectors corresponding to different sectors of the hilbert space , partitioned by a superoperator has a flaw as outlined below . | in ref . @xcite , section 2.4(impossibility of superposing states belonging to different coherent sectors ) , the authors construct a proof using a superoperator g that commutes with all observables of the system .
the eigenstates of operator g are of the form @xmath0 where @xmath1 and @xmath2 are labels to distinguish different eigenvectors all of which are non - degenerate , i.e. @xmath3 if @xmath4 . + in the next section(2.4 ) , they construct a vector @xmath5 = @xmath6 superposing the eigenstates of g and claim that since g commutes with every other observable , @xmath5 should be an eigenstate of every operator in a complete set of commuting observables of the system .
+ now we know that if @xmath7 and @xmath8 are eigenvectors of an operator g , then @xmath9 is not an eigenvector unless they have the same eigenvalue since : @xmath10 ( for some @xmath11 ) unless a = b , which is not true in general and definitely not true for the hermitian operators with non - degenerate eigenvalues in the paper .
it is true that if g commutes with every other observable and has non - degenerate eigenvalues , then all such obervables share the same eigenvectors .
but @xmath5 itself is not an eigenvector of g as proved above . and
so @xmath5 can not be an eigenvector of other observables too .
to summarize , the proof of superselection rules proposed is thus is not valid for the case discussed by the authors . whether a similar proof can be constructed is still open .
10 c cisneros et al 1998 _ eur .
j. phys _ 19 237 , doi:10.1088/0143 - 0807/19/3/005 , limitations on the superposition principle : superselection rules in non - relativistic quantum mechanics . | @xcite section 2.4(impossibility superposing states sectors authors construct proof superoperator g commutes observables eigenstates g form @xmath0 @xmath1 @xmath2 labels distinguish eigenvectors non - degenerate @xmath3 next section(2.4 construct vector @xmath5 = @xmath6 superposing eigenstates g g commutes with observable @xmath5 eigenstate every operator observables @xmath7 @xmath8 eigenvectors operator g @xmath9 not eigenvector unless same eigenvalue since @xmath10 @xmath11 a = b not true hermitian operators non - degenerate eigenvalues if g commutes with observable non - degenerate eigenvalues all share same eigenvectors @xmath5 not eigenvector g not eigenvector other observables proof of superselection rules proposed not valid for case discussed authors similar proof open cisneros et al 1998 eur . j. phys _ 19 237 doi:10.1088/0143 - 0807/19/3/005 limitations superposition principle superselection rules non - relativistic quantum mechanics | 0.07507 | 432 | 240 | , Saving $0.0 in GPT-4. |
55,135 | * abstract * 0.2 in we study the mobility of an overdamped particle in a periodic potential tilted by a constant force .
the mobility exhibits a stochastic resonance in inhomogeneous systems with space dependent friction coefficient .
the result indicates that the presence of oscillating external field is not essential for the observability of stochastic resonance , at least in the inhomogenous medium .
= 9.4 in = 6.5 in = 0.3 in = 0.1 in = 0.6 in = 0.0 cm 0.3 in | stochastic resonance ( s.r ) is an important phenomena with considerable implications in all branches of science @xcite .
the enhanced response of a nonlinear system to input signals at the expense of and as a function of noise is termed as stochastic resonance .
it is generally accepted that sr can be observed provided certain essential conditions are fulfilled .
attempts are being continually made to reduce the number and stringency of these constraints for the realization of the phenomenon .
the simplest and the minimal ( currently accepted ) conditions under which conventional sr can be observed are , the presence of a ) a bistable system , b ) a tunable gaussian white noise and c ) a time varying periodic signal ( force ) .
recently , hu @xcite suggested that the last ingredient may hopefully be replaced by a constant force and , by implication , sr may be observed , e.g , in the drift velocity ( mobility ) of an overdamped brownian particle in a tilted periodic potential as a function of noise strength .
unfortunately the suggestion was proved to be incorrect @xcite . however , marchesoni , @xcite by analyzing the work of risken @xcite , observed that sr can be observed in the drift velocity of brownian particles in a tilted periodic ( washboard ) potential only in the underdamped situation where friction acts as surrogate to the external periodic drive . in the present work
we show that sr can be observed in the mobility of even an overdamped brownian particle in a tilted periodic potential ( without the presence of oscillating field ) but in the presence of a space - dependent periodic friction coefficient ( i.e. , in an inhomogeneous medium ) .
the space dependence of friction coefficient does not affect the equilibrium properties such as the equilibrium probability distribution .
however , it does affect the dynamical ( nonequilibrium ) properties of the system ( such as the relaxation rates ) .
the space dependence of friction @xmath0 can be microscopically modeled through the nonlinear ( space - dependent ) coupling between the particle degrees of freedom and the thermal bath ( characterized by its equilibrium temperature ) @xcite . in this work
we find that the mobility of overdamped particles in a sinusoidal potential subject to a sinusoidal friction coefficient but with a phase difference @xmath1 shows a peak as a function of noise strength ( temperature of the bath ) in the presence of only a constant force field @xmath2 . by properly choosing @xmath1
we obtain this sr behaviour in the mobility ( defined as drift velocity divided by @xmath2 ) even for a small constant external force field ( when the barrier for the particle motion in the tilted periodic potential remains finite ) .
the motion of an overdamped particle , in a periodic potential @xmath3 and subjected to a space - dependent friction coefficient @xmath0 and an additional constant force @xmath2 , at temperature @xmath4 , is described by the fokker - planck equation @xcite .
@xmath5\ ] ] one can calculate @xcite the probability current @xmath6 , for the potential function @xmath3 with @xmath7 , as @xmath8 where @xmath9 and @xmath10 . the mobility ( defined @xcite as the ratio of the drift velocity , @xmath11 divided by the applied force @xmath2 ) , @xmath12 .
we take @xmath13 and @xmath14 , with @xmath15 .
clearly , @xmath16 as @xmath17 , but the mobility @xmath18 remains finite as @xmath19 , for nonzero temperature @xmath4 .
however , for @xmath20 as @xmath21 , @xmath22 . also , as @xmath23
however , for given @xmath24 and @xmath1 , in order to find the variation of @xmath18 at intermediate @xmath4 and @xmath2 , one needs to evaluate the double integral in the denominator of eqn .
( 2 ) numerically @xcite .
the variation of mobility @xmath18 in the parameter space of @xmath25 provides a very rich structure where the phase lag @xmath1 plays an important role .
however , in this paper we report only the variation of @xmath18 with temperature ( noise strength ) @xmath4 at a few carefully selected values of @xmath26 in order to highlight the observability of s.r . from eqn .
( 2 ) one can find that , even though @xmath27 , except for phase lag @xmath28 or @xmath29 , but yet @xmath30 .
therefore , we need to explore only @xmath31 . also in order to observe sr the choice of larger values of @xmath24 is found to be more appropriate .
so we take @xmath32 and explore the full range of @xmath33 $ ] .
we find that there is a range of values of @xmath1 within which sr is observed .
for example , for @xmath34 , sr is observed but the peaks are very broad . and for this @xmath1 , sr is more prominent for @xmath35 than @xmath36 . for smaller
@xmath1 it is harder to observe the peaks as they become still broader .
however , as @xmath1 is increased the peaks become sharper .
fig.1 shows the mobility @xmath18 ( in dimensionless units @xmath37 ) as a function of @xmath4 ( in dimensionless units ) for @xmath38 and @xmath39 . here
the peaks are larger for @xmath35 than for @xmath36 .
2 , shows the mobility @xmath18 as a function of @xmath4 for @xmath40 , @xmath41 .
it can be seen that the peaks are almost flat for @xmath42 , but are prominent for the intermediate temperature range . from fig .
3 , we do not observe s.r for any @xmath31 .
but the figure prominently shows that for @xmath36 and for small @xmath4 , the mobility decreases instead of increasing with temperature .
this in itself is a novel feature .
the above mentioned features result from a subtle combined effect of the periodic space dependent friction and the periodic potential in the presence of a constant applied force .
the effect can be observed only when there is a phase difference between the potential and the frictional profile ; the phase difference makes the mobility asymmetric with respect to the reversal of the field @xmath2 .
the mobility shows many other interesting features as a function of other parameters , @xmath43 and @xmath1 .
we observe the occurrence of stochastic resonance in the the mobility of an overdamped brownian particle in a sinusoidal potential tilted by a constant force and subjected to a gaussian white noise but , of course , in an inhomogeneous system with space - dependent friction coefficient .
thus the time dependent external oscillating field , which is generally considered as an essential ingredient for the observability of sr can be replaced by a constant force field concomitant with a space - dependent(periodic ) friction coefficient of a spatially extended periodic system .
we would like to point out here that the correct high friction langevin equation in the space dependent frictional medium involves a multiplicative noise along with a temperature drift term @xcite .
m. c. m thanks the institute of physics , bhubaneswar , for financial assistance and hospitality .
m. c. m and a. m. j acknowledge partial financial support from the b. r. n. s project , d. a. e , india . | stochastic resonance ( s.r ) is important phenomena implications in all branches science enhanced response nonlinear system to input signals at expense noise termed stochastic resonance accepted sr can be observed essential conditions fulfilled attempts to reduce constraints for simplest minimal conditions conventional sr are bistable system tunable gaussian white noise time varying periodic signal ( force ) suggested last ingredient may be replaced by constant force sr be observed in drift velocity of overdamped brownian particle in tilted periodic potential function of noise strength suggestion incorrect marchesoni observed sr can be observed in drift velocity of brownian particles in tilted potential only in underdamped situation where friction acts surrogate to external periodic drive present work sr can be observed in mobility of overdamped brownian particle in tilted periodic potential without oscillating field but space - dependent periodic friction coefficient inhomogeneous medium space dependence of friction coefficient affect equilibrium properties dynamical nonequilibrium properties system relaxation ratesspace dependence of friction @xmath0 modeled through nonlinear coupling between particle degrees freedom thermal bath equilibrium temperature @xcite mobility of overdamped particles in sinusoidal potential friction coefficient phase difference peak function of noise strength temperature bath constant force field @xmath2 choosing @xmath1 behaviour in mobility drift velocity divided by @xmath2 even for small constant external force field barrier particle motion tilted potential finite motion of overdamped particle in periodic potential @xmath3 space - dependent friction coefficient additional constant force @xmath2 at temperature @xmath4 described by fokker - planck equation @xcite calculate probability current @xmath6 for potential function @xmath3 with @xmath7 mobility ratio of drift velocity divided by applied force @xmath2 @xmath13 @xmath14 @xmath15 mobility @xmath18 remains finite as @xmath19 for nonzero temperature @xmath4 @xmath20@xmath23 @xmath24 @xmath1 find variation of @xmath18 at intermediate @xmath4 @xmath2 evaluate double integral in denominator eqn @xcite variation of mobility @xmath18 in parameter space @xmath25 provides rich structure phase lag @xmath1 important paper report variation of @xmath18 with temperature noise strength ) @xmath4 at selected values @xmath26 highlight observability s.r. from eqn @xmath27 phase lag @xmath28 @xmath29 @xmath30 explore @xmath31 observe sr larger values @xmath24 appropriate @xmath32 explore full range of @xmath33 range of values of @xmath1 sr observed @xmath34 sr observed peaks broad sr prominent for @xmath35 than @xmath36 smaller @xmath1 harder to observe peaks broader @xmath1 increased peaks sharper fig.1 shows mobility @xmath18 function of @xmath4 for @xmath38 @xmath39 peaks larger for @xmath35 than @xmath36 mobility @xmath18 function of @xmath4 for @xmath40 @xmath41 peaks almost flat for @xmath42 prominent for intermediate temperature rangefig . 3 observe s.r @xmath31 shows @xmath36 @xmath4 mobility decreases with temperature novel feature features result from effect periodic space dependent friction periodic potential constant force effect observed phase difference between potential frictional profile makes mobility asymmetric reversal field @xmath2 mobility shows features function parameters @xmath43 @xmath1 observe stochastic resonance in mobility overdamped brownian particle sinusoidal potential tilted by constant force subjected gaussian white noise inhomogeneous system with space - dependent friction coefficient time dependent external oscillating field essential sr replaced by constant force field space - dependent(periodic friction coefficient spatially extended periodic system correct high friction langevin equation space dependent frictional medium involves multiplicative noise temperature drift term @xcite thanks institute of physics bhubaneswar for financial assistance acknowledge partial financial support from b. r. n. s project d e india | 0.266726 | 1,597 | 798 | , Saving $0.0 in GPT-4. |
55,163 | missing web pages ( pages that return the @xmath0 `` page not found '' error ) are part of the browsing experience .
the manual use of search engines to rediscover missing pages can be frustrating and unsuccessful .
we compare four automated methods for rediscovering web pages .
we extract the page s title , generate the page s lexical signature ( ls ) , obtain the page s tags from the bookmarking website ` delicious.com ` and generate a ls from the page s link neighborhood .
we use the output of all methods to query internet search engines and analyze their retrieval performance .
our results show that both lss and titles perform fairly well with over @xmath1 uris returned top ranked from yahoo!. however , the combination of methods improves the retrieval performance .
considering the complexity of the ls generation , querying the title first and in case of insufficient results querying the lss second is the preferable setup .
this combination accounts for more than @xmath2 top ranked uris .
= 10000 = 10000 information search and retrieval | inaccessible web pages and `` 404 page not found '' responses are part of the web browsing experience . despite guidance for how to create `` cool uris '' that do not change @xcite there are many reasons why uris or even entire websites break @xcite .
however , we claim that information on the web is rarely completely lost , it is just missing . in whole or in part , content is often just moving from one uri to another .
it is our intuition that major search engines like google , yahoo ! and msn live (
our experiments were conducted before microsoft introduced bing ) , as members of what we call the web infrastructure ( wi ) , likely have crawled the content and possibly even stored a copy in their cache .
therefore the content is not lost , it `` just '' needs to be rediscovered .
the wi , explored in detail in @xcite , also includes ( besides search engines ) non - profit archives such as the internet archive ( ia ) or the european archive as well as large - scale academic digital data preservation projects e.g. , citeseer and nsdl .
it is commonplace for content to `` move '' to different uris over time .
figure [ fig : ht06 ] shows two snapshots as an example of a web page whose content has moved within one year after its creation .
figure [ fig : ht06_1 ] shows the content of the original uri of the hypertext @xmath3 conference as displayed in 12/2009 .
the original uri clearly does not hold conference related content anymore .
our suspicion is that the website administrators did not renew the domain registration and therefore enabling someone else to take over .
however , the content is not lost .
it is now available at a new uri as shown in figure [ fig : ht06_2 ] . in this paper
we investigate the retrieval performance of four methods that can be automated and together with the wi used to discover missing web pages .
these methods are : 1 .
lexical signatures ( lss ) typically the @xmath4-@xmath5 most significant keywords extracted from a cached copy of the missing page that capture its `` aboutness '' 2 .
the title of the page the two underlying assumptions here are : * web pages have descriptive titles * titles only change infrequently over time 3 . social bookmarking tags terms suggested by internet users on ` delicious.com ` when the page was bookmarked 4 .
link neighborhood lss ( lnls ) a ls generated from the pages that link to the missing page ( inlinks ) and not from a cached copy of the missing page .
figure [ fig : flow_diagram ] displays the scenario how the four methods of interest can automatically be applied for the discovery of a missing page .
the occurrence of an @xmath0 error is displayed in the first step .
search engine caches and the ia will consequently be queried with the uri requested by the user . in case
older copies of the page are available they can be offered to the user . if the user s information need is satisfied , nothing further needs to be done ( step ( @xmath6 ) ) .
if this is not the case we need to proceed to step ( @xmath7 ) where we extract titles , try to obtain tags about the uri and generate lss from the obtained copies .
the obtained terms are then queried against live search engines .
the returned results are again offered to the user and in case the outcome is not satisfying more sophisticated and complex methods need to be applied ( step ( @xmath4 ) ) .
search engines can be queried to discover pages linking to the missing page .
the assumption is that the aggregate of those pages is likely to be `` about '' the same topic . from this link neighborhood
a ls can be generated . at this point
the approach is the same as the ls method , with the exception that the ls has been generated from a link neighborhood and not a cached copy of the page itself .
the important point of this scenario is that it works while the user is browsing and therefore has to provide results in real time .
queries against search engines can be automated through apis but the generation of lss needs to be automated too . as an example
let us look at how the methods would be applied to the web page ` www.nicnichols.com ` .
the page is about a photographer named nic nichols .
table [ tab : data_example ] displays all data we obtained about the page using the four methods .
.data obtained from ` www.nicnichols.com ` [ cols= " < , < " , ] the contrast of total title length in number of characters and rank is shown in figure [ fig : title_clenvsrank ] .
while the title length varies greatly between @xmath8 and @xmath9 characters we only see @xmath10 uris with a title length greater or equal to @xmath11 and only three uris with more than @xmath12 characters in their title .
figure [ fig : title_clenvsrank ] does not reveal an obvious pattern between number of characters and rank returned for a title but very short titles ( less than @xmath13 characters ) do not seem to perform well .
a title length between @xmath13 and @xmath14 characters is most common and the ranks seem to be better in the range of @xmath13 to @xmath15 characters total .
figure [ fig : title_sw_meanlenvsrank ] depicts on the left the mean number of characters per title term and their retrieval performance .
it seems that terms with an average of @xmath4 , @xmath16 or @xmath5 characters seem to be most suitable for well performing query terms . on the bottom right end of the barplot we can see two titles that have a mean character length per term of @xmath17 and @xmath18 .
since such long words are rather rare they perform very well .
the observation of stop word frequency in the titles and their performance is not surprising . as shown on the right in figure [ fig : title_sw_meanlenvsrank ] titles with more than a couple of stop words seem to harm the performance .
the intuition is that search engines filter stop words from the query ( keep in mind , these are non - quoted titles ) and therefore it makes sense that for example the title with @xmath19 stop words does not return its uri within the top @xmath11 ranks . for completeness we removed all stopwords from the titles and analyzed their retrieval performance in dependence of the new title length .
the results are shown in figure [ fig : title_nosw_lenvsrank ] .
as expected we see more titles with fewer terms performing slightly better than the original titles .
this result indicates that the performance of the method using the web page s titles can still be improved .
further analysis of the best combination of methods with titles without stop words remains for future work .
our main aspect of future work is the implementation of the system described in the flow diagram of figure [ fig : flow_diagram ] .
the system will operate as a browser plugin and will trigger once the user encounters a @xmath0 `` page not found '' error .
it will provide all of the introduced methods to rediscover the missing page and since the discovery process happens in an automated fashion , the system can provide the user with the results in real - time while she is browsing the internet .
we have shown in @xcite that lss evolve over time and consequently lose some of its retrieval strength .
here we are arguing that titles of web pages are a powerful alternative to lss .
the next logical step is to investigate the evolution and possible decay of titles over time .
our intuition is that titles do not decay quite as quickly as lss do since the actual content of a web page ( a headline , sentence or paragraph ) presumably changes more frequently than its general topic which is what the title is supposed to represent .
our set of obtained tags is limited .
it remains for future work to investigate the retrieval performance of tags in a large scale experiment .
it also would be interesting to see what the term overlap between tags , titles and lss is since all three methods are generated on different grounds .
our method to generate lnlss may not be optimal .
we chose to use inlink pages only and created a bucket of all neighborhood terms per uri .
the lnlss are based on this bucket .
it remains to be seen whether outlink pages actually can contribute to the retrieval performance and other methods than the bucket of terms are preferable .
it is possible that our neighborhood is too big since it includes the entire neighboring page .
a page that links to many pages ( hub ) may have a diffuse `` aboutness '' .
hence we are going to restrict the content gained from the neighborhood to the link anchor text of the inlink pages .
in this paper we evaluate the retrieval performance of four methods to discover missing web pages .
we generate a dataset of uris by randomly sampling uris from ` dmoz.org ` and assume these pages to be missing .
we generate lss from copies of the pages , parse the pages titles , obtain tags of the uris from the bookmarking website ` delicious.com ` and generate lss based on link neighborhood .
we use the three major search engines google , yahoo ! and msn live to acquire mandatory document frequency data for the generation of the lss .
we further query all three search engines for all our methods and combine methods to improve the retrieval performance .
we are able to recommend a setup of methods and see one search engine performing best in most of our experiments .
it has been shown in related work that lss can perform well for retrieving web pages .
our results confirm these findings , for example more than two - thirds of our uris have been returned as the top result when querying @xmath4- and @xmath5-term lss against the yahoo ! search engine api .
they also lead us to the claim that titles of web pages are a strong alternative to lss .
almost @xmath20 of the uris have been returned as the top result from the google search engine api when queried with the ( non - quoted ) title .
however , our results show that a combination of methods performs best .
querying the title first and then using the @xmath4-term lss for all remaining undiscovered uris against yahoo ! provided the overall best result with @xmath21 of top ranked uris and another @xmath22 in the top @xmath13 ranks .
the combination @xmath5-term ls , title , @xmath4-term ls returned @xmath23 of the uris in the top ranks but since lss are more expensive to generate than titles , we recommend the former combination of methods . a good strategy , based on our results , is to query the title first and if the results are insufficient generate and query lss second . yahoo ! returned the best results for all combination of methods and thus seems to be the best choice even though google returned better results when querying the title only .
this work is supported in part by the library of congress . | inaccessible web pages 404 page not found responses part of web browsing experience despite guidance for cool uris change @xcite many reasons why uris websites break @xcite claim information web rarely completely lost just missing content often moving from one uri to another intuition major search engines like google yahoo ! msn live members of web infrastructure wi likely crawled content possibly stored copy in cache content not lost needs to rediscovered wi includes non - profit archives internet archive european archive large academic digital data preservation projects citeseer nsdl commonplace for content to move to different uris over time figure [ fig ht06 ] shows web page content moved within one year after creation figure ] shows content of original uri of hypertext @xmath3 conference displayed in 12/2009 original uri not hold conference related content suspicion website administrators not renew domain registration someone else to take over content not lost now available at new uri in figure_2 paper investigate retrieval performance of four methods to discover missing web pages methods are lexical signatures @xmath4-@xmath5 significant keywords extracted from cached copy missing page 2.title of page two assumptions are web pages have descriptive titles titles change infrequently over time social bookmarking tags suggested by users on delicious.com page bookmarked link neighborhood lss generated from pages to missing page not cached copy figure displays four methods of interest applied for discovery of missing page @xmath0 error displayed first step search engine caches ia queried with uri requested user older copies available offered to user if user s information need satisfied nothing further step @xmath6 ) if not proceed to step ( @xmath7 ) extract titles obtain tags uri generate lss from obtained copies obtained terms queried against live search engines returned results offered to user outcome not more sophisticated complex methods applied ( step @xmath4 ) search engines queried to discover pages linking to missing page assumption is aggregate pages likely about same topic from link neighborhood ls generated approach same as ls method ls generated from link neighborhood not cached copy page works while user browsing results in real time queries against search engines automated through generation of lss needs automated example methods applied to web page ` www.nicnichols.com `page about photographer nic nichols table data_example displays data using four methods.data obtained from www.nicnichols.com contrast of title length number of characters rank shown in figure title_clenvsrank title length varies between @xmath8 and @xmath9 characters only @xmath10 uris with title length greater or equal to @xmath11 three uris with more than @xmath12 characters pattern between number characters rank short titles less than @xmath13 characters perform well title length between @xmath13 and @xmath14 characters common ranks better in range @xmath13 to @xmath15 characters figure title_sw_meanlenvsrank depicts mean number of characters per title term retrieval performance terms with average of @xmath4 @xmath16 or @xmath5 characters suitable for well performing query terms two titles mean character length per term of @xmath17 and @xmath18 long words rare perform well stop word frequency in titles performance not surprising titles with more than stop words harm performance search engines filter stop words from title with @xmath19 stop words return within top @xmath11 ranksremoved stopwords from titles analyzed retrieval performance new title length results shown in figure title more titles with fewer terms performing slightly better than original titles performance method using web page s titles can be improved further analysis of combination methods with titles without stop words remains for future work main aspect future implementation of system in flow diagram system as browser plugin trigger user encounters page not found '' error methods to rediscover missing page discovery automated system results in real - time shown lss evolve over time lose retrieval strength arguing titles of web pages powerful alternative to lss investigate evolution decay of titles over time titles do not decay as quickly as lss content web page changes more frequently than general topic title set of obtained tags limited for future work to investigate retrieval performance of tags in large scale experiment interesting to see term overlap between tags titles lss method to generate lnlss may not be optimal chose to use inlink pages only created bucket of neighborhood terms per uri lnlss based on this bucketwhether outlink pages contribute retrieval performance other methods than bucket terms preferable possible neighborhood big includes entire neighboring page page many pages may diffuse aboutness restrict content from neighborhood to link anchor text inlink pages evaluate retrieval performance of four methods to discover missing web pages generate dataset uris from dmoz.org assume pages missing generate lss from copies pages parse titles obtain tags from delicious.com generate lss based on link neighborhood use search engines google yahoo ! msn live document frequency data query all three search engines for methods combine methods to improve retrieval performance recommend setup of methods one search engine best lss well for retrieving web pages results confirm two - thirds of uris returned top result @xmath4- @xmath5-term lss against yahoo ! api titles pages strong alternative to lss almost @xmath20 uris returned top result from google non - quoted title results show combination of methods performs best querying title first then using @xmath4-term lss for undiscovered uris against yahoo ! best result with @xmath21 top ranked uris @xmath22 top @xmath13 rankscombination @xmath5-term ls , title @xmath4-term ls returned @xmath23 uris top ranks lss expensive than titles recommend former combination strategy query title first if results insufficient lss second yahoo ! returned best results combination methods best choice google returned better results title only work supported by library of congress | 0.357047 | 2,326 | 1,131 | , Saving $0.1 in GPT-4. |
55,179 | we illustrate the stochastic method for solving the schwinger - dyson equations in large-@xmath0 quantum field theories described in http://arxiv.org/abs/1009.4033[_arxiv:1009.4033_ ] on the example of the gross - witten unitary matrix model . in the strong - coupling limit
, this method can be applied directly , while in the weak - coupling limit we change the variables from compact to noncompact ones in order to cast the schwinger - dyson equations in the stochastic form .
this leads to a new action with an infinite number of higher - order interaction terms .
nevertheless , such an action can be efficiently handled .
this suggests the way to apply the method of http://arxiv.org/abs/1009.4033[_arxiv:1009.4033_ ] to field theories with @xmath1 field variables as well as to effective field theories in the large-@xmath0 limit . | in many physically interesting situations direct monte - carlo sampling of the path integral of the theory is not an efficient simulation method .
probably the most notable example in hadronic physics is qcd at finite chemical potential .
another example are quantum field theories in the large-@xmath0 limit , since the computational complexity of monte - carlo algorithms tends to infinity in this limit , and one always has to rely on extrapolations from finite @xmath0
. one of the ways to overcome these difficulties is to expand the desired path integral in powers of some coupling constants , and to sum up the obtained series stochastically .
this is the basic idea of the diagrammatic monte - carlo @xcite .
this method , however , can not be straightforwardly applied to field theories with continuous variables , since in such cases the perturbative series are typically only asymptotic @xcite .
for field theories with a non - abelian internal symmetry group , such as @xmath1 , each diagram contains also some power of @xmath0 , and the sums over the terms leading in @xmath0 are typically convergent @xcite .
this salient feature of large-@xmath0 quantum field theories was used in @xcite to construct a stochastic algorithm for summing over all planar diagrams in perturbative expansions .
this algorithm is based on the stochastic solution of the schwinger - dyson ( sd ) equations in the factorized form , so that they are interpreted as the equations for the stationary probability distribution of a certain random process .
the correlators of the field variables are thus estimated as probability distributions of some random variables , similarly to the `` worm algorithm '' @xcite .
once the sd equations can be cast in the stochastic form @xmath2 where @xmath3 schematically denote the field correlators and the coefficients @xmath4 , @xmath5 and @xmath6 satisfy the inequality @xmath7 , one can construct a `` nonlinear '' random process for which @xmath3 is the stationary probability distribution @xcite . however , when the field variables are the elements of @xmath1 , sd equations can be rewritten as ( [ random_process_eq ] ) only in the strong coupling limit . on the other hand ,
the continuum limit of such theories typically corresponds to the weak - coupling limit . in this paper
we consider the simplest nontrivial model defined by the integral over @xmath1 group in the limit @xmath8 , namely , the gross - witten unitary matrix model @xcite .
we demonstrate that the method of @xcite can be used to simulate this model in both the strong- and the weak - coupling regimes . in the latter case
, one has to change the variables to the noncompact ones , which are more adequate for the weak - coupling expansion .
the partition function of the gross - witten model @xcite is @xmath9 and the observables of interest are @xmath10 .
the sd equations for the model ( [ gw_pf ] ) in terms of the observables @xmath11 are : @xmath12 by a variable redefinition @xmath13 one can transform these equations into the stochastic form ( [ random_process_eq ] ) if @xmath14 . according to @xcite ,
such equations describe the stationary distribution of the random process operating on the stack of the elements of the form @xmath15 , where @xmath16 is an integer labeling the observables @xmath17 .
@xmath18 is proportional to the difference of the probabilities to find the elements @xmath19 and @xmath20 at the top of the stack . at each step of
the random process one performs at random one of the following steps : create : : : with probability @xmath21 push a new element @xmath22 to the stack .
join : : : with probability @xmath23 pop the two elements @xmath24 and @xmath25 and push a new element @xmath26 .
increase : : : with probability @xmath27 increase the topmost element in the stack by one .
decrease : : : with probability @xmath28 decrease the topmost element in the stack by one and change its sign .
restart : : : otherwise restart with a stack containing a single element @xmath22 . the dependence of the observables @xmath11 on @xmath29 is illustrated on fig .
[ fig : gw_sc_seq ] .
the data were obtained for @xmath30 iterations of the random process , which took several seconds on a standard 2 ghz cpu . in agreement with the exact solution @xcite , only @xmath31 is nonzero and behaves as @xmath32 , while all other observables are equal to zero within error range . it should be noted that the element @xmath15 with @xmath33 are generated by our random process , but their contributions to @xmath17 at @xmath33 cancel on average , and as a result @xmath34 for @xmath33 .
a similar stochastic strong - coupling solution can be easily constructed for other models with @xmath1 field variables , for example , for the large-@xmath0 @xmath1 non - abelian lattice gauge theory . in this case
the desired random process will be a minor modification of the stochastic solution of the weingarten model @xcite .
however , such a strong - coupling solution is not very interesting , since the continuum physics should emerge in the weak - coupling limit .
at sufficiently small @xmath29 the equations ( [ gw_sd_eq ] ) can not be transformed to the stochastic form by a simple rescaling of observables . in order to obtain a perturbative solution in this case ,
let us introduce the new hermitian matrix variables @xmath35 via the relation @xmath36 , where the value of @xmath37 will be given later .
first we have to express the haar measure on @xmath1 in terms of @xmath35 .
it can be found from the invariant metric form @xmath38 : @xmath39 now one can include the haar measure ( [ un_inv_metric ] ) into the action for the variables @xmath35 and express integrals over @xmath40 as integrals over @xmath35 .
the new action can be expanded in powers of @xmath35 , thus leading to the hermitian matrix model with an infinite number of terms in the potential : @xmath41 now we set @xmath42 so that the quadratic term in the action ( [ newvar_action ] ) becomes @xmath43 .
it is interesting that another parameterizations of @xmath1 group manifold by hermitian matrices ( e.g. the exponential @xmath44 ) can in general lead to multi - trace terms in the action .
the sd equations for the new action ( [ newvar_action ] ) involve the observables @xmath45 .
again , we rescale the variables as @xmath46 to cast the sd equation in the stochastic form ( [ random_process_eq ] ) : @xmath47 following @xcite , one can devise a random process which samples the unknowns @xmath18 .
the configuration space of the process is again the stack containing positive integers which label the observables @xmath48 and signs . at each time step one
performs at random one of the following actions : create : : : with probability @xmath49 push a new element @xmath22 to the stack .
join : : : with probability @xmath23 pop the two elements @xmath24 and @xmath25 and push a new element @xmath50 .
increase : : : with probability @xmath51 increase the topmost element in the stack by one .
decrease : : : with probability @xmath52 decrease the topmost element @xmath15 in the stack by @xmath53 and change its sign if @xmath54 is even .
restart : : : otherwise restart with a stack containing a single element @xmath22 . in order to calculate the observables @xmath17
, one should expand them in powers of @xmath35 : @xmath56 .
@xmath57 can be then found from the averages @xmath58 w.r.t .
the above described random process . here
@xmath54 is the topmost element in the stack and @xmath59 is the corresponding sign .
the observables @xmath11 are compared with the exact solution @xcite on fig .
[ fig : gw_wc_seq ] for @xmath60 and for @xmath61 . in order to illustrate the efficiency of the algorithm , on fig .
[ fig : gw_wc_pt ] we plot the contributions of different @xmath54 ( positive and negative separately ) to @xmath62 at @xmath60 .
solid lines are the exact values @xmath11 .
it should be noted that since @xmath63 , each such contribution corresponds to the term of order @xmath64 in the weak - coupling expansion of @xmath11 .
@xmath31 is saturated almost completely by the lowest - order term , while for @xmath65 large contributions of different @xmath54 cancel to yield a numerically small value .
contributions of negative @xmath54 are numerically small in both cases .
thus the strength of the `` sign problem '' for our random process actually depends on the choice of observables .
we also note that no critical slowing down near @xmath66 was observed neither in the weak- nor in the strong - coupling phases .
this work was supported by grants rfbr 09 - 02 - 00338 , 10 - 02 - 09484-mob_z , a grant for scientific schools no .
nsh-679.2008.2 , and by personal grants from the `` dynasty '' foundation and from the fair - russia research center ( frrc ) . | situations direct monte - carlo sampling efficient simulation notable example hadronic physics qcd at finite chemical potential quantum field theories in large-@xmath0 limit computational complexity monte - carlo algorithms on extrapolations from finite @xmath0 expand desired path integral in powers coupling constants sum up obtained series stochastically basic idea diagrammatic monte - carlo @xcite method applied to field theories with continuous variables perturbative series asymptotic field theories non - abelian internal symmetry group @xmath1 each diagram contains power of @xmath0 sums over terms convergent large-@xmath0 used @xcite stochastic algorithm for summing planar diagrams in perturbative expansions algorithm based on stochastic solution schwinger - dyson ) equations factorized form interpreted stationary probability distribution random process correlators of field variables estimated as probability distributions of random variables worm algorithm '' @xcitesd equations stochastic form @xmath2 @xmath3 denote field correlators coefficients @xmath4 satisfy inequality @xmath7 construct nonlinear random process @xmath3 stationary probability distribution @xcite field variables elements @xmath1 sd equations rewritten as random_process_eq ] strong coupling limit continuum limit corresponds weak - coupling limit simplest nontrivial model integral @xmath1 group limit @xmath8 gross - witten unitary matrix model @xcite method simulate strong- weak - coupling regimes change variables to noncompact ones for weak - coupling expansion partition function gross - witten model @xcite is @xmath9 observables interest @xmath10 sd equations observables @xmath11 @xmath12 variable redefinition @xmath13 transform equations into stochastic form [ random_process_eq ] ) @xmath14 equations describe stationary distribution random process stack elements form @xmath15 @xmath16 integer labeling observables @xmath17 @xmath18 proportional to difference probabilities find elements @xmath19 and @xmath20 top of stackstep random process performs create @xmath21 push new element @xmath22 stack join @xmath23 pop elements @xmath24 and @xmath25 push new element @xmath26 increase @xmath27 increase topmost element one decrease @xmath28 decrease topmost element one change sign restart with stack single element @xmath22 dependence observables @xmath11 on @xmath29 illustrated fig . data obtained for @xmath30 iterations seconds standard 2 ghz cpu @xmath31 nonzero behaves as @xmath32 other observables equal to zero range @xmath15 with @xmath33 generated by random process contributions to @xmath17 @xmath33 cancel average @xmath34 for @xmath33 similar stochastic strong - coupling solution constructed for other models with @xmath1 field variables large-@xmath0 non - abelian lattice gauge theory desired random process minor modification of stochastic solution weingarten model @xcite strong - coupling solution not interesting continuum physics in weak - coupling limit small @xmath29 equations not transformed stochastic form by rescaling observablesperturbative solution introduce new hermitian matrix variables @xmath35 via relation @xmath36 value @xmath37 given later express haar measure on @xmath1 in @xmath35 found from invariant metric form @xmath38 @xmath39 include haar measure un_inv_metric into action variables @xmath35 express integrals over @xmath40 @xmath35 new action expanded in powers @xmath35 hermitian matrix model infinite terms set @xmath42 quadratic term action becomes @xmath43 parameterizations of @xmath1 group by hermitian matrices exponential @xmath44 lead to multi - trace terms action sd equations for new action involve observables @xmath45 rescale variables as @xmath46 cast sd equation in stochastic form random_process_eq @xmath47 devise random process unknowns @xmath18 configuration space stack positive integers observables @xmath48 signs step one performs actions create probability @xmath49 push new element @xmath22 join @xmath23 pop elements @xmath24 @xmath25 push new element @xmath50 increase probability @xmath51 topmost element stack by onedecrease probability @xmath52 decrease topmost element @xmath15 by @xmath53 change sign if @xmath54 even restart with stack single element @xmath22 calculate observables @xmath17 expand powers @xmath35 : @xmath56 @xmath57 found from averages @xmath58 random process @xmath54 topmost element @xmath59 corresponding sign observables @xmath11 compared with solution @xcite @xmath60 @xmath61 plot contributions @xmath54 to @xmath62 @xmath60 solid lines exact values @xmath11 since @xmath63 each contribution corresponds to term @xmath64 weak - coupling expansion @xmath11 @xmath31 saturated by lowest - order term @xmath65 large contributions @xmath54 yield small value contributions negative @xmath54 small both strength sign problem depends on choice observables no critical slowing down near @xmath66 observed weak- strong - coupling phases work supported by grants rfbr 09 - 02 - 00338 10 - 02 - 09484-mob_z grant scientific schools-679.2008.2 personal grants dynasty '' foundation fair - russia research center | 0.315704 | 2,121 | 1,112 | , Saving $0.1 in GPT-4. |
55,185 | we present the results of second epoch simultaneous multifrequency observations of 45 sources out of the 55 in the sample of high frequency peakers ( hfp ) by dallacasa et al .
( 2000 ) .
seven of the stellar sources ( i.e. 16% of the total number of sources and 26% of the stellar fraction ) no longer show a convex spectrum and are thus likely to be flat spectrum quasars in an outburst phase during the first epoch of observations .
the rest frame peak frequency distribution of hfp quasars extends up to @xmath0 ghz while that of hfp galaxies is confined to @xmath1 ghz .
imaging at 1.465 ghz and 1.665 ghz has revealed extended emission for 31% of sources , a substantially larger fraction than found for gps sources ( @xmath2 ) . in the case of hfp quasars , but not of hfp galaxies , extended emission is associated with strong variability , consistent with a core - jet structure .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | the compact steep spectrum ( css ) and gigahertz peaked spectrum sources ( gps ) are two classes of intrinsically compact objects ( linear size @xmath3 1 kpc ) defined on the basis of their spectral properties : the overall shape is convex with the position of the turnover between 100 mhz ( css ) and few ghz ( gps ) and the spectral index at high frequencies is steep ( see odea 1998 ) .
the currently favoured model ( the _ youth scenario _ ) relates the small linear size of gps / css sources to their age , implying that they are the progenitors of extended radio sources .
both the spectral analysis ( murgia et al . 1999 ) and the dynamical studies of the separation speed of the hot spots ( owsianik & conway 1998 ; owsianik , conway , & polatidis 1998 ) provide strong evidence in favour of this hypothesis indicating age values of @xmath4 years .
observational studies of the population of gps / css sources have led to the discovery of a correlation between the radio turnover frequency ( @xmath5 ) and the projected angular size ( @xmath6 ) ( odea & baum 1997 ; fanti et al .
this relationship is expected from synchrotron self absorption , where @xmath7 , although free - free absorption could also play a role ( bicknell , dopita , & odea 1997 ) ) . in the youth scenario
this means that the youngest objects have the highest turnover frequencies . objects with turnover frequencies above 5 ghz , which is the limit of gps samples , would represent smaller/_younger _ sources .
we call them high frequency peakers ( hfp )
. one of the disadvantages of this approach is the contamination by beamed radio sources like blazars which possess radio spectra peaking above a few ghz as the result of self - absorbed synchrotron emission from the jet base .
two epochs of simultaneous multifrequency vla observations have been carried out to investigate the reliability of the selection criterium based on the spectral properties . here
we present the results of the second epoch of observations and we compare them with those of the first epoch ( see tinti et al . , in preparation ) .
the complete sample of candidate bright hfps has been obtained by dallacasa et al .
sources with @xmath8 300 mjy and inverted spectra ( slope steeper than -0.5 , s@xmath9 ) were selected in the area of the 87 gb catalogue ( 4.9 ghz ) by cross correlating it with the nvss catalogue ( 1.4 ghz ) .
the sample was then `` cleaned '' from variable flat spectrum sources by means of simultaneous multifrequency vla observations at 1.365 , 1.665 , 4.535 , 4.985 , 8.085 , 8.485 , 14.96 and 22.46 ghz . the resulting sample consist of 55 sources with radio spectra peaking at observed frequencies ranging from a few ghz to 22 ghz .
the selection criterium is independent of optical identification and redshift . as a result
both galaxies and quasars are present in the sample , which is composed of 20% galaxies , 65% stellar objects and 15% still unidentified objects . for comparison ,
galaxies and quasars are almost equally represented in gps samples and most css sources are galaxies .
this confirms that the fraction of galaxies , as opposed to quasars , decreases in samples selected at increasing turnover frequency ( fanti et al .
1990 ; odea 1998 ; stanghellini et al . 1998 ) .
additional observations in the same bands were carried out with the vla for 45 sources of the original sample to study the variability in both flux density and spectral shape . moreover
this subsample has been imaged at 1.465 and 1.665 ghz by means of sensitive and longer observations in order to determine if there is any extended emission surrounding the compact source .
in order to estimate the peak flux density and frequency of the hfp sources , we have fitted the simultaneous radio spectra at the two epochs using an hyperbolic function . a considerable fraction of sources ( 16% ) , identified as stellar objects ( i.e. 26% of such objects ) ,
do not show a convex spectrum at the second epoch and are flat over the full frequency range sampled by our observations .
therefore they should be classified as flat spectrum sources , plausibly in an outburst state in the first epoch . among the sources which preserve the convex spectral shape , hfp quasars
are found to be more variable in radio flux density than galaxies ( see figure 1 ) . in figure 2
, the hfps with known redshifts are plotted together with the css sample of fanti et al .
( 1990 ) , the gps samples of stanghellini et al .
( 1998 ) and snellen et al .
( 1998 ) in the turnover frequency - linear size plane .
the line is the expected correlation ( in the source frame ) in the case of synchrotron self absorption ( fanti et al .
the distribution of hfp sources is consistent with the correlation obeyed by the other classes . in gps samples
it has been found that quasars typically peak at higher frequencies than galaxies ( snellen et al .
1997 ; stanghellini et al .
peak frequencies in the bright hfp sample are on average about a factor 5 higher than in gps samples but still the galaxies cover a range of values lower than the quasars do ( figure 2 , right panel ) .
within our sample 14 ( 31% ) sources show extended emission ( 2 galaxies , 9 quasars and 3 objects without optical identification ) on scales ranging from 14 to 55 arcsec .
stanghellini et al .
( 1990 ) and baum et al .
( 1990 ) found a fraction of 10% of gps sources displaying diffuse and faint extended emission on arcsecond scales near / or around the core .
the youth scenario and the presence of radio emission on scales of tens of kpc could be reconciled by the hypothesis of recurrency proposed by baum et al .
( 1990 ) : the extended emission can be the relic of a previous large scale radio source while the peaked spectrum is due to the young central component that is propagating out admist the relic of the previous epoch of activity .
eight out of the nine quasars with extended emission have strongly variable spectra while the only two extended galaxies in the sample both have non variable and sharply peaked spectra .
the recurrent model seems to be suitable for radio galaxies , while radio quasars can be more easily explained as beamed objects whose jet is pointing towards the observer and whose radio emission is dominated by a single compact component .
milli - arcsecond images at different frequencies can distinguish between a core - jet structure , typically associated to blazars , and the cso - like structure expected for young or recurrent objects .
the fraction of quasars compared to galaxies is substantially higher in our sample than in samples of gps sources . a substantial fraction ( 31% ) of our hfp sources
show extended emission which , in the case of quasars , is associated in most cases with strong variability . on the other hand ,
the two galaxies with extended emission are non variable .
this adds to the body of evidence that gps galaxies and quasars likely represent a different phenomenon and that for quasars the beamed emission from compact components i d dominant .
spectral selection relying on high turnover frequencies is an easier method for selecting candidate young sources than the morphological selection .
however the contamination of spectrally selected samples by doppler boosted objects increases with turnover frequency .
we plan to complete the analysis and the identification of the sample using high resolution observations to study the mas morphology .
vlba observations , at two different frequencies in the optically thin part of the spectrum , have been already carried out and the analysis of the data is in progress .
baum , s.a . ,
odea , c.p . , de bruyn , a.g . , & murphy , d.w . 1990 , , 232 , 19 bicknell , g. , dopita , m. a. , & odea , c. p. 1997 , , 485 , 112 dallacasa , d. , stanghellini , c. , cetonza , m. , & fanti , r. 2000 , , 363 , 887 fanti r. , fanti c. , schilizzi r.t . , spencer , r.e . ,
nan rendong , parma , p. , van breugel , w.j.m . ,
& venturi , t. 1990 , , 231 , 333 fanti , c. , & fanti , r. 2002,in issues in unification of active galactic nuclei , asp conference series 258 , ed r. maiolino , a. marconi & n. nagar ( san francisco : asp ) , 261 murgia , m. , fanti , c. , fanti , r. , gregorini , l. , klein u. , mack , k - h . , & vigotti , m. 1999 , , 345,769 odea , c.p . , & baum , s. a. 1997 , , 113 , 148 odea , c.p .
1998 , , 110 , 493 owsianik , i. , & conway , j.e .
1998 , , 337 , 69 owsianik , i. , conway , j.e . , &
polatidis , a.g .
1998 , , 336 , l37 snellen i. a. g. 1997 , phd thesis , leiden observatory snellen , i.a.g . , schilizzi , r.t .
, de bruyn a.g . , et al . , 1998 , _
a&as _ , * 131 * , 435 stanghellini , c. , baum , s.a . ,
odea , c.p . , & morris , g. b. 1990 , , 233 , 379 stanghellini , c. , odea , c.p .
, dallacasa , d. , baum , s.a . ,
fanti , r. , fanti , c. 1998 , , 131 , 303 stanghellini c. 2003 , pasa , 20 , 118s | compact steep spectrum css ) and gigahertz peaked spectrum sources gps ) are classes compact objects ( linear size @xmath3 1 kpc ) defined spectral properties shape is convex turnover between 100 mhz css and few ghz spectral index at high frequencies steep odea 1998 ) model youth scenario relates small linear size of gps css sources to age progenitors of extended radio sources spectral analysis murgia 1999 dynamical studies of separation speed hot spots provide evidence hypothesis age values of @xmath4 years observational studies gps css correlation between radio turnover frequency @xmath5 and projected angular size ( @xmath6 ) 1997 relationship expected from synchrotron self absorption youth scenario youngest objects have highest turnover frequencies objects turnover frequencies above 5 ghz limit of gps samples represent smaller/_younger sources them high frequency peakers ( hfp ) contamination by beamed radio sources like blazars peaking above few ghz self - absorbed synchrotron emissiontwo epochs multifrequency vla observations selection criterium spectral properties present results second epoch with first epoch tinti complete sample bright hfps obtained by dallacasa et al sources with @xmath8 300 mjy inverted spectra selected 87 gb catalogue ( 4.9 ghz ) with nvss catalogue ( 1.4 ghz ) sample cleaned from variable flat spectrum sources multifrequency vla observations at 1.365 1.665 4.535 4.985 8.085 8.485 14.96 22.46 ghz resulting sample 55 sources radio spectra at frequencies to 22 ghz selection criterium independent of optical identification redshift both galaxies quasars present in sample 20% galaxies 65% stellar objects 15% unidentified objects galaxies quasars almost equally represented in gps samples most css sources are galaxies fraction of galaxies decreases in samples at increasing turnover frequency fanti 1990 1998 additional observations same bands for 45 sources original sample study variability in flux density spectral shape subsample imaged at 1.465 and 1.665 ghz extended emission surrounding compact sourceestimate peak flux density frequency hfp sources fitted simultaneous radio spectra at two epochs hyperbolic function sources ( 16% ) stellar objects 26% show convex spectrum at second epoch flat over full frequency range classified as flat spectrum sources plausibly in outburst state in first epoch convex spectral shape hfp quasars more variable in radio flux density than galaxies figure 1 figure 2 hfps with redshifts plotted with css sample of fanti et gps samples stanghellini snellen turnover frequency - linear size plane line is expected correlation synchrotron self absorption distribution of hfp sources consistent with correlation other classes quasars peak at higher frequencies than galaxies peak frequencies hfp sample average factor 5 higher than gps samples galaxies cover values lower than quasars figure 2 sample 14 ( 31% ) sources show extended emission 2 galaxies 9 quasars 3 objects without optical identification on scales 14 to 55 arcsec stanghellini baum found 10% of gps sources displaying diffuse faint extended emission on arcsecond scales near coreyouth scenario radio emission tens of kpc by hypothesis of recurrency baum et al 1990 extended emission relic of previous scale radio source peaked spectrum due to young central component propagating eight nine quasars with extended emission have variable spectra two extended galaxies non variable peaked spectra recurrent model for radio galaxies radio quasars as beamed objects jet pointing towards observer radio emission dominated by single compact component milli - arcsecond images distinguish core - jet structure blazars cso - like structure young recurrent objects fraction of quasars compared to galaxies higher in our sample than gps sources 31% ) hfp sources show extended emission strong variability two galaxies with extended emission non variable gps galaxies quasars represent different phenomenon for quasars beamed emission from compact components dominant spectral selection high turnover frequencies easier young sources morphological selection contamination samples by doppler boosted objects increases with turnover frequency plan to complete analysis identification using high resolution observations morphology vlba observations at two frequencies carried out analysis in progress baum , s.aodea c.p de bruyn murphy 1990 232 , 19 bicknell dopita odea c. 1997 485 , 112 dallacasa stanghellini cetonza fanti 2000 363 , 887 fanti r fanti c. schilizzi r spencer nan rendong parma van breugel.j venturi 1990 231 , 333 fanti c. fanti r. 2002 unification active galactic nuclei asp conference series 258 maiolino marconi nagar san francisco asp ) 261 murgia fanti c. r gregorini klein u mack vigotti m 1999 345,769 odea c.p baum s 1997 113 , 148 odea.p 1998 110 , 493 owsianik conway j.e . 1998 337 , 69 polatidis a.g 1998 336 , l37 snellen 1997 phd thesis leiden observatory snellen schilizzi r.t de bruyn a.get al 1998 a&as 131 435 stanghellini c baum s.a odea c.p . morris 1990 233 , 379 stanghellini odea c.p dallacasa d baum s.a fanti 1998 131 , 303 stanghellini 2003 20 , 118s | 0.354572 | 2,293 | 1,175 | , Saving $0.1 in GPT-4. |
55,195 | we show the results of microsecond resolution radio data analysis focused on flux measurements of single pulses of psr b1133 + 16 .
the data were recorded at 4.85 ghz and 8.35 ghz with 0.5 ghz and 1.1 ghz bandwidth , respectively , using radio telescope effelsberg ( mpifr ) .
the most important conclusion of the analysis is , that the strongest single pulse emission at 4.85 ghz and 8.35 ghz contributes almost exclusively to the trailing part of the leading component of the pulsar mean profile , whereas studies at lower frequencies report that the contribution is spread almost uniformly covering all phases of the pulsar mean profile .
we also estimate the radio emission heights to be around 1%2% of the light cylinder radius which is in agreement with previous studies . additionally these observations allowed us to add two more measurements of the flux density to the psr b1133 + 16 broadband radio spectrum covering frequencies from 16.7 mhz up to 32 ghz .
we fit two different models to the spectrum : the broken power law and the spectrum based on flicker noise model , which represents the spectrum in a simpler but similarly accurate way .
[ firstpage ] pulsars : general pulsars : individual : b1133 + 16 | pulsar radio emission is still not explained in details but it is believed to originate close to the pulsar surface ( @xcite and references therein ) .
the analysis of 62 mean profiles of 23 pulsars at different frequencies regarding aberration and retardation effects yielded the estimation of the emission height being below 1500 km , but most probably being of the order of 500 km above the pulsar surface .
the advance in theoretical understanding of the emission mechanism and conditions in the magnetosphere have been driven mainly by the observations . recently
many observations have been concentrated on observing single pulses which carry detailed information about the physics of radio emission .
analysis of single pulse high resolution time series can show a few interesting properties i.e. giant and bright pulses , subpulse drift , nulling , microstructure , etc .
observations of single pulses for most pulsars may be carried out mainly at lower frequencies because pulsars are weak radio sources at high frequencies which is clearly seen in their steep spectra .
the pulsar spectrum in general can be described by a power law @xmath0 , where @xmath1 is the spectral index . the average spectral index for 266 pulsars for frequency spread from 0.4 to 23 ghz is @xmath2 @xcite .
.basic properties of psr b1133 + 16 @xcite .
[ cols= " < , < " , ]
the analysis of psr b1133 + 16 single pulses is a process that needs a certain amount of care .
first off all , it is important to take into account different observational and technical effects that can affect recorded data , especially with very high time resolution .
the effects that are presented in this paper play huge role and alter the data significantly . understanding of such effects and their influence on the data recording process is important for proper data reduction .
some of the effects are not visible in the mean profiles but only in single pulses data .
the mean profiles of psr b1133 + 16 at 4.85 ghz and 8.35 ghz consist of two components ( fig .
[ fig : mean.profiles ] ) .
the second component is emitted almost exclusively by low intensity individual pulses .
on the other hand , the first component is seen in single pulses regardless of their intensity except for the cases when it is not present at all .
however , lower intensity emission contributes mostly to the leading part of the first component whereas higher intensity single pulses contribute mainly to its trailing part ( fig . [
fig : intensity.ranges ] ) which was also reported by @xcite .
the results of analysis of 4.85 ghz and 8.35 ghz data are consistent with previous studies by @xcite at 430 mhz but studies of b0329 + 54 @xcite show entirely opposite behaviour without full explanation .
this inconsistency requires further studies of other pulsars single pulses to explain this effect .
we show , in contradiction to studies at lower frequencies by @xcite , who report an almost uniform spread of single pulse maxima , that the maximum emission of b1133 + 16 single pulses at 4.85 and 8.35 ghz contributes almost exclusively to the trailing part of the leading component of the mean profile .
our result is consistent with the behaviour at 341 , 626 , 1412 and 4850 hmz mentioned by @xcite and extends the studies up to 8.35 ghz .
radio emission arises close to the pulsar surface at the distances of around 65 stellar radii at frequencies of 4.85 ghz and 8.35 ghz .
weaker emission , which contributes to the leading part of the leading components , comes in earlier phases which suggests that originates in magnetosphere further from the pulsar surface than more energetic emission .
our calculations shows , that the difference of the emission heights for stronger and weaker emission is of the order of a few stellar radii , amounting to a change of 1%2% of the emission height , which is consistent with previous estimations @xcite .
there are 60 mean flux measurements in the literature of psr b1133 + 16 that are known to us .
they span a very wide radio frequency range from 16.7 mhz up to 32 ghz . to reproduce the spectrum we fitted two different models : the broken power
law model and one based on flicker noise model of pulsar radio emission @xcite .
surprisingly , the model proposed by @xcite is not widely used in the literature although it reproduces the pulsar spectrum comparably well to the power - law model .
future high time resolution observations might be useful to verify nano pulse emission model . due to the fact , that the pulsar radio emission is weaker at higher frequencies , giant pulses are the ones that can allow us to study closely their structure . in our both data samples
there is roughly one per cent of bright pulses that are at least ten times stronger than mean flux and their microstructure is clearly visible .
the presented results are based on the observations with the 100m telescope of the mpifr ( max - planck - institut fr radioastronomie ) at effelsberg .
this work has been supported by polish national science centre grants dec-2011/03/d / st9/00656 ( kk , as ) , dec-2012/05/b / st9/03924 ( om ) and dec-2011/02/a / st9/00256 ( jd ) . data analysis and figures were partly prepared using r @xcite . | pulsar radio emission not explained believed close to pulsar surface analysis of 62 mean profiles of 23 pulsars at frequencies emission height below 1500 km probably 500 km above pulsar surface advance theoretical understanding emission mechanism conditions driven by observations observations concentrated on single pulses information physics radio emission analysis of single pulse high resolution time series properties giant bright pulses subpulse drift nulling microstructure observations of single pulses pulsars at lower frequencies pulsars weak radio sources at high frequencies in steep spectra pulsar spectrum described by power law @xmath0 @xmath1 spectral index average spectral index for 266 pulsars for 0.4 to 23 ghz is @xmath2 @xcite properties of psr b1133 + 16 @xcite analysis of psr b1133 + 16 single pulses needs care observational technical effects recorded data high time resolution effects alter data understanding effects important for data reduction effects not visible in mean profiles single pulses data mean profiles of psr b1133 + 16 at 4.85 ghz and 8.35 ghz consist of two components ( figsecond component emitted by low intensity pulses first component seen in single pulses intensity lower intensity emission contributes to leading part first component higher intensity pulses contribute trailing part fig intensity reported by @xcite results 4.85 8.35 ghz consistent with studies @xcite at 430 mhz studies b0329 + 54 show opposite behaviour inconsistency requires further studies maximum emission of b1133 + 16 pulses at 4.85 8.35 ghz contributes to trailing part leading component consistent with behaviour at 341 626 1412 4850 hmz @xcite extends studies up to 8.35 ghz radio emission arises close to pulsar surface around 65 stellar radii at frequencies 4.85 8.35 ghz weaker emission contributes to leading part in earlier phases originates in magnetosphere further from pulsar surface difference emission heights for stronger weaker emission few stellar radii change of 1%2% emission height consistent with previous estimations @xcite 60 mean flux measurements in literature psr b1133 + 16 span radio frequency range 16.7 mhz up to 32 ghzreproduce spectrum fitted two models broken power law model flicker noise model pulsar radio emission @xcite model @xcite not widely used reproduces pulsar spectrum well future high time resolution observations might verify nano pulse emission model pulsar radio emission weaker at higher frequencies giant pulses study structure data samples roughly one per cent bright pulses ten times stronger than mean flux microstructure visible results based on observations 100m telescope mpifr effelsberg work supported by polish national science centre grants dec-2011/03/d st9/00656 dec-2012/05/b / st9/03924 dec-2011/02/a st9/00256 data analysis figures prepared using @xcite | 0.188021 | 1,178 | 600 | , Saving $0.0 in GPT-4. |
55,196 | we present new transit observations of the transiting exoplanet tres-3b obtained in the range 2009 2011 at several observatories .
the orbital parameters of the system were redetermined and the new linear ephemeris was calculated .
we performed numerical simulations for studying the long - term stability of orbits . | tres-3b is one of the more massive transiting extrasolar planets .
the planetary system consists of a nearby g - type dwarf and a massive hot jupiter with an orbital period of 1.3 days .
it was discovered by ( * ? ? ?
* odonovan et al . ( 2007 ) ) and a discovery - quality light curve has also been obtained by the superwasp survey ( ( * ? ? ?
* collier cameron et al . 2007 ) ) .
follow - up transit photometry has been presented by ( * ? ? ?
* sozzetti et al .
( 2009 ) ) and ( * ? ? ?
* gibson et al . (
all observations used in this study were carried out at the several observatories : star lesn ( slovakia ) , toru center for astronomy ( poland ) , michael adrian observatory ( germany ) , university observatory jena ( germany ) and vihorlat observatory ( slovakia ) .
we used telescopes with diameters of the primary mirrors in the range of 30120 cm and optical ccd - cameras ( ri - bands ) . in order to have a homogeneous dataset we have used the composition of 560 data points from two nights obtained at one observatory in the same filter ( see fig . 1 ) .
to obtain an analytical transit lc we assumed the quadratic limb darkening law .
the limb darkening coefficients @xmath0 and @xmath1 were linearly interpolated from ( * ? ? ?
* claret ( 2000 ) ) for the following stellar parameters : @xmath2 = 5650 k , @xmath3 ( g ) = 4.4 and [ fe / h ] = -0.19 ( based on the results of sozzetti et al .
finally , the orbital parameters of the system were redetermined : @xmath4 = 0.1819(20 ) , @xmath5 = 0.1785(70 ) , @xmath6 = 80.9(7 ) , @xmath7 = 0.876@xmath8 @xmath9 and @xmath10 = 1.551@xmath11 @xmath12 . to estimate the uncertainties of parameters we have used the monte carlo simulation method . based on the transits obtained at other observatories we calculated following new linear ephemeris for tres-3b : @xmath13 = 2454538.5807(1 ) + e @xmath14 1.30618595(15 ) . in this part of our work
, we investigate the gravitational influence of tres-3b on a potential second planet in the system .
* gibson et al . ( 2009 ) ) claimed , that their data are sensitive enough to probe small mass planets on circular orbits near the 2:1 mean - motion resonance ( mmr ) with tres-3b .
we performed numerical simulations for studying the stability of orbits and checking their chaotic behavior using the method of maximum eccentricity ( e.g. ( * ? ? ? * dvorak et al .
we have generated @xmath15 massless particles to represent small planets ( earth - like ) in this system .
the long - term stability plot in @xmath16 plane showing the maximum eccentricity after 500 years ( about 140 000 revolutions of tres-3b around the parent star ) is shown on fig.[f1 ] ( right plot ) .
we found that the region from @xmath17au to @xmath18au is almost completely depleted , excepted the regions near 1:1 and 3:4 mmrs .
the region near 2:1 mmr is richly populated , but the particles have relatively high eccentricities and inclinations , thus the stability and also the detection using the ttv method is questionable .
more stable regions are beyond the 1:3 mmr , where the gravitational influence of the tres-3b is weak
. more detailed study of the long - term stability in this system ( especially near 2:1 mmr ) will be presented in the future work . + * acknowledgements * this work has been supported by grant , vega no .
2/0078/10 , 2/0074/09 , 2/0094/11 .
pd received the support from apvv grant lpp-0049 - 06 and lpp-0024 - 09 .
tk thanks the grant ga r gd205/08/h005 .
gm acknowledges iuventus plus grant ip2010 023070 .
sr thanks the german national science foundation ( dfg ) for support in project ne 515 / 33 - 1 . | tres-3b massive transiting extrasolar system nearby g - type dwarf massive hot jupiter orbital period 1.3 days discovered by odonovan et al 2007 discovery - quality light curve obtained by superwasp survey collier cameron 2007 follow - up transit photometry presented by sozzetti et al . 2009 gibson et al . observations at observatories star lesn toru center astronomy michael adrian university observatory jena vihorlat observatory used telescopes 30120 cm optical ccd - cameras homogeneous dataset used 560 data points from two nights one observatory same filter fig . 1 transit lc assumed quadratic limb darkening law limb darkening coefficients @xmath0 @xmath1 linearly interpolated from claret 2000 for stellar parameters @xmath2 = 5650 k @xmath3 g = 4.4 fe / h = -0.19 results sozzetti orbital parameters redetermined @xmath4 = 0.1819(20 ) @xmath5 = 0.1785(70 ) @xmath6 = 80.9(7 ) @xmath7 = 0.876@xmath8 @xmath10 = 1.551uncertainties used monte carlo simulation method transits observatories calculated linear ephemeris for tres-3b @xmath13 = 2454538.5807(1 ) + @xmath14 1.30618595(15 ) gravitational influence tres-3b on potential second planet gibson et al data sensitive probe small mass planets orbits near 2:1 mean - motion resonance with tres-3b performed numerical simulations stability chaotic behavior maximum eccentricity generated @xmath15 massless particles small planets long - term stability plot @xmath16 plane maximum eccentricity after 500 years 140 000 revolutions tres-3b star fig.[f1 region from @xmath17au to @xmath18au depleted excepted near 1:1 3:4 mmrs region near 2:1 mmr richly populated particles high eccentricities inclinations stability detection questionable stable regions beyond 1:3 mmr gravitational influence tres-3b weak detailed study - stability near 2:1 mmr future work work supported by grant vega 2/0078/10 2/0074/09 2/0094/11 received support from apvv grant lpp-0049 - 06 lpp-0024 - 09thanks grant/h005 acknowledges iuventus grant 023070 thanks german national science foundation support project 515 / 33 - 1 | 0.180467 | 1,028 | 553 | , Saving $0.0 in GPT-4. |
55,206 | accurate to the first order in the uniform angular velocity , the general relativity frame dragging effect of the moments of inertia and radii of gyration of two kinds of neutron stars are calculated in a relativistic @xmath0 model .
the calculation shows that the dragging effect will diminish the moments of inertia and radii of gyration . | as we knows , neutron stars may be created in the aftermath of the gravitational collapse of the core of a massive star at the end of its life , or may be created when the mass of a accreting white dwarf exceed the chandrasekhar limit .
the typical character of neutron stars may be described as following : spinning rapidly , often making several hundred rotations per second ; having a small stature with a radius @xmath1 of @xmath2 km and a heavy avoirdupois with a mass @xmath3 on the order of 1.5 solar masses @xmath4 ; and having a very compact central density as high as several times the nuclear saturation density .
neutron stars are one of the densest forms of matters in the observable universe .
+ one of the important global property of neutron star is the moment of inertia @xmath5 .
eary estimates of the moment of inertia by the energy -loss rate from pulsars spanned a wide range of @xmath5 @xcite , and several researchers have given a lower bound on the moment of inertia of the pulsar as @xcite @xmath6 .
recently , kalogera@xcite etc . also studied the bound of the moment of inertia . in order to investigate the general relativistic frame dragging effect on the moments of inertia of neutron stars ,
the moments of inertia of the non - rotating and rotating at the kepler frequency neutron stars will be studied in this work .
+ here we adopt the metric signature - + + + , g = c=1 .
in relativity , the space - time geometry of a static spherical symmetric star can be described as @xmath7 and the space - time geometry of a rotating star in equilibrium is described by a stationary and axisymmetric metric of the form @xmath8 where @xmath9 is the angular velocity of the local inertial frame and is proportional to the star s rotational frequency @xmath10 , which is the uniform angular velocity of the star relative to an observer at infinity .
+ from the @xmath11 component of einstein field equations , accurate to the first order in @xmath10 , one gets@xcite @xmath12 where@xmath13^{\frac{1}{2}}$ ] , @xmath14 , which denotes the angular velocity of the fluid relative to the local inertial frame .
the boundary conditions are imposed as @xmath15 at the center , @xmath16 , where @xmath17 is chosen arbitrarily . integrating eq.(3 )
outward from the center of the star , one would get the function @xmath18 . outside the star , from eq.(3 )
one has @xmath19 , where @xmath20 is the total angular mementum of the star , which takes the form@xcite @xmath21 , where @xmath1 is the surface radius of the star . according to the traditional definition , we have @xmath22 where @xmath5 is the moment of inertia . at the neutron star s surface , in term of eqs.(3 ) and
( 4 ) , one can get @xmath23 from the ( 0,0 ) and ( 1,1 ) components of einstein field equations to the static spherically symmetric star , one gets@xcite @xmath24 @xmath25 accurate to the first order in @xmath10 , substituting eqs .
( 6 ) and ( 7 ) into eq .
( 5 ) , one has@xcite @xmath26 this is just the approximate expression of the moment of inertia accurate to the first order in @xmath10 . to the non - rotating neutron stars , @xmath27 . as a first order approximation
, it is means that there is only consideration of the frame dragging effect and without consideration of the deformation caused by the rotation@xcite . according to eq.(8 ) , we can estimate that , to the same neutron star , the moment of inertia of the non - rotating star will be bigger than that of the rotating star , which will be testified in our later numerical calculation .
this is just the frame dragging effect to the moment of inertia .
approximating in week field , according to eqs .
( 6 ) and ( 7 ) , one gets @xmath28 in newtonian theory , @xmath29 , so from eq.(5 ) , we have @xmath30 this is just the calculating expression of the moment of inertia of a spherical symmetric star in newtonian theory .
+ similar to the newtonian theory , in general relativity , we also define the radius of gyration of a spherical symmetric star as @xmath31
there are several models to deal with the superdense matters , such as non - relativistic models , relativistic field theoretical models@xcite . to different models ,
the fractions of particles in the superdense matters are different , and then the bulk properties of superdense matters are different , that is , the eos of them are different@xcite . in this work ,
two kind of eos will be employed@xcite , one is for the traditional neutron stars ( tns ) , in which @xmath32 are the main elements ; the other is for the hyperon stars ( hs ) , in which @xmath33 are the main elements . the eos will be considered in the relativistic @xmath0 model@xcite .
fig.(1 ) shows the eos of the ths and hs .
+ by using above eos and eqs.(8 ) and ( 11 ) , the moments of inertia and radii of gyration of neutron stars were calculated .
fig.(2 ) presents the moment of inertia of non - rotating and rotating ( at the kepler frequency ) neutron stars as a function of central density .
it is clear that , because of the frame dragging effect , as a first order approximation , the moments of inertia of the non - rotating tns ( or hs ) are bigger than that of the corresponding rotating stars at the same central density . from this figure , one can also see that , with the same central density , as the star rotating at its kepler frequency , the frame dragging effect on the moment of inertia of tns is bigger than that of hs .
the reason of this is that the eos of tns is stiffer than the eos of hs , so there is a bigger kepler frequency of tns at the same central density , and then corresponds a bigger frame dragging effect . in order to show the frame dragging effect of the moments of inertia entirely , we present them in a three - dimension figure , fig.3 just presents the moments of inertia of tns as a function of the central densities and the central angular velocities relative to the local inertial frame .
+ fig.4 gives the radii of gyration of non - rotating and rotating ( at the kepler frequency ) neutron stars as a function of central density . from this figure
one can see that , to the same central density , there is a big gap between the lines of the non - rotating and rotating stars , that is to say , there is a remarkable frame dragging effect to the radii of gyration , and as the central density increasing , the effect increases .
we also present the radii of gyration of tns in a three - dimension figure , see fig.5 .
+ from our calculation , one can see that to a star at the same central density , the frame dragging effect will diminish its moment of inertia and radius of gyration , and as the star rotates faster , the dragging effect will become more obviously .
the project supported by national natural science foundation of china ( grant no . of 10275099 , 10175096 ) .
+ 1 m. ruderman , _ ann .
astrophys . _
* 10 * , 427 ( 1972 ) . v. trimble and m. rees ,
_ astrophys .
lett . _ * 5 * , 93 ( 1970 ) . g. borner and j. m. cohen , _ astrophys .
j. _ * 185 * , 959 ( 1973 ) .
v. kalogera and d. psaltis , _ phys .
d _ * 61*,024009 ( 2002 ) .
j. b. hartle , _ ap . j._*150*,1005 ( 1967 ) .
s. chandrasekhar , _ ap . j._*140*,417 ( 1964 )
. j. l. fredman j. r. ipser and l. parker , _ ap .
j._*304*,115 ( 1986 ) . | neutron stars created gravitational collapse core massive star or mass accreting white dwarf chandrasekhar limit typical character neutron stars spinning rapidly several hundred rotations per second small stature radius @xmath2 km heavy mass 1.5 solar masses compact central density times nuclear saturation density neutron stars densest matters in observable universe important global property neutron star moment of inertia estimates inertia energy -loss rate pulsars spanned wide range @xmath5 @xcite researchers lower bound moment inertia pulsar as @xcite @xmath6 kalogera@xcite studied bound moment of inertia investigate relativistic frame dragging effect on moments of inertia neutron stars non - rotating rotating at kepler frequency neutron stars studied adopt metric signature - +, g = c=1 space - time geometry of static spherical symmetric star described as @xmath7 geometry rotating star in equilibrium stationary axisymmetric metric form @xmath8 @xmath9 is angular velocity of local inertial frame proportional to star s rotational frequency @xmath10 uniform star observer@xmath11 einstein field equations accurate first order @xmath10 gets@xcite @xmath12@xmath13 @xmath14 angular velocity fluid local inertial frame boundary conditions @xmath15 center @xmath16 @xmath17 chosen arbitrarily integrating eq.(3 ) center star function @xmath18 outside star eq.(3 ) @xmath19 @xmath20 total angular mementum star @xmath21 @xmath1 surface radius star traditional definition @xmath22 @xmath5 moment inertia neutron star s surface eqs.(3 ) ( 4 ) @xmath23 ( 1,1 ) einstein field equations static spherically symmetric star@xcite @xmath24 @xmath25 first order @xmath10 substituting eqs ( 6 ) ( 7 ) eq . ( 5 )@xcite @xmath26 approximate expression moment of inertia first order @xmath10 non - rotating neutron stars @xmath27 first order approximation consideration frame dragging effect deformation rotation eq.(8 estimate same neutron star moment of inertia non - rotating star bigger than rotating star testified later numerical calculation frame dragging effect moment inertia approximating week field eqs ( 6 ) ( 7 ) gets @xmath28 newtonian theory @xmath29 from eq.(5 ) @xmath30 calculating expression moment of inertia spherical symmetric star newtonian theory general relativity define radius of gyration spherical symmetric star as @xmath31 several models deal superdense matters non - relativistic relativistic field theoretical models fractions of particles different bulk properties eos different two eos employed traditional neutron stars @xmath32 main elements hyperon stars @xmath33 main elements eos considered in relativistic @xmath0 model fig.(1 ) shows eos of ths and hs using eos eqs.(8 ) ( 11 ) moments of inertia radii of gyration of neutron stars calculated fig.(2 ) presents moment of inertia non - rotating rotating neutron stars function of central densityclear frame dragging effect moments of inertia of non - rotating tns or hs ) bigger than rotating stars at same central density same central density star rotating at kepler frequency frame dragging effect on moment inertia of tns bigger than hs eos of tns stiffer than hs bigger kepler frequency at same central density bigger frame dragging effect present in three - dimension figure fig.3 moments of inertia of tns function of central densities central angular velocities local inertial frame fig.4 radii of gyration of non - rotating and rotating neutron stars function central density same central density big gap between lines non - rotating and rotating stars remarkable frame dragging effect to radii of gyration central density increasing effect increases present radii of gyration of tns in three - dimension figure fig.5 star at same central density frame dragging effect moment of inertia and radius of gyration star rotates faster dragging effect project supported by national natural science foundation of china ( grant no 10275099 , 10175096 ) ruderman ,astrophys 10 427 1972 ) trimble m. rees astrophys 5 93 ( 1970 ). borner m. cohen astrophys 185 959 ( 1973 ) kalogera d. psaltis 61*,024009 ( 2002 ) b. hartle j*150*,1005 ( 1967 ) chandrasekhar*140*,417 ( 1964 ) fredman ipser l. parker*304*,115 ( 1986 ) | 0.313281 | 1,827 | 935 | , Saving $0.1 in GPT-4. |
55,232 | we modify the penna model for biological aging , which is based on the mutation - accumulation theory , in order to verify if there would be any evolutionary advantage of triploid over diploid organisms .
we show that this is not the case , and that usual sex is always better than that involving three individuals . | the reasons why sexual reproduction appeared , hundreds of million years ago , are still under investigation today @xcite .
one of the reasons commonly used to justify sex is the higher genetic diversity it produces if compared to asexual reproduction . on the other hand ,
sex is closely related to aging : all sexual species present aging , which starts as soon as reproduction starts @xcite .
because of its ability to address both questions in a single context , the penna model for biological aging @xcite has been extensively used in recent years to study why sex evolved ( for a review see @xcite ) . s martins and moss de oliveira , for instance , have shown that the higher diversity of sexual populations can prevent extinction if a catastrophe damaging the genome of the individuals ( such as exposure to radiation ) occurs @xcite .
diversity was also the key issue discussed by s martins @xcite in his simulation of the competition of sexual and asexual varieties of a single species , coevolving under the assault of parasitic infestation .
the density of parasites was shown to drive a transition between asexual and sexual orders , in agreement with the so - called red queen hypothesis for the evolution of sex @xcite ) .
the above mentioned results show that the model provides sexual varieties with survival advantages , due to their larger genetic diversity , over the asexual ones , whenever they compete in a changing environment .
nevertheless , in the absence of competition and a mutant environment , the asexual populations resulting from simulations of the standard penna model are always much larger , since they generate twice as many offspring for the same population size .
this effect is not overcome by the need , in diploid individuals , for homozygose if the harmful allele is to become active in loci for which it is not dominant . without catastrophes or parasites ,
if a sexual and an asexual population are allowed to evolve under the same immutable environment , competing for the same food resources , the asexual one will win and the sexual one disappears ( due to the verhulst logistic factor explained in section two ) . only very recently has a new ingredient been introduced @xcite into the penna model allowing larger sexual populations to be obtained , with sizes comparable to the asexual ones .
we now use this modified penna model to address a tantalizing issue : since a diploid sexual population has the upper hand when competing against an asexual one due do the diversity generated by the use of genetic material coming from two different parents , why does not nature enhance this effect by allowing the genome of the offspring to benefit from three different templates ? is the need for homozygose mentioned above enough to overcome the burden of using three individuals to generate one offspring ? to provide answers to these questions , we compare two different kinds of sexual populations : one involving the mating of two diploid individuals as parents ( normal sexual reproduction ) and the other involving three triploid individuals .
the paper has the following structure : in section 2 we describe the model , first for diploids and then for triploids , in section 3 we present our results and in section 4 , our conclusions .
each individual of the population is represented by a `` chronological genome '' , which consists of two bit - strings of 32 bits ( 32 loci or positions ) each , that are read in parallel .
one string contains the genetic information inherited from the mother and the other , from the father .
each position of the bit - strings is associated to a period of the individual s life , which means that each individual can live at most for 32 periods ( `` years '' ) .
each step of the simulation corresponds to reading one new position of all individuals genomes .
diseases are represented by bits 1 .
if an individual has two bits 1 in the @xmath0-th position of both bit - strings ( homozygote ) , it will start to suffer the effects of a disease at his @xmath0-th year of life .
if the individual is homozygous with two bits zero , no disease appears in that age .
if the individual is heterozygous in that position , he will become sick only if that locus is one for which the harmful allele is dominant .
the dominant loci are randomly chosen at the beginning of the simulation and remain fixed . if the actual number of accumulated diseases reaches a threshold @xmath1 , the individual dies . in order to avoid an exponential increase of the population and to introduce a dispute for food and space ,
a logistic verhulst factor is used .
every time step and for each individual , a random number between zero and one is generated .
this number is compared with @xmath2 , where @xmath3 is the actual size of the population and @xmath4 is the carrying capacity . if the random number is smaller than @xmath5 , the individual dies , independent of its age or genome . for a discussion of other alternatives of implementation of this mean - field - like interaction between individuals of the population
we direct the reader to ref .
@xcite . if a female succeeds in surviving until the minimum reproduction age @xmath6 , it generates @xmath7 offspring every period until death .
the female chooses randomly a male to mate , with age also greater than or equal to @xmath6 .
the offspring genome is constructed from the parents ones ; first , the strings of the mother are cut in two at a randomly chosen position ( `` crossing '' ) and two complementary pieces , one from each string , are recombined to generate the female gamete ( one string of 32 bits ) . @xmath8
deleterious mutations are then randomly introduced .
the same process occurs with the father s genome and the union of the two resulting gametes forms the new genome .
the sex of the baby is randomly chosen , with equal probability .
deleterious mutation means that if a bit 0 is randomly chosen in the parent s genome , it is set to 1 in the offspring genome . however ,
if a bit already set to 1 is randomly chosen , it remains 1 in the offspring genome ( no back mutations ) .
the description given above corresponds to the original sexual version of the penna model @xcite .
the new ingredient , mentioned in section 1 and introduced by s martins and stauffer @xcite , consists in assuming that harmful mutation reduces the survival probability .
thus , at each iteration , or `` year , '' each individual survives with probability @xmath9 if it has a total of @xmath10 harmful mutations ( taking into account dominant positions ) in it s full genome ( it is killed if a random number is tossed that is smaller than the survival probability ) .
@xmath11 is a parameter of the simulation , fixed from the start . to summarize ,
an individual may now die for any one of three reasons : i ) randomly , due to the verhulst logistic factor ; ii ) if its actual number of accumulated diseases reaches the limit @xmath1 ; iii ) due to its survival probability being too small . in this case , we assume that mating involves three triploid individuals ( two males and one female or vice - versa ) .
the chronological genomes consist of three bit - strings that are read in parallel .
homozygous positions are those with three equal bits at homologous loci .
harmful mutations are active only if there are three bits 1 at that same position , or at a heterozygous locus at which harmful mutations are dominant .
only females generate offspring .
there are random crossing and recombination to produce the offspring genome ( see fig .
1 ) , and deleterious mutations are randomly introduced in each of the three gametes .
the baby is a male or a female , with equal probability . the first question we want to investigate relates to the competition between the benefit provided by a triploid genome and the effort involved in a mating that needs three individuals , instead of two , to generate the offspring .
the benefit is the fact that , for triploids , mutations that happen in loci where the harmful allele is not dominant need to appear in all three bit - strings to become active .
the curves presented below correspond to the average of the results obtained for 20 different populations ( 20 different initial seeds for the random number generator ) , using the following parameters : initial population = 10,000 ; maximum population size @xmath12 individuals ; maximum number of genetic diseases @xmath13 ; minimum reproduction age @xmath14 ; birth rate @xmath15 ; mutation rate @xmath16 per bit - string ( or gamete ) ; number of dominant positions @xmath17 ; decrease in survival probability @xmath18 ; total number of monte carlo steps = 800,000 ; in figure 2 we present the time evolution of a diploid population ( upper curve ) and of two triploid populations , one for which reproduction involves the mating of one male and two females ( central curve ) and the other corresponding to the mating of two males and one female ( lower curve ) . in figure 3 we present the corresponding dimensionless survival rates @xmath19 as a function of age @xmath20 , @xmath21 , where @xmath22 is the number of individuals with age @xmath23 and @xmath24 is the number of individuals with age @xmath20 . from these figures
we can see that the diploid population is not only larger than the other two , but also presents a higher survival probability
. however , it may be not enough to guarantee that the diploid population is better than the triploid ones .
a second important measure , as mentioned in the introduction , is the genetic diversity of the populations .
it has already been shown that the survival probability of a sexual population is the same as that of a diploid asexual population that reproduces by meiotic parthenogenesis .
however , the larger genetic diversity of the sexual population may prevent it from extinction if , for example , exposure to radiation occurs [ 5 ] .
the genetic diversity is calculated by measuring the hamming distance , in this case defined by the number of different loci ( bits ) between the genomes , for all pairs of individuals .
the probability distribution of these distances is obtained by making a histogram of the fraction of pairs , out of all possible pairs in the population , that present a given hamming distance , normalized by its maximum possible value ( 64 for diploids and 96 for triploids ) .
figure 4 shows the resulting distributions for the diploid and triploid populations .
it is clear that the diploid population presents both a larger mean distance between pairs , indicated roughly by the position of the peak of the distribution , and a larger variance , measured by the width at half the maximum height of the curves .
this assures a larger diversity within its genome space for the diploid populations , and will give them the upper hand when competing against the triploids under the pressure of a rapidly mutating environment .
it is worth while mentioning that the results are essentially the same if we allow a double crossing of the triploid genome during reproduction , and there is no benefit for the triploids to ensure that the offspring have their genetic material gathered from all three parents .
we generalized the sexual version of the penna model for biological aging to simulate also triploid populations . in these populations individuals
present three sets of homologous chromosomes .
a harmful mutation needs to appear in all of them to become active or to be in a locus where the harmful allele is dominant .
we showed that normal diploid sexual populations have higher survival rates and are larger than the triploid ones , for the same carrying capacity of the environment .
the genetic diversities of the populations , measured by the hamming distance distribution , were also compared .
the result shows that the diploid population presents a higher genetic diversity than the triploid ones .
we may thus conclude that , concerning survival probabilities , population sizes and diversity , usual sex is better than that involving three individuals .
sexual diploid populations would be favored in direct competition with triploids either in a stable or a mutating environment , and we claim that this is the reason for normal sex to have been chosen by evolution as the dominant reproduction strategy .
these results mean that the fact that triploids need mutations to appear in all three sets of homologous cromossomes to become effective is not enough to overcome the effort of a mating involving three individuals , and that more is not necessarily better insofar as genetic diversity is concerned . 99 s. dasgupta , physica a 298 ( 2001 ) 465 . l. partridge and n.h .
barton , nature 362 ( 1993 ) 305 .
penna , j. stat .
78 ( 1995 ) 1629 .
s. moss de oliveira , p.m.c . de oliveira and d. stauffer , _ evolution , money , war and computers _ ,
teubner , leipzig ( 1999 ) .
s martins and s. moss de oliveira , int .
c 9 ( 1998 ) 421 .
s martins , phys .
e 61 ( 2000 ) 2212 .
b. wuethrich , science 281 ( 1998 ) 1982 .
s martins and d. stauffer , physica a 294 ( 2001 ) 191 . j.s .
s martins and s. cebrat , theor .
119 ( 2000 ) 156 . | reasons sexual reproduction appeared hundreds million years ago still under investigation today sex is higher genetic diversity to asexual reproduction sex related to aging all sexual species present aging starts reproduction starts both questions penna model for biological aging used to study why sex evolved martins oliveira shown higher diversity of sexual populations prevent extinction if exposure to radiation diversity key issue discussed by s martins in simulation competition of sexual and asexual varieties under parasitic infestation density of parasites transition between asexual and sexual orders agreement with red queen hypothesis for evolution of sex results show model provides sexual varieties survival advantages due to larger genetic diversity over asexual in changing environment in absence of competition mutant environment asexual populations from simulations standard penna model larger generate twice as many offspring for same population size effect not overcome by need in diploid individuals for homozygose if harmful allele active in loci not dominantwithout catastrophes parasites if sexual and asexual population evolve under same environment competing for food resources asexual sexual disappears due to verhulst logistic factor section two recently new ingredient introduced into penna model larger sexual populations sizes comparable to asexual modified penna model address issue diploid sexual population upper hand against asexual due diversity by genetic material from two parents why nature enhance effect genome offspring benefit from three different templates ? need for homozygose enough overcome burden of using three individuals one offspring ? compare two sexual populations one mating two diploid individuals normal three triploid individuals paper structure section 2 model diploids triploids section 3 results section 4 conclusions each individual represented by `` chronological genome two bit - strings of 32 bits read in parallel one string contains genetic information from mother other father each position associated to period individual s life each individual can live most for 32 periods each step simulation corresponds to reading one new position genomes diseases represented by bits 1 .individual has two bits 1 in @xmath0-th position ( homozygote ) effects disease at @xmath0-th year life if homozygous with two bits zero no disease if heterozygous sick only if harmful allele dominant dominant loci randomly chosen remain fixed if accumulated diseases reaches threshold @xmath1 individual dies avoid increase population dispute for food space logistic verhulst factor used random number between zero and one generated compared with @xmath2 @xmath3 is size population @xmath4 carrying capacity if random number smaller than @xmath5 individual dies independent of age or genome discussion alternatives reader to ref . @xcite female until minimum reproduction age @xmath6 generates @xmath7 offspring until death female chooses randomly male to mate with age greater than or equal to @xmath6 offspring genome constructed from parents ones strings mother cut in two two pieces recombined to generate female gamete 32 bits @xmath8 deleterious mutations randomly introduced same process with father s genome union of two gametes forms new genome sex baby randomly chosen equal probabilitydeleterious mutation bit 0 randomly chosen in parent s genome set to 1 in offspring genome if bit set to 1 chosen remains 1 offspring genome no back mutations description corresponds to original sexual version penna model @xcite new ingredient martins stauffer harmful mutation reduces survival probability each iteration individual survives if @xmath10 harmful mutations in full genome killed if random number smaller than survival probability @xmath11 parameter of simulation fixed from start individual may die for three randomly to verhulst logistic factor if accumulated diseases reaches limit @xmath1 survival probability too small mating involves three triploid individuals two males one female chronological genomes three bit - strings read parallel homozygous positions with three equal bits at homologous loci harmful mutations active if three bits 1 at same position or heterozygous locus females generate offspring random crossing recombination produce offspring genome fig . 1 deleterious mutations randomly introduced in each three gametes baby male or female equal probabilityfirst question competition between benefit triploid genome and effort mating three individuals instead two offspring benefit is for triploids mutations in harmful allele not dominant need appear in all three bit - strings to active curves correspond average results for 20 populations parameters initial population = 10,000 maximum population size @xmath12 maximum genetic diseases @xmath13 minimum reproduction age @xmath14 birth rate @xmath15 mutation rate @xmath16 per bit - string number dominant positions @xmath17 decrease survival probability @xmath18 total monte carlo steps = 800,000 ; figure 2 time evolution of diploid population and two triploid populations one reproduction mating one male two females other two males one female figure 3 dimensionless survival rates @xmath19 function of age @xmath20 @xmath21 @xmath22 age @xmath23 and @xmath24 age @xmath20 diploid population larger presents higher survival probability not enough to guarantee diploid population better than triploid second important measure genetic diversity of populations survival probability of sexual population same as diploid asexual population by meiotic parthenogenesislarger genetic diversity of sexual population may prevent extinction if exposure to radiation genetic diversity calculated by measuring hamming distance defined by number different loci between genomes for all pairs probability distribution distances obtained by histogram of fraction of pairs given hamming distance normalized by maximum value ( 64 for diploids 96 for triploids figure 4 shows distributions for diploid and triploid populations diploid population presents larger mean distance between pairs larger variance measured by width at half maximum height curves assures larger diversity genome space for diploid populations upper hand competing against triploids rapidly mutating environment results same if double crossing of triploid genome during reproduction no benefit for triploids offspring genetic material from all three parents generalized sexual version of penna model for biological aging to simulate triploid populations three homologous chromosomes harmful mutation needs in all normal diploid sexual populations have higher survival rates larger than triploid for same carrying capacity genetic diversities measured by compared shows diploid population presents higher genetic diversity than triploidconclude survival population sizes diversity usual sex better than three individuals diploid populations competition triploids stable or mutating environment normal sex dominant reproduction strategy results triploids need mutations three cromossomes not enough mating three more not better genetic diversity dasgupta 298 ( 2001 ) 465 partridge barton nature 362 ( 1993 ) 305 penna j stat . 78 ( 1995 ) 1629 moss de oliveira oliveira stauffer evolution money war computers teubner leipzig ( 1999 ) martins oliveira int . c 9 ( 1998 ) 421 martins phys . 61 ( 2000 ) 2212 wuethrich science 281 ( 1998 ) 1982 martins stauffer physica 294 ( 2001 ) 191 martins cebrat theor . 119 ( 2000 ) 156 | 0.440646 | 2,843 | 1,413 | , Saving $0.1 in GPT-4. |
55,235 | very recently a new enhancement around 4.05 gev was observed by belle experiment . in this short note , we discuss some possible assignments for this enhancement , i.e. @xmath0 and @xmath1 molecular state . in these two assignments
, @xmath2 can decay into @xmath3 with comparable branching ratio with that of @xmath4 .
thus one suggests high energy experimentalists to look for @xmath2 in @xmath3 channel .
furthermore one proposes further experiments to search missing channels @xmath5 , @xmath6 and especially @xmath7 and @xmath8 , which will be helpful to distinguish @xmath0 and @xmath1 molecular state assignments for this new enhancement .
very recently belle collaboration observed an enhancement with mass @xmath9 mev and width @xmath10 mev besides confirming @xmath11 by studying initial state radiation ( isr ) process @xmath12 @xcite .
belle experiment also indicated that a fit using two interfering breit - wigner shapes describes the data better than one that uses only the @xmath11 @xcite . in this work
, we named this new structure as @xmath2 . recently a series of observations of charmonium like states @xmath13 , @xmath14 , @xmath15 @xcite is challenging our understanding for non - perturbative qcd . at present how to understand
this new structure is one of intriguing and challengeable topics . in this short note , we are dedicated to the discussion of the possible interpretations for @xmath2 . | in the known charmonium states listed in particle data book , only the mass of @xmath16 is close to that of @xmath2 @xcite . at present @xmath16
is usually considered as the candidate for @xmath0 .
the central value of width of @xmath2 is larger than that of @xmath16 around 100 mev .
however , due to the large error given by belle experiment , the mass and width of this new enhancement are consistent with that of @xmath16 . for @xmath2 ,
belle experiment also gave @xmath17 ev and @xmath18 ev corresponding to two solutions in fitting the data @xcite .
as the candidate of @xmath0 , the decay width of @xmath19 is @xmath20 kev @xcite . using the above values , we can roughly estimate @xmath21=5.8\times 10^{-3}$ ] and @xmath22 for the above two solutions if @xmath2 is @xmath0 state .
due to the large experimental error , the central value of the former one is not contradict the upper limit of the branching ratio of @xmath23 ( @xmath24<4\times 10^{-3}$ ] ) though the former one is slightly larger than the upper limit of the branching ratio of @xmath23 . at present , only @xmath4 are reported by belle @xcite .
if @xmath2 is @xmath0 , @xmath25 $ ] is comparable with @xmath21 $ ] . thus @xmath2 can be found in @xmath3 channel .
although at present the experiments did not give the measurement for @xmath26 , the transition of @xmath0 to lower states @xmath27 @xmath28 with two pions being emitted can be solved by the qcd multipole expansion ( qcdme ) method proposed by gottfried , yan and kuang @xcite , which is depicted by fig . [ me ] . in a recent work @xcite , ke et al .
calculated the transitions of @xmath29 , and obtain @xmath30=589.91\;\rm{kev},\\ & & \gamma[\psi(3s)\to \psi(2s)\pi\pi]=14.96\;\rm{kev } , % & & b[\psi(3s)\to j/\psi\pi^+\pi^-]=**,\\ % & & b[\psi(3s)\to j/\psi\pi^0\pi^0]=**,\\ % & & b[\psi(3s)\to \psi(2s)\pi^+\pi^-]=**,\\ % & & b[\psi(3s)\to \psi(2s)\pi^0\pi^0]=**,\end{aligned}\ ] ] by adopting the cornell potential @xmath31 @xcite and @xmath30=12.38\;\rm{kev},\\ & & \gamma[\psi(3s)\to \psi(2s)\pi\pi]=8.84\;\rm{kev } , % & & b[\psi(3s)\to j/\psi\pi^+\pi^-]=**,\\ % & & b[\psi(3s)\to j/\psi\pi^0\pi^0]=**,\\ % & & b[\psi(3s)\to \psi(2s)\pi^+\pi^-]=**,\\ % & & b[\psi(3s)\to \psi(2s)\pi^0\pi^0]=**,\end{aligned}\ ] ] by adopting a modified cornell potential which includes a spin - related term @xcite @xmath32 where @xmath33 and @xmath34 is the zero - point energy ( for more detail , see ref .
the above numerical results by two potential show that there exists large uncertainty for the estimate of @xmath35 by qcdme method , which is indicated in ref
however the estimate of @xmath36 without spin - related term is consistent with that with spin - related term .
if we trust the estimate of @xmath36 by qcdme method , it is hopeful to search @xmath2 in @xmath37 channel in future experiments .
furthermore , if @xmath2 is @xmath0 , we know that @xmath38 is not its main decay channel .
@xmath2 can mainly decay into @xmath5 and @xmath6 .
due to the fact that @xmath2 is of wide decay width with about 200 mev , @xmath2 can also decay into @xmath1 through its mass tail .
there has been a long history about the molecular structure of hadrons .
to explain some phenomena which are hard to find natural interpretations in the canonical framework , people have tried to search for new structure beyond it .
the molecular structure is one of the possible candidates . because the mass of @xmath2 is close to the threshold of @xmath1 , and @xmath2 is of about 200 mev wide width , thus @xmath2 can be assumed as a @xmath1 molecular state . in the history , okun and
voloshin studied the interaction between charmed mesons and proposed possibilities of the molecular states involving charmed quarks @xcite .
rujula , geogi and glashow once suggested @xmath16 as a @xmath1 molecular state @xcite . in ref .
@xcite , dubynskiy and voloshin proposed that there exists a possible new resonance at the @xmath1 threshold . because @xmath2 is observed along with @xmath11 which is of @xmath39 , thus the most possible quantum number of @xmath2 is @xmath39 .
furthermore @xmath2 must be a p - wave @xmath1 . at present
one can not use the experimental information to determine the quantum number @xmath40 of @xmath2 .
thus @xmath2 can be isosinglet state with @xmath41 or isovector state with @xmath42 .
if @xmath2 is a @xmath1 molecular state , @xmath2 falls apart into @xmath1 by its mass tail , which is depicted in fig .
[ fallapart ] . in the following
we will discus its other possible decay modes .
\(i ) @xmath2 as an isoscalar @xmath1 molecular state by the @xmath1 recattering effect , @xmath2 with @xmath43 can decay into @xmath44 , @xmath45 and @xmath46 by the mechanism depicted in fig .
[ eta ] , and into @xmath47 by fig .
[ omega ] . here
@xmath48 can be also replaced as @xmath49 and @xmath50 .
+ + + by the same mechanism , @xmath2 also decay into @xmath51 and @xmath52 by exchanging @xmath53 and @xmath54 mesons between @xmath55 and @xmath56 .
in fact , as secondary decay , the branching ratio of @xmath57 is comparable with that of @xmath58 . because @xmath59 and @xmath60 dominantly decay into @xmath61 , thus , according to the isospin symmetry , one can roughly estimate @xmath62}{b[y(4008)\to j/\psi\pi^+\pi^-]}\sim\frac{1}{2}.\end{aligned}\ ] ] furthermore the decay mechanism depicted by fig .
[ eta ] can be test in further experiments by analyzing the @xmath61 invariant mass spectrum .
if this mechanism is correct , the @xmath61 invariant mass distribution should show the signature of @xmath59 or @xmath63 .
the branching ratio of @xmath64 is almost 89.1% , thus @xmath65 to @xmath66 is overwhelming . @xmath7 and @xmath8 are expected as special and main decay modes of @xmath2 . meanwhile @xmath65 also decay into @xmath67 and @xmath68 with the branching ratio @xmath69 and @xmath70 respectively @xcite .
thus @xmath71 , @xmath72 , @xmath73 , @xmath74 are important decay modes for @xmath2 .
the typical decay modes of @xmath2 with the assignment of @xmath1 molecular state ( @xmath75 ) mainly include @xmath76 , @xmath77 , @xmath7 , @xmath72 , @xmath71 , @xmath78 , @xmath74 , @xmath73 , @xmath5 , @xmath6 .
as one of the main decay modes , @xmath79 should be seen if @xmath2 is a @xmath1 molecular state with @xmath43 .
however there exists the difficulty to distinguish @xmath80 in the experiment . here
index @xmath81 of @xmath82 can be @xmath83 .
\(ii ) @xmath2 as an isovector @xmath1 molecular state for isovector @xmath1 molecular state , @xmath2 can decay into @xmath84 , @xmath85 and @xmath86 , which are depicted in fig .
[ rho ] .
+ + + + + the branching ratio of @xmath87 is almost @xmath88 @xcite .
thus the typical decay modes of @xmath2 as an isovector @xmath1 molecular state are @xmath84 , @xmath71 and @xmath73 .
besides these decays , of course @xmath2 can also decay into @xmath5 and @xmath6 .
because @xmath38 is forbidden for an isovector @xmath1 molecular state , thus one can exclude the assignment of isovector @xmath1 molecular state for @xmath2 .
in the above sections , we discuss possible assignments for @xmath2 : @xmath0 and @xmath1 molecular state . in these two possible pictures ,
one finds that the branching ratio of @xmath89 is comparable with that of @xmath4 .
thus one suggests further experiments to search @xmath2 in @xmath3 invariant mass distribution .
how to distinguish these two assignments becomes a key problem . in the following
we will illustrate the differences of @xmath2 decays for two assignments , which will be helpful to distinguish @xmath0 and @xmath1 molecular state pictures : \(1 ) search for @xmath5 , @xmath6 decay channels
. if @xmath2 is @xmath0 , @xmath5 , @xmath6 are main decay channels . if @xmath2 is a @xmath1 molecular state , @xmath5 , @xmath6 , as the secondary decay modes , are comparable with @xmath7 and @xmath90 .
thus one suggests experiments to search these missing decay channels .
\(2 ) search for @xmath91 and @xmath90 decay channels . in the picture of @xmath1 molecular state ,
@xmath7 and @xmath92 are main decay modes . however , as @xmath0 , besides decaying to @xmath5 , @xmath93 and @xmath1 , @xmath2 mainly decays into @xmath77
. it will be a decisive factor to distinguish @xmath0 and @xmath1 molecular state assignments if the @xmath94 can be found in further experiments .
one strongly urges our experimental colleague to design more accurate experiments to find @xmath94 .
we thank h.w . ke for useful communication about their work .
we also thank prof .
zhu for interesting discussions and useful suggestions .
this project was supported by the national natural science foundation of china under grants 10421503 , 10625521 10705001 , and the china postdoctoral science foundation under grant no 20060400376 .
belle collaboration , s.k .
coi et al . , phys .
lett . * 91 * , 262001 ( 2003 ) .
cdf collaboration , d. acosta et al .
lett . * 93 * , 072001 ( 2004 ) .
d0 collaboration , v.m .
abazov et al .
lett . * 93 * , 162002 ( 2003 ) .
babar collaboration , b. aubert et al .
, phys . rev . *
d 71 * , 071103 ( 2005 ) ; belle collaboration , k. abe et al .
, arxiv : hep - ex/0505037 .
babar collaboration , b. aubert , phys . rev .
* d 74 * , 071101 ( 2006 ) .
belle collaboration , g. gokhroo et al .
* 97 * , 162002 ( 2006 ) .
babar collaboration , talk given by p. grenier in moriond qcd 2007 , 17 - 24 march , http://moriond.in2p3.fr/qcd/2007/sundayafternoon/ + grenier.pdf .
cdf collaboration , a. abulencia et al .
* 96 * , 102002 ( 2006 ) .
belle collaboration , k. abe et al .
, arxiv : hep - ex/0505038 .
cdf collaboration , a. abulencia , phys .
lett * 98 * , 132002 ( 2007 ) .
babar collaboration , b. aubert et al .
lett . * 95 * , 142001 ( 2005 ) .
cleo collaboration , t. e. coan et al.,phys .
lett . * 96 * , 162003 ( 2006 ) .
cleo collaboration , q. he et al .
d 74 * , 091104 ( 2006 ) .
belle collaboration , k. abe et al .
, arxiv : hep - ex/0612006 .
k. gottfried , phys .
lett . * 40 * , 598(1978 ) ; y.p .
kuang and t.m .
yan , phys .
* d 24 * , 2874(1981 ) ; y.p .
kuang , front .
china * 1 * , 19(2006 ) ; t.m .
yan , phys .
* d 22 * , 1652(1980 ) ; y.p .
kuang , y.p .
yi and b. fu , phys . rev . * d 42 * , 2300(1990 ) . | charmonium states mass @xmath16 close to @xmath2 @xcite @xmath16 candidate for @xmath0 central value width @xmath2 larger @xmath16 around 100 mev large error belle experiment mass width consistent with @xmath16 belle experiment gave @xmath17 ev @xmath18 ev two solutions candidate @xmath0 decay width @xmath19 is @xmath20 kev @xcite estimate @xmath21=5.8\times 10^{-3}$ ] @xmath22 for solutions if @xmath2 is @xmath0 state large experimental error central value former upper limit branching ratio @xmath23 @xmath24<4\times 10^{-3}$ ] former slightly larger @xmath4 reported by belle @xcite if @xmath2 is @xmath0 @xmath25 comparable with @xmath21 $ @xmath2 found in @xmath3 channel experiments measurement for @xmath26 transition of @xmath0 to lower states @xmath27 @xmath28 two pions solved by qcd multipole expansion method gottfried yan kuang depicted fig . calculated transitions of @xmath29 @xmath30=589.91\rm{kev}[\psi(3s) \psi(2s)\pi\pi]=14.96[\psi(3s)\psi\pi^+\pi^-]=**[\psi(3s)\psi\pi^0\pi^0]=**[\psi(3s)\to \psi(2s)\pi^+\pi^-]=**)\pi^0\pi^0] adopting cornell potential @xmath31 @xcite @xmath30=12.38[\psi(3s)\to \psi(2s)\pi\pi]=8.84[\psi(3s) j/\psi\pi^+\pi^-]=**[\psi(3s)\psi\pi^0\pi^0]=**[\psi(3s) \psi(2s)\pi^+\pi^-]=**\psi(3s) \psi(2s)\pi^0\pi^0]=** adopting modified cornell potential spin - related term @xcite @xmath32 @xmath33 @xmath34 zero - point energy large uncertainty estimate @xmath35 qcdme method estimate @xmath36 without spin - related term consistenttrust estimate @xmath36 qcdme search @xmath2 @xmath37 channel future experiments if @xmath2 is @xmath0 @xmath38 not main decay channel @xmath2 decay into @xmath5 @xmath6 @xmath2 wide decay width 200 mev decay into @xmath1 mass tail long history molecular structure hadrons explain phenomena new structure molecular structure mass @xmath2 close to threshold @xmath1 200 mev wide width assumed @xmath1 molecular state okun voloshin studied interaction charmed mesons proposed molecular states quarks rujula geogi glashow suggested @xmath16 @xmath1 molecular state dubynskiy voloshin proposed possible new resonance at @xmath1 threshold @xmath2 observed with @xmath11 @xmath39 possible quantum number @xmath2 @xmath39 @xmath2 must be p - wave @xmath1 use experimental information determine quantum number @xmath40 @xmath2 @xmath2 can be isosinglet state with @xmath41 or isovector state @xmath42@xmath2 @xmath1 molecular state falls into @xmath1 mass tail depicted in fig . discus decay modes @xmath2 isoscalar @xmath1 molecular state recattering effect @xmath2 @xmath43 decay into @xmath44 @xmath46 mechanism fig . eta @xmath47 fig . omega @xmath48 replaced as @xmath49 @xmath50 @xmath2 into @xmath51 @xmath52 exchanging @xmath53 @xmath54 mesons between @xmath55 @xmath56 secondary branching ratio of @xmath57 comparable with @xmath58 @xmath59 @xmath60 decay into @xmath61 estimate @xmath62 decay mechanism fig . [ eta ] analyzing @xmath61 invariant mass spectrum @xmath61 invariant mass distribution show signature of @xmath59 or @xmath63 branching ratio of @xmath64 89.1% @xmath65 to @xmath66 overwhelming @xmath7 @xmath8 special main decay modes of @xmath2@xmath65 decay into with branching ratio @xmath69 @xcite @xmath71 important decay modes for @xmath2 typical decay modes of @xmath2 with @xmath1 state include @xmath76 @xmath79 seen if @xmath2 is @xmath1 molecular state with @xmath43 difficulty to distinguish @xmath80 @xmath81 of @xmath82 can be @xmath83 @xmath2 as isovector @xmath1 molecular state can decay into @xmath84 depicted in fig . branching ratio of @xmath87 is almost @xmath88 @xcite typical decay modes of @xmath2 are @xmath84 @xmath71 @xmath73 @xmath2 can decay into @xmath5 @xmath6 @xmath38 forbidden for isovector @xmath1 state exclude assignment of isovector @xmath1 molecular state for @xmath2 discuss possible assignments for @xmath2 : @xmath0 @xmath1 molecular statetwo pictures branching ratio @xmath89 comparable with @xmath4 suggests experiments search @xmath2 in @xmath3 invariant mass distribution distinguish assignments key problem illustrate differences @xmath2 decays two assignments distinguish @xmath0 @xmath1 state pictures search for @xmath5 @xmath6 decay channels @xmath2 main decay channels @xmath1 secondary decay modes comparable with @xmath7 @xmath90 suggests experiments search missing decay channels search for @xmath91 @xmath90 decay channels @xmath1 @xmath7 @xmath92 main decay modes @xmath0 @xmath2 decays into @xmath77 decisive distinguish @xmath0 @xmath1 state assignments if @xmath94 found in experiments urges colleague design accurate experiments find @xmath94 thank h.w . ke communication prof . zhu discussions suggestions project supported by national natural science foundation china grants 10421503 10625521 10705001 china postdoctoral science foundation grant no 20060400376 collaboration s coi et al .91 262001 2003 ) cdf collaboration acosta 93 072001 2004 ) collaboration abazov 93 162002 2003 ) babar collaboration aubert . 71 071103 ( 2005 ) belle collaboration k abe hep - ex/0505037 babar collaboration aubert . 74 071101 ( 2006 ) belle collaboration g gokhroo 97 162002 ( 2006 ) babar collaboration talk p grenier moriond qcd 2007 17 - 24 march.in2p3.fr grenier.pdf cdf collaboration abulencia 96 102002 ( 2006 ) belle collaboration k abe hep - ex/0505038 cdf collaboration abulencia 98 132002 ( 2007 ) babar collaboration aubert 95 142001 ( 2005 ) cleo collaboration coan 96 162003 ( 2006 ) cleo collaboration 74 091104 ( 2006 ) belle collaboration k. abe hep - ex/0612006 k gottfried phys . 40 * 598(1978 ) y kuang yan phys 24 * 2874(1981 ) kuang frontchina 1 19(2006 t.m. yan . 22 1652(1980 ) y.p . kuang yi fu . rev. 42 2300(1990 ) | 0.492809 | 3,182 | 1,720 | , Saving $0.1 in GPT-4. |
55,251 | an equivalence is proven between the riemann hypothesis and the speed of convergence to @xmath0 of the probability that two independent random variables following the same geometric distribution are coprime integers , when the parameter of the distribution goes to 0 .
what is the probability that two random integers are coprime ?
since there is no uniform probability distribution on @xmath1 , this question must be made more precise . before going into probabilistic considerations
, we would like to remind a seminal result of dirichlet @xcite : the set @xmath2 of ordered pairs of coprime integers has an asymptotic density in @xmath3 which is equal to @xmath4 .
the proof is not trivial and can be found in the book of hardy and wright ( @xcite , theorem 332 ) , but if one admits the existence of a density @xmath5 , its value can be easily obtained . indeed , the set @xmath3 is the disjoint union of @xmath2 , @xmath6 , @xmath7 , which have respective density @xmath5 , @xmath8 , @xmath9 , , hence @xmath10 a natural choice for picking two integers at random is to draw two random variables @xmath11 and @xmath12 independently according to the uniform distribution on @xmath13 and let @xmath14 goes to @xmath15 . in this case , denoting by @xmath16 the classical euler totient function , the probability that @xmath11 and @xmath12 are coprime equals @xmath17 as was shown by mertens @xcite in 1874 .
the sharpest error term in this problem up to now was obtained in 1963 by walfisz @xcite who replaced the @xmath18 term of mertens by @xmath19 .
one can quickly realize that the error term can not boil down to @xmath20 .
indeed , the passage from @xmath14 to @xmath21 perturbates the average by a term of order @xmath22 . a striking result by montgomery @xcite shows that the fluctuations are actually at least of order @xmath23 .
recently , kaczorowski and wiertelak considered in @xcite a smoothing of the sequence above which corresponds to another family of probabilities .
namely , for a given @xmath14 , the former uniform distribution on @xmath24 is replaced by the probability distribution putting a weight @xmath25 to the couple @xmath26 for all integers @xmath27 satisfying @xmath28 , up to a normalization factor .
notice that , under this distribution , the two random variables @xmath11 and @xmath12 have the same distribution but are not independent anymore .
simultaneously , they iterated this smoothing procedure in @xcite , and smartly recovered montgomery s result . in this paper
, we consider a natural parametrized family of probability distributions for @xmath11 and @xmath12 that smoothes even more the fluctuations with respect to the parameter . for @xmath29 ,
the random variables @xmath11 and @xmath12 are drawn independently according to the geometric distribution of parameter @xmath30 : @xmath31 = ( 1-e^{-\beta})e^{-\beta(n-1)}.\ ] ] the asymptotic regime corresponding to large integers is obtained by taking @xmath32 . for this distribution , we are able to obtain a direct connection with the riemann hypothesis .
let @xmath33 be two independent random variables following the geometric distribution of parameter @xmath30 .
the riemann hypothesis holds if and only if , for all @xmath34 , as @xmath35 goes to @xmath36 , @xmath37 = \frac{6}{\pi^2 } + \frac{6}{\pi^2}\beta + o(\beta^{\frac32-\epsilon}).\ ] ] before we start with the proof , let us make a comment .
the smoothing " we have chosen here is stronger than the smoothing of @xcite we described above .
however , one could think about an even stronger one , corresponding to independent variables @xmath11 and @xmath12 both following the zeta distribution of parameter @xmath38 , i.e. @xmath39={\mathbf p}[y = n]=\frac1{\zeta(\alpha)}\frac1{n^\alpha}.\ ] ] in this case , @xmath40 is distributed according to the zeta distribution of parameter @xmath41 .
this explicit example shows that a too strong smoothing erases all oscillations of the probability of coprimality and makes the dependence on the localization of the zeroes of the zeta function vanish .
now , we turn to the proof of the theorem .
we begin by expressing the probability that @xmath11 and @xmath12 are coprime as a function of @xmath35 : @xmath42 = \sum_{(x , y ) \in \mathcal p } { \mathbf p}[x = x , y = y ] = ( e^{\beta}-1)^2 \sum_{(x , y ) \in \mathcal p } e^{-\beta(x+y)}.\ ] ] we see therefore that the function @xmath43 defined for all @xmath44 by @xmath45 plays a key role in the analysis of this probability . in particular , condition is equivalent to the following expansion of @xmath46 as @xmath35 goes to @xmath36 : @xmath47 in the sequel , we will work with instead of . | the starting point of the proof is the following relation between the function @xmath48 and the riemann zeta function . for all complex number @xmath49 such that @xmath50 , @xmath51 let us first consider the case of a real number @xmath52 .
by using the fubini - tonelli theorem , @xmath53 to simplify this series , we observe that each @xmath54 can be written as @xmath55 with @xmath56 and @xmath57 in a unique way , leading to the identity @xmath58 since @xmath52 , the left - hand side is easily expressed in terms of the zeta function : @xmath59 which ends the proof of .
the extension of to the complex domain @xmath50 follows the same steps , except that permutations of the order of summation are justified by the fubini theorem .
now , let us assume that holds .
our strategy is to show that the function @xmath60 defined for all @xmath49 with @xmath50 by @xmath61 has a holomorphic continuation to the domain @xmath62 .
this fact would indeed prevent @xmath63 from having zeros with real part larger than @xmath64 , since such zeros would correspond to poles of @xmath60 in the same region .
the celebrated functional equation ( theorem 2.1 in @xcite ) satisfied by @xmath63 shows that its zeros in the critical strip @xmath65 are symmetrically distributed with respect to the vertical line @xmath66 .
hence , the function @xmath63 would not vanish in the region @xmath67 either .
the holomorphic continuation is obtained by an integral representation of @xmath60 based on formula in the lemma .
namely , for all @xmath49 with @xmath50 , @xmath68 the estimate for @xmath46 in the neighbourhood of @xmath36 implies that the first integral defines a holomorphic function of @xmath49 in the domain @xmath69 .
the second integral defines a holomorphic function of @xmath49 in the whole complex plane because @xmath70 when @xmath35 goes to @xmath15 .
formula shows that the function @xmath71 is the mellin transform of the function @xmath48 in the domain @xmath50 .
an application of the mellin inversion formula leads therefore to the following integral representation of @xmath46 for all @xmath44 and @xmath72 , @xmath73 let @xmath34 and @xmath74 .
recall here that the zeta function has a unique holomorphic continuation to @xmath75 , and that @xmath76 is a simple pole with residue @xmath76 . assuming that riemann s hypothesis holds , the only pole of the function @xmath71 in the region @xmath77 lies at @xmath78 with residue @xmath79 .
we can therefore apply the residue theorem to show that @xmath80 taking @xmath81 in the mellin inversion formula above , we see that the integral on @xmath82 $ ] tends to @xmath46 as @xmath83 tends to @xmath15 . in order to deal with the other three integrals
, we are going to use the following estimates as @xmath84 : * as a consequence of the complex stirling formula , the @xmath85 function has exponential decay along vertical lines . in particular ,
@xmath86}|\gamma(\sigma+it)| = o(|t|^{\frac 52}e^{-\frac \pi 2 |t|}).\ ] ] * the zeta function has polynomial growth along vertical lines ( see section 5.1 of @xcite )
. there exists @xmath87 such that @xmath88 * the inverse zeta function has sub - polynomial growth along vertical lines in the region @xmath77 as a consequence of riemann s hypothesis ( see theorem 14.2 in @xcite ) . for all @xmath89 , @xmath90}\frac{1}{|\zeta(\sigma + it)| } = o(|t|^\alpha).\ ] ] we can see with these bounds that both integrals along the horizontals segments @xmath91 $ ] and @xmath92 $ ] tend to @xmath36 as @xmath93 .
similarly , the integral along the vertical segment @xmath94 $ ] is bounded by @xmath95 . | starting point proof relation between function @xmath48 riemann zeta function complex number @xmath49 consider case real number @xmath52 fubini - tonelli theorem @xmath54 written as @xmath55 @xmath56 @xmath57 identity @xmath58 @xmath52 left - hand side expressed zeta function @xmath59 ends proof extension complex domain @xmath50 follows same steps permutations order summation justified by fubini theorem assume holds strategy show function @xmath60 defined @xmath49 @xmath61 holomorphic continuation to domain @xmath62 @xmath63 zeros part larger than @xmath64 correspond to poles @xmath60 region functional equation theorem 2.1 satisfied @xmath63 shows zeros in critical strip @xmath65 symmetrically distributed vertical line @xmath66 function @xmath63 vanish in region @xmath67 holomorphic continuation obtained by integral representation of @xmath60 formula @xmath49 estimate for @xmath46 @xmath36 implies first integral defines holomorphic function of @xmath49 in domain @xmath69second integral defines holomorphic function @xmath49 complex plane @xmath70 @xmath35 goes to @xmath15 function @xmath71 mellin transform @xmath48 domain @xmath50 mellin inversion formula leads integral representation @xmath46 for @xmath44 @xmath73 @xmath34 zeta function unique holomorphic continuation @xmath75 @xmath76 simple pole with residue @xmath76 riemann s hypothesis holds only pole @xmath71 region @xmath77 at @xmath78 with residue @xmath79 residue theorem show @xmath80 @xmath81 mellin inversion formula integral @xmath82 tends to @xmath46 @xmath83 tends @xmath15 other three integrals estimates @xmath84 complex stirling formula @xmath85 function has exponential decay along vertical lines @xmath86 zeta function has polynomial growth vertical lines section @xmath87 @xmath88 inverse zeta function has sub - polynomial growth vertical lines region @xmath77 riemann s hypothesis theorem 14.2@xmath89 @xmath90\frac{1|\zeta(\sigma + it) =(|^\alpha). integrals horizontals segments @xmath91 @xmath92 tend to @xmath36 @xmath93 integral vertical segment @xmath94 bounded @xmath95 | 0.180108 | 950 | 514 | , Saving $0.0 in GPT-4. |
55,255 | let @xmath0 denote a closed nonorientable surface of genus @xmath1 .
for @xmath2 the mapping class group @xmath3 is generated by dehn twists and one crosscap slide ( @xmath4homeomorphism ) or by dehn twists and a crosscap transposition .
margalit and schleimer observed that dehn twists have nontrivial roots .
we give necessary and sufficient conditions for the existence of roots of crosscap slides and crosscap transpositions . | let @xmath5 be a connected , nonorientable surface of genus @xmath1 with @xmath6 boundary components and @xmath7 punctures , that is a surface obtained from a connected sum of @xmath1 projective planes @xmath0 by removing @xmath6 open disks and specifying the set @xmath8 of @xmath7 distinguished points in the interior of @xmath0 .
if @xmath6 or / and @xmath7 equals zero , we omit it from notation . the _ mapping class group _ @xmath9 consists of isotopy classes of self homeomorphisms @xmath10 fixing boundary components pointwise and such that @xmath11 .
the mapping class group @xmath12 of an orientable surface is defined analogously , but we consider only orientation preserving maps .
if we allow orientation reversing maps , we obtain the _ extended mapping class group _ @xmath13 . by abuse of notation
, we identify a homeomorphism with its isotopy class . in the orientable case ,
the mapping class group @xmath14 is generated by dehn twists @xcite .
as for nonorientable surfaces , lickorish proved that dehn twists alone do not generate @xmath3 , @xmath2 .
this group is generated by dehn twists and one crosscap slide ( @xmath4homeomorphism ) @xcite .
a presentation for @xmath3 using these generators was obtained by stukow @xcite .
this presentation was derived from the presentation given by paris and szepietowski @xcite , which used as generators dehn twists and yet another homeomorphisms of nonorientable surfaces , so called crosscap transpositions .
margalit and schleimer discovered a surprising property of dehn twists : in the mapping class group of a closed , connected , orientable surface @xmath15 of genus @xmath2 , every dehn twist has a nontrivial root @xcite .
it is natural to ask if crosscap slides and crosscap transpositions also have a similar property .
the main goal of this paper is to prove the following : in @xmath3 a nontrivial root of a crosscap transposition [ or a crosscap slide ] exists if and only if @xmath16 or @xmath17 and the complement of the support of this crosscap transposition [ or crosscap slide ] is orientable .
let @xmath18 be a nonorientable surface of genus @xmath2 .
let @xmath19 and @xmath20 be two simple closed curves in @xmath21 intersecting in one point , such that @xmath19 is two sided and @xmath20 is one sided .
a regular neighborhood of @xmath22 is homeomorphic to the klein bottle with a hole denoted by @xmath23 .
a convenient model of @xmath23 is a disk with 2 crosscaps , see figure [ uy ] . in this figure
shaded disks represent crosscaps , thus the boundary points of these disks are identified by the antipodal map . a _ crosscap transposition _ @xmath24 specified by @xmath20 and @xmath19 is a homeomorphism of @xmath23 which interchanges two crosscaps keeping the boundary of @xmath23 fixed @xcite .
it may be extended by the identity to a homeomorphism of @xmath21 .
if @xmath25 is the dehn twist about @xmath19 ( with the direction of the twist indicated by small arrows in figure [ uy ] ) , then @xmath26 is a _
crosscap slide _ of @xmath20 along @xmath19 , that is the effect of pushing @xmath20 once along @xmath19 keeping the boundary of @xmath23 fixed .
note that @xmath27 .
if @xmath1 is even , then the complement of @xmath23 in @xmath28 can be either a nonorientable surface @xmath29 or an orientable surface @xmath30 , therefore on surfaces of even genus two conjugacy classes of crosscap slides and crosscap transpositions exist .
represent @xmath28 as a connected sum of @xmath1 projective planes and let @xmath31 be one sided circles that correspond to crosscaps as in indicated in figure [ fig_surface ] . by abuse of notation
, we identify @xmath32 with the corresponding crosscap .
.,scaledwidth=80.0% ] if @xmath33 are two sided circles indicated in the same figure , then for each @xmath34 by @xmath35 we denote the dehn twist about @xmath36 , the crosscap transposition @xmath37 , and the crosscap slide @xmath38 , respectively .
a full presentation for @xmath3 is given in @xcite . among others ,
the following relations hold in @xmath3 : * @xmath39 for @xmath40 , * @xmath41 for @xmath42 , * @xmath43 , * @xmath44 and hence @xmath45 for @xmath46 it is straightforward to check that relations ( r1)(r3 ) imply * @xmath47 geometrically @xmath48 is a cyclic rotation of @xmath31 and @xmath49 is a cyclic rotation of @xmath50 around @xmath51 . in particular , * @xmath52 in @xmath53 .
we also have the following chain relation between dehn twists ( proposition 4.12 of @xcite ) : if @xmath54 is even and @xmath55 is a chain of simple closed curves in a surface @xmath56 , such that the boundary of a closed regular neighborhood of their union is isotopic to @xmath57 , then * @xmath58 .
[ rem : odd ] automorphisms of @xmath59 induced by crosscap transpositions and crosscap slides have determinants equal to @xmath60 , so if a root of a crosscap slide or a crosscap transposition exists , it must be of odd degree .
let @xmath23 be a subsurface of @xmath28 that is a klein bottle with one boundary component @xmath61 and which contains @xmath51 and @xmath62 ( figure [ fig_surface ] ) .
in particular @xmath63 .
let @xmath65 be such that @xmath66 . by relations ( r6 ) and ( r1 ) , @xmath67 analogously , by relations ( r6 ) , ( r1 ) and ( r4 ) , @xmath68 .
let @xmath65 be such that @xmath71 . by relations ( r6 ) and ( r1 ) , @xmath72 analogously , by relations ( r6 ) , ( r1 ) and ( r4 ) , @xmath73 .
suppose now that crosscap transposition @xmath75 and crosscap slide @xmath76 are supported in a klein bottle with a hole @xmath23 such that @xmath77 is orientable . if @xmath78 is a chain of two sided circles in @xmath77 , then by relation ( r7 ) , @xmath79 analogously , @xmath80 .
crosscap slides and a crosscap transpositions are primitive in @xmath82 because @xcite @xmath83 [ rem : order6 ] it is known that the mapping class group @xmath85 is hyperelliptic @xcite and has the central element @xmath86 such that @xmath87 is the extended mapping class group @xmath88 of a sphere with 4 punctures .
two upper subscripts mean that we have four punctures on the sphere , but one of them must be fixed .
this implies @xcite that the maximal finite order of an element in @xmath88 is 3 , and hence the maximal finite order of an element in @xmath85 is 6 .
moreover , each two rotations of order 3 in @xmath88 are conjugate , which easily implies that each two elements of order 6 in @xmath85 are conjugate .
the details of the proof of the last statement are completely analogous to that used in @xcite , hence we skip them . the same conclusion can be obtained also purely algebraically : it is known @xcite that @xmath89 and the maximal finite order of an element in @xmath90 is 6 .
moreover , there is only one conjugacy class of such elements in @xmath90 for details see for example theorem 2 of @xcite .
we will show that crosscap transpositions do not have nontrivial roots in @xmath85 .
suppose that @xmath91 exists such that @xmath92 , where @xmath93 ( see remark [ rem : odd ] ) .
then @xmath94 by remark [ rem : order6 ] , @xmath95 .
moreover , by relation ( r7 ) , @xmath96 hence @xmath97 is conjugate to @xmath98 .
this contradicts remark [ rem : odd ] , because dehn twists induce automorphisms of @xmath99 with determinant equal to 1 and @xmath100 . in the case of a crosscap slide
the argument is completely analogous , hence we skip the details .
if @xmath103 is nonorientable , then @xmath61 cuts @xmath104 into two klein bottles with one boundary component : @xmath23 and @xmath105 .
moreover , as was shown in appendix a of @xcite , @xmath106 if @xmath107 exists such that @xmath92 and @xmath93 ( see remark [ rem : odd ] ) , then @xmath108 in particular , @xmath97 commutes with @xmath109 and @xmath110 by proposition 4.6 of @xcite , up to isotopy of @xmath104 , @xmath111 . because @xmath112 does not interchange two sides of @xmath61 and does not reverse the orientation of @xmath61 , @xmath97 has exactly the same properties . therefore
, we can assume that @xmath97 is composed of maps of @xmath23 and @xmath105 .
moreover @xmath113 interchanges @xmath51 and @xmath62 and does not interchange @xmath114 and @xmath115 , hence @xmath116 but then @xmath117 which is a contradiction , because dehn twists about disjoint circles generate a free abelian group ( proposition 4.4 of @xcite ) . in the case of a crosscap slide
the argument is completely analogous , hence we skip the details .
margalit and schleimer observed in @xcite that if @xmath118 , then roots of elementary braids in @xmath119 exist .
the main theorem implies slightly stronger version of that result . by proposition 2.4 of @xcite
, there is a monomorphism @xmath123 which is induced by blowing up each puncture to a crosscap . in particular , this monomorphism sends elementary braids to crosscap transpositions .
moreover , all roots of crosscap transpositions constructed in the proof of the main theorem are elements of @xmath124 . | @xmath5 connected nonorientable surface @xmath1 @xmath6 boundary components punctures obtained @xmath1 projective planes removing @xmath6 disks specifying @xmath8 @xmath7 points interior @xmath0 @xmath6 @xmath7 zero omit notation mapping class group @xmath9 isotopy classes homeomorphisms @xmath10 fixing boundary components pointwise mapping class group @xmath12 orientable surface defined analogously orientation preserving maps orientation reversing maps extended mapping class group @xmath13 identify homeomorphism isotopy class mapping class group @xmath14 generated dehn twists @xcite dehn twists generate @xmath3 @xmath2 group generated dehn twists crosscap slide @xmath4homeomorphism @xcite presentation @xmath3 generators stukow @xcite derived paris szepietowski @xcite generators dehn twists homeomorphisms nonorientable surfaces crosscap transpositions margalit schleimer property dehn twists closed connected orientable surface @xmath15 @xmath2 dehn twist nontrivial root @xciteask if crosscap slides transpositions similar property goal prove in @xmath3 nontrivial root crosscap transposition exists if @xmath16 or @xmath17 complement orientable @xmath18 nonorientable surface @xmath2 @xmath19 @xmath20 closed curves in @xmath21 intersecting point @xmath19 two sided one sided regular neighborhood @xmath22 homeomorphic to klein bottle hole by @xmath23 model @xmath23 disk with 2 crosscaps figure shaded disks represent crosscaps boundary points identified by antipodal map crosscap transposition @xmath24 @xmath20 @xmath19 homeomorphism of @xmath23 interchanges two crosscaps boundary fixed extended by identity homeomorphism @xmath21 @xmath25 dehn twist about @xmath19 small arrows @xmath26 crosscap slide of @xmath20 along @xmath19 pushing @xmath20 along @xmath19 keeping boundary @xmath23 fixed @xmath1 even complement of @xmath23 in @xmath28 nonorientable surface @xmath29 or orientable surface @xmath30 on surfaces even genus two classes of crosscap slides transpositions existrepresent @xmath28 sum @xmath1 projective planes @xmath31 one sided circles crosscaps figure identify @xmath32 corresponding crosscap=80.0% @xmath33 two sided circles @xmath34 @xmath35 denote dehn twist @xmath36 crosscap transposition @xmath37 crosscap slide @xmath38 full presentation @xmath3 @xcite relations hold @xmath3 @xmath39 @xmath40 @xmath41 @xmath42 @xmath43 @xmath44 @xmath45 for @xmath46 relations ( r1)(r3 ) imply @xmath47 @xmath48 cyclic rotation @xmath31 @xmath49 cyclic rotation @xmath50 @xmath51 @xmath52 in @xmath53 chain relation between dehn twists proposition 4.12 @xmath54 even @xmath55 chain closed curves surface @xmath56 closed regular neighborhood isotopic to @xmath57 * @xmath58 automorphisms @xmath59 crosscap transpositions slides determinants equal to @xmath60 root crosscap slide transposition exists odd degree@xmath23 subsurface @xmath28 klein bottle boundary component @xmath61 contains @xmath51 @xmath62 figure @xmath63 @xmath65 @xmath66 relations ( r6 ) ( r1 ) @xmath67 r4 @xmath68 @xmath65 @xmath71 relations ( r6 ) r1 ) @xmath72 r6 r1 r4 ) @xmath73 crosscap transposition @xmath75 crosscap slide @xmath76 supported klein bottle hole @xmath23 @xmath77 orientable @xmath78 chain two sided circles @xmath77 relation ( r7 ) @xmath79 @xmath80 crosscap slides transpositions primitive in @xmath82 @xcite @xmath83 order6 mapping class group @xmath85 hyperelliptic central element @xmath86 @xmath87 extended mapping class group @xmath88 sphere 4 punctures two upper subscripts four punctures sphere one fixed implies maximal finite order element @xmath88 3 maximal order @xmath85 6 two rotations order 3 in @xmath88 conjugate implies two elements order 6 @xmath85 conjugate details proof last statement analogous @xcite skipconclusion algebraically known @xcite @xmath89 maximal finite order element @xmath90 is 6 one conjugacy class in @xmath90 theorem 2 @xcite crosscap transpositions nontrivial roots in @xmath85 @xmath91 exists @xmath92 @xmath93 odd @xmath94 @xmath95 relation r7 @xmath96 @xmath97 conjugate to @xmath98 contradicts remark dehn twists induce automorphisms of @xmath99 determinant equal to 1 @xmath100 crosscap slide argument analogous @xmath103 nonorientable @xmath61 cuts @xmath104 into two bottles boundary component @xmath23 and @xmath105 appendix a @xcite @xmath106 @xmath107 exists @xmath92 @xmath93 @xmath108 @xmath97 commutes with @xmath109 @xmath110 proposition 4.6 @xcite isotopy of @xmath104 @xmath111 @xmath112 interchange sides @xmath61 reverse orientation @xmath61 @xmath97 same properties assume @xmath97 composed of maps of @xmath23 and @xmath105@xmath113 interchanges @xmath51 interchange @xmath114 @xmath116 @xmath117 contradiction dehn twists disjoint circles generate free abelian group proposition 4.4 @xcite crosscap slide argument analogous skip details margalit schleimer observed if @xmath118 roots elementary braids in @xmath119 exist main theorem implies stronger version proposition 2.4 monomorphism @xmath123 induced by blowing puncture crosscap monomorphism sends elementary braids to crosscap transpositions roots crosscap transpositions elements @xmath124 | 0.351172 | 2,422 | 1,352 | , Saving $0.1 in GPT-4. |
55,263 | so - called `` buchert averaging '' is actually a coarse - graining procedure , where fine detail is `` smeared out '' due to limited spatio - temporal resolution . for technical reasons , ( to be explained herein ) , `` averaging ''
is not really an appropriate term , and i shall consistently describe the process as a `` coarse - graining '' . because einstein gravity is nonlinear the coarse - grained einstein tensor is typically not equal to the einstein tensor of the coarse - grained spacetime geometry . the discrepancy can be viewed as an effective " stress - energy . to keep otherwise messy technical issues firmly under control ,
i shall work with conformal - flrw ( cflrw ) cosmologies .
these cflrw - based models are particularly tractable , and are also particularly attractive observationally : the cmb is not distorted . in this cflrw context
one can prove some rigorous theorems regarding the interplay between buchert coarse - graining , tracelessness of the effective stress - energy , and the classical energy conditions . | the cosmological back - reaction problem continues to generate considerable ( and sometimes quite heated ) debate .
@xcite there is significant disagreement as to just how one should split the universe into `` smooth background '' plus `` local perturbations '' , and yet more disagreement as to whether or not these perturbations remain small .
i shall work within the framework of conformally - flrw ( cflrw ) cosmologies , @xcite where the mathematics and physics are both firmly under control .
@xcite these cflrw cosmologies are simply generic flrw spacetimes distorted by a ( possibly non - perturbatively large ) conformal factor .
cosmographically @xcite this is the unique non - perturbative distortion that can be applied to flrw cosmologies without grossly modifying the cmb .
@xcite in counterpoint , when working in this cflrw framework , buchert s coarse - graining procedure simplifies tremendously .
the contribution to the effective stress - energy due to coarse graining can be explicitly calculated .
this coarse - graining contribution to the effective stress - energy need not be traceless , though it will typically satisfy the classical energy conditions .
( further details are in preparation .
let @xmath0 represent a flrw spacetime . adopt conformal time for the flrw . then : @xmath1 here @xmath2 is translation invariant in both space and ( conformal ) time . on this flrw space - time
, define coarse - graining by @xmath3 here @xmath4 is any non - negative translation - invariant normalized kernel in both space and ( conformal ) time .
the high symmetries of flrw are essential to this construction .
in this flrw framework , coarse - graining the gradient of a scalar is the same as taking the gradient of the coarse - grained scalar : @xmath5 \ ; \phi(y ) \;\d^4 y \\ & = & + \oint [ \nabla_x f(x - y)]\ ; \phi(y ) \ ; \d^4 y \\ & = & \nabla_x \oint f(x - y)\ ; \phi(y ) \;\d^4 y \\ & = & \nabla_x \langle \phi(x ) \rangle.\end{aligned}\ ] ] translation invariance of the kernel is crucial to establishing this result . now calculate : @xmath6(x - y ) \ ; \ ; \phi(y ) \;\d^4 y \\ & \neq & \langle \phi(x ) \rangle\end{aligned}\ ] ] so you have to be very careful . since @xmath7 , then if you insist on using the word `` averaging '' to describe the coarse - graining process , it follows that `` the average of the average is not the average '' .
this is at gross variance with the normal use of the word `` averaging '' , and for this reason i shall eschew this term . furthermore : @xmath8 more precisely @xmath9 for ensemble averages , or indeed any situation where the normal rules of averaging apply , that second line vanishes .
but it does not vanish for a coarse - graining process .
for these reasons the phrase `` coarse - graining '' is more appropriate than the word `` averaging '' . while one can
not now naively assert @xmath10 based on the usual `` averaging '' arguments , this inequality can instead be derived from positivity of the coarse - graining kernel plus an appeal to the cauchy
schwarz inequality : @xmath11
consider first a conformal deformation of a given metric : @xmath12 .
we first note a number of utterly standard results .
ricci tensor : @xmath13 - g_{ab } [ \nabla^2\phi+2|\nabla\phi|^2].\ ] ] ricci scalar : @xmath14.\ ] ] einstein tensor : @xmath15 + g_{ab } [ 2\nabla^2\phi+|\nabla\phi|^2].\ ] ] now apply ( generic ) coarse - graining .
coarse - grain the einstein tensor : @xmath16.\ ] ] coarse - grain the conformal mode : @xmath17.\ ] ] coarse - grain the conformally deformed metric holding the base metric fixed : @xmath18 \nonumber \\ & & + g_{ab } \left[\nabla^2\ln\langle e^{2\phi}\rangle+{1\over4}|\nabla\ln\langle e^{2\phi}\rangle|^2\right].\end{aligned}\ ] ] this so far holds for any generic coarse - graining process , details as yet unspecified .
( we have not even used any assumed translation invariance for the kernel in the coarse - graining process . )
for cflrw space - time smearing this simplifies considerably .
coarse - grain the einstein tensor : @xmath19.\ ] ] coarse - grain the conformal mode : @xmath17.\ ] ] coarse - grain the metric : @xmath18 \nonumber \\ & & + g_{ab } \left[\nabla^2\ln\langle e^{2\phi}\rangle+{1\over4}|\nabla\ln\langle e^{2\phi}\rangle|^2\right].\end{aligned}\ ] ] combining the above , for cflrw coarse - graining we have : @xmath20 + g_{ab } [ \langle|\nabla\phi|^2\rangle - |\nabla\langle\phi\rangle|^2 ] .
\nonumber \\ & & \end{aligned}\ ] ] this is a specific , ( very high symmetry ) , case of buchert coarse - graining , one where all the technical details are now fully under control .
the coarse - grained spacetime `` sees '' the effective stress - energy @xmath21 the part of the `` effective stress tensor '' due to cflrw coarse - graining is : @xmath22 - g_{ab } [ \langle|\nabla\phi|^2\rangle - |\nabla\langle\phi\rangle|^2].\ ] ] note that this vanishes ( as it should ) in the limit where the smearing kernel becomes a delta function ( so that coarse - graining is switched off ) . for the trace of this buchert contribution to the effective stress tensor we have @xmath23
this need not be zero , and generically is in fact nonzero .
since the first term is positive and the last three terms are negative , the trace can easily be arranged to be either positive or negative .
in contrast , this effective stress energy typically will satisfy many of the other classical energy conditions . for instance , for null vectors , using the flrw symmetries to assert @xmath24 , we have : @xcite @xmath25 so the null energy condition ( nec ) is satisfied for this form of coarse - graining . for timelike vectors , we have : @xcite @xmath26 use a tetrad decomposition with @xmath27 being the timelike vector @xmath28 of interest , then @xmath29 this is a positive definite euclidean signature `` metric '' . now using the flrw symmetries to assert @xmath30 , we have @xmath31
so the weak energy condition ( wec ) is satisfied for this form of coarse - graining . for the strong energy condition ( sec ) consider the trace - reversed quantity @xcite @xmath32.\end{aligned}\ ] ] then for timelike vectors @xmath33 so the strong energy condition ( sec ) is satisfied for this form of coarse - graining .
in a cosmological setting , on large enough regions , one might ( approximately ) hope : @xmath35 then @xmath36,\ ] ] implying @xmath37.\ ] ] thus @xmath38 this agrees with the sec calculation .
the `` effective @xmath34 parameter '' is @xmath39 this contribution to stress - energy always has @xmath40 .
when studying buchert coarse - graining , working with cflrw spacetimes is a great technical help .
specifically , a flrw background provides you with a very nicely controlled version of buchert coarse - graining , and cflrw coarse - graining satisfies some nice differential identities .
indeed cflrw coarse - graining yields an explicit and tractable effective stress - energy .
the effective stress energy is not necessarily traceless .
furthermore for this form of coarse - graining the classical energy conditions are satisfied by the effective stress - energy .
finally i emphasize that buchert coarse - graining @xmath41 ensemble average ; several key properties are radically different .
s. r. green and r. m. wald , arxiv:1506.06452 [ gr - qc ] .
t. buchert and s. rsnen , _ ann . rev
. nucl . part .
* 62 * ( 2012 ) 57 .
g. f. r. ellis , _ class .
_ * 28 * ( 2011 ) 164001 .
t. buchert , _ class .
_ * 28 * ( 2011 ) 164007 , [ arxiv:1103.2016 [ gr - qc ] ] .
d. l. wiltshire , _ class .
_ * 28 * ( 2011 ) 164006 .
t. buchert , m. carfora , g.f.r .
ellis , e.w .
kolb , m.a.h .
maccallum , + j.j .
ostrowski , s. rsnen , b.f .
roukema , l. andersson , a.a .
coley , + and d.l .
wiltshire , _ class .
* 32 * ( 2015 ) 215021 .
t. buchert , a. a. coley , h. kleinert , b. f. roukema , and d. l. wiltshire , + arxiv:1512.03313 [ astro-ph.co ] . j. j. ostrowski and b. f. roukema , arxiv:1512.02947 [ gr - qc ] . m. visser , _ gen .
* 37 * ( 2005 ) 1541 , [ gr - qc/0411131 ] .
m. visser , _ science _ * 276 * ( 1997 ) 88 , [ arxiv:1501.01619 [ gr - qc ] ] .
m. visser , _ phys .
* d56 * ( 1997 ) 7578 , [ gr - qc/9705070 ] .
m. visser , gr - qc/9710010 .
m. visser and c. barcel , doi:10.1142/9789812792129_0014 gr - qc/0001099 . | cosmological back - reaction problem debate significant disagreement universe into smooth background local perturbations perturbations remain small work within conformally - flrw cosmologies mathematics and physics under control cosmologies are generic flrw spacetimes distorted by large conformal factor unique non - perturbative distortion applied to flrw cosmologies without modifying cmb buchert s coarse - graining procedure simplifies contribution to effective stress - energy due to coarse graining calculated graining contribution need not be traceless classical energy conditions details in preparation @xmath0 represent flrw spacetime adopt conformal time for flrw @xmath1 @xmath2 is translation invariant in space and conformal time define coarse - graining by @xmath3 @xmath4 is non - negative translation - invariant normalized kernel in space and conformal time high symmetries of flrw essential to constructionflrw framework coarse - graining gradient scalar same as gradient coarse - grained scalar @xmath5 \phi(y )\d^4 y \\ = + \oint\nabla_x f(x - y)] \phi(y ) \d^4 y =\nabla_x f(x - y)\phi(y )\d^4 y =_x\phi(x )\end{aligned} translation invariance kernel crucial result calculate : @xmath6(x - y ) \phi(y ) \d^4 y \\ \neq & \langle \phi(x ) \rangle\end{aligned} careful @xmath7 `` averaging '' coarse - graining process average average not average variance with normal use averaging eschew term @xmath8 @xmath9 for ensemble averages normal rules averaging second line vanishes for coarse - graining process phrase `` coarse - graining '' more appropriate than averaging @xmath10 inequality derived from positivity coarse - graining kernel plus appeal to cauchy schwarz inequality : @xmath11 consider conformal deformation of metric : @xmath12 note standard results ricci tensor @xmath13 - g_ \nabla^2\phi+2|\nabla\phi|^2] ricci scalar @xmath14. einstein tensor @xmath15 + g_{ab 2\nabla^2\phi+|\phi|^2] apply generic coarse - graining - einstein tensor @xmath16. conformal mode @xmath17. deformed metric base metric @xmath18 \nonumber g_{ab\nabla^2\ln\langle e^{2\phi}\rangle+{1\over4}|^2\right].{aligned holds generic coarse - graining process details unspecified used assumed translation invariance kernel cflrw space - time smearing simplifies coarse - grain einstein tensor @xmath19. conformal mode @xmath17. @xmath18 \nonumber g_{ab[\nabla^2\ln\langle e^{2\phi}\rangle+{1\over4}^2\right].\end{aligned} combining cflrw coarse - graining @xmath20 + g_{ab } [ \langle|\nabla\phi|^2\rangle - \rangle|^2 ]{aligned specific high symmetry case of buchert coarse - graining technical details under control coarse - grained spacetime sees effective stress - energy @xmath21 effective stress tensor @xmath22 - g_{ab } [ \langle|\nabla\phi|^2\rangle -|^2] vanishes in limit smearing kernel becomes delta function coarse - graining switched off trace of buchert contribution to effective stress tensor @xmath23 not zero nonzero first term positive last three terms negative trace can be arranged positive or negative effective stress energy classical energy conditions null vectors : @xcite @xmath25 null energy condition satisfied coarse - graining timelike vectors : @xcite @xmath26 use tetrad decomposition @xmath27 timelike vector @xmath28 of interest @xmath29 positive definite euclidean signature metricusing flrw symmetries assert @xmath30 @xmath31 weak energy condition satisfied coarse - graining strong energy condition sec consider trace - reversed quantity @xcite @xmath32 timelike vectors @xmath33 strong energy condition satisfied coarse - graining cosmological setting large regions hope @xmath35 @xmath36 implying @xmath37 @xmath38 agrees sec calculation effective @xmath34 parameter @xmath39 contribution stress - energy @xmath40 studying buchert coarse - graining cflrw spacetimes help flrw background controlled buchert coarse - graining cflrw coarse - graining satisfies differential identities yields effective stress - energy not traceless classical energy conditions satisfied effective stress - energy buchert coarse - graining @xmath41 ensemble average key properties different s. r. green r. m. wald arxiv:1506.06452 t. buchert s. rsnen rev nucl part 62 2012 ) 57 f. r. ellis class 28 2011 ) 164001 t. buchert class 28 2011 ) 164007 arxiv:1103.2016 d. l. wiltshire class28 2011 ) 164006 t buchert carfora ellis kolb maccallum j ostrowski rsnen roukema andersson coley d.l wiltshire class 32 2015 ) 215021 t. buchert coley kleinert roukema d l wiltshire arxiv:1512.03313 astro-ph.co ostrowski f roukema arxiv:1512.02947 gr - qc m. visser gen 37 2005 ) 1541 gr - qc/0411131 visser science 276 1997 ) 88 arxiv:1501.01619 gr - qc visser d56 1997 ) 7578 gr - qc/9705070 visser qc/9710010 c. barcel doi:10.1142/9789812792129_0014 gr - qc/0001099 | 0.351733 | 2,425 | 1,282 | , Saving $0.1 in GPT-4. |
55,272 | spectral analyses of h - deficient post - agb stars have shown that a small group of four extremely hot objects exists which have almost pure he absorption - line spectra in the optical .
these are classified as o(he ) stars . for their evolution
there are two scenarios : they could be the long - sought hot successors of rcrb stars , which have not been identified up to now .
if this turns out to be true , then a third post - agb evolutionary sequence is revealed , which is probably the result of a double - degenerate merging process .
an alternative explanation might be that o(he ) stars are post early - agb stars .
these depart from the agb just before they experience their first thermal pulse ( tp ) which will then occur as a late thermal pulse ( ltp ) .
this would be a link to the low - mass he - enriched sdo stars and low - mass , particularly he - rich pg1159 stars . | 95% of all stars end their lives as white dwarfs .
about 20% of the hot post - agb stars are h deficient .
most of these are the result of a ( very ) late he - shell flash , but the evolutionary status of a fraction of about @xmath0% of the hottest h - deficient stars , namely the o(he ) stars @xcite , is as yet unexplained .
o(he ) stars form a small group of four extremely hot post - agb stars which exhibit absorption - line spectra dominated by heii . for a detailed introduction see @xcite and @xcite .
( v)ltp evolutionary models can explain the observed he / c / o abundances in wolf - rayet and pg1159 stars ( see , e.g. , * ? ? ?
* ) , but they have never reproduced he - dominated surface abundances .
recently , miller bertolami & althaus ( 2006 ) presented a `` numerical test '' where they increased the mass - loss rate of a ltp star artificially in order to `` blow away '' the h from the stellar surface .
however , this is in contradiction with radiation - driven wind theory @xcite and , if at all , valid for low - mass o(he ) stars only ( see below ) . as an alternative , a third post - agb
evolutionary sequence might exist and the o(he ) stars could be the successors of the rcrb stars which are relatively cool ( @xmath1 around 10000k ) stars with he - dominated atmospheres , too . if this is true , we can expect similar metal abundances . in order to investigate on these in o(he ) stars , we have performed fuv observations of all four objects with the far ultraviolet spectroscopic explorer ( fuse ) .
previous spectral analyses of o(he ) stars were based on optical , uv ( international ultraviolet explorer , iue ) , and x - ray ( rosat ) observations @xcite . the photospheric parameters are summarized in table [ tab : php ] . [
tab : php ] lcr@.lr@.lr@.lr@.lr@.l & & & & & & + & & & + lotr4 & 120 & 5&5 & 0&5 & 0&004 & 0&001 & 0&008 + hs1522 + 6615 & 140 & 5&5 & 0&1 & 0&003 & + hs2209 + 8229 & 100 & 6&0 & 0&2 & + k1 - 27 & 105 & 6&5 & 0&2 & 0&005 & 0&005 & + majority rcrb & & & 0&0001 & 0&010 & 0&004 & 0&005 + v854 cen & & & 0&5 & 0&030 & 0&0003 & 0&003 + since high - resolution and high - s / n uv spectra are a prerequisite for a reliable determination of metal abundances in hot stars , we have applied for observations with the hubble space telescope ( hst ) and the space telescope imaging spectrograph ( stis )
. we were awarded observational time in cycle 13 .
unfortunately , stis encountered a failure during that cycle on aug 4 , 2004 while our first observations were scheduled on aug 9 . a fuse campaign in 2002 ( cycle 3 , program c178 , exposure time about 40ksec ) was more successful .
for the analysis of the obtained fuv spectra we used tmap , the tbingen nlte model - atmosphere package @xcite , and calculated fully - line blanketed models in order to identify weak photospheric lines . in the course of this analysis
, it turned out that the spectra are strongly contaminated by interstellar absorption and the s / n was not sufficient for our purpose .
the mass - loss rates could be determined from the ovi @xmath21031.9 , 1037.6 resonance doublet @xcite and appear in agreement with radiation - driven wind theory .
one o(he ) star ( hs1522 + 6615 ) was included in a re - observation campaign ( cycle 7 , program u103 , exposure time about 4ksec ) and our own re - observation proposal ( cycle 8 , program h024 , 204ksec in order to improve the s / n ) was accepted .
these observations were scheduled for summer 2007 the fuse failure on july 12 , 2007 prevented them .
thus , at the moment , we can not determine the photospheric metal abundances of the o(he ) stars reliably .
the group of four known o(he ) stars divides into two sub - groups , two high - mass ( @xmath3 , the central star ( cs ) of the planetary nebula ( pn ) lotr4 and hs1522 + 6615 ) and two low - mass ( @xmath4 , cspn of k1@xmath527 and hs2209 + 8829 ) members .
its is possible that different evolutionary scenarios are valid for each group .
the low - mass o(he ) stars may be so - called post early - agb stars , i.e. they experienced the first thermal pulse ( tp ) after their departure from the asymptotic giant branch ( agb ) .
@xcite calculated an evolutionary track for a @xmath6 post early - agb star which closely matches the positions of the cspn of k1@xmath527 and of hs2209 + 822 in the @xmath7 plane ( fig .
[ fig : teffg ] ) . due to an artificially increased mass - loss rate ( ten times higher than predicted by radiation - driven wind theory )
, a strong h deficiency is achieved .
spectral analysis of the fuse spectra has shown that the mass - loss rates of o(he ) stars are not higher than predicted by radiation - driven wind theory @xcite and therefore , any change of the surface composition due to a stellar wind is unlikely . moreover , for the h abundance in these o(he ) stars , presently only upper limits are known ( tab . [
tab : php ] ) .
therefore a further , detailed investigation on a distinct `` sdo(he ) @xmath8 o(he ) @xmath8 do white dwarf '' evolutionary channel ( cf .
* ) is necessary .
however , it is worth to note that the do - type cspn pg1034 + 001 ( cf . * ? ? ?
[ fig : teffg ] ) may be a descendant from low - mass o(he ) or pg1159 stars .
the high - mass o(he ) stars appear to follow the `` normal '' born - again scenario @xcite where a ( very ) late tp causes the h - deficiency .
however , it is worthwhile to note that both known objects , the cspn of lotr4 as well as hs1522 + 6615 , show remaining h in their spectra ( tab . [
tab : php ] ) .
since their surface gravity @xmath9 is relatively low , gravitational settling during their following evolution might turn them into h - rich ( da ) white dwarfs .
therefore , it is unclear whether the extremely hot do white dwarf kpd0005 + 5106 ( fig .
[ fig : teffg ] , cf .
werner et al .
these proceedings ) is a successor of high - mass o(he ) stars .
an alternative o(he ) scenario is that of double - degenerate mergers .
similar h / he surface compositions suggest that o(he ) stars are the progeny of rcrb stars @xcite and they follow a `` rcrb @xmath8 o(he ) @xmath8 do white dwarf '' sequence .
available fuse spectra of o(he ) stars do not show isolated metal lines and thus , allow to give upper limits for metal abundances only .
iron - group abundances are probably solar .
is supported by the _
german astrophysical virtual observatory _
project of the german federal ministry of education and research ( bmbf ) under grant 05ac6vtb .
e.r . is supported by dfg grant we1312/30@xmath51 .
is supported by the fuse project , funded by nasa contract nas5@xmath532985 .
blcker , t. 1995 , a&a , 299 , 755 herwig , f. , blcker , t. , langer , n. , & driebe , t. 1999 , a&a , 349 , l5 iben , i. jr . ,
kaler , j. b. , truran , j. w. , & renzini , a. 1983 , apj , 264 , 605 mndez , r. h. 1991 , iau symp .
145 , kluwer , dordrecht , p. 375
miller bertolami , m. m. , & althaus , l. g. 2006 , a&a , 454 , 845 pauldrach , a. , puls , j. , kudritzki , r. p. , mndez , r. h. , & heap , s. r. 1988 , a&a , 207 , 123 rao , n. k. , & lambert , d. l. 1996 , the asp conference series vol .
96 ( san francisco : asp ) , p. 39
rauch , t. , kppen , j. , & werner , k. 1994 , a&a , 286 , 543 rauch , t. , kppen , j. , & werner , k. 1996 , a&a , 310 , 613 rauch , t. , dreizler , s. , & wolff , b. 1998 , a&a , 338 , 651 rauch , t. , reiff , e. , werner , k. , herwig , f. , koesterke l. , & kruk j. w. 2006 , in : _ astrophysics in the far ultraviolet , five years of discovery with fuse _ , eds . g. sonneborn , h. w. moos , b .-
andersson , the asp conference series vol .
348 ( san francisco : asp ) , p. 194
werner , k. , dreizler , s. , & wolff , b. 1995 , a&a , 298 , 567 werner , k. , deetjen , j. l. , dreizler , s. , nagel , t. , rauch , t. , & schuh , s. l. , 2003 , in : _ stellar atmosphere modeling _ , eds . i. hubeny , d. mihalas , k. werner , the asp conference series vol . 288
( san francisco : asp ) , p. 31 | 95% stars as white dwarfs 20% hot post - agb stars h deficient result of late he - shell flash evolutionary status of hottest h - deficient stars o(he ) stars unexplained o(he ) stars form group four hot post - agb stars absorption - line spectra dominated by heii see @xcite evolutionary models explain he / c / o abundances in wolf - rayet pg1159 stars never reproduced he - dominated surface abundances miller bertolami & althaus 2006 increased mass - loss rate of ltp star artificially to blow away h from stellar surface with radiation - driven wind theory valid for low - mass o(he ) stars only third post - agb evolutionary sequence might exist o(he ) stars could be successors of rcrb stars cool stars with he - dominated atmospheres if expect similar metal abundances performed fuv observations of all four objects with far ultraviolet spectroscopic explorer previous spectral analyses based on optical uv x - ray observationsphotospheric parameters summarized in table tab php lotr4 120 5&5 0 0&004 0&001 0&008 hs1522 6615 140 5&5 0&1 0&003 hs2209 8229 100 6&0 0&2 k1 - 27 105 6&5 0&2 0&005 rcrb 0&0001 0&010 0&004 0&005 v854 0&5 0&030 0&0003 0&003 high - resolution high - s / n uv spectra prerequisite for metal abundances in hot stars applied for observations hubble space telescope imaging spectrograph awarded time cycle 13 stis failure aug 4 2004 first observations aug 9 fuse campaign 2002 cycle 3 c178 successful analysis fuv spectra used tmap tbingen nlte model - atmosphere package @xcite calculated fully - line blanketed models identify weak photospheric lines spectra contaminated by interstellar absorption s / n not sufficient mass - loss rates determined from ovi @xmath21031.9 1037.6 resonance doublet @xcite radiation - driven wind theoryo(he ) star ( hs1522 + 6615 ) included in re - observation campaign cycle 7 program u103 exposure time 4ksec our re - observation proposal cycle 8 program h024 204ksec improve s / n accepted observations scheduled summer 2007 fuse failure july 12 2007 prevented determine photospheric metal abundances stars reliably four o(he ) stars into two sub - groups two high - mass @xmath3 + 6615 two low - mass @xmath4@xmath527 hs2209 + 8829 different evolutionary scenarios valid for group low - mass o stars post early - agb stars first thermal pulse departure branch calculated evolutionary track for @xmath6 post early - agb star matches positions cspn of k1@xmath527 hs2209 + 822 in @xmath7 plane fig to artificially increased mass - loss rate ten higher strong h deficiency achieved spectral analysis mass - loss rates stars not higher than radiation - wind theory change surface composition due to stellar wind unlikelyh abundance in o(he ) stars upper limits known further investigation on sdo @xmath8 o white dwarf evolutionary channel necessary do - type cspn pg1034 + 001 may be descendant from low - mass o(he ) or pg1159 stars high - mass o(he stars follow normal born - again scenario @xcite late tp causes h - deficiency known objects cspn of lotr4 hs1522 + 6615 show remaining h in spectra surface gravity @xmath9 low gravitational settling might turn into h - rich white dwarfs unclear whether hot do white dwarf kpd0005 + 5106 successor of high - mass o(he ) stars alternative scenario double - degenerate mergers compositions suggest o(he ) stars progeny of rcrb stars @xcite follow rcrb @xmath8 o(he ) @xmath8 do white dwarf '' sequence fuse spectra of o(he stars show isolated metal lines upper limits for metal abundances only iron - group abundances probably solar supported german astrophysical virtual observatory project federal ministry education research grant 05ac6vtb dfg grant we1312/30@xmath51 fuse project nasa contract nas5@xmath532985 blcker t 1995 a&a 299 , 755 herwig blcker langer driebe 1999 a&a 349 , l5 iben kaler truran renzini 1983 264 , 605 mndez r. 1991 145 kluwer dordrecht p. 375 miller bertolami althaus 2006 a&a 454 , 845 pauldrach puls kudritzki mndez r heap s. 1988 a&a 207 , 123 rao lambert d.. 1996 asp conference series vol 96 san francisco p. 39 rauch t. kppen j werner k 1994 a&a 286 , 543 rauch 1996 310 , 613 rauch dreizler wolff b. 1998 a&a 338 , 651 rauch reiff werner kherwig f koesterke kruk j 2006 astrophysics far ultraviolet five years discovery fuse eds sonneborn moos andersson asp conference series vol 348 san francisco asp p. 194 werner k. dreizler s wolff 1995 a&a 298 567 werner deetjen j. dreizler nagel t. rauch schuh s 2003 stellar atmosphere modeling eds hubeny d. mihalas werner asp conference series vol 288 san francisco asp p. 31 | 0.354316 | 2,393 | 1,251 | , Saving $0.1 in GPT-4. |
55,277 | the advent of deep , wide , accurate , digital photometric surveys exemplified by the sloan digital sky survey ( sdss ) has had a profound impact on studies of the milky way . in the past decade
, we have transitioned from a scarcity to an ( over)abundance of precise , well calibrated , observations of stars over a large fraction of the galaxy .
the avalanche of data will continue throughout this decade , culminating with gaia and lsst .
this new reality will necessitate changes in methodology , habits , and expectations both on the side of the large survey projects as well as the astrophysics community at large .
we argue , based on the experience with sdss , that surveys should release data as early and often as possible incorporating incremental improvements in each subsequent release , as opposed to holding off for a single , big , final release .
the scientific community will need to reciprocate by performing analyses and ( re - analyses ) appropriate to the current fidelity of the released data , understanding that these are continually evolving and improving products . | in the past decade , the data from the sloan digital sky survey ( sdss ) have drammatically enhanced our picture of the milky way .
this includes comprehensive galactic structure studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , discoveries of overdensities and tidal streams ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) as well as discoveries of a significant population of ultra faint milky way satellites ( e.g. * ? ? ?
* ; * ? ? ?
* and others ) .
an example from @xcite given in figure [ clumps ] shows a detection of two low - contrast disk overdensities in stellar number density maps constructed from sdss observations of @xmath0 million stars . the sdss is a digital photometric and spectroscopic survey .
it has covered a contiguous region of about 7,600 deg@xmath1 centered on the north galactic cap , a smaller but deeper area in the galactic south ( @xmath2 deg@xmath1 ) , and approximately @xmath3deg@xmath1 of imaging for the sloan extension for galactic understanding and exploration ( segue ) . more generally , it is an example of the kind of deep , wide and accurate surveys that will be increasingly common in this decade .
a non - exhaustive list includes pan - starrs ps1 @xcite that has already surpassed the sdss in terms of covered area in less than 6 months after the start of its science mission , skymapper ( keller et al . 2005 ) , currently in commissioning , the large sky area multi - object fibre spectroscopic telescope ( lamost ; * ? ? ?
* ) , the dark energy survey ( des ; * ? ? ?
* ) , and of course gaia @xcite and lsst @xcite .
all these , including gaia , share the common thread of aiming to produce and publish large ( @xmath4 objects ) and wide area datasets , a challenge the sdss has already faced and successfully tackled .
the lessons learned from sdss are many . here we limit ourselves to only two specific areas of immediate interest to gaia and its users : deciding when and what to publish , and appropriately approaching the published data .
the sdss has published the first public release of its data ( the `` early data release '' ; edr ) in june 2001 @xcite .
it consisted of roughly @xmath5 deg@xmath6 of imaging data , @xmath7 follow - up spectra ( about 5% of the planned totals ) and a photometric and astrometric catalog of @xmath814 million objects . by today
s sdss collaboration standards , the data included in this release would be considered substandard , even embarrassing . to illustrate this point , at the time of the edr even the true filter bandpasses were not known
, the photometric calibration was at 3 - 5% level ( as opposed to current @xmath9% ) , and the output formats and tables in the data distribution system were far from being finalized . in spite of these deficiencies ,
the edr was a major success ( and never criticized as premature ) .
first , it demonstrated that the survey was collecting and processing data ( an important milestone for a project 10 years in the making ) .
secondly , the release generated invaluable feedback from the community which was incorporated into , and significantly strenghtened the subsequent data releases .
most importantly , the data described in the edr produced substantial scientific returns , including ( of importance to galactic astronomy ) the discovery of the monoceros stream @xcite . at the time of writing of this contribution , the edr paper has been cited 1215 times .
the sdss has continued with releases on a 12 - 18 month schedule , with the latest ( 8@xmath10 ) data release planned for december 2010 .
besides adding more area , every new release involves reprocessing of _ all previously published data _ to correct problems identified in the older releases , as well as to benefit from major improvements in the processing software ( for example , the inclusion of an improved photometric calibration algorithm ; * ? ? ?
* ) . in sdss experience , many of the problems reported by users were very subtle and practically the only way to discover them was to perform cutting - edge " science analysis and then critically examine the results .
the sdss is far from alone in this approach .
the rave survey @xcite , as well as ukidss @xcite , follow the same strategy .
the planned large synoptic survey telescope ( lsst ) will follow a similar yearly release schedule .
this `` release early , release often '' approach is also well known in open source software development @xcite . the early data releases result in better communication and feedback from the wider astrophysical community , improve the quality of the published datasets , and reduce the overall `` time to science '' .
we feel the same model is applicable to gaia .
in addition to helping the gaia project team discover and correct problems as they appear , early releases will surely generate a significant amount of follow - up science , enable synergies with ground based surveys such as lsst ( see section [ lsst ] ) , as well as benefit the project in terms of education and public outreach .
the paradigm described above puts an important onus of responsibility on the users to take into account the evolving nature of the datasets . as will be noted elsewhere in this volume ( see the contribution by hogg ) , the only `` final '' product a survey can ever deliver are the _
raw images_. everything else , including the catalogs , is a _ derived product _ and subject to change as the understanding of the instrument , the processing algorithms , or the underlying assumptions ( priors ) evolve . to the best of our knowledge , no survey ever had all of these requirements well known early .
however , the benefits of the early and regular access to data as they are collected by far outweigh difficulties associated with evolving datasets .
the end users need to be aware of these caveats , especially when making use of the early data releases that will not have the full quality and reliability traditionally expected from a published catalog . of course , while certainly a potential source of frustration , good understanding of ( any ! ) dataset and its limits significantly improves the quality and longevity of the resulting science . again ,
early and frequent data releases not only help users to understand the survey error bars " , but perhaps more importantly help the project team to improve them before it is too late to fix the problem " .
, for a g2v star ( it is assumed that @xmath11 , where @xmath12 is the gaia s broad - band magnitude ) . in the top panel ,
the curve marked `` sdss '' corresponds to a single sdss observation .
the red curves correspond to gaia ; the long - dashed curve shows a single _ transit _ accuracy , and the dot - dashed curve the end of mission accuracy ( assuming 70 transits ) .
the blue curves correspond to lsst ; the solid curve shows a single _ visit _ accuracy , and the short - dashed curve shows accuracy for co - added data ( assuming 230 visits in the @xmath13 band ) .
the curve marked `` sdss - poss '' in the middle panel shows accuracy delivered by the proper motion catalog of @xcite . in the middle and bottom panels ,
the long - dashed curves correspond to gaia , and the solid curves to lsst .
note that lsst will smoothly extend gaia s error vs. magnitude curves four magnitudes fainter .
the assumptions used in these computations are described in eyer et al .
( in prep . ) .
] the large synoptic survey telescope , ( lsst ; * ? ? ?
* ) , will be a large , wide - field ground - based system designed to obtain multiple images covering the sky visible from its location at cerro pachn , chile . the current baseline design , with an 8.4 m ( 6.7 m effective ) primary mirror , a 9.6 deg@xmath6 field of view , and a 3,200 megapixel camera , will allow about 10,000 square degrees of sky to be covered using pairs of 15-second exposures in two photometric bands every three nights on average .
the survey area will include 30,000 deg@xmath6 with @xmath14 , and will be imaged multiple times in six bands , @xmath15 , covering the wavelength range 3201050 nm .
about 90% of the observing time will be devoted to a deep - wide - fast survey mode which will observe a 20,000 deg@xmath6 region about 1000 times in six bands during anticipated 10 years of operations .
these data will result in databases including about 20 billion objects .
lsst will produce a massive and exquisitely accurate photometric and astrometric dataset for about 10 billion milky way stars .
the coverage of the galactic plane will yield data for numerous star - forming regions , and the @xmath16 band data will penetrate through the interstellar dust layer .
photometric metallicity measurements will be available for about 200 million main - sequence f / g stars which will sample the halo to distances of 100 kpc @xcite .
no other existing or planned survey will provide such a massive and powerful dataset to study the outer halo ( including gaia which is flux limited at @xmath17 , and pan - starrs which will not have the @xmath18 band ) .
the lsst in its standard surveying mode will be able to detect rr lyrae and classical novae out to 400 kpc , and hence explore the extent and structure of the halo out to half the distance to m31 .
all together , the lsst will enable studies of the stellar distribution beyond the presumed edge of the galactic halo , of their metallicity distribution throughout most of the halo , and of their kinematics beyond the thick disk / halo boundary @xcite . in the context of gaia
, the lsst can be thought of as its deep complement .
a comparison of lsst and gaia performance is given in figure [ lsstvsgaia ] .
gaia will provide an all - sky catalog with unsurpassed trigonometric parallax , proper motion and photometric measurements to @xmath19 for about @xmath20 stars .
lsst will extend this map to @xmath21 over half of the sky , detecting about @xmath22 stars .
because of gaia s superb astrometric and photometric quality , and lsst s significantly deeper reach , the two surveys are highly complementary : gaia will map the milky way s disk with unprecedented detail , and lsst will extend this map all the way to the halo edge ( eyer et al .
, in prep ) .
mj wishes to acknowledge the support for this work provided by nasa through hubble fellowship grant hf-51255.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
i acknowledges support by nsf grant ast 07 - 07901 , and by nsf grant ast 05 - 51161 to lsst for design and development activity .
99 belokurov , v. , et al .
2006 , apjl , 642 , l137 belokurov , v. , et al .
2007 , apj , 654 , 897 bond , n. a. , et al .
2010 , apj , 716 , 1 carollo , d. , et al .
2007 , nature , 450 , 1020 flaugher , b. 2005 , international journal of modern physics a , 20 , 3121 ivezi , . ,
sesar , b. , juri , m. , et al . 2008a , apj , 684 , 287 ( i08 ) ivezi , . ,
tyson , j.a . ,
allsman , r , et al . 2008b , arxiv:0805.2366 juri , m. , ivezi , .
, brooks , a. , et al .
2008 , apj , 673 , 864 ( j08 ) kaiser , n. , et al .
2002 , proceedings of the spie , 4836 , 154 lawrence , a. , et al .
2007 , mnras , 379 , 1599 lsst science collaborations , et al .
2009 , `` lsst science book , v2.0 '' , arxiv:0912.0201 munn , j. a. , et al .
2004 , aj , 127 , 3034 newberg h.j . ,
yanny , b. , rockosi , c. , et al .
2002 , apj , 569 , 245 padmanabhan , n. , et al .
2008 , apj , 674 , 1217 perryman , m. a. c. , et al .
2001 , a&a , 369 , 339 raymond , e. s. , `` the cathedral and the bazaar '' , http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/ar01s04.html stoughton , c. , lupton , r.h . , bernardi , m. , et al .
2002 , aj , 123 , 485 su , d. q. , cui , x. , wang , y. , & yao , z. 1998 , proc .
spie , 3352 , 76 tyson , j. a. 2002 , proceedings of the spie , 4836 , 10 willman , b. , et al .
2005 , apjl , 626 , l85 zwitter , t. , et al .
2008 , aj , 136 , 421 | past decade data from sloan digital sky survey sdss ) enhanced picture milky way includes comprehensive galactic structure studies discoveries of overdensities tidal streams discoveries significant population ultra faint milky way satellites example from @xcite in figure shows two low - contrast disk overdensities in stellar number density maps from sdss observations of @xmath0 million stars sdss is digital photometric spectroscopic survey covered contiguous region of about 7,600 deg@xmath1 north galactic cap smaller deeper area in galactic south ( @xmath2 deg@xmath1 ) approximately @xmath3deg@xmath1 of imaging for sloan extension for galactic understanding exploration example of deep , wide accurate surveys increasingly common in this decade non - exhaustive list includes pan - starrs ps1 @xcite surpassed sdss covered area in less than 6 months after mission skymapper large sky area multi - object fibre spectroscopic telescope dark energy survey gaia @xcite and lsst @xcite share aiming to produce publish large ( @xmath4 objects wide area datasets challenge sdss faced successfully tackledlessons learned from sdss many limit to two areas gaia : deciding when to publish approaching published data sdss published first public release early data release ; edr in june 2001 @xcite consisted of @xmath5 deg@xmath6 imaging data @xmath7 follow - up spectra 5% of planned totals photometric and astrometric catalog of @xmath814 million objects s sdss standards data substandard embarrassing edr true filter bandpasses not known photometric calibration at 3 - 5% level current @xmath9% ) output formats tables data distribution system from finalized edr major success demonstrated survey collecting processing data important milestone for project 10 years release generated feedback community incorporated into strenghtened subsequent data releases data edr produced substantial scientific returns including discovery of monoceros stream @xcite edr paper cited 1215 times sdss continued with releases on 12 - 18 month schedule latest ( 8@xmath10 ) data release planned for december 2010adding more area new release involves reprocessing previously published data to correct problems in older releases from major improvements in processing software ( for improved photometric calibration algorithm in sdss experience problems reported were subtle only to discover was perform cutting - edge science analysis critically examine results sdss alone approach rave survey @xcite ukidss @xcite follow same strategy planned large synoptic survey telescope ( lsst ) similar yearly release schedule release early , release often '' approach known in open source software development @xcite early data releases result in better communication feedback astrophysical improve quality published datasets reduce time to science same model applicable to gaia helping gaia project team discover correct problems early releases generate follow - up science enable synergies with surveys lsst benefit project education public outreach paradigm puts onus responsibility on users to evolving nature of datasets only final product survey are raw images everything else including catalogs is derived product subject to change as understanding instrument processing algorithms assumptions evolve no survey had all these requirements known early benefits of early regular access to data outweigh difficulties evolving datasets end users of caveats especially early data releases full quality reliability from published catalog understanding of dataset limits improves quality longevity science early frequent data releases help understand survey error bars project team improve before late g2v star assumed @xmath11 @xmath12 gaia s broad - band magnitude top panel curve sdss single sdss observation red curves to gaia long - dashed curve single _ transit _ accuracy dot - dashed end of mission accuracy 70 transits blue curves to lsst solid curve single _ visit _ accuracy short - dashed curve accuracy for co - added data 230 visits in @xmath13 band curve sdss - poss '' shows accuracy by motion catalog @xcite middle bottom long - dashed curves correspond to gaia solid curves to lsst lsst extend gaia s error vs. magnitude curves four magnitudes fainter assumptions computations described in eyer et al .large synoptic survey telescope large wide - field ground system multiple images cerro pachn chile baseline design 8.4 m ( 6.7 m mirror 9.6 deg@xmath6 field of view 3,200 megapixel camera 10,000 square degrees sky 15-second exposures two bands every three nights survey area 30,000 deg@xmath6 @xmath14 imaged multiple times in six bands @xmath15 wavelength range 3201050 nm 90% observing time deep - wide - fast survey mode 20,000 deg@xmath6 region 1000 times in six bands during 10 years operations data databases 20 billion objects lsst photometric astrometric dataset for 10 billion milky way stars coverage galactic plane data star - forming regions @xmath16 band data interstellar dust layer photometric metallicity measurements for 200 million main - sequence f / g stars sample halo distances 100 kpc @xcite no other survey massive dataset outer halo gaia @xmath17 pan - starrs lsst standard mode rr lyrae classical novae to 400 kpc structure halo half distance to m31lsst studies stellar distribution beyond galactic halo metallicity distribution halo kinematics beyond disk / halo boundary @xcite gaia lsst its deep complement comparison lsst gaia in figure [ lsstvsgaia ] gaia all - sky catalog trigonometric parallax motion photometric measurements to @xmath19 @xmath20 stars lsst map to @xmath21 half sky detecting @xmath22 stars gaia s astrometric photometric quality lsst s deeper reach surveys complementary gaia milky way s disk detail lsst map to halo edge ( eyer et al . prep support nasa through hubble fellowship grant hf-51255.01-a space telescope science institute operated association of universities for research astronomy nasa contract nas 5 - 26555 support by nsf grant 07 - 07901 grant 05 - 51161 to lsst for design development belokurov . 2006 , 642 , l137 2007 654 , 897 bond 2010 , 716 1 carollo 2007 450 , 10202005 international journal modern physics 20 , 3121 ivezi sesar juri 2008a apj 684 , 287 ( i08 ) ivezi tyson allsman 2008b arxiv:0805.2366 juri ivezi brooks 2008 apj , 864 ( j08 ) kaiser 2002 proceedings spie , 4836 , 154 lawrence a 2007 379 , 1599 lsst science collaborations 2009 lsst science book v2.0 arxiv:0912.0201 munn j. a. 2004 127 , 3034 newberg yanny rockosi c 2002 apj 569 , 245 padmanabhan n 2008 apj 674 , 1217 perryman 2001 369 , 339 raymond cathedral bazaar '' http://www.catb.org/writings/cathedral-bazaar stoughton lupton bernardi m 2002 aj 123 , 485 su cui wang y yao z 1998 proc spie , 3352 , 76 tyson j. a2002 proceedings spie 4836 10 willman b. 2005 apjl 626 l85 zwitter 2008 136 421 | 0.504227 | 3,053 | 1,576 | , Saving $0.1 in GPT-4. |
55,281 | generalizing a covariant framework previously developed , it is shown that confinement insures that meson @xmath0 decay amplitudes vanish when both quarks are on - shell .
regularization of singularities in a covariant linear potential associated with nonzero energy transfers ( i.e. @xmath1 ) is improved .
macroft ettcom | even for a simple system such as a quark - antiquark ( referred to collectively as `` quarks '' ) bound state , color confinement implies that the matrix element for meson decay , @xmath2 , must vanish whenever this decay is kinematically possible .
this trivial statement can be realized by two possible mechanisms .
either * i ) * the quark propagators are free of timelike mass poles , as is usually assumed in euclidean metric based studies @xcite , or * ii ) * the vertex function of the bound state * vanishes * when both quarks are on - shell . in an earlier work
@xcite it was assumed that the quark propagators have mass poles , but it was not shown that relativistic generalization of the linear confining potential guaranteed the correct vanishing of the vertex function . in this work it is shown that the vanishing of the vertex function is a general feature of any confining interaction , and that * insisting on the correct nonrelativistic limit leads naturally to the second option . *
it was shown earlier@xcite that the nonrelativistic linear interaction @xmath3 can be expressed in momentum space by @xmath4-(2\pi)^3\delta^3(\vec{q } ) c , \label{nrpotential}\ ] ] where @xmath5 in coordinate space @xmath6 and hence the delta function subtraction cancels the infinite @xmath7 term , insuring that the limit @xmath8 exists and that the potential confines .
it also insures that the linear potential vanishes at the origin .
in fact , the delta function subtraction could be used to construct confining potentials for any monotonically increasing @xmath9 for which @xmath10 for some @xmath11 .
for example , after the delta function subtraction even the coulomb potential @xmath12 would result in a confining interaction . using the linear potential eq .
( [ nrpotential ] ) the schrdinger equation for two quarks of equal mass @xmath13 becomes @xmath14\psi(\vec{p})=-\int \frac{d^3k}{(2\pi)^3 } v_a(\vec{p}-\vec{k})\,\left[\psi(\vec{k})-\psi(\vec{p})\,\right]+ c\psi(\vec{p})\ , , \label{schroedinger}\ ] ] where @xmath15 is the binding energy .
a seemingly natural candidate for a relativistic generalization of the linear interaction ( [ nrpotential ] ) is to use the bethe - salpeter equation with a kernel given by @xmath16-(2\pi)^3\delta^4(q ) c\ , , \ ] ] where @xmath17 in eq .
( [ va ] ) .
unfortunately this form does not have the correct nonrelativistic limit .
therefore we rephrase the question : can we find a covariant equation that reduces to the schrdinger equation with linear interaction in the nonrelativistic limit ? in order to motivate the relativistic equation , start by defining the schrdinger _
function @xmath18 @xmath19\psi(\vec{p})\ , .\nonumber\ ] ] the schrdinger equation with linear interaction is then @xmath20 + mc\frac{\phi(\vec{p } ) } { \left(\vec{p}^2-me\right)}\ , .
\label{seq2}\end{aligned}\ ] ] introducing the four - vectors @xmath21 , @xmath22 , and @xmath23 , with @xmath24 and @xmath25 ( so that @xmath26 ) , the relativistic generalization of ( [ seq2 ] ) which reduces to it in the nonrelativistic limit ( when @xmath27 ) is @xmath28\nonumber\\ & + & 2mc\frac{\phi(p)}{m^2-(p - p)^2}\ , .
\label{geq}\end{aligned}\ ] ] = 2.0 in = 0.5 in this is the gross equation ( see fig .
[ gross.fig ] ) and the discussion shows that this is the most natural relativistic equation obtained from a generalization of the nonrelativistic schrdinger equation for confined particles . in a previous application of the gross equation@xcite
the kernel ( [ va ] ) , with @xmath17 , was used .
this choice is undesirable for use with the two channel version of this equation , where the mass shell constraints which fix @xmath29 introduce singularities for non - zero momentum transfers ( i.e. @xmath30 , but @xmath31 ) . to correct this problem the kernel @xmath32
is defined to be @xmath33 advantages of this form are many : * i ) * singularities are restricted to @xmath34 , * ii ) * interaction strength _
does not _ depend on the bound state momentum @xmath21 in the rest frame , * iii ) * it has the correct nonrelativistic dependence on @xmath35 , and * iv ) * the ultraviolet regularization used previously@xcite is no longer needed .
it is sufficient to consider the case when the constant term @xmath36 . if the mass of the bound state @xmath37
, then there exists a value of three momentum @xmath38 when both quarks can be on - shell . in this case @xmath39 , and the subtraction term in eq .
( [ geq ] ) appears to be singular .
this singularity is not cancelled by the first term , and hence , if the equation is to have a solution , the vertex function must be zero at @xmath40 . in particular , as @xmath41 , @xmath42 and the vertex function would have an infinite discontinuity at @xmath40 unless it were zero there .
we conclude that @xmath43 , when @xmath44 ( as illustrated in fig .
[ conf.fig ] ) .
= 2.0 in = 0.5 in a numerical confirmation of this result is shown in fig .
[ pistar.fig ] . in the actual calculations to be presented in an upcoming paper , the bound state equation
is averaged over positive and negative energy contributions and symmetrized by picking up the pole contributions of both constituents .
the average over positive and negative energy contributions leads to a two channel equation with a two component vertex function . in fig .
[ pistar.fig ] the solid line is the ( large ) component in which the off - shell quark has positive energy and the dashed line is the ( small ) component in which the off - shell quark has negative energy .
since a physical decay must produce two positive energy quarks , only the large component must have the `` confinement '' node .
[ pistar.fig ] shows the vertex functions for an excited pion with mass @xmath45 gev composed of two quarks with masses @xmath46 mev .
the large vertex function must have _ two _ nodes : one due to the excitation and another at @xmath47 gev@xmath48 , exactly where both particles are on - shell .
-0.3 in -0.5 in
the relativistic equation obtained from a generalization of the nonrelativistic schrdinger equation for confined particles is of the gross equation type rather than the bethe - salpeter equation type .
the confinement mechanism arises not from the lack of quark mass poles , but through the vanishing of the vertex function when both quarks are on their positive energy mass - shell .
because of this confinement mechanism , the vertex functions ( ground or excited state ) have one additional node if the bound state is heavy enough .
this work was supported in part by the us department of energy under grant no .
de - fg02 - 97er41032 . | simple system quark - antiquark bound state color confinement implies matrix element for meson decay @xmath2 vanish whenever decay possible statement realized by two mechanisms quark propagators free of timelike mass poles or vertex function vanishes when both quarks on - shell earlier assumed quark propagators have mass poles not shown relativistic generalization linear confining potential guaranteed correct vanishing vertex function vanishing vertex function general feature confining interaction insisting correct nonrelativistic limit leads to second option nonrelativistic linear interaction @xmath3 expressed in momentum space by @xmath4-(2\pi)^3\delta^3( @xmath5 coordinate space @xmath6 delta function subtraction cancels infinite @xmath7 term limit @xmath8 exists potential confines insures linear potential vanishes at origin delta function subtraction construct confining potentials for monotonically increasing @xmath9 @xmath10 @xmath11 after delta function subtraction coulomb potential @xmath12 in confining interaction linear potential eqschrdinger equation two quarks equal mass @xmath13 becomes @xmath14\psi(\vec{p})=-\int \frac{d^3k}{(2\pi)^3(\vec{p}-\vec{k})(-{p} c\psi(\vec{p}) @xmath15 binding energy candidate relativistic generalization linear interaction bethe - salpeter equation kernel @xmath16-(2\pi)^3\delta^4(q ) @xmath17 eq form correct nonrelativistic limit question covariant equation schrdinger equation linear interaction nonrelativistic limit ? motivate relativistic equation schrdinger _ function @xmath18 @xmath19\psi(\vec{p}) schrdinger equation with linear interaction @xmath20 + mc\frac{\phi(\vec{p } )(\vec{p}^2-me\right){seq2}{aligned introducing four - vectors @xmath21 @xmath24 @xmath26 ) relativistic generalization seq2 reduces nonrelativistic limit @xmath27 ) @xmath28\nonumber\\ 2mc\frac{\phi(p)}{m^2-(p - p)^2}\label{geq}\end{aligned} = 2.0 = 0.5 gross equation fig gross natural relativistic equation generalization nonrelativistic schrdinger equation for confined particles previous application equation kernel ( [ va ] ) @xmath17 used undesirable two channel version equation mass shell constraints @xmath29 introduce singularities for non - zero momentum transfers @xmath30 @xmath31 correct problem kernel @xmath32 defined @xmath33 advantages singularities restricted to @xmath34 interaction strength depend on bound state momentum @xmath21 correct nonrelativistic dependence on @xmath35 ultraviolet regularization used no longer needed consider case constant term @xmath36 mass bound state @xmath37 value of three momentum @xmath38 both quarks on - shell @xmath39 subtraction term in eq .geq ] ) singular not cancelled by first term equation solution vertex function must be zero at @xmath40 @xmath41 @xmath42 vertex function infinite discontinuity at @xmath40 unless zero @xmath43 @xmath44 fig . . = 2.0 in = 0.5 numerical confirmation shown in fig ..fig calculations bound state equation averaged over positive negative energy contributions symmetrized by pole contributions average leads to two channel equation two component vertex function solid line large component off - shell quark positive energy dashed line small quark negative energy physical decay must two positive energy quarks large component confinement '' node vertex functions for excited pion with mass @xmath45 gev two quarks with masses @xmath46 mev large vertex function must two nodes one due to excitation another at @xmath47 gev@xmath48 both particles on - shell -0.3 in -0.5 in relativistic equation for confined particles gross equation type bethe - salpeterconfinement mechanism arises lack quark mass poles but vanishing vertex function when quarks on positive energy mass shell vertex functions have additional node if state heavy work supported by us department of energy grant - fg02 - 97er41032 | 0.309799 | 1,678 | 890 | , Saving $0.0 in GPT-4. |
55,287 | the accuracy of calculations of atomic rydberg excitations can not be judged by the usual measures , such as mean unsigned errors of many transitions .
we show how to use quantum defect theory to ( a ) separate errors due to approximate ionization potentials , ( b ) extract smooth quantum defects to compare with experiment , and ( c ) quantify those defects with a few characteristic parameters .
the particle - particle random phase approximation ( pp - rpa ) produces excellent rydberg transitions that are an order of magnitude more accurate than those of time - dependent density functional theory with standard approximations .
we even extract reasonably accurate defects from the lithium rydberg series , despite the reference being open - shell .
our methodology can be applied to any rydberg series of excitations with 4 transitions or more to extract the underlying threshold energy and characteristic quantum defect parameters .
our pp - rpa results set a demanding challenge for other excitation methods to match . | an accurate description of electronic excited states has always been a major goal for theoretical and computational chemists .
nowadays , linear - response time - dependent density functional theory ( lr - tddft ) @xcite has become the standard workhorse for practical applications , because of the favorable balance between accuracy and computational cost . however , there are a variety of well - established limitations of tddft with the standard semilocal functionals @xcite .
these challenges include double excitations , charge transfer ( ct ) excitations , and rydberg excitations@xcite .
although many of these problems can be ameliorated by using range - separated @xcite or more accurate asymptotic - corrected response kernels @xcite , the results are not always satisfactory , and a more careful examination beyond the usual tabulations of mean absolute errors is required to check for an accurate rydberg series . quantum defect ( qd ) theory@xcite has long been applied as a test for atomic rydberg excitations@xcite . in the long history of quantum mechanics , the qd was originally introduced to empirically describe weak electron penetration effects relative to strong shielding effect @xcite . later , systematic derivations were given , creating a non - empirical theory @xcite .
the qd can also be applied to describe many molecular rydberg states @xcite . in the present work ,
we restrict ourselves to atomic rydberg series .
qds are often given as lists , with the index corresponding to the principal quantum number .
but the crucial feature of qds is that they are smoothly varying functions of the energy .
in fact , they even merge continuously with scattering phase shifts across the ionization threshold@xcite .
we use the smoothness feature to create our procedure for extracting qds from lists of excitation energies . in earlier work ,
van faassen and burke @xcite analyzed the singlet and triplet spin - states of the s- and p- series for atomic helium , beryllium and neon as test cases for standard tddft .
although excitation energy errors were no more than a few millihartree , the delicate qd results exposed their limitations . with a ground - state dft calculation that yields a potential with the correct asymptotic decay (
although see ref .
@xcite for a way around even this restriction ) , qds that were reasonably but not highly accurate were usually found with semilocal approximations to the kernel in tddft .
later , the d series in be was shown to be very poorly described by tddft@xcite .
recently , two of us developed the particle - particle random phase approximation ( pp - rpa ) and the particle - particle tamm - dancoff approximation ( pp - tda ) theory for calculation of challenging excitations such as double excitations , charge transfer excitations , rydberg excitations , and excitations in diradicals .
good excitation energies were typically found with a relatively low @xmath0 computational cost , where @xmath1 is the number of virtual orbitals.@xcite thus pp - rpa and pp - tda are promising methods that complement standard lr - tddft , especially for these challenging cases . in the present work ,
we develop qd analysis to more stringently test these methods , and compare them to tddft .
.[beexci ] first 11 excitation energies in be @xmath2p series . for transitions to @xmath3 and higher ,
the values have been extracted from the fits to the associated qds given later ( for alda , this begins at @xmath4 ) . [ cols="^,^,^,^",options="header " , ] the fact that pp - rpa and pp - tda are able to describe charge transfer and rydberg excitations is often attributed to their coulomb and exchange kernels which are asymptotically correct . however
, the orbital energies from the two - electron deficient reference also play a vital role .
although the coulomb and exchange kernel is the same , the pp - rpa and pp - tda with dft references substantially underestimate the rydberg excitation energies @xcite as a result of the poor kohn - sham orbital energies , making it meaningless to further look into the qd , just as in the tddft case . because pp - rpa is only a first - order approximation to the adiabatic connection - pairing matrix fluctuation theory
, it remains sensitive to the properties of the reference calculation .
a well - behaved reference , such as hf in this atomic rydberg excitation case , is needed for pp - rpa and pp - tda to produce meaningful qds . however , for low - energy excitations in molecules , dft references lead to significantly better results.@xcite
in this paper , we have developed a very general approach for extracting both threshold energies and qds from limited series of rydberg excitations .
we have demonstrated that our procedure can be used to extract extremely accurate threshold energies ( errors in the 0.1 mh range ) , and maximum errors of order 0.01 for qds .
we have shown why measures like mean unsigned errors in collections of transition frequencies are not useful for rydberg series , and how to parametrize and quantify errors in qd s .
we find that pp - rpa with a hf reference greatly outperforms tddft with a local approximation to the exchange - correlation kernel .
the results reported here should become the benchmark for approximate calculations of rydberg series .
our qd extraction procedure can be applied to other series , such as excitions in solids .
any quantum chemical method for excitations should be measured against our pp - rpa results for these atoms .
we suspect they will be difficult be beat with the same level of generality and low computational cost .
y.y . appreciates the support as part of the center for the computational design of functional layered materials , an energy frontier research center funded by the u.s .
department of energy , office of science , basic energy sciences under award de - sc0012575 .
k.b . was supported by the u.s department of energy ( doe ) , of- fice of science , basic energy sciences ( bes ) under award de - fg02 - 08er46496 .
w.y . was supported by the national science foundation ( che-1362927 ) .
m. e. casida , time - dependent density functional response theory of molecular systems : theory , computational methods , and functionals , in _ recent developments and applications of modern density functional theory _ , edited by j.
seminario , volume 4 of _ theoretical and computational chemistry _
, pages 391 439 , elsevier , 1996 . | accurate description of electronic excited states major goal for chemists linear - response time - dependent density functional theory ) @xcite standard for applications balance between accuracy computational cost limitations of tddft with standard semilocal functionals @xcite challenges include double excitations charge transfer excitations rydberg excitations problems range - separated @xcite or accurate asymptotic - corrected response kernels results not always satisfactory careful examination beyond required for accurate rydberg series quantum defect ( qd ) theory@xcite applied test for atomic rydberg excitations introduced describe weak electron penetration effects strong effect derivations non - empirical theory molecular rydberg states present work restrict to atomic rydberg series qds given as lists index to principal quantum number smoothly varying functions of energy merge continuously with scattering phase shifts across ionization threshold smoothness feature procedure for extracting qds from lists excitation energies van faassen and burke analyzed singlet triplet spin - states of s- p- series for atomic helium beryllium neon for standard tddftexcitation energy errors few millihartree qd results exposed limitations ground - state dft calculation potential correct asymptotic decay qds not accurate found with semilocal approximations to kernel in tddft d series in be poorly described by tddft@xcite developed particle - particle random phase approximation tamm - dancoff approximation - tda theory for challenging excitations double charge transfer rydberg diradicals good excitation energies found low @xmath0 computational cost @xmath1 number virtual orbitals pp - rpa pp - tda promising methods standard lr - tddft challenging cases develop qd analysis to test methods compare to tddft first 11 excitation energies in be @xmath2p series transitions to @xmath3 higher values extracted from fits associated qds begins at @xmath4 pp - rpa pp - tda describe charge transfer rydberg excitations to coulomb exchange kernels asymptotically correct orbital energies from two - electron deficient reference vital rolecoulomb exchange kernel same pp - rpa pp - tda with dft references underestimate rydberg excitation energies poor kohn - sham orbital energies meaningless qd tddft case pp - rpa first - order approximation adiabatic connection - pairing matrix fluctuation theory sensitive to properties reference calculation well - behaved reference hf rydberg needed for pp - rpa pp - tda produce qds low - energy excitations dft references better results developed general approach for extracting threshold energies qds from rydberg excitations accurate threshold energies errors 0.1 mh range maximum errors 0.01 for qds shown mean unsigned errors transition frequencies not useful for rydberg series parametrize quantify errors qd pp - rpa with hf reference outperforms tddft local approximation exchange - correlation kernel results benchmark for calculations rydberg series qd extraction procedure applied to other series solids quantum chemical method excitations measured against pp - rpa results difficult beat generality low computational cost support center for computational design of functional layered materials fundeddepartment energy office science basic energy sciences award de - sc0012575 supported u.s department energy science basic energy sciences - fg02 - 08er46496 national science foundation-1362927 casida time - dependent density functional response theory molecular systems computational methods recent developments applications modern density functional theory j. seminario volume 4 theoretical computational chemistry pages 391 439 elsevier 1996 | 0.275504 | 1,423 | 706 | , Saving $0.0 in GPT-4. |
55,298 | the aim of this work is the derivation of two approximated expressions for the two dimensional gaussian q - function , @xmath0 .
these expressions are highly accurate and are expressed in closed - form .
furthermore , their algebraic representation is relatively simple and therefore , convenient to handle both analytically and numerically .
this feature is particularly useful for two reasons : firstly because it renders the derived expressions useful mathematical tools that can be utilized in numerous analytic performance evaluation studies in digital communications under fading ; secondly because the two dimensional gaussian q - function is neither tabulated nor a built - in function in popular mathematical software packages such as @xmath1 , @xmath2 and @xmath3 . | @xmath4 + special functions are invaluable mathematical tools in natural sciences and engineering .
their wide use in analytical performance evaluation studies in wireless communications often allows the derivation of closed - form relationships for vital performance measures such as channel capacity and probability of error .
in addition , the majority of special functions are built - in functions in popular mathematical software packages . as such , the algebraic representation of any related expressions as well as their computation have been significantly simplified .
+ it is recalled that special functions play a significant role in the area of digital communications [ 11]-[13 ] and [ 16]-[19 ] and the references therein .
such a special function is also the two dimensional gaussian q - function .
it is denoted as @xmath0 and is constituted by the two - dimensional gaussian integral , [ 1]-[5 ] .
alternative integral representations for @xmath0 along with performance bounds were reported in [ 6]-[10 ] .
furthermore , an exact infinite series representation , which is expressed in terms of the incomplete gamma function and elementary functions , was derived in [ 11]-[12 ] .
nevertheless , it is noted that this series is the only explicit relationship for the two dimensional gaussian q - function since @xmath0 is neither expressed in terms of other elementary and/or special functions , nor it is included as a built - in function in popular mathematical software packages such as @xmath1 , @xmath2 and @xmath3 . as a consequence ,
both its computation and analytical tractability are not as straightforward as in the case of most special functions in communication theory .
+ motivated by this , the aim of this work is the derivation of two tight approximations for the @xmath0 function .
the high accuracy of the offered expressions is validated through comparisons with results obtained from numerical integrations as well as by the identity @xmath5 .
an important feature of the proposed approximations is that they are expressed in closed - form and have a rather simple algebraic representation . as a result
, they are expected to be useful in various analytical studies relating to the performance evaluation of digital communications over fading environments .
the two dimensional gaussian q - function is defined as : @xmath6 according to [ 2 ] , the above relationship can be also represented in the more desired craig form as follows , @xmath7 for the special case that @xmath8 , equation @xmath9 reduces to [ 2 ] , @xmath10 whereas for the case that @xmath11 , it reduces to the product of the one - dimensional gaussian q - functions of its arguments , namely , @xmath12 an exact series representation for @xmath0 was reported in [ 11]-[12 ] , namely , @xmath13
it was mentioned above that the only explicit representation of @xmath0 is the infinite series in @xmath14 .
the algebraic form of this series is relatively simple .
however , the fact that it has an infinite form typically raises convergence issues .
more specifically , the number of terms required for truncating the series adequately is sufficiently related to its parameters .
this is evident by the fact that for small values of @xmath15 , @xmath16 or @xmath17 , only a few number of terms are required to truncate the series with small error . on the contrary , as as the value of @xmath16 and @xmath17 increases , the numbers of terms required for a small truncation error also increase . as a consequence ,
the required computation time increase as well .
+ an effective mean to resolve such issues is the derivation accurate approximations that are expressed in closed - form and are also analytically tractable .
such approximated expression can be ultimately derived with the aid of two approximations for the one dimensional q - function , @xmath18 , that were proposed in [ 11]-[13 ] , namely , @xmath19 and @xmath20 the behaviour and high accuracy of @xmath21 and @xmath22 with respect to @xmath18 are illustrated in figure @xmath23 .
this can be also justified by figure @xmath24 which depicts the level of the involved absolute error and absolute relative error , namely , @xmath25 and @xmath26 a closed - form approximation for the @xmath0 function can be derived as follows : the double integral in @xmath27 can be alternatively expressed as , [ 11 ] , @xmath28dv\ ] ] consequently , by integrating once by part the inner integral and carrying out some straightforward algebraic manipulations , the following equivalent representation is deduced , @xmath29 thus , it is evident that an explicit expression for @xmath30 is subject to analytic evaluation of the semi - infinite integral , @xmath31 where @xmath32 . to this effect , by substituting @xmath21 into @xmath33 , one obtains @xmath34 which after algebraic manipulations yields the following closed - form expression , @xmath35 where , @xmath36 and @xmath37 it is recalled here that the @xmath0 function is not a built - in function in mathematical software packages such as @xmath1 , @xmath2 and @xmath3 .
however , the following identity holds by definition , @xmath38 . as a consequence ,
the accuracy of equation @xmath39 is assessed by means of the this identity in figures @xmath40 and @xmath41 .
one can clearly observe the excellent agreement between the plotted curves for almost all cases .
this is also evident by the level of the involved absolute and absolute relative which for all cases is less than @xmath42 . by following the same methodology , a second closed - form approximation for the two dimensional q - function can be derived . to this end , by making the necessary transformation of variables in @xmath22 and then substituting in @xmath33 , the following relationship is deduced , @xmath43 @xmath44 @xmath45 evidently , the derivation of a closed - form expression for @xmath46 is subject to analytic evaluation of the three involved integrals .
notably , these integrals have the same algebraic form as the @xmath47 .
therefore , after some straightforward algebraic manipulation , one obtains the following closed - form expression , the behaviour of @xmath48 is illustrated in figures @xmath49 and @xmath50 along with the @xmath18 function .
one can observe the excellent agreements between the theoretical and approximated results .
this agreement is also evident by the level of the involved absolute error and absolute relative error which is less than @xmath51 for almost all parametric values .
@xmath52 @xmath53 @xmath54 where , @xmath55 @xmath56 @xmath57 @xmath58 @xmath59 and @xmath60
in this work , we derived two novel closed - form approximations for the two dimensional gaussian q - function , @xmath0
. two critical features of these expressions are their high accuracy and their relatively simple algebraic form .
these features ultimately render them rather convenient to handle both analytically and numerically .
the latter is particularly important since the two dimensional gaussian q - function is not a built - in function in popular mathematical software packages . as a consequence , the derived expression can be considered a useful mathematical tool in various analytical studies in natural sciences and engineering , in general , and particularly in applications in communication theory relating to digital communications over fading channels .
j. g. proakis , _ digital communications _ , 3rd ed .
new york : mcgraw - hill , 1995 . m. k. simon and m. -s .
alouni , _ digital communication over fading channels _
, new york : wiley , 2005 m. k. simon , _ probability distributions involving gaussian random variables : a handbook for engineers scientists and mathematicians _ , springer , 2006 s. park and s.h .
cho , _ sep performance of coherent mpsk over fading channels in the presence of phase / quadrature error and i - q gain mismatch _ , ieee trans .
1088 - 1091 , jul . 2005 .
m.k . simon and m
alouini , _ a unified approach to the performance analysis of digital communications over generalized fading channels _ , proc .
ieee , vol .
9 , sep . 1998 , pp . 1860 - 1877
m.k . simon and m
alouini , _ a unified approach to the performance analysis of digital communications over generalized fading channels _ , proc .
ieee , vol .
1998 , pp . 1860 - 1877
m. k. simon , _ a simpler form of the craig representation for the two - dimensional joint gaussian q - function _ , ieee commun . lett .
, vol . 6 , no .
49 - 51 , feb .
s. yousefi and b. holmes , _ a simple form for the two - dimensional q - function suitable for performance evaluation of communication systems _ , in proc . ieee vtc05 spring , vol .
2 , stockholm , sweden , may 2005 , pp .
1091 - 1095 s. park and u. j. choi , _ a generic craig form for the two - dimensional gaussian q - function _ , etri j. , vol .
516 - 517 , aug .
2007 g. t. f. de abreu , _ jensen - cotes upper and lower bounds on the gaussian q - function and related functions _
, ieee trans .
on commun , vol .
3328 - 3338 , nov .
2009 p. c. sofotasios , _ on special functions and composite statistical distributions and their applications in digital communications over fading channels _ , ph.d dissertation , university of leeds , uk , 2010 p. c. sofotasios , s. freear , _ novel expressions for the one and two dimensional gaussian q - functions
_ , in proc .
ieee icwits , aug./sep .
2010 , hawaii , usa p. c. sofotasios , s. freear , _ novel expressions for the marcum and one dimensional q - functions _ , in proc .
@xmath61 iswcs , sep .
2010 , york , uk i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series , and products _ , @xmath61 ed .
new york : academic , 2007 .
m. abramowitz and i. a. stegun , _ handbook of mathematical functions with formulas , graphs , and mathematical tables .
_ , new york : dover , 1974 . | @xmath4 special functions invaluable mathematical tools in natural sciences engineering use in performance wireless communications allows derivation of closed - form relationships for performance measures channel capacity probability of error majority special functions built - in in mathematical software packages algebraic representation computation simplified special functions significant role in digital communications [ 11]-[13 ] 16]-[19 ] special function is two dimensional gaussian q - function denoted as @xmath0 constituted by two - dimensional gaussian integral [ 1]-[5 ] alternative integral representations for performance bounds reported in [ 6]-[10 ] exact infinite series representation derived in [ 11]-[12 ] series only explicit relationship for two dimensional gaussian q - function @xmath0 expressed in built - in in software packages computation analytical tractability not straightforward most communication theory aim of work is derivation of two tight approximations for @xmath0 function high accuracy of expressions validated through comparisons with numerical integrations identity @xmath5 proposed approximations expressed closed - form simple algebraic representation expected useful in analytical studies performance evaluation digital communications fading environments two dimensional gaussian q - function defined as @xmath6 2 ] relationship represented craig form, @xmath7 @xmath8 equation @xmath9 reduces to [ 2 ] @xmath10 @xmath11 reduces to product one - gaussian q - functions @xmath12 exact series representation for @xmath0 reported in [ 11]-[12 ] @xmath13 only explicit representation of @xmath0 infinite series in @xmath14 algebraic form series simple infinite form raises convergence issues number of terms required for truncating series related to parameters small values of @xmath15 @xmath16 @xmath17 few terms required truncate small error value @xmath16 @xmath17 increases terms required for small truncation error increase required computation time increase effective mean resolve issues derivation accurate approximations expressed closed - form analytically tractableapproximated expression derived two approximations for one dimensional q - function @xmath18 proposed in 11-[13 ] @xmath19 @xmath20 behaviour accuracy of @xmath21 @xmath22 illustrated in figure @xmath23 justified by figure @xmath24 absolute relative error @xmath25 and @xmath26 closed - form approximation for @xmath0 function derived double integral in @xmath27 expressed as @xmath28dv integrating inner integral algebraic manipulations equivalent representation deduced , @xmath29 explicit expression for @xmath30 subject to analytic evaluation of semi - infinite integral @xmath31 substituting @xmath21 into @xmath33 obtains @xmath34 yields closed - form expression @xmath35 @xmath36 @xmath37 @xmath0 function not built - in in mathematical software packages identity holds , @xmath38 accuracy of equation @xmath39 assessed identity in figures @xmath40 and @xmath41 agreement between plotted curves for all cases evident by level involved absolute relative less than @xmath42following same methodology second closed - form approximation for two dimensional q - function derived variables in @xmath22 substituting in @xmath33 relationship deduced @xmath43 @xmath45 derivation closed - form expression for @xmath46 subject to analytic evaluation of three integrals integrals same algebraic form as @xmath47 manipulation obtains closed - form expression behaviour of @xmath48 illustrated in figures @xmath49 @xmath50 @xmath18 function agreements between theoretical approximated results absolute error relative error less than @xmath51 for all parametric values @xmath52 @xmath60 derived two closed - form approximations for two dimensional gaussian q - function @xmath0 high accuracy simple algebraic form convenient to handle analytically numerically two gaussian q - function not built - in in popular mathematical software packages derived expression useful mathematical tool in analytical studies natural sciences engineering communication theory digital communications j. g. proakis digital communications 3rd ed new york mcgraw - hill 1995 m. k. simon m. -salouni digital communication fading channels new york wiley 2005 simon probability distributions gaussian random variables handbook engineers scientists mathematicians springer 2006 park cho performance mpsk fading channels phase quadrature error i - q gain mismatch ieee trans 1088 - 1091 jul 2005 simon alouini unified approach performance analysis digital communications fading channels ieee vol 9 sep 1998 pp 1860 - 1877 simon performance digital communications fading channels 1998 1860 - 1877 simpler form two - gaussian q - function commun lett vol. 6 no 49 - 51 feb yousefi b holmes simple form two - q - function performance evaluation communication systems ieee vtc05 spring vol 2 stockholm sweden may 2005 pp 1091 - 1095 park choi generic form two - dimensional gaussian q - function etri j. vol 516 - 517 aug 2007 de abreu jensen - cotes upper lower bounds gaussian q - function related functions commun vol 3328 - 3338 nov 2009sofotasios special functions composite statistical distributions applications digital communications ph.d dissertation university leeds uk 2010 sofotasios freear novel expressions one two dimensional gaussian q - functions icwits aug./sep. 2010 hawaii usa sofotasios freear novel expressions marcum one dimensional q - functions @xmath61 sep 2010 york uk gradshteyn ryzhik table integrals series products @xmath61 new york academic 2007 abramowitz stegun handbook mathematical functions formulas mathematical tables new york dover 1974 | 0.356565 | 2,325 | 1,178 | , Saving $0.1 in GPT-4. |
55,302 | in this note , we prove that every open primary basic semialgebraic set is stably equivalent to the realization space of a neighborly simplicial polytope .
this in particular provides the final step for mnv s proof of the universality theorem for simplicial polytopes . | mnv s universality theorem was a fundamental breakthrough in the theory of oriented matroids and convex polytopes .
it states that the realization spaces of oriented matroids and polytopes , i.e. the spaces of point configurations with fixed oriented matroid / face lattice , can be arbitrarily complex .
it comes in four flavours : [ thm : universality ] let @xmath0 be a primary basic semialgebraic set defined over @xmath1 , then [ it : uom ] there is an * oriented matroid * of rank @xmath2 whose realization space is stably equivalent to @xmath0 , and [ it : up ] there is a * polytope * whose realization space is stably equivalent to @xmath0 ; if moreover @xmath0 is open , then [ it : uuom ] there is a * uniform oriented matroid * of rank @xmath2 whose realization space is stably equivalent to @xmath0 , and [ it : usp ] there is a * simplicial polytope * whose realization space is stably equivalent to @xmath0 .
mnv announced this theorem in 1985 @xcite and published a sketch of the proof in 1988 @xcite .
a more detailed reasoning can be found in his thesis @xcite ( in russian ) .
shor @xcite simplified a key step in mnv s line of reasoning for part and .
moreover , part of theorem [ thm : universality ] was later elaborated upon by richter - gebert @xcite and gnzel @xcite , who proved the stronger _
universal partition theorem _ for oriented matroids . using lawrence extensions to rigidify the face lattices ,
it is easy to prove part from part @xcite@xcite . here
, a face lattice is called _ rigid _ if it uniquely determines the oriented matroid defined by its vertices . additionally , theorem [ thm : universality ] was generalized greatly by richter - gebert , who proved that already @xmath3-dimensional polytopes are universal @xcite@xcite . for a proof of part , in contrast , only mnv s original papers are available , apart of some preliminary results of sturmfels @xcite and bokowski guedes de oliveira @xcite .
moreover , mnv s elaborations for this case in @xcite@xcite are specially concise and , in our opinion , incomplete . hence , we think part of theorem [ thm : universality ] , although widely believed to be true , should be considered open until now .
it is important to stress that , despite the wrong common belief , lawrence extensions _ can not _ be used to deduce the universality theorem for simplicial polytopes .
we use a different approach to rigidify matroids , namely one based on a result of shemer proving rigidity of neighborly polytopes @xcite .
in particular , we establish here that _
neighborly _ polytopes , i.e. @xmath4-polytopes with a complete @xmath5-skeleton , are universal .
even more , we obtain this result with even - dimensional polytopes .
since all neighborly polytopes of even dimension are simplicial , this provides a proof of theorem [ thm : universality ] .
[ thm : unin ] every open primary basic semialgebraic set defined over @xmath1 is stably equivalent to the realization space of some neighborly @xmath6-dimensional polytope on @xmath7 vertices .
this is in contrast to the case of cyclic polytopes , who have trivial realization spaces ( * ? ?
* example 5.1 ) .
this holds more generally for all totally sewn and gale sewn polytopes @xcite@xcite , of which cyclic polytopes are a particular case .
these are large families of neighborly polytopes . indeed ,
the number of gale sewn polytopes with @xmath8 vertices in a fixed dimension @xmath4 is asymptotically at least @xmath9 , which is the current best lower bound for the number of _ all _ polytopes @xcite .
[ thm : sewn ] the ( homogenized ) realization space of any even - dimensional neighborly @xmath4-polytope on @xmath8 vertices constucted with the extended sewing or the gale sewing construction is contractible , and in fact an open , piecewise smooth @xmath10-ball . encouraged by the work of richter - gebert
, we make the daring conjecture that universality holds even when we restrict to neighborly @xmath3-dimensional polytopes .
every open primary basic semialgebraic set defined over @xmath1 is stably equivalent to the realization space of some neighborly ( and hence simplicial ) @xmath3-polytope .
a universality conjecture for simplicial @xmath3-polytopes is supported by the existence of simplicial @xmath3-polytopes without the isotopy property , that is , with disconnected realization space @xcite .
the following idea provides further motivation for our conjecture .
altshuler and steinberg proved that vertex figures of neighborly @xmath3-polytopes are always @xmath2-dimensional stacked polytopes @xcite .
despite the fact that realization spaces of stacked polytopes are trivial , their oriented matroids can be complicated : notice that if @xmath11 is any planar point configuration , then there exists a stacked @xmath2-polytope @xmath12 with a distinguished vertex @xmath13 such that the contraction of @xmath13 in @xmath12 coincides with @xmath11 .
now , realization spaces of planar point configurations are universal , and it is conceivable that these realization spaces can be captured by constructing neighborly @xmath3-polytopes having them as edge contractions , combined with the fact that all neighborly 4-polytopes are rigid .
the _ realization space _ @xmath14 of an oriented matroid @xmath11 is the set of vector configurations that share the same oriented matroid , i.e. if @xmath11 is of rank @xmath4 , and @xmath15 is the ground set of @xmath11 , then @xmath16 similarly , if @xmath12 is a polytope in @xmath17 , and @xmath18 is its vertex set , then we define its ( homogenized ) realization space as @xmath19 hence , the realization space @xmath20 of a polytope is the union of the realization spaces of all oriented matroids whose ( las vergnas ) face lattice ( i.e. the dual to the lattice of positive cocircuits ) coincides with the face lattice of @xmath12 , see ( * ? ? ?
a _ basic semialgebraic set _ in @xmath17 is the set of solutions to a finite number of rational polynomial equalities and inequalities ; it is called _ primary _ if all the inequalities in its definition are strict . a basic semialgebraic set
@xmath21 is a
_ stable projection _ of a basic semialgebraic set @xmath22 if , for the projection @xmath23 , we have that @xmath24 and that for every @xmath25 , the fiber @xmath26 is the relative interior of a non - empty polyhedron defined by equalities and strict inequalities that depend polynomially on @xmath27 .
two basic semialgebraic sets @xmath28 and @xmath29 are _ rationally equivalent _ if there is a homeomorphism @xmath30 such that @xmath31 and @xmath32 are rational functions . two basic semialgebraic sets @xmath28 and @xmath29 are _ stably equivalent _ if they belong to the same equivalence class generated by stable projections and rational equivalences .
we denote the stable equivalence of @xmath28 and @xmath29 as @xmath33 .
we refer to @xcite@xcite for more detailed definitions of these concepts . a technique that pervades our proofs is the use of _ lexicographic extensions _
( see ( * ? ? ? * section 7.2 ) ) .
when @xmath11 is realized by a vector configuration @xmath0 , then its lexicographic extension by @xmath34 $ ] , where @xmath35 are elements of @xmath0 and @xmath36 are signs , is realized by adjoining to @xmath0 the vector @xmath37 for any @xmath38 small enough ( cf .
figure [ fig : lex ] ) .
the following lemma is straightforward , see also ( * ? ? ?
* lemma 8.2.1 and proposition 8.2.2 ) .
[ lem : lex ] let @xmath11 denote any oriented matroid , and let @xmath39 $ ] denote a lexicographic extension of @xmath11 . then the projection induced by deletion @xmath40)\ \ & \longrightarrow\ \ { \mathscr{r}}(m)\\ x\ \ & \longmapsto\
\ x_{|e(m)}\end{aligned}\ ] ] is surjective , and its fibers are ( polynomially parametrized ) polyhedra of dimension @xmath41\}$ ] .
we start with an observation of mnv / shor concerning universality of uniform oriented matroids @xcite@xcite , which is proven using constructible oriented matroids and a substitution technique from @xcite@xcite .
the key step of the proof of theorem [ thm : unin ] is to use a construction of kortenkamp , who proved that every @xmath4-dimensional point configuration of at most @xmath44 points appears as a face figure of a neighborly polytope @xcite . for larger point configurations ,
this is still an open problem , first asked by perles .
for the proof , recall that in oriented matroid theory a polytope is called _ rigid _ if the face lattice of @xmath12 determines the oriented matroid @xmath11 spanned by the vertices of @xmath12 ( see ( * ? ? ?
* section 6.6 ) ) .
in particular , for all rigid polytopes , we have @xmath45 . by a theorem of kortenkamp ( * ? ? ?
* theorem 1.2 ) , every realizable oriented matroid of rank @xmath2 can be extended to the gale dual of an even - dimensional neighborly polytope by performing @xmath8 lexicographic extensions , obtaining a rank @xmath2 matroid @xmath48 on @xmath7 elements . by lemma [ lem : lex ]
we obtain @xmath49 let @xmath12 be the face lattice of the gale dual of @xmath50 .
now , oriented matroid duality preserves realization spaces and , by ( * ? ? ?
* theorem 2.10 ) and ( * ? ? ?
* theorem 4.2 ) , every neighborly polytope of even dimension is rigid .
hence , @xmath51 let @xmath12 be a polytope with a flag of faces @xmath52 , and define @xmath53 .
a point @xmath54 is said to be _ sewn _ onto @xmath12 through @xmath55 if it realizes the lexicographic extension of @xmath12 by @xmath56 $ ] , where these sets represent their elements in any order .
shemer @xcite proved that if @xmath12 is even dimensional and neighborly and @xmath54 is sewn through a _ universal flag _ then @xmath57 is also neighborly . here , a _ universal flag _ of @xmath12 is a flag consisting of faces in every odd dimension , and such that the quotients of @xmath12 by these faces are still neighborly .
this is extended in @xcite by relaxing the condition of being a universal flag to containing a universal subflag .
this technique of generating neighborly polytopes iteratively sewing starting from a cyclic polytope is called _ the ( extended ) sewing construction _ ( see @xcite and ( * ? ? ?
* section 3 ) for details ) .
similarly , let @xmath11 be the oriented matroid of a neighborly @xmath4-polytope @xmath12 , with dual @xmath58 , and let @xmath59 be an oriented matroid whose dual is obtained by doing first a lexicographic extension in general position of @xmath58 by @xmath60 $ ] followed by a lexicographic extension by @xmath61 $ ] .
then @xmath59 is also the oriented matroid of a neighborly polytope ( of dimension @xmath62 ) .
this operation is called _ gale sewing _ , and the neighborly polytopes obtained by repeating this procedure from a polygon or a @xmath2-polytope are called _ gale sewn _ ( cf .
@xcite ) .
cyclic polytopes arise as a special case of these constructions : they can be obtained by repeatedly extended sewing from a simplex , as well as by gale sewing from a polygon or certain stacked @xmath2-polytope @xcite .
observe first that , since these are even - dimensional neighborly polytopes , they are rigid and therefore it suffices to argue at the level of realization spaces of oriented matroids ( instead of polytopes ) .
moreover , up to a change to the dual in gale sewing that does not affect the realization space , both constructions are based on performing a sequence of lexicographic extensions starting on a configuration with trivial realization space .
hence , at the oriented matroid level , their realizations spaces are open and contractible .
indeed , in each step the fibers of the deletion map are open polyhedra ( lemma [ lem : lex ] ) .
even more , since these fibers are piecewise polynomially ( and hence piecewise smoothly ) parametrized , then the whole realization space is a piecewise smooth ball .
* acknowledgements * we want to thank gnter ziegler for his insightful comments on a previous version of this manuscript .
also , we wish to thank nikolai mnv , jrgen richter - gebert and bernd sturmfels for helpful discussions concerning the history of the universality theorem , and the state of the universality theorem for simplicial polytopes in particular . finally , we want to thank ivan izmestiev for translating part of n. mnv s doctoral thesis , and hiroyuki miyata for sparking our interest in this problem .
a. bjrner , m. las vergnas , b. sturmfels , n. white , and g. m. ziegler , _ oriented matroids _ , second ed . ,
encyclopedia of mathematics and its applications , vol .
46 , cambridge university press , cambridge , 1999 .
to3em , _ the topology of configuration varieties and convex polytopes varieties _
thesis , st .
petersburg state university , st .
petersburg , ru , 1986 , 116 pages , http://www.pdmi.ras.ru/~mnev/mnev_phd1.pdf[pdmi.ras.ru/~mnev/mnevphd1.pdf ] .
to3em , _ the universality theorems on the classification problem of configuration varieties and convex polytopes varieties .
_ , topology and geometry , rohlin semin .
1984 - 1986 , lect .
notes math .
1346 , 527 - 543 ( 1988 ) . , 1988 .
jrgen richter - gebert , _ the universality theorems for oriented matroids and polytopes _ , advances in discrete and computational geometry ( south hadley , ma , 1996 ) , contemp . math .
223 , amer .
soc . , providence , ri , 1999 , pp .
269292 . p.
w. shor , _ stretchability of pseudolines is @xmath64-hard _ , applied geometry and discrete mathematics the victor klee festschrift ( p. gritzmann and b. sturmfels , eds . ) , dimacs series in discrete mathematics and theoretical computer science , vol . 4 , amer
providence ri , 1991 , pp . 531554 . | mnv s universality theorem breakthrough in theory oriented matroids convex polytopes states realization spaces of matroids polytopes fixed matroid face lattice arbitrarily complex comes in four flavours : @xmath0 primary basic semialgebraic set over @xmath1 oriented matroid @xmath2 realization space equivalent to @xmath0 polytope * realization space equivalent to @xmath0 if @xmath0 open uniform oriented matroid of @xmath2 realization space equivalent to @xmath0 * simplicial polytope * realization space equivalent to @xmath0 mnv announced theorem in 1985 published sketch proof in 1988 detailed reasoning in his thesis @xcite simplified key step in mnv s line reasoning for theorem elaborated by richter - gebert gnzel proved stronger universal partition theorem for oriented matroids using lawrence extensions to rigidify face lattices easy to prove part from part @xcite face lattice rigid if determines oriented matroid defined by vertices theorem universality generalized by richter - gebert proved @xmath3-dimensional polytopes are universalproof part mnv s original papers available preliminary results sturmfels bokowski guedes de oliveira mnv s elaborations @xcite concise incomplete part theorem universality ] considered open lawrence extensions used deduce universality theorem for simplicial polytopes use different approach rigidify matroids based result shemer proving rigidity of neighborly polytopes @xcite establish neighborly polytopes @xmath4-polytopes with complete @xmath5-skeleton universal obtain result with even - polytopes all neighborly polytopes simplicial proof of theorem universality open primary semialgebraic set over @xmath1 equivalent to realization space neighborly @xmath6-dimensional polytope on @xmath7 vertices contrast to cyclic polytopes trivial realization spaces holds for all totally sewn gale sewn polytopes @xcite cyclic polytopes particular case large families neighborly polytopes number of gale sewn polytopes with @xmath8 vertices in fixed dimension @xmath4 is asymptotically at least @xmath9 current best lower bound for number of all polytopes @xcitesewn realization space even - dimensional @xmath4-polytope on @xmath8 vertices extended gale sewing contractible open smooth @xmath10-ball richter - gebert universality holds neighborly @xmath3-dimensional polytopes open semialgebraic set over @xmath1 equivalent to realization space neighborly simplicial @xmath3-polytope universality conjecture for simplicial @xmath3-polytopes supported by without isotopy property disconnected realization space @xcite altshuler steinberg vertex figures of neighborly @xmath3-polytopes are @xmath2-dimensional stacked polytopes realization spaces stacked polytopes trivial oriented matroids complicated @xmath11 planar point configuration stacked @xmath2-polytope @xmath12 with distinguished vertex @xmath13 contraction coincides with @xmath11 realization spaces of planar point configurations universal captured by constructing neighborly @xmath3-polytopes edge contractions neighborly 4-polytopes rigid realization space @xmath14 of oriented matroid @xmath11 is vector configurations same oriented matroid@xmath11 rank @xmath4 @xmath15 ground set @xmath11 @xmath16 similarly @xmath12 polytope @xmath17 @xmath18 vertex set define realization space as @xmath19 realization space @xmath20 polytope union realization spaces oriented matroids face lattice coincides with face lattice @xmath12 basic semialgebraic set @xmath17 solutions to finite rational polynomial equalities inequalities primary if inequalities strict basic semialgebraic set @xmath21 stable projection of basic semialgebraic set @xmath22 projection @xmath23 @xmath24 @xmath25 fiber @xmath26 relative interior non - empty polyhedron defined equalities strict inequalities depend polynomially on @xmath27 semialgebraic sets @xmath28 @xmath29 rationally equivalent if homeomorphism @xmath30 @xmath31 @xmath32 rational functions @xmath28 @xmath29 stably equivalent if same equivalence class stable projections equivalences denote stable equivalence of @xmath28 @xmath29 as @xmath33 refer to @xcite for detailed definitions technique use lexicographic extensions section 7.2@xmath11 realized by vector configuration @xmath0 lexicographic extension @xmath34 @xmath35 elements @xmath0 @xmath36 signs adjoining to @xmath0 vector @xmath37 for @xmath38 small figure lemma straightforward lemma 8.2.1 proposition 8.2.2 @xmath11 oriented matroid @xmath39 lexicographic extension of @xmath11 projection induced by deletion @xmath40)\longrightarrow\mathscr{aligned surjective fibers are polynomially parametrized polyhedra of dimension @xmath41 observation of mnv / shor universality of uniform oriented matroids @xcite proven using constructible oriented matroids substitution technique from @xcite proof theorem construction of kortenkamp @xmath4-dimensional point configuration of @xmath44 points appears as face figure of neighborly polytope @xcite larger point configurations open problem asked by perles oriented matroid theory polytope called rigid if face lattice of @xmath12 determines oriented matroid @xmath11 spanned by vertices @xmath12 section 6.6 for rigid polytopes @xmath45 .theorem kortenkamp 1.2 oriented matroid @xmath2 extended to gale dual even - dimensional neighborly polytope @xmath8 lexicographic extensions obtaining rank @xmath2 matroid @xmath48 on @xmath7 elements lemma obtain @xmath49 @xmath12 face lattice gale dual @xmath50 oriented matroid duality preserves realization spaces theorem 2.10 theorem 4.2 neighborly polytope even dimension rigid @xmath51 let @xmath12 polytope with flag faces @xmath52 define @xmath53 point @xmath54 sewn onto @xmath12 through @xmath55 lexicographic extension @xmath12 @xmath56 sets represent elements any order @xcite @xmath12 even dimensional neighborly @xmath54 sewn through universal flag @xmath57 neighborly universal flag @xmath12 faces every odd dimension quotients @xmath12 faces neighborly extended @xcite relaxing condition universal flag containing universal subflag technique generating neighborly polytopes cyclic polytope extended sewing construction see @xcite section 3 details@xmath11 oriented matroid neighborly @xmath4-polytope @xmath12 dual @xmath58 @xmath59 oriented matroid dual obtained lexicographic extension @xmath58 by @xmath60 $ followed extension @xmath61 $ @xmath59 oriented matroid of neighborly polytope dimension @xmath62 ) operation called _ gale sewing _ neighborly polytopes obtained procedure from polygon or @xmath2-polytope called _ gale sewn _ cyclic polytopes special case obtained by repeatedly extended sewing from simplex gale sewing from polygon or stacked @xmath2-polytope @xcite even - dimensional neighborly polytopes rigid argue at level realization spaces of oriented matroids change dual in gale sewing affect realization space constructions based on lexicographic extensions starting configuration trivial realization space matroid level realizations spaces open contractible fibers of deletion map open polyhedra fibers piecewise polynomially smoothly parametrized whole realization space piecewise smooth ball gnter ziegler for insightful comments on previous version manuscriptnikolai mnv jrgen richter - gebert bernd sturmfels discussions universality theorem simplicial polytopes thank ivan izmestiev translating mnv s doctoral thesis hiroyuki miyata interest a. bjrner m. las vergnas. n. white. m. ziegler oriented matroids second ed encyclopedia mathematics applications vol 46 cambridge university press 1999 topology configuration varieties convex polytopes thesis st . petersburg state university 1986 116 pages/~mnev_phd1.pdf universality theorems classification problem configuration varieties convex polytopes topology geometry rohlin semin 1984 - 1986 notes math . 1346 527 - 543 1988 jrgen richter - gebert universality theorems oriented matroids polytopes advances discrete computational geometry south hadley 1996 math . 223 amer . soc . providence 1999 269292 w. shor stretchability pseudolines @xmath64-hard applied geometry discrete mathematics victor klee festschriftgritzmann eds dimacs series discrete mathematics computer science vol 4 amer providence 1991 pp 531554 | 0.570301 | 3,588 | 1,902 | , Saving $0.1 in GPT-4. |
55,303 | we describe a sensitive targeted search campaign at parkes to find faint young radio pulsars using the center beam of the 20-cm multibeam receiver .
the high frequency of the receiver mitigates scattering effects which can dominate profiles of distant pulsars at low frequencies and prevent detection .
the targets included several composite supernova remnants ( snrs ) , two fast x - ray pulsars , long - period anomalous x - ray pulsars , and a soft gamma repeater .
the sensitivity limits in this search are significantly better than those previously reported for these objects .
we have discovered three new radio pulsars , two of which are not associated with the target objects .
the third pulsar , psr j1713@xmath03949 , is possibly associated with snr g347.3@xmath00.5 ; the estimated ages and distances of the two objects are consistent , but a more significant positional coincidence must be established with the central x - ray source in the snr through pulsar timing to confirm or refute the association .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | the highly successful parkes multibeam pulsar survey ( pm survey ) and other recent surveys have used the 20-cm multibeam receiver on the parkes telescope for pulsar searches ( e.g. , manchester et al .
2001 ; crawford et al .
apart from the advantages of the multiple beams and good raw sensitivity of this receiver , deleterious pulse scattering and dispersion smearing effects are greatly reduced at the high observing frequency .
this instrument is therefore good for finding fast , distant pulsars which may have been missed in previous low - frequency surveys .
we have used the multibeam receiver and the parkes telescope to search young objects for radio pulsars .
the list of objects is shown in table 1 and includes supernova remnants ( snrs ) , anomalous x - ray pulsars ( axps ) , fast x - ray pulsars , and a soft gamma repeater ( sgr ) . in the search
, we used the center beam of the multibeam receiver at a center frequency of 1374 mhz .
integration times of 16800 s were typically used with 0.25 ms sampling and 288 mhz of bandwidth split into 96 channels . during processing
, data were dedispersed at trial dispersion measures ( dms ) ranging from 0 to 3000 pc @xmath1 , frequency channels were summed , and each resulting time series was filtered then fourier transformed to get an amplitude spectrum for each dm trial .
candidate periods were identified in the spectra , and the original data were dedispersed and folded at these periods .
the flux sensitivity limits of the searches in all cases are @xmath2 mjy at 1374 mhz ( see table 1 ) .
these limits were estimated using the survey sensitivity modeling described in crawford ( 2000 ) and manchester et al .
( 2001 ) , and they took into account estimates of dm , scattering , and sky temperature , as well as instrumental effects . for those objects for which no _ a priori _ period was known ,
@xmath3 ms was assumed in the sensitivity calculations .
the flux sensitivity limits in all cases are better than those previously reported for these objects
. however , they depend on assumptions about pulsed duty cycle , period , dm , and scattering effects .
we obtained upper limits at 1374 mhz for most objects ( table 1 ) , and discovered three new radio pulsars in the search .
these discoveries are presented in table 2 and are described below .
lccccc target object & period & distance & dm & @xmath4 & result / psr + & ( s ) & ( kpc ) & ( pc @xmath1 ) & ( mjy ) & + snr g0.9 + 0.1 & ? & 10 @xmath5 3 & @xmath6 775 & 0.18 & j1747@xmath02802 + snr g21.5@xmath00.9 & ?
& 5.5 @xmath5 0.5 & @xmath6 360 & 0.16 & @xmath0 + snr g266.2@xmath001.2 & ?
& @xmath7 1 & @xmath7 85 & 0.11 & @xmath0 + snr g327.1@xmath01.1 & ? & 6.5 @xmath5 1 & @xmath6 330 & 0.07 & @xmath0 + snr g347.3@xmath00.5 & ? & 6 @xmath5 1 & @xmath6 450 & 0.06 & j1713@xmath03949 + a0538@xmath066 & 0.069 & @xmath6 50 & @xmath6 100 & 0.04 & @xmath0 + ax j0043@xmath0737 & 0.087 & @xmath6 57 & @xmath6 100 & 0.04 & @xmath0 + 1e 1048@xmath05937 & 6.456 & @xmath6 10 ? & @xmath6 500 ? & 0.13 & @xmath0 + ax j1845@xmath00258 & 6.971 & @xmath6 10 ?
& @xmath6 960 ? & 0.13 & j1844@xmath00256 + 1e 1841@xmath0045 & 11.774 & @xmath6 7 & @xmath6 560 & 0.25 & @xmath0 + rxs j170849@xmath0400910 & 11 & @xmath6 10 ? &
@xmath6 800 ? & 0.22 & @xmath0 + sgr 0526@xmath066 & 8 & @xmath6 50 & @xmath6 100 & 0.14 & @xmath0 + lccc psr & j1713@xmath03949 & j1747@xmath02802 & j1844@xmath00256 + @xmath8 ( ms ) & 392 & 2780 & 273 + @xmath9 & @xmath6 100 kyr & @xmath6 20 myr & @xmath6 300 kyr + dm ( pc @xmath1 ) & 337 @xmath5 3 & 795 @xmath5 10 & 820 @xmath5 3 + distance ( kpc ) & 5.0 @xmath5 0.2 & 11.4 @xmath5 0.1 & 8.8 @xmath5 0.5 + s / n@xmath10 & 33.4 & 10.7 & 22.9 + target object & snr g347.3@xmath00.5 & snr g0.9 + 0.1 & ax j1845@xmath00258 + association ? & possible & no & no + psr j1713@xmath03949 has a period @xmath8 = 392 ms and was discovered while searching the x - ray point source 1wgaj1713.4@xmath03949 at the center of snr g347.3@xmath00.5 ( see figure 1 ) .
this central source is believed to be a neutron star associated with the snr ( slane et al .
the pulsar s dm of 337 pc @xmath1 indicates a distance of @xmath11 kpc using the taylor & cordes ( 1993 ) dm - distance model .
this is consistent with the estimated snr distance of @xmath12 kpc ( slane et al .
a measured period derivative obtained from observations of the pulsar taken a year apart indicates a characteristic age of @xmath13 kyr . this value is larger than the snr age of @xmath14 kyr suggested by slane et al.(1999 )
however , these ages could still be consistent given the uncertainties in both of the age estimates and the possibility that @xmath15 might not reflect the true pulsar age .
the positional coincidence of psr j1713@xmath03949 with 1wgaj1713.4@xmath03949 also suggests an association with the snr , but the uncertainty in the pulsar s position is large ( @xmath16 , the discovery beam radius ) . a timing solution for the pulsar will provide arcsecond positional accuracy for the pulsar as well as a more accurate age estimate .
this will enable us to establish a more significant positional coincidence with 1wgaj1713.4@xmath03949 which can directly confirm or refute an association between psr j1713@xmath03949 and snr g347.3@xmath00.5 .
psr j1747@xmath02802 is coincident with snr g0.9 + 0.1 and has a period @xmath8 = 2780 ms and dm = 795 pc @xmath1 with a large scattering tail ( @xmath17 100 ms ) .
this pulsar was also discovered in the pm survey , and timing results are reported in detail by morris et al .
the pulsar is old ( @xmath18 myr ) and is clearly not associated with snr g0.9 + 0.1 ( @xmath19 1 - 7 kyr ) .
psr j1844@xmath00256 was discovered in a search of axp ax j1845@xmath00258 .
its @xmath8 = 273 ms and dm = 820 pc @xmath1 with a large scattering tail ( @xmath17 40 ms ) .
the period derivative obtained from observations taken a year apart indicates that the pulsar is much older ( @xmath20 kyr ) than the target axp ( @xmath21 kyr ) , which is associated with snr g29.6 + 0.1 ( gaensler , gotthelf , & vasisht 1999 ) . the pulsar s age and
observed period indicate that there is no association with ax j1845@xmath00258 .
we have used the 20-cm multibeam receiver at parkes to conduct targeted pulsar searches of young objects .
our upper limits on radio pulsations from these objects are the best yet obtained , but they depend on assumptions about pulsed duty cycle , scattering effects , and dm .
three new pulsars were discovered in the search ; two of the pulsars ( psrs j1747@xmath02802 and j1844@xmath00256 ) are not associated with the target objects , while psr j1713@xmath03949 might be associated with snr g347.3@xmath00.5 .
the agreement in age and distance suggests a possible association .
however , a more significant positional coincidence must be established between psr j1713@xmath03949 and the central x - ray source 1wgaj1713.4@xmath03949 in snr g347.3@xmath00.3 to confirm or refute this conclusion . | successful parkes multibeam pulsar survey recent surveys used 20-cm multibeam receiver for pulsar searches 2001 advantages multiple beams sensitivity deleterious pulse scattering dispersion smearing effects reduced at high frequency instrument good for finding fast distant pulsars missed used multibeam receiver parkes telescope to search young objects for radio pulsars list objects table 1 includes supernova remnants anomalous x - ray pulsars fast x - ray pulsars soft gamma repeater used center beam receiver at frequency 1374 mhz integration times 16800 s 0.25 ms sampling 288 mhz bandwidth split into 96 channels processing data dedispersed at trial dispersion measures 0 to 3000 pc frequency channels summed time series filtered fourier transformed amplitude spectrum for candidate periods identified spectra original data dedispersed folded at periods flux sensitivity limits are @xmath2 mjy at 1374 mhz table 1 limits estimated using survey sensitivity modeling crawford 2000 manchester 2001 estimates dm scattering sky temperature instrumental effectsobjects no period known @xmath3 ms assumed calculations flux sensitivity limits better depend assumptions pulsed duty cycle period dm scattering effects obtained upper limits 1374 mhz objects discovered three new radio pulsars discoveries table 2 target object period distance dm @xmath4 result psr kpc @xmath1 mjy snr g0.9 0.1 @xmath5 3 @xmath6 775 0.18 j1747@xmath02802 snr g21.5@xmath00.9 5.5 @xmath5 0.5 @xmath6 360 0.16 @xmath0 snr g266.2@xmath001.2 @xmath7 1 85 0.11 @xmath0 snr g327.1@xmath01.1 6.5 @xmath5 1 330 0.07 g347.3@xmath00.5 6 @xmath5 1 @xmath6 450 0.06 j1713@xmath03949 a0538@xmath066 0.069 @xmath6 50 @xmath6 100 0.04 j0043@xmath0737 0.087 @xmath6 57 @xmath6 100 0.04 1048@xmath05937 6.456 @xmath6 10 500 0.13 j1845@xmath00258 10 960 0.13 j1844@xmath00256 1841@xmath0045 11.774 @xmath6 7 @xmath6 560 0.25 j170849@xmath0400910 11 @xmath6 10 800 0.22 0526@xmath066 8 @xmath6 50 @xmath6 100 0.14 j1713@xmath03949 j1747@xmath02802 j1844@xmath00256 @xmath8 392 2780 273 @xmath9 @xmath6 100 @xmath6 20 300 337 @xmath5 3 795 10 820 distance 5.0 0.2 11.4 0.1 8.8 0.5 33.4 10.7 22.9 target object g347.3@xmath00.5 g0.9 0.1 j1845@xmath00258 associationj1713@xmath03949 period @xmath8 = 392 ms discovered x - ray source 1wgaj1713.4@xmath03949 center of snr g347.3@xmath00.5 central source neutron star snr pulsar s dm 337 pc @xmath1 distance @xmath11 kpc consistent with estimated snr distance @xmath12 kpc measured period derivative observations pulsar age @xmath13 kyr larger than snr age @xmath14 kyr slane. ages could consistent uncertainties age estimates reflect true pulsar age positional coincidence j1713@xmath03949 with 1wgaj1713.4@xmath03949 suggests association snr uncertainty pulsar s position large timing solution arcsecond positional accuracy accurate age estimate significant positional coincidence with 1wgaj1713.4@xmath03949 refute association snr g347.3@xmath00.5 psr j1747@xmath02802 coincident with snr g0.9 + 0.1 period @xmath8 = 2780 ms dm = 795 pc @xmath1 large scattering tail @xmath17 100 ms pulsar discovered in pm survey timing results reported by morris et al .pulsar old @xmath18 not associated with snr g0.9 + 0.1 @xmath19 1 - 7 kyr psr j1844@xmath00256 discovered j1845@xmath00258 @xmath8 = 273 ms dm = 820 pc large scattering tail @xmath17 40 ms period derivative pulsar older @xmath20 kyr than target axp @xmath21 kyr associated snr g29.6 + 0.1 age period no association with j1845@xmath00258 used 20-cm multibeam receiver pulsar searches upper limits radio pulsations depend pulsed duty cycle scattering effects dm three new pulsars two j1747@xmath02802 j1844@xmath00256 not associated target objects j1713@xmath03949 might associated snr g347.3@xmath00.5 agreement age distance possible association positional coincidence between psr j1713@xmath03949 central x - ray source 1wgaj1713.4@xmath03949 snr g347.3@xmath00.3 | 0.311722 | 2,207 | 1,257 | , Saving $0.1 in GPT-4. |
55,311 | we consider the most accurate tests of bound state qed , precision theory of simple atoms , related to the hyperfine splitting in light hydrogen - like atoms .
we discuss the hfs interval of the @xmath0 state in muonium and positronium and of the @xmath1 state in hydrogen , deuterium and helium-3 ion .
we summarize their qed theory and pay attention to involved effects of strong interactions .
we also consider recent optical measurements of the @xmath1 hfs interval in hydrogen and deuterium . | light simple atoms are basically described by quantum electromagnetic theory . quantum electrodynamics ( qed )
is well established and in particular it covers all interactions of leptons ( electrons and muons ) and photons .
such a lepton - photon theory is obviously incomplete because even pure leptonic systems are not free of hadronic effects which enter through virtual hadronic intermediate states .
effects of strong interactions can not be calculated _ ab initio _ and additional experimental data and/or phenomenological models are needed .
here we consider qed tests with hyperfine splitting in light hydrogen - like atoms paying attention to both : basic _ ab initio _ qed theory and relatively small , but most uncertain , hadronic contributions .
an application of qed to the bound state problem , bound state qed , is much more complicated than ordinary qed and it deserves serious tests .
some of such tests are significant for the determination of fundamental constants and in particular of the fine structure constant @xmath2 , which may be obtained from the hyperfine structure ( hfs ) interval in muonium ( see reviews @xcite for more detail ) . the hfs interval in hydrogen and some other light atoms has been known for a while with an experimental accuracy at the level of a part in @xmath3 .
meanwhile , the related theory suffers from uncertainties of the nuclear structure effects at one - ppm level . here
, we consider a few possibilities to perform qed tests going far beyond this level of accuracy .
problems in accurate calculations of the proton or nuclear structure effects drew attention to studies of pure leptonic atoms such as a bound system of a positive muon and an electron , the muonium .
in contrast to the hydrogen atom , the nucleus , a muon , is free of effects of strong interactions .
nevertheless , those effects enter through hadronic vacuum polarization . that sets an ultimate limit on any qed tests with muonium .
uncertainties of the qed theory and of calculations of the hadronic effects are presented in @xcite .
muonium is of metrological interest due to determination of the fine structure @xmath2 , muon - to - electron mass ratio @xmath4 and some other fundamental constants @xcite .
a calculation of the hadronic effects @xcite is similar to those for the anomalous magnetic moment of the muon @xcite .
it is based on the low energy @xmath5 data which accuracy , although fast improving , is still behind the measurement of the muon @xmath6 value @xcite .
the current difference between the experimental value and its theoretical prediction differs from zero by almost three standard deviations @xcite .
attempts to increase the accuracy of the prediction by adding data on the @xmath7 lepton decays revealed one more possible deviation from the standard model expectation @xcite . [ cols="<,<,<,<,<",options="header " , ] + we summarize state - of - the - art in the precision tests of the bound state qed theory of the hyperfine structure in table [ t : exp ] .
the theoretical accuracy is limited by our ability to calculate higher - order radiative , recoil and radiative recoil effects ( see review @xcite for more detail ) .
the higher - order contributions crucial for the uncertainty are related to the same diagrams and thus all tests listed in the table are really competitive .
theory and experiment are generally in good agreement .
there is a minor discrepancy for positronium up to approximately 3 standard deviations , but statistically that is acceptable if the tests as a whole are considered .
the work was supported in part by dfg and rfbr ( under grant # 06 - 02 - 16156 ) .
j. r. sapirstein , e. a. terray and d. r. yennie , phys .
d29 ( 1984 ) 2290 ; + a. czarnecki , s. i. eidelman , and s. g. karshenboim , phys .
d65 ( 2002 ) 053004 ; + s. g. karshenboim and v. a. shelyuto , phys .
b517 , ( 2001 ) 32 ; + s. i. eidelman , s. g. karshenboim , and v. a.shelyuto , can . j. phys . 80 ( 2002)1297 . | atoms described by quantum electromagnetic theory quantum electrodynamics qed ) established covers interactions of leptons photons lepton - photon theory incomplete pure leptonic systems not free of hadronic effects effects strong interactions calculated additional experimental data models needed consider qed tests with hyperfine splitting in light hydrogen - atoms to basic qed theory hadronic contributions application of qed to bound state problem more complicated deserves serious tests significant for determination fundamental constants fine structure constant @xmath2 from hyperfine structure hfs interval in muonium hfs interval in hydrogen light atoms known with experimental accuracy at level part in @xmath3 theory from uncertainties of nuclear structure effects at one - ppm level consider possibilities to perform qed tests beyond accuracy problems in calculations proton nuclear structure effects attention to studies of pure leptonic atoms of positive muon electron nucleus muon free of effects strong interactions effects enter through hadronic vacuum polarization on qed tests with muonium uncertainties of qed theory calculations of hadronic effects presented in @xcite muonium metrological interest due to fine structure @xmath2 muon - to - electron mass ratio @xmath4 fundamental constants @xcite calculation hadronic effects similar to anomalous magnetic moment muon based on low energy @xmath5 data behind measurement muon @xmath6 value current difference experimental value theoretical prediction differs three standard deviations attempts increase accuracy adding data @xmath7 lepton decays revealed possible deviation from model summarize precision tests bound state qed theory hyperfine structure table theoretical accuracy limited by higher - order radiative recoil effects review higher - order contributions crucial for uncertainty related to same diagrams all tests competitive theory experiment agreement minor discrepancy for positronium 3 standard deviations statistically acceptable work supported by dfg rfbr grant # 06 - 02 - 16156 ) j r. sapirstein e a. terray d. r. yennie, phys . d29 ( 1984 ) 2290 ; a. czarnecki eidelman karshenboim phys . d65 ( 2002 ) 053004 ; karshenboim shelyutob517 2001 ) 32 eidelman karshenboim v a.shelyuto . 80 2002)1297 . | 0.182281 | 930 | 450 | , Saving $0.0 in GPT-4. |
55,312 | we present a new hamiltonian partitioning which converges an arbitrary number of states of interest in the effective hamiltonian to the full configuration interaction limits simultaneously .
this feature is quite useful for the recently developed model space quantum monte carlo .
a dual partitioning ( dp ) technique is introduced to avoid the intruder state problem present in the previous eigenvalue independent partitioning of coope .
the new approach is computationally efficient and applicable to general excited states involving conical intersections .
we present a preliminary application of the method to model systems to investigate the performance . | the quantum monte carlo ( qmc ) methods in configuration space have been increasing the popularity in molecular electronic structure studies in recent years.@xcite the majority of these approaches are on the basis of the projection , ( ) = e^-(h - e)(0 ) , [ eq : proj ] that converges to the lowest eigen function with a nonzero overlap with the initial wavefunction @xmath0 after a sufficiently long time interval @xmath1 . several extensions for excited states
have been also proposed.@xcite one promising approach among them is the model space qmc ( msqmc ) based on the effective hamiltonian formalism .
we shall outline the main features of msqmc in what follows . , [ eq : itmc ] where @xmath4 and @xmath5 are the projection operators onto the the model space ( p - space ) and its orthogonal component ( q - space ) , respectively , and the subscripts of the hamiltonian mean the individual contributions like @xmath6 .
msqmc treats the model space amplitude @xmath7 deterministically by the diagonalization of effective hamiltonian , and the imaginary time evolution of @xmath8 for a fixed @xmath9 is calculated as = - ( h_qq - e)_q()-h_qp_p.[eq : itetc ] we have recently formulated the ite for the energy dependent partitioning ( edp),@xcite = -(*h*_qq-*i*_qqe)*t*_qp -*h*_qp,[eq : itm_tqp_edp ] as a sufficiency condition of ( [ eq : itetc ] ) , where @xmath10 is the transfer matrix connecting the configuration interaction ( ci ) coefficients in the p- and q - spaces as @xmath11 , and @xmath12 is the identity matrix .
the stationary solution of ( [ eq : itm_tqp_edp ] ) , ( * h*_qq-*i*_qqe)*t*_qp + * h*_qp=0,[eq : loewdin ] corresponds to the lwdin partitioning@xcite applicable to arbitrary electronic states in the vicinity of @xmath13 ( even with degeneracy ) .
the dimension of @xmath10 can be sizable for a full ci ( fci ) problem , and the partitioning technique is usually worthless for determining the exact solutions .
however , the population dynamics of walkers allows us to treat @xmath10 stochastically and to integrate the contribution , @xmath14 , to the effective hamiltonian on - the - fly.@xcite note the ite ( [ eq : itm_tqp_edp ] ) provids a general framework to solve the linear system of equations ( [ eq : loewdin ] ) in a stochastic manner .
the msqmc in edp is robust and in principle capable of treating arbitrary excited states as has been demonstrated in the calculation of potential energy curves of diatomic molecules , recently.@xcite one of the shortcomings of edp is that independent qmc processes are required for electronic states with different target energies .
moreover , the computational cost for each of the qmc processes grows linearly with the dimension of the model space . in the previous paper,@xcite we noted there
exists an alternative choice of the eigenvalue independent partitioning ( eip ) in msqmc where all solutions of the effective hamiltonian converge to the fci limits simultaneously.@xcite nevertheless , eip seemingly suffers from the intruder state problems as likewise in the usual state - universal multi - reference approaches . in this communication
, we develop a new partitioning to solve all of these points at issue .
this work was partly supported by the strategic programs for innovative research ( spire ) and flagship2020 as the priority issue 5 ( development of new fundamental technologies for high - efficiency energy creation , conversion / storage and use ) of mext , and the computational materials science initiative ( cmsi ) .
we are also grateful for the computer resources through the hpci system research project ( project i d : hp150228 , hp150278 ) .
99 s. zhang , h. krakauer , _ phys .
_ , * 90 * , 136401 ( 2003 )
. w. a. al - saidi , s. zhang , h. krakauer , _ j. chem . phys . _ , * 127 * , 144101 ( 2007 ) .
y. ohtsuka , s. nagase , _ chem .
_ , * 463 * , 431 ( 2008 ) .
y. ohtsuka , s. nagase , _ chem .
_ , * 485 * , 367 ( 2010 ) . g. h. booth , a. j. w. thom , a. alavi , _ j. chem .
_ , * 131 * , 054106 ( 2009 )
. d. cleland , g. h. booth , a. alavi , _
phys . _ , * 132 * , 174104 ( 2010 ) .
f. r. petruzielo , a. a. holmes , h. j. changlani , m. p. nightingale , c. j. umrigar , _ phys .
_ , * 109 * , 230201 ( 2012 ) . g. h. booth , g. k .- l .
_ , * 137 * , 191102 ( 2012 ) .
s. ten - no , _ j. chem .
_ , * 138 * , 164126 ( 2013 ) . j. p. coe , m. j. paterson , _ j. chem
_ , * 139 * , 154103 ( 2013 ) .
a. humeniuk , r. mitri , _ j. chem .
_ , * 141 * , 194104 ( 2014 )
. n. s. blunt , s. d. smart , g. h. booth , a. alavi , _ j. chem .
_ , * 143 * , 134117 ( 2015 ) .
lwdin , _ j. chem .
_ , * 19 * , 1396 ( 1951 ) .
lwdin , _ j. mol . spec .
_ , * 14 * , 112 ( 1964 ) .
y. ohtsuka , s. ten - no , _ j. chem .
_ , * 143 * , 214107 ( 2015 ) . j. a. r. coope , _ mol
. phys . _ * 18 * , 571 ( 1970 ) .
j. a. r. coope , d. w. sabo , _ j. comput
. phys . _ * 23 * , 404 ( 1977 ) .
d. mukhopadhyay , s. mukhopadhyay , r. chaudhuri , d. mukherjee , _ theor .
. acta _ * 80 * , 441 ( 1991 ) , and references therein .
w. j. hehre , r. ditchfield , j. a. pople , _ j. chem
. phys . _ * 56 * , 2257 ( 1972 ) .
t. j. lee , j. e. rice , g. e. scuseria , h. f. schaefer , _ theor .
_ , * 75 * , 81 ( 1989 ) . c. overy , g. h. booth , n. s. blunt , j. j. shepherd , d. cleland , a. alavi , ali ,
_ j. chem .
_ , * 141 * , 244117 ( 2014 ) . | quantum monte carlo methods increasing in molecular electronic structure studies majority approaches projection = e^-(h - e)(0 ) converges to lowest function nonzero overlap with initial wavefunction @xmath0 after long time interval extensions for excited states proposed promising approach model space qmc ) based on effective hamiltonian formalism features msqmc @xmath4 and @xmath5 projection operators onto model space p - q - subscripts hamiltonian mean individual contributions like msqmc treats model space amplitude deterministically by diagonalization hamiltonian imaginary time evolution of @xmath8 for fixed @xmath9 calculated as = - ( h_qq - e)_q()-h_qp_p formulated for energy dependent partitioning = -(*h*_qq-*i*_qqe)*t*_qp -*h*_qp sufficiency condition @xmath10 transfer matrix connecting configuration interaction coefficients in p- and q - spaces @xmath12 identity matrixstationary solution of eq : itm_tqp_edp h*_qq-_qqe*t*_qp + h*_qp=0, : loewdin ] corresponds to lwdin partitioning applicable to arbitrary electronic states vicinity @xmath13 with degeneracy dimension of @xmath10 sizable for full problem partitioning technique worthless for determining exact solutions population dynamics of walkers treat @xmath10 stochastically integrate contribution @xmath14 to effective hamiltonian eq : itm_tqp_edp ] framework to solve linear equations eq : loewdin msqmc in edp robust treating arbitrary excited states demonstrated in calculation potential energy curves of diatomic molecules edp independent qmc processes required for electronic states with different target energies computational cost grows linearly with dimension model space alternative eigenvalue independent partitioning ( eip ) in msqmc solutions hamiltonian converge to fci limits eip suffers from intruder state problems new partitioning to solve pointswork supported programs innovative research spire flagship2020 priority issue 5 technologies high - efficiency energy creation conversion storage mext computational materials science initiative grateful computer resources hpci system research project hp150228 hp150278 s. zhang h. krakauer phys. 90 136401 ( 2003 ) al - saidi s zhang h krakauer chem 127 144101 ( 2007 ) y ohtsuka s. nagase chem 463 431 ( 2008 ) ohtsuka nagase chem 485 367 ( 2010 ) h. booth w thom a. alavi j chem 131 054106 ( 2009 ) d. cleland g h. booth a. alavi phys 132 174104 ( 2010 ) petruzielo. holmes. changlani. nightingale. umrigar phys 109 230201 ( 2012 ) h. booth 137 191102 ( 2012 ) s. ten - no j. chem 138 164126 ( 2013 ) j. p. coe m j. paterson j. chem 139 154103 ( 2013 ) a. humeniuk rmitri j. chem 141 194104 ( 2014 ) blunt. smart h booth a alavi 143 134117 ( 2015 ) lwdin j chem 19 1396 ( 1951 ) lwdin mol spec 14 112 ( 1964 ) ohtsuka ten - no chem 143 214107 ( 2015 ) j a. r. coope mol 18 571 ( 1970 ) coope w sabo comput phys 23 404 ( 1977 ) d. mukhopadhyay s mukhopadhyay chaudhuri d. mukherjee theor 80 * 441 ( 1991 ) references j. hehre. ditchfield j a. pople chem 56 * 2257 ( 1972 ) j. lee j e. rice scuseria f. schaefer theor 75 * 81 ( 1989 ) c. overy h. booth blunt j. shepherd d. cleland a. alavi ali j. chem 141 244117 ( 2014 ) | 0.26941 | 1,669 | 876 | , Saving $0.0 in GPT-4. |
55,324 | we discuss dynamical mass measurements of high-@xmath0 galaxies with the next generation space telescope ( ngst ) . in particular , we review some of the observational limits with the current instrument / telescope generation , we discuss the redshift limits and caveats for absorption and emission lines studies with ngst , and the existence of suitable targets at high redshift .
we also briefly summarize strengths and weaknesses of proposed ngst instruments for dynamical studies . | during this meeting we have heard of mass measurements at redshifts that were out of the realm of possibilities just a few years ago .
however , several obstacles stand between us and the future when dynamical mass measurements will be routinely feasible at @xmath1 . among them ,
the lack of spatial resolution and the atmospheric emission / absorption at infrared ( ir ) wavelengths ( see e.g. the contribution by m. franx ) .
the next generation space telescope ( ngst ) with its unique combination of large collecting area , superb spatial resolution , and low background will provide a major contribution to the extension of mass measurements at @xmath1 . here
we discuss mass measurements with ngst , focusing on dynamical measurements ( lensing and stellar mass measurements are discussed elsewhere in these proceedings , see e.g. the contribution by h. ferguson ) . as for any dynamical mass measurement , a spatial scale and a velocity scale
are needed . in the following
, we will assume that spatial scales are easily measurable even though this is not the case for sub - galactic clumps at very high redshift .
our focus in this contribution will be on emission and absorption line measurements ( sec . 2 and 3 ) of velocity scales .
in particular , we will briefly review the limits of what is feasible with the current technology and present some detailed simulations on the capabilities of ngst for mass measurements .
finally , we will discuss the implications of the choice of ngst - instrumentation for mass measurements .
3 even for the most massive progenitors of l@xmath2 galaxies .
right panel : correlation between available resolving power and maximum redshift for which a the internal kinematics of a spiral galaxy can be resolved . for relatively large objects at high redshift ngst is superior to even a 30 m ground based telescope .
, title="fig:",scaledwidth=40.0% ] 3 even for the most massive progenitors of l@xmath2 galaxies .
right panel : correlation between available resolving power and maximum redshift for which a the internal kinematics of a spiral galaxy can be resolved .
for relatively large objects at high redshift ngst is superior to even a 30 m ground based telescope .
, title="fig:",scaledwidth=43.0% ]
kinematical measurements based on emission lines are easy to carry out but hard to interpret .
this is because the gas may be far from equilibrium and thus gas kinematics may not be telling us anything about the mass of the host galaxy .
one way to overcome this difficulty is to have access to two - dimensional kinematical information and this is indeed what was needed to solve the same problem in the context of mass measurement in nearby spiral galaxies using hi @xcite and black hole mass measurements from the kinematics of nuclear gas disks @xcite .
ngst presents several advantages in this context because of its high angular resolution .
its uninterrupted wavelength coverage guarantees that strong emission lines will be available over a large range of redshifts .
the low background redwards of 2.5 @xmath3 m will guarantee high sensitivity for measurements using h@xmath4 at z@xmath5 3 . on the basis of the visibility of
h@xmath4 one can argue that for objects at z@xmath6 3 a detailed 2d kinematic mapping can be carried out from the ground using an adaptive optics fed integral field spectrograph on an 8-meter class telescope . at z@xmath53 oxygen lines
are still accessible from the ground but may be suppressed by the expected low metallicity of most objects at that redshift .
thus , the availability h@xmath4 and the sensitivity of ngst make it the ideal instrument at z@xmath53 .
in addition to the availability of lines another issue is the resolving power needed to carry out a measurement . in a nutshell ,
faint galaxies will on average have lower mass and smaller internal velocities , requiring higher resolving power to be studied .
since on average we expect galaxy mass to decrease with redshift , one will need progressively higher resolving power to study galaxies at increasingly high redshift .
this intuitive result is illustrated in figure [ fig : sim ] where we show , in the left panel , simulated ngst spectra for a typical milky way progenitor ( obj=13 ) at z=3 , 5 , and 7 and for a bright milky way progenitor ( obj=9 ) at z=7 .
these models have been obtained with a merging tree code @xcite .
the resolving power needed to measure internal kinematics is in the range 6,000 to 16,000 . in the right panel
we show a relation connecting the available resolving power to the highest redshift that can be probed .
this plot has been obtained by assuming the validity of the tully - fisher relation and by requiring that for the given exposure time and resolving power sufficient signal - to - noise is achieved to carry out the measurement .
the optimal emission lines is used for each redshift @xcite .
it is clear that for low mass objects ngst is not competitive with large ground based telescopes .
however , ngst is superior to 30 m class ground based telescopes for massive objects at @xmath7 .
, ground - based limits : evolution of the fundamental plane of e / s0 galaxies from @xmath8 to @xmath9 ( from @xcite ) .
the average offset of the intercept of field galaxies from the local fp relation as a function of redshift ( large filled pentagons ) , is compared to the offset observed in clusters ( open squares ) .
see @xcite for references , description , and details . ]
stellar absorption lines are a very good probe of velocity scales for several reasons : _
i ) _ a large fraction of galaxies do not have emission lines , and therefore absorption lines are the only way to go ; _ ii ) _ stars are a good tracer of the velocity distribution of the system , as opposed to emission lines coming from hii regions which might not be in dynamical equilibrium with the galaxy ; _ iii ) _ stellar absorption lines kinematics tend to suffer less from the effects of interstellar absorption , local motions , and winds then emission lines kinematics ; _ iv ) _ if streaming motions are not significant as in massive e / s0 galaxies , interesting dynamical constraints can be gathered without the need for spatially resolved information .
however , stellar absorption lines have to be present , and stellar populations need to be in dynamical equilibrium , in order to obtain meaningful information . typical optical absorption lines ( e. g. the mg triplet ) need of order @xmath10 gyr to develop in stellar populations , while the time scales can be relatively shorter for other frequently used lines such as the near - ir ca triplet .
on the one hand this is good , because it guarantees that if such lines are present then the stellar populations are almost certainly old enough that the system is in equilibrium . on the other hand ,
it is worth asking the question of whether systems with stellar absorption lines exist at high redshift . in the following
we will illustrate with a few examples the current limits of ground based measurements , we will show why we believe there are interesting targets beyond these limits , and we will explore the feasibility of such measurements with ngst .
recent studies of the evolution of the fundamental plane ( fp ) with redshift show that e / s0 galaxies do not undergo major structural changes from @xmath11 to the present ( e. g. @xcite ) . such measurements , based on a combination of spectroscopy ( velocity dispersion ) and imaging ( photometric structural parameters ) provide important information on the evolution of the internal structure of e / s0 galaxies and their stellar populations ( see also the contribution by g. illingworth ) . for example , @xcite inferred from the evolution of the fp that cluster e / s0 have old stellar populations , while @xcite used this technique to show that secondary episodes of star formation are common in massive field e / s0 at @xmath12 .
it would be interesting to extend such measurements to higher redshifts , where we know that some old e / s0 exist , at least in the range @xmath13 ( @xcite ; see figure [ fig : highzes0 ] ) .
however , long integrations on large telescopes are needed to push such studies significantly beyond @xmath14 because e / s0 become very faint in the optical and sky emission lines severely limit what can be done in the ir .
for example , @xcite measured the central velocity dispersion of the lens galaxy in the gravitational lens system mg2016 + 112 ( @xmath15 ) by integrating 8.5hrs with the echellette spectrograph imager ( esi ) at the keck ii telescope .
given the excellent sensitivity and resolution of esi ( that allows for a good removal of sky lines in the red / ir ) , it seems likely that with the current generation of 8 - 10 m class telescopes such measurements will be limited to bright e / s0 at @xmath16 . as shown in the right panel of fig .
[ fig : highzes0 ] , the redshift range @xmath13 would be easily within reach with ngst ( see also @xcite ) , and even objects beyond @xmath17 could be targeted , if they exist . based on calculations similar to the one used in fig . 1 ( right panel ) , adopting the fp instead of the tully - fisher relation as luminosity - velocity scaling law , at @xmath1 ngst will be significantly better than a 30-m telescope , especially for large masses .
recent studies suggest that old stellar populations indeed exist at @xmath7 .
for example @xcite found that the colors of lyman-@xmath4 selected galaxies at @xmath18 are consistent with those of an old stellar population ( @xmath19 gyr of age )
responsible for the observed red colors with a sprinkle of starbursting population
responsible for the observed emission lines
( see also @xcite ) . :
targets for ngst . left : spectral energy distribution of and extremely red object with e / s0 morphology at @xmath20 ( from @xcite ) .
right : comparison of the performance of ngst , a 10 m ground based telescope and a 30 m ground based telescope.,title="fig:",scaledwidth=40.0% ] : targets for ngst . left : spectral energy distribution of and extremely red object with e / s0 morphology at @xmath20 ( from @xcite ) .
right : comparison of the performance of ngst , a 10 m ground based telescope and a 30 m ground based telescope.,title="fig:",scaledwidth=43.0% ] so far , we only discussed absorption line kinematics in the absence of spatially resolved information .
however , spatially resolved kinematics have been very important in our study of the local universe . for example it has provided crucial constraints on the internal structure of e / s0 ( e. g. @xcite ) , and hubble space telescope ( hst )
subarcsecond resolution spectroscopy has been used to prove the existence of black holes at the center of massive e / s0 .
unfortunately , the small aperture of hst and the seeing - limited spatial resolution of ground based telescopes have so far basically confined spatially resolved kinematics to the local universe .
as an example of the current observational limits , @xcite have recently started a kinematic survey of gravitational lenses . under the best observing conditions ( seeing @xmath21 ) with long integration times at the keck - ii telescope they were able to measure velocity dispersion profiles extended beyond the effective radius for lens galaxies at @xmath22 .
although the extended profiles provide valuable and unique information on the mass distribution , a higher spatial resolution is needed to study in detail the mass distribution within the central kiloparsecs . ngst with its combination of large collecting area and superb spatial resolution offers the tremendous opportunity to explore spatially resolved kinematics at cosmological distances . with a typical resolution
@xmath23 ngst will be able to study in detail the mass distribution of e / s0 and its evolution with redshift .
both emission and absorption line studies of the kinematics of high redshift galaxies require a resolving power of at least 3,000 , i.e. somewhat higher than what is currently in the baseline for ngst .
the need for high resolving power is stronger for emission line measurements since they can be carried out on intrinsically fainter objects . because of this reason there is a strong rationale for a two - dimensional spectroscopic capability at r@xmath243,000 .
this capability could be in the form of an integral field spectrograph @xcite if budget , volume , and weight constraints allow it .
alternatively , it could be provided by a multi - object spectrograph based on a configurable slit array provided the configuration time is short enough .
such a spectrograph would allow one to step the slit over resolved objects to be mapped in two dimensions while integrating on fainter sources @xcite .
we acknowledge financial support by nasa through a grant from stsci ( hst - ar-09222 ) , which is operated by aura , under nasa contract nas5 - 26555 .
j. w. mackenty , m. stiavelli : a multi - object spectrometer using micro mirror arrays. in : _ imaging the universe in three dimensions : astrophysics with advanced multi - wavelength imaging devices _ , ed . w. van bruegel and j. bland - hawthorn , asp , vol .
195 , pp . 443 ( 2000 ) m. stiavelli : internal structure of high - z galaxies with ngst. in : _ the next generation space telescope : science drivers and technological challenges , 34th lige astrophysics colloquium , belgium , 15 - 18 june 1998 _ , ed . by p.
benvenuti , p. madau ( esa publications 1998 ) pp .
71 t. treu : the fundamental plane of elliptical galaxies at low and high redshift. in : _ the next generation space telescope : science drivers and technological challenges , 34th lige astrophysics colloquium , belgium , 15 - 18 june 1998 _ , ed . by p.
benvenuti , p. madau ( esa publications 1998 ) pp . 255 | meeting heard mass measurements at redshifts possibilities few years ago obstacles future dynamical mass measurements feasible at @xmath1 lack of spatial resolution atmospheric emission / absorption at infrared wavelengths franx next generation space telescope ngst ) large collecting area spatial resolution low background major contribution to extension mass measurements at @xmath1 discuss mass measurements with ngst focusing on dynamical measurements lensing stellar mass measurements discussed h ferguson dynamical mass measurement spatial scale velocity scale needed assume spatial scales easily measurable not for sub - galactic clumps at high redshift focus on emission absorption line measurements velocity scales review limits feasible with current technology present detailed simulations on capabilities ngst for mass measurements discuss implications of choice ngst - instrumentation for mass measurements massive progenitors l@xmath2 galaxies correlation between resolving power maximum redshift internal kinematics spiral galaxy large objects at high redshift ngst superior to 30 m ground based telescope massive progenitors l@xmath2 galaxiespanel correlation between resolving power maximum redshift internal kinematics spiral galaxy large objects high redshift ngst superior to 30 m ground telescope kinematical measurements emission lines easy hard to interpret gas far from equilibrium gas kinematics mass host galaxy access to two - dimensional kinematical information needed problem mass measurement in nearby spiral galaxies using hi @xcite black hole mass measurements nuclear gas ngst advantages high angular resolution uninterrupted wavelength coverage strong emission lines over large range redshifts low background redwards of 2.5 @xmath3 m high sensitivity for measurements h@xmath4 at z@xmath5 3 visibility h@xmath4 objects at z@xmath6 3 detailed 2d kinematic mapping from ground using adaptive optics fed integral field spectrograph 8-meter class telescope at z@xmath53 oxygen lines accessible suppressed by low metallicity objects redshift availability h@xmath4 sensitivity of ngst ideal instrument at z@xmath53 availability issue resolving power needed measurement faint galaxies lower mass smaller internal velocities higher resolving powergalaxy mass decrease with redshift higher resolving power study galaxies high redshift illustrated in figure simulated ngst spectra typical milky way progenitor=13 at z=3 5 7 bright milky way progenitor=9 at z=7 models obtained with merging tree code @xcite resolving power internal kinematics 6,000 to 16,000 relation connecting resolving power to highest redshift plot obtained tully - fisher relation requiring exposure time power sufficient signal - to - noise measurement optimal emission lines used for each redshift @xcite low mass objects ngst not competitive with large ground telescopes superior to 30 m class ground telescopes for massive objects at @xmath7 evolution fundamental plane of / galaxies from @xmath8 to @xmath9 average offset intercept field galaxies from local fp relation redshift compared to offset in clusters see @xcite for references description detailsstellar absorption lines good probe velocity scales reasons large fraction of galaxies emission lines absorption lines only way go stars good tracer velocity distribution emission lines from regions stellar absorption lines suffer less from effects interstellar absorption local motions winds emission lines if streaming motions not significant in massive e / s0 galaxies dynamical constraints can be gathered without spatially resolved information stellar absorption lines present stellar populations need dynamical equilibrium obtain meaningful information typical optical absorption lines mg triplet need order @xmath10 gyr to develop in stellar populations time scales shorter for other lines near - ir ca triplet guarantees if lines present stellar populations old enough system in equilibrium worth asking question systems with stellar absorption lines exist at high redshift illustrate current limits of ground based measurements show interesting targets beyond limits explore feasibility of measurements with ngst recent studies evolution fundamental plane with redshift show e / s0 galaxies undergo major structural changes from @xmath11 to presentmeasurements spectroscopy imaging photometric provide information on evolution e / s0 galaxies stellar populations g. illingworth @xcite inferred cluster e / s0 have old stellar populations @xcite secondary episodes star formation common in massive field e / s0 at @xmath12 to extend measurements to higher redshifts old e / s0 exist in range @xmath13 long integrations on large telescopes needed studies beyond @xmath14 e / s0 faint in optical sky emission lines limit ir @xcite measured central velocity dispersion of lens galaxy gravitational lens system mg2016 + 112 ( @xmath15 ) integrating 8.5hrs with echellette spectrograph imager esi at keck ii telescope sensitivity resolution of esi removal sky lines current generation 8 - 10 m class telescopes measurements limited to bright e / s0 at @xmath16 redshift range @xmath13 within reach with ngst @xcite objects beyond @xmath17 could be targeted calculations similar to in fig .adopting fp tully - fisher relation as luminosity - velocity scaling law at @xmath1 ngst better than 30-m telescope especially for large masses studies suggest old stellar populations exist at @xmath7 found colors of lyman-@xmath4 galaxies at @xmath18 consistent with old stellar population @xmath19 for red colors starbursting population for emission lines targets for ngst spectral energy distribution of extremely red object with e / s0 morphology at @xmath20 comparison performance of ngst 10 m 30 m telescope targets for ngst spectral energy distribution extremely red object with e / s0 morphology at @xmath20 comparison performance ngst 10 30 discussed absorption line kinematics spatially resolved information spatially resolved kinematics important in study local universe provided constraints on internal structure of e / s0 hubble telescope existence of black holes at center of e / s0 small aperture of hst - limited spatial resolution of ground telescopes confined spatially resolved kinematics to local universe@xcite started kinematic survey gravitational lenses best conditions @xmath21 long integration times - telescope velocity dispersion profiles extended beyond effective radius lens galaxies at @xmath22 extended profiles provide information mass distribution higher spatial resolution needed central kiloparsecs ngst large collecting area superb spatial resolution opportunity explore spatially resolved kinematics at cosmological distances typical resolution @xmath23 mass distribution of e / s0 evolution with redshift emission absorption line studies high redshift galaxies require resolving power 3,000 higher than baseline ngst need high resolving power stronger for emission line measurements fainter objects strong rationale for two - dimensional spectroscopic capability at r@xmath243,000 integral field spectrograph @xcite alternatively multi - object spectrograph configurable slit array configuration time short slit over resolved objects integrating on fainter sources financial support nasa grant from stsci hst - ar-09222 operated by aura nasa contract nas5 - 26555 j. w. mackenty , m. stiavelli multi - object spectrometer using micro mirror arraysimaging universe three dimensions astrophysics multi - wavelength imaging devices van bruegel bland - hawthorn vol 195 pp 2000 stiavelli structure high - z galaxies next generation space telescope science drivers challenges 34th lige astrophysics colloquium belgium 15 - 18 june 1998 p. benvenuti madau 1998 71 treu fundamental plane elliptical galaxies low high redshift next generation space telescope science challenges 34th colloquium 15 18 june 1998 benvenuti madau pp 255 | 0.50709 | 3,068 | 1,515 | , Saving $0.1 in GPT-4. |
55,326 | an overview of the optical and near - ir observations of grbs in the afterglow era is presented .
they have allowed to a better understanding of the underlying physics as well as to constraint the progenitor models . | _ bsax _ made possible to detect the first x - ray afterglow following grb 970228 whose precise localization ( 1@xmath0 ) led to the discovery of the first optical transient ( or optical afterglow , oa ) associated to a grb [ 1,2 ] .
the oa was afterwards found on earlier images [ 3,4 ] and the light curve exhibited a power - law ( pl ) decay f @xmath1 t@xmath2 with @xmath3 = 1.1 [ 5 ] , thus confirming the prediction of the relativistic blast - wave model [ 6 ] .
pl declines have been measured for 26 oas in 1997 - 2000 yielding values in the range 0.8 @xmath4 @xmath3 @xmath4 2.3 with @xmath5 = 1.35 .
a break deviating from a pl decay was first observed in the grb 990123 light curve at t@xmath6 + 1.5 @xmath7 ( @xmath8 1.5 days after the high energy event ) and it was interpreted as the presence of a beamed outflow with a half opening angle @xmath9 @xmath8 0.1 [ 7 - 9 ] , i.e. reducing the inferred energy by a factor of @xmath8 @xmath9@xmath10/4 .
further breaks have been reported in another 5 grbs : grb 990510 [ 10 - 11 ] , grb 991208 [ 12 ] , grb 991216 [ 13,14 ] , grb 000301c [ 15,16 ] and grb 000926 [ 17,18 ] .
there are six possible explanations for the observed breaks : i ) sideways expansion of the jet caused by the swept - up matter [ 19,20 ] although the effect might be negligible for @xmath9 @xmath11 0.1 [ 21 ] ; ii ) when the jet material propagates in an uniform density medium and the observer sees the edge of the jet , with @xmath3 increasing by @xmath8 0.7 [ 22 ] ; iii ) when the jet material propagates in a medium with a pl density profile and the observer sees the edge of the jet , @xmath3 increases by a factor of @xmath12 [ 22 ] with @xmath13 the bulk lorentz factor ; iv ) when the jet material propagates within a pre - ejected stellar wind ( @xmath14 @xmath1 r@xmath15 ) and the observer sees the edge of the jet , @xmath3 increases by @xmath8 0.4 [ 22 ] ; v ) when in both the relativistic and non - relativistic cases ( if @xmath14 is high ) the inverse compton scattering is important , the light curves can be flattened or steepened [ 21 ] ; and vi ) when the transition from the relativistic phase to the non - relativistic phase of an isotropic blastwave takes place in a dense medium ( @xmath14 @xmath8 10@xmath16 @xmath17 ) [ 23 ] .
rapid fading ( @xmath3 @xmath11 2.0 ) has been observed in 3 grbs : grb 980326 [ 24 ] , grb 980519 [ 25,26 ] and grb 991208 [ 12 ] .
two possible causes can explain such behaviour : the synchrotron emission during the mildly relativistic and non - relativistic phases [ 23 ] and the interaction of a spherical burst with a pre - burst wolf - rayet star wind [ 27,28 ] . for grb 970508 ,
a `` plateau '' ( @xmath3 = 0 ) was observed between t@xmath6 + 3 @xmath18 and t@xmath6 + 1 @xmath7 [ 29,30 ] .
the optical light curve reached a peak in two days [ 31,32 ] and was followed by a pl decay f @xmath1 t@xmath19 .
the `` plateau '' has been explained by several plasmoids with different fluxes occurring at different times [ 33 ] .
another `` plateau '' was detected in the near - ir light curve of grb 971214 between t@xmath6 + 3 @xmath18 and t@xmath6 + 7 @xmath18 [ 34 ] .
flux fluctuations are expected due to inhomogeneities in the surrounding medium as a consequence of interstellar turbulence or by variability and anisotropy in a precursor wind from the grb progenitor [ 35 ] .
however , short - term variability was found neither in grb 970508 [ 30 ] nor in grb 990510 [ 11 ] . in grb 000301c ,
the high variability observed at optical wavelengths [ 14,15 ] can be due to several reasons : i ) refreshed shock effects ; ii ) energy injection by a strongly magnetic millisecond pulsar born during the grb [ 36 ] ; iii ) an ultra - relativistic shock in a dense medium rapidly evolving to a non - relativistic phase [ 37 ] ; and iv ) a gravitational microlens [ 38 ] .
a peculiar type ib / c supernova ( sn 1998bw ) was found in the error box for the soft grb 980425 [ 39 ] coincident with a galaxy at z = 0.0085 , but this sn / grb relationship is still under debate . in any case , `` sn - like '' bumps have been detected in other grbs : grb 970228 [ 40,41 ] , grb 970508 [ 42 ] , grb 980326 [ 43,44 ] , grb 980703 [ 45 ] , grb 991208 [ 12 ] and grb 000418 [ 46 ] .
there are some alternative explanations for the existence of such a bump in the oa light curves : i ) scattering of a prompt optical burst by 0.1 m@xmath20 dust beyond its sublimation radius 0.1 - 1 pc from the burst , producing an echo after 20 - 30 @xmath7 [ 47 ] ; ii ) delayed energy injection by shell collision [ 48 ] ; and iii ) an axially symmetric jet surrounded by a less energetic outflow [ 49 ] .
but this is certainly not the case for all grbs : grb 990712 provided the first firm evidence that an underlying sn was not present [ 50 ] . 23.9 and 24.3 respectively .
the dashed - lines are the pure oas contributions to the total fluxes .
the solid lines ( only shown here after t @xmath11 5 @xmath7 ) are the combined fluxes ( oa plus underlying galaxy on each case ) .
upper limits are for grb 991208 . adapted from
[ 12 ] . ]
grb 970508 was the clue to the distance : optical spectroscopy obtained during the oa maximum brightness allowed a direct determination of a lower limit for the redshift ( @xmath21 @xmath22 0.835 ) , implying d @xmath22 4 gpc ( for h@xmath6 = 65 km s@xmath23 mpc@xmath23 ) and e @xmath22 7 @xmath24 10@xmath25 erg .
it was the first proof that grb sources lie at cosmological distances [ 51 ] .
the flattening of the decay at t@xmath6 + 100 @xmath7 [ 30,52 ] revealed the contribution of a constant brightness source -the host galaxy- seen in late - time imaging at t@xmath6 + 1 @xmath26 .
the 15 grb redshifts measured so far are in the range 0.430 @xmath27 z @xmath27 4.50 [ 53 ] with @xmath28 = 1.5 and they were derived either from absorption lines in the oa spectrum , from the ly-@xmath3 break , or from emission lines arising in the host galaxy . as synchrotron radiation under favourable conditions can be up to 70 % polarized , the first polarimetric observations were attempted for grb 990123 , but only an upper limit was established ( @xmath29 @xmath4 2.3% ) [ 54 ] . polarized optical emission was first detected in grb 990510 ( @xmath29 = 1.7 @xmath30 0.2% ) by means of observations performed at t@xmath6 + 18.5 @xmath18 [ 55 ] , t@xmath6 + 21 @xmath18 and t@xmath6 + 43 @xmath18 [ 56 ] .
this confirmed the synchrotron origin of the blast wave itself and represented another case for a jet - like outflow [ 11 ] .
further polarization measurements were carried out in grb 990712 during a 1 @xmath7 time interval ( at t@xmath6 + 10 @xmath18 , t@xmath6 + 17 @xmath18 and t@xmath6 + 35 @xmath18 ) . in that case , the polarization angle did not vary significantly but the degree of polarization was not constant [ 18 ] which can be explained by a laterally expanding jet [ 57 ] . see also [ 58,59 ] .
significant early optical emission may arise from the reverse shock [ 27,60 ] , i.e. strong optical / near - ir flashes accompanying gamma - ray emission should be a generic characteristic ( at least for typical grbs , with e @xmath8 10@xmath31 erg and @xmath14 @xmath8 1 @xmath17 [ 61 ] ) .
such observations will allow : i ) to derive @xmath13 by the relative timing of optical and @xmath32-ray emission , ii ) to pinpoint the process by which the shells responsible for the external shock arise , and iii ) to constraint the environment .
rotse [ 62 ] achieved the detection simultaneously to the grb of the bright optical emission from grb 990123 [ 63 ] : the most luminous object ever recorded ( fig .
1 ) , with m@xmath33 = @xmath3436 ( peaking at m@xmath33 = 8.9 ) , implying that at least some subsets of grbs do exhibit variable optical emission as violent as the gamma - ray variations .
however , upper limits ( in the range r = 4 - 15 ) were derived for prompt optical emission of a dozen of bursts by means of rotse and other experiments , like lotis [ 64 ] , tarot [ 65 ] , bootes [ 66,67 ] , eon [ 68 ] and concam [ 69 ] .
thus , bright optical counterparts are uncommon .
why most optical flashes are not detected ?
due to several reasons : i ) lack of deeper coverage , given the wide grb luminosity function [ 70 ] ; ii ) fireballs in low - density environments ( @xmath14 @xmath35 1 @xmath17 ) would not be expected to produce strong prompt emission ; iii ) the reverse shock energy is radiated at a non - optical frequency , with the synchrotron peak frequency @xmath36 @xmath37 @xmath38 or @xmath36 @xmath35 @xmath38 [ 71 ] ; and iv ) highly absorbed grb by dust in their host galaxies .
the first `` dark '' event ( grb 970828 ) was detected as a fading x - ray source [ 72 ] , although no optical counterpart down to m@xmath39 = 24 at t@xmath6 + 4 @xmath18 was detected [ 73 ] .
at least in another three cases ( grb 981226 , grb 990506 and grb 001109 ) , radiotransients were detected without accompanying optical / ir transients . for grb 000210 ,
the x ray position ( 1@xmath40 accuracy ) coincides with a faint galaxy [ 74 ] . about 40% of the grbs with x - ray counterparts
do not show oas , and this could be due to : i ) intrinsic faintness because of a low ambient medium ; ii ) lyman limit absorption in high redshift galaxies ( @xmath21 @xmath11 7 ) ; and iii ) high absorption in a dusty enviroment : if grbs are tightly related to star - formation , a substantial fraction of them should occur in highly obscured regions . )
compared to the detection of optical counterparts in the afterglow era ( since 1997 ) .
it is clearly seen that the fact the no optical afterglows were discovered prior to the bsax launch was just a matter of bad luck ( i.e. the @xmath8 10 events for which prompt deep follow - up optical searches were performed seemed to be either `` rapidly - fading '' or `` dark '' grbs ) . ]
the first optical / near - ir counterparts have been found for @xmath8 30 precisely localized grbs in 1997 - 2000 although they should have been discovered prior to the _ bsax _ launch ( fig .
2 ) . in any case
, only the population of grbs with durations of few seconds has been explored ( see [ 76 ] for a more extensive review ) .
short bursts lasting less than 1 s , that follow the @xmath343/2 slope in the log n - log s diagram ( in contrast to the longer bursts ) remain to be detected at longer wavelengths .
future missions should be able to address some of the issues still to be solved , i.e. prompt optical and near - ir observations should be persued ! | bsax first x - ray afterglow grb 970228 localization 1@xmath0 first optical transient grb oa found images light curve power - law pl decay @xmath1@xmath2 @xmath3 = 1.1 confirming relativistic blast - wave model pl declines measured 26 oas 1997 - 2000 values 0.8 @xmath4 @xmath3 2.3 @xmath5 = 1.35 break deviating pl decay observed in grb 990123 light curve t@xmath6 + 1.5 @xmath7 @xmath8 1.5 days after high energy event interpreted beamed outflow half opening angle @xmath9 @xmath8 0.1 reducing inferred energy factor @xmath8@xmath10/4 further breaks reported in 5 grbs grb 990510 grb 991208 991216 grb 000301c grb 000926six explanations for breaks : i sideways expansion jet by swept - up matter effect negligible for @xmath9 @xmath11 0.1 ii jet material propagates in uniform density medium observer @xmath3 increasing by @xmath8 0.7 medium pl density profile @xmath3 increases by @xmath12 @xmath13 bulk lorentz factor iv jet propagates pre - ejected stellar wind @xmath3 increases by @xmath8 0.4 v relativistic non - relativistic cases @xmath14 high inverse compton scattering important light curves flattened or steepened vi transition from relativistic to non - relativistic phase of isotropic blastwave in dense medium rapid fading ( @xmath3 @xmath11 2.0 ) observed in 3 grbs 980326 980519 991208 two causes synchrotron emission during relativistic non - relativistic phases interaction of spherical burst with pre - burst wolf - rayet star wind for grb 970508 `` plateau '' ( @xmath3 = 0 ) observed between t@xmath6 + 3 @xmath18 and@xmath6 + 1 @xmath7optical light curve peak in two days followed by pl decay f @xmath1 t@xmath19 plateau explained by several plasmoids different fluxes times plateau in near - light curve grb 971214 between t@xmath6 + 3 and + 7 @xmath18 flux fluctuations expected due to inhomogeneities surrounding medium interstellar turbulence or variability anisotropy in precursor wind grb short - term variability in grb 970508 nor grb 990510 grb 000301c high variability at optical wavelengths due to refreshed shock effects energy injection by magnetic millisecond pulsar ultra - relativistic shock medium non - relativistic phase gravitational microlens peculiar type ib / c supernova ( sn 1998bw ) found in error box for soft grb 980425 coincident with galaxy at z = 0.0085 relationship under debate sn - like bumps detected in other grbs grb 970228 970508 grb 980326 980703 991208 grb 000418alternative explanations bump in oa light curves scattering optical burst 0.1 m@xmath20 dust beyond sublimation radius - 1 echo after 20 - 30 @xmath7 delayed energy injection by shell collision axially symmetric jet less energetic outflow not all grbs grb 990712 evidence underlying sn present 23.9 24.3 dashed - lines pure oas contributions fluxes solid lines after 5 @xmath7 combined fluxes oa plus galaxy upper limits for grb 991208 grb 970508 clue to distance optical spectroscopy maximum brightness lower limit redshift ( @xmath21 @xmath22 0.835 ) implying @xmath22 4 gpc@xmath6 = 65 km 7 @xmath24 10@xmath25 first proof grb sources at cosmological distances flattening decay at t@xmath6 + 100 @xmath7 contribution constant brightness source host galaxy late - time imaging at t@xmath6 + 1 @xmath2615 grb redshifts measured 0.430 @xmath27 z 4.50 @xmath28 = 1.5 derived from absorption lines oa spectrum ly-@xmath3 break emission lines host galaxy synchrotron radiation up to 70 % polarized first polarimetric observations attempted for grb 990123 upper limit established ( @xmath29 @xmath4 2.3% ) polarized optical emission detected in grb 990510 ( @xmath29 = 1.7 @xmath30 0.2% ) observations at t@xmath6 + 18.5 @xmath18 + 21 + 43 confirmed synchrotron origin blast wave jet - like outflow further polarization measurements in grb 990712 1 @xmath7 time interval + 10 + 17 + 35 polarization angle vary degree not constant explained by laterally expanding jet significant early optical emission from reverse shock strong optical / near - ir flashes gamma - ray emission generic characteristic typical grbs @xmath8@xmath31 1 @xmath17observations allow derive @xmath13 timing optical @xmath32-ray emission pinpoint process shells external shock arise constraint environment rotse bright optical emission from grb 990123 most luminous object recorded fig . 1 m@xmath33 = @xmath3436 peaking = 8.9 ) implying grbs exhibit variable optical emission gamma - ray variations upper limits r = 4 - 15 ) derived for prompt optical emission bursts by rotse other experiments like lotis tarot bootes eon concam bright optical counterparts uncommon most optical flashes not detected due to reasons lack of deeper coverage fireballs in low - density environments not produce strong prompt emission reverse shock energy radiated at non - optical frequency synchrotron peak frequency @xmath36 @xmath37 highly absorbed grb by dust in host galaxies first dark event ( grb 970828 ) detected as fading x - ray source no optical counterpart down to m@xmath39 = 24 at t@xmath6 + 4 @xmath18 detectedthree cases ( grb 981226 990506 001109 radiotransients detected without optical / ir transients for grb 000210 x ray position coincides with faint galaxy 40% grbs with x - ray counterparts show oas due to intrinsic faintness low ambient medium lyman limit absorption in high redshift galaxies high absorption in dusty enviroment if grbs related to star - formation fraction should occur in highly obscured regions compared to detection optical counterparts in afterglow era since 1997 no optical afterglows discovered prior to bsax launch bad luck @xmath8 10 events rapidly - fading or dark '' grbs first optical / near - ir counterparts found for @xmath8 30 precisely localized grbs in 1997 - 2000 should have been discovered prior to bsax launch only population of grbs with durations few seconds explored 76 short bursts less than 1 s @xmath343/2 slope detected at longer wavelengths future missions should address issues prompt optical and near - ir observations | 0.437966 | 3,008 | 1,533 | , Saving $0.1 in GPT-4. |
55,334 | we present a lattice algorithm specifically designed for some classical applications of lattice reduction .
the applications are for lattice bases with a generalized knapsack - type structure , where the target vectors are boundably short . for such applications ,
the complexity of the algorithm improves traditional lattice reduction by replacing some dependence on the bit - length of the input vectors by some dependence on the bound for the output vectors .
if the bit - length of the target vectors is unrelated to the bit - length of the input , then our algorithm is only linear in the bit - length of the input entries , which is an improvement over the quadratic complexity floating - point lll algorithms . to illustrate the usefulness of this algorithm
we show that a direct application to factoring univariate polynomials over the integers leads to the first complexity bound improvement since 1984 .
a second application is algebraic number reconstruction , where a new complexity bound is obtained as well . | lattice reduction algorithms are essential tools in computational number theory and cryptography .
a lattice is a discrete subset of @xmath0 that is also a @xmath1-module .
the goal of lattice reduction is to find a ` nice ' basis for a lattice , one which is near orthogonal and composed of short vectors .
since the publication of the 1982 lenstra , lenstra , lovsz @xcite lattice reduction algorithm many applications have been discovered , such as polynomial factorization @xcite and attacking several important public - key cryptosystems including knapsack cryptosystems @xcite , rsa under certain settings @xcite , and dsa and some signature schemes in particular settings @xcite .
one of the important features of the lll algorithm was that it could approximate the shortest vector of a lattice in polynomial time .
this is valuable because finding the exact shortest vector in a lattice is provably np - hard @xcite . given a basis @xmath2 which satisfies @xmath3 , the lll algorithm has a running time of @xmath4 using classical arithmetic .
recently there has been a resurgence of lattice reduction work thanks to nguyen and stehl s @xmath5 algorithm @xcite which performs lattice reduction in @xmath6)$ ] cpu operations .
the primary result of @xmath5 was that the dependence on @xmath7 is only quadratic allowing for improvement on applications using large input vectors . * the main result : * many applications of lll ( see the applications section below ) involve finding a vector in a lattice whose norm is known to be small in advance . in such cases it can be more efficient to reduce a basis of a sub - lattice which contains all targeted vectors than reducing a basis of the entire lattice . in this paper
we target short vectors in specific types of input lattice bases which we call knapsack - type bases .
the new algorithm introduces a search parameter @xmath8 which the user provides .
this parameter is used to bound the norms of targeted short vectors . to be precise :
the specifications of our algorithm are as follows .
it takes as input a knapsack - type basis @xmath11 of a lattice @xmath12 with @xmath13 @xmath14 and a search parameter @xmath8 ; it returns a _ reduced basis generating a _ sub - lattice @xmath15 such that if @xmath16 and @xmath17 then @xmath18 .
_ _ the schnhage algorithm is not widely implemented because of its impracticality . for most polynomials
, @xmath19 is much smaller than @xmath20 .
our main algorithm will reduce the all - coefficients matrix with a competitive practical running time , but constructing the matrix itself will require more hensel lifting than seems necessary in practice . in @xcite a similar switch - complexity bound to section [ psect ] is given on a more practical factoring algorithm .
the problem of finding a minimal polynomial from an approximation of a complex root was attacked in @xcite using lattice reduction techniques using knapsack - type bases . for an extensive treatment
see @xcite .
[ algnumrecon ] suppose we know @xmath21 bits of precision of a complex root @xmath22 of an unknown irreducible polynomial , @xmath23 , where the degree of @xmath24 is @xmath25 and its maximal coefficient has absolute value @xmath26 .
algorithm [ mainalg ] can be used to find @xmath23 in @xmath27 cpu operations .
this new complexity is an improvement over the @xmath5 algorithm which would use @xmath28 cpu operations to reduce the same lattice .
although , one can prove a better switch - complexity with a two - column knapsack matrix by using ( * ? ? ?
* lem . 2 ) to bound the determinant of the lattice as @xmath29 and thus the potential function from @xcite is @xmath30 , leading to a switch complexity of @xmath31 ( posed as an open question in @xcite ) . using this argument the complexity for @xmath5
is reduced to @xmath32 .
* acknowledgements . *
we thank damien stehl , nicolas brisebarre , and valrie berth for many helpful discussions .
also ivan morel for introducing us to h - lll .
this work was partially funded by the lareda project of the agence nationale de la recherche , it was also supported in part by a grant from the national science foundation .
it was initiated while the second author was hosted by the laboratoire dinformatique de robotique et de microlectronique de montpellier ( lirmm ) . | lattice reduction algorithms essential in computational number theory cryptography lattice subset of @xmath0 @xmath1-module goal reduction find nice basis for lattice near orthogonal composed short vectors since 1982 lenstra lovsz @xcite lattice reduction algorithm applications discovered polynomial factorization attacking public - key cryptosystems including knapsack cryptosystems rsa dsa signature schemes lll algorithm shortest vector of lattice in polynomial time finding exact shortest vector hard basis @xmath2 satisfies @xmath3 lll algorithm has running time of @xmath4 using classical arithmetic resurgence of lattice reduction work thanks to nguyen stehl s @xmath5 algorithm performs reduction in cpu operations dependence on @xmath7 quadratic improvement on applications using large input vectors applications of lll involve finding vector in lattice norm small efficient to reduce basis of sub - lattice targeted vectors than reducing basis entire lattice target short vectors in specific input lattice bases knapsack - type bases new algorithm introduces search parameter @xmath8 user provides norms of targeted short vectorsspecifications algorithm takes input knapsack - type basis @xmath11 lattice @xmath12 @xmath13 @xmath14 search parameter @xmath8 returns reduced basis sub - lattice @xmath15 @xmath16 @xmath17 @xmath18 schnhage algorithm not implemented impracticality @xmath19 smaller than @xmath20 main algorithm all - coefficients matrix competitive running time constructing matrix more hensel lifting @xcite similar switch - complexity bound section practical factoring algorithm problem finding minimal polynomial from complex root attacked lattice reduction techniques knapsack - type bases extensive treatment see @xcite @xmath21 precision complex root @xmath22 unknown irreducible polynomial @xmath23 degree @xmath24 @xmath25 maximal coefficient value @xmath26 algorithm find @xmath23 in @xmath27 cpu operations new complexity improvement over @xmath5 algorithm @xmath28 cpu operations reduce lattice better switch - complexity two - column knapsack matrix bound determinant lattice as @xmath29 potential function @xcite @xmath30 switch complexity of @xmath31 open questionargument complexity @xmath5 reduced to @xmath32 acknowledgements thank damien stehl nicolas brisebarre valrie berth discussions ivan morel introducing h - lll work funded lareda project agence nationale recherche supported grant national science foundation initiated second author hosted laboratoire dinformatique robotique microlectronique montpellier | 0.184662 | 1,004 | 506 | , Saving $0.0 in GPT-4. |
55,338 | in this talk we present a model of the universe in which dark energy is modelled explicitely with both a dynamical quintessence field and a cosmological constant .
our results confirm the possibility of a collapsing universe ( for a given region of the parameter space ) , which is advantageous for an adequate formulation of both perturbative quantum field and string theories .
we have also reproduced the measurements of modulus distance from supernovae with good accuracy .
# 1@xmath0#1 | from 1998 to date several important discoveries in the astrophysical sciences have being made , which have given rise to the so called new cosmology @xcite . amongst its more important facts we may cite : - flat , critical density accelerating universe - early period of rapid expansion ( inflation ) - density inhomogeneities produced from quantum fluctuations during inflation - composition : 2/3 dark energy ; 1/3 dark matter ; 1/200 brigh stars - matter content : @xmath1 nonbarionic dark matter ; @xmath2 baryons
, @xmath3 neutrinos - @xmath4 - @xmath5 - @xmath6 it is a fact that the standard lcdm model , though rather simple from the theoretical point of view , can accomodate most of today s astrophysical data .
however , it still has several open questions ( see , for instance , @xcite ) , one of them being : could other ( yet unknown ) models fit the data equally well ?
alternative models should obviously consider the main components of the universe .
there is much concern about unveiling the dark part of our universe , which implies that we do nt lack candidates .
so , for dark matter we have neutrinos , axions , neutralinos ; for dark energy : the cosmological constant , scalar fields ( for example , quintessence ) , cosmic field defects , etc . so far ,
most models of dark energy have a rather phenomenological character , though a few proposals concerning the possible role of the dark energy field in the context of fundamental physics have appeared @xcite .
the recent dark energy scalar field research has several interesting features ( see @xcite for an extense review ) .
many models have attractor or tracker behaviour , allowing , for a wide range of initial conditions , a subdominant field energy density at high redshifts ( radiation and matter dominated eras ) . in the simplest versions , scalar fields models of dark energy
have a scalar field kinetic term , and the scalar field is coupled only to itself and gravity .
so , the scalar field part of the model is fully characterized by the scalar field potential , with some broad constraints on the initial conditions for the field , if the attractor behaviour is realized .
many different potentials have being used ( see reviews @xcite ) : [ cols="<,<,<",options="header " , ] in this talk i want to call the attention to exponential potentials , which have being often discarded on fine tunig arguments or ( the simplest exponential ) because they can not produce the wanted transition from subdominant to dominant energy density ( @xcite ) . however , as shown in @xcite , they have proved useful in describing several features in the history of the universe , from radiation decoupling to nowadays .
also , several authors have recently pointed out that the degree of fine tuning needed in these scenarios is no more than in others usually accepted @xcite .
especially interesting results are obtained if we model dark energy using both a scalar field and a cosmological constant .
the cosmological constant can be incorporated into the quintessence potential as a constant which shifts the potential value , especially , the value of the minimum of the potential , where the quintessence field rolls towards .
conversely , the height of the minimum of the potential can also be regarded as a part of the cosmological constant .
usually , for separating them , the possible nonzero height of the minimum of the potential is incorporated into the cosmological constant and then set to be zero .
the cosmological constant can be provided by various kinds of matter , such as the vacuum energy of quantum fields and the potential energy of classical fields and may also be originated in the intrinsic geometry .
so far there is no sufficient reason to set the cosmological constant ( or the height of the minimum of the quintessence potential ) to be zero , especially when the ultimate fate of our universe is more sensitive to the presence of the cosmological constant ( or the nonzero height of the minimum of the quintessence potential ) than any other matter content , even though the cosmological constant may be extremely tiny and undetectable at all in present time ( @xcite . in particular , some mechanisms to generate a negative cosmological constant have been pointed out , in the context of spontaneous symmetry breaking @xcite .
we consider a model consisting of a three - component cosmological fluid : matter , scalar field ( quintessence with an exponential potential ) and a negative cosmological constant .
we point out that we model dark energy with both the quintessence field and the negative cosmological constant , resulting possitive our effective cosmological constant , in agreement with experimetal data @xcite .
matter means barionic + cold dark matter , with no pressure , and the scalar field is minimally coupled and noninteracting with matter , so the action is : @xmath7 where @xmath8 is the cosmological constant , @xmath9 is the lagrangian for the matter degrees of freedom and the lagrangian for the quintessense field is given by @xmath10 this model can not be used from the very beginning of the universe , but only since decoupling of radiation and dust .
thus , we do nt take into account inflation , creation of matter , nucleosynthesis , etc .
we apply the same technique of adimensional variables we used in @xcite ( this allows to determine the integration constants without additional assumptions ) .
we use the dimensionless time variable @xmath11 , where @xmath12 is the cosmological time and @xmath13 is the present value of the hubble parameter . in this case
@xmath14 is the scale factor .
then we have that , at present @xmath15 @xmath16 considering a homogeneous and isotropic universe , and using the experimental fact of a spatially flat universe @xcite , the field equations derivable from ( 2.1 ) are @xmath17 @xmath18 and @xmath19 where the dot means derivative in respet to @xmath20 and , @xmath21 @xmath22 with @xmath23 - the present density of matter , @xmath24 and @xmath25 - a generic constant .
applying the noether symmetry aproach @xcite , it can be shown that the new variables we should introduce to simplify the field equations are the same used in @xcite : @xmath26 and @xmath27 in these variables the field eqs .
( 2.4 - 2.6 ) may be written as the following pair of equations @xmath28 and @xmath29 where : @xmath30 combining of eqs .
( 2.11 ) and ( 2.12)yields @xmath31 and @xmath32 the solutions of the equations ( 2.14 ) and ( 2.15 ) are found to be @xmath33 and @xmath34 where @xmath35 and @xmath36 are the integration constants . in finding the integration constants we use the equations ( 2.3 ) and field equations evaluated at @xmath37 . finally ,
using @xmath38 and the ansatz @xmath39 where @xmath40 is a positive real number , then the above integration constants can be written in the following way : @xmath41 @xmath42 @xmath43}^{(\pm)}=\frac{\{\sqrt{3}-[\pm]\sqrt{1+q_0 - 1.5\omega{m_0}}\}}{% \sqrt{-3\;\omega_\lambda}}u_2^{(\pm)},\ ] ] and @xmath44}^{(\pm)}=\frac{2-\sqrt{-\omega_\lambda}\;v_2^{(\pm)}u_{1\;[% \pm]}^{(\pm)}}{\sqrt{-\omega_\lambda}\;u_2^{(\pm)}},\ ] ] respectively .
effective quintessence potential @xmath45 from field equations ( 2.4 ) or ( 2.5 ) and equations ( 2.7 ) and ( 2.13 ) can be written @xmath46 so the ansatz ( 2.18 ) establishes an interesting relationship between the value @xmath47 of the exponential potential @xmath48 and the value @xmath49 towards which @xmath45 asymptotes .
other ansatze could have been taken , however , this one notably simplifies equations ( for instance , eq.(2.17 ) ) .
also , as we will see later , many of the cosmological parameters result independent of @xmath40 , avoiding fine tuning respect to this parameter .
since @xmath50 should be real ( see equation ( 2.21 ) ) then , the following constrain on the present value of the deceleration parameter follows after making a detailed study , it was determined that the only relevant cosmological magnitude that has a sensible dependence on parameter @xmath40 is the state parameter @xmath55 .
we used @xmath56 and @xmath57 .
figure 1 shows the evolution of the scale factor for @xmath58 . it can be shown both algebraically and graphically that the evolution of the universe is independent of @xmath40 , but not writen for the sake of simplicity .
this results favour the formulation of both quantum field and string theories .
a breakdown of perturbative quantum field theory in spacetimes with accelerated expansion is known to occur @xcite.on the other hand , an eternally accelerating universe seems to be at odds with string theory , because of the impossibility of formulating the s - matrix . in a desitter space
the presence of an event horizon , signifying causally disconnected regions of space , implies the absence of asymptotic particle states which are needed to define transition amplitudes .
it is also interesting that sen and sethi , using an ansatz for the scale factor that produces future deceleration , obtain from field equations that the quintessence potential should be a double exponential plus a constant @xcite .
figure 2 shows the behaviour of the deceleration parameter as function of the redshift z for the same values of the parameters .
this figure shows an early stage of deceleration and a current epoch of acceleration . a transition from an accelerated phase to a decelerated one
is seen approximately for z = 0.5 .
we appreciate an increase of the deceleration parameter upon increasing the value of z. this points at a past epoch in the evolution when gravity of the dark energy was attractive . as follows from figure 1 , aceleration is not eternal : in the future @xmath59 again , which gives rise to the collapse .
figure 3 shows the evolution of the state parameter of the effective quintessence field @xmath60 .
it s noticeable that the effective quintessence field has state parameter @xmath61 near @xmath62 today , which means that its behaviour is similar to the pure cosmological constant , as a vacuum fluid . if we are to explain the very desirable for today s cosmology recent and future deceleration obtained in our model , it s important to look at the dynamical quintessence field .
we see that in the recent past @xmath63 , which implies that quintessence field behaved ( or simply was ) like ordinary atractive matter , giving rise to the logical deceleration . in the future
this will happen again ( @xmath64 , with the consequent deceleration .
now we proceed to analyze how our solution reproduces some experimental results . with this purpose , in fig .
4 we plot the distance modulus @xmath65 vs redshift z , calculated by us and the one obtained with the usual model with a constant @xmath66 term .
the relative deviations are of about 0.5@xmath67 .
je - an gu and w - y .
p. hwang , hep - th/0102179 ; hep - th/0105133 .
g efstathiou et al , astro - ph/0109152 .
r. cardenas , t. gonzales , o. martin , i. quiros and d. villegas , astro - ph/0112316 ( to appear in gen .
rel . grav . ) . p.
de bernardis et al , nature*404 * , 955 ( 2000 ) . | 1998 to date important discoveries in astrophysical sciences rise to new cosmology facts flat critical density accelerating universe early period rapid expansion inflation density inhomogeneities from quantum fluctuations during inflation composition : 2/3 dark energy ; 1/3 dark matter ; 1/200 brigh stars matter content : @xmath1 nonbarionic dark matter ; @xmath2 baryons @xmath3 neutrinos @xmath4 @xmath5 @xmath6 standard lcdm model accomodate most today s astrophysical data open questions could other models fit data well ? alternative models consider main components universe concern about unveiling dark part of universe candidates for dark matter neutrinos axions neutralinos dark energy cosmological constant scalar fields cosmic field defects most models of dark energy phenomenological character few proposals possible role dark energy field in fundamental physics appeared recent dark energy scalar field research interesting features many models have attractor or tracker behaviour subdominant field energy density at high redshiftsscalar fields models of dark energy have scalar field kinetic term scalar field coupled to itself gravity scalar field part model characterized by scalar field potential constraints on initial conditions for field if attractor behaviour realized different potentials used ( see reviews @xcite ) cols=" attention to exponential potentials often discarded on fine tunig arguments produce transition from subdominant to dominant energy density ( @xcite useful in describing features in history universe from radiation decoupling to nowadays authors degree fine tuning needed in scenarios no more than in accepted @xcite interesting results model dark energy using scalar field and cosmological constant cosmological constant incorporated into quintessence potential shifts potential value especially minimum potential field towards height of minimum potential part of cosmological constant possible nonzero height of minimum potential incorporated into cosmological constant set to be zero cosmological constant provided by various matter vacuum energy quantum potential energy classical fields originated in intrinsic geometryno sufficient reason to set cosmological constant zero fate universe sensitive to than other matter content constant tiny undetectable in present time mechanisms to generate negative cosmological constant pointed out spontaneous symmetry breaking consider model three - component cosmological fluid matter scalar field negative cosmological constant model dark energy with quintessence field negative cosmological constant effective cosmological constant agreement with experimetal data @xcite matter means barionic + cold dark matter no pressure scalar field minimally coupled noninteracting with matter action is @xmath7 @xmath8 is cosmological constant @xmath9 lagrangian for matter degrees freedom lagrangian quintessense field given by @xmath10 model not used from beginning universe only since decoupling of radiation and dust account inflation creation of matter nucleosynthesis apply same technique of adimensional variables in @xcite determine integration constants without assumptions use dimensionless time variable @xmath11 @xmath12 is cosmological time @xmath13 present value hubble parameter @xmath14 is scale factor @xmath15 @xmath16 homogeneous isotropic universe spatially flat universe @xcite field equations from ( 2.1 ) are @xmath17 dot derivative @xmath20 @xmath23 present density matter @xmath24 @xmath25 generic constant noether symmetry @xcite new variables equations same @xcite @xmath26 @xmath27 field eqs ( 2.4 - 2.6 ) equations @xmath28 @xmath29 @xmath30 combining eqs ( 2.11 2.12)yields @xmath31 @xmath32 solutions equations ( 2.14 ) ( 2.15 ) @xmath33 @xmath34 @xmath35 @xmath36 integration constants integration constants use equations ( 2.3 ) field equations @xmath37using @xmath38 ansatz @xmath39 @xmath40 positive number integration constants written @xmath41\pm)}=\frac{\sqrt{3}-[\pm]\sqrt{1+q - 1.5\omega {-3\omega_\lambda}}u_2^{(\pm) @xmath44(\pm)}=\frac{2-\sqrt{-\omega_\lambda}_2{(\pm)}u{1[% \pm]-\omega\lambda_2^{\pm quintessence potential @xmath45 field equations 2.4 2.5 ) 2.7 ) 2.13 ) written @xmath46 ansatz ( 2.18 ) establishes relationship value @xmath47 exponential potential @xmath48 value @xmath49 @xmath45 asymptotes simplifies equations eq.(2.17 cosmological parameters independent of @xmath40 avoiding fine tuning @xmath50 real equation ( 2.21 ) constrain on present value deceleration parameter follows only relevant cosmological magnitude dependence on @xmath40 state parameter @xmath55 used @xmath56 @xmath57figure 1 shows evolution scale factor for @xmath58 evolution universe independent of @xmath40 not results favour quantum field string theories breakdown of perturbative quantum field theory in accelerated expansion @xcite eternally accelerating universe at odds with string theory impossibility of formulating s - matrix event horizon disconnected regions implies absence asymptotic particle states transition amplitudes sen and sethi ansatz future obtain quintessence potential should be double exponential plus constant @xcite figure 2 shows behaviour deceleration parameter function redshift z for same values shows early stage deceleration current epoch of acceleration transition from accelerated phase to decelerated seen approximately for z = 0.5 increase deceleration parameter upon increasing value z points past epoch when gravity dark energy attractive aceleration not eternal future @xmath59 gives rise to collapse figure 3 shows evolution state parameter effective quintessence field @xmath60 has state parameter @xmath61 near @xmath62 behaviour similar to cosmological constant vacuum fluid explain desirable recent and future deceleration model important to look at dynamical quintessence fieldrecent past @xmath63 quintessence field behaved ordinary matter logical deceleration future @xmath64 consequent deceleration analyze solution experimental results fig . 4 distance modulus @xmath65 vs redshift z calculated obtained usual model constant @xmath66 term relative deviations 0.5@xmath67 - an gu. hwang hep - th/0102179 hep th/0105133 efstathiou et al astro - ph/0109152 r. cardenas t. gonzales o. martin. quiros d. villegas astro - ph/0112316 de bernardis et al nature*404 *, 955 ( 2000 ) | 0.4341 | 2,657 | 1,369 | , Saving $0.1 in GPT-4. |
55,339 | polarization is a useful probe to investigate the geometries and dynamics of outflow in bal qsos .
we perform a monte - carlo simulation to calculate the polarization produced by resonant and electron scattering in balr .
we find : 1)a rotated and funnel - shaped outflow is preferred to explain many observed polarization features . 2)the resonant scattering can contribute a significant part of nv emission line in some qsos . | broad absorption lines ( bals ) are observed in the uv spectra of about 10 - 20% qsos and taken as an important indicator for the central engine of qsos .
bal carries only information of the bal region along the line of sight to the continuum source while the global information is contained in the scattered polarized light .
bal qsos are the only highly polarized population among radio quiet qsos .
ogle et al .
( 1999 ) studied the polarization of 36 bal qsos and found : position angle(pa ) rotation in the bals , blueshifted bals in polarized flux relative to that in total flux spectra and deficit of polarized flux redward of the bals . in order to explain these features we perform a monte - carlo simulation to calculate resonant and electron scattering for radial outflows of two different geometries : equatorial and funnel - shaped outflows with and without rotational velocity
we refer the readers to wang et al.(2005 , 2007 ) for details .
we present our results in fig 1(left ) for rotated outflow with geometry similar to elvis(2000 , his figure 3 ) . in an axisymmetric model , pa rotation in the bals can be produced only when the outflow carries angular momentum .
the pa rotation increases with the angular velocity as well as the subtending angle of the flow . in order to produce a pa rotation @xmath0 10observed in a few bal qsos ,
subtending angle of the outflow should be larger than 25 .
similar requirement is imposed to explain the red deficit of polarized flux observed in some bal qsos .
if the subtending angle of outflow is larger than 25the polarization plane of back - scattered light is perpendicular to that of continuum and the polarization vectors of the light redward of the bal cancel , giving a deficit of polarized flux(wang et al .
2007 ) . assuming the electron scattering region(for example the shielding gas , murray et al .
1995 , proga et al .
2000 ) is cospatial with the outflow , we calculate the effect of electron scattering to the polarization in bals . because of rotation of the outflow and similar size of two scattering regions , the electron scattered light fill the trough at @xmath1 , where @xmath2 is the rotational velocity of outflow at the launching radius , and the bal in polarized flux is blueshifted relative to that in the total flux , just as shown in fig .
1(right ) .
we also find resonantly scattered light will contribute a significant part of nv line that is widely used to determine the metal abundance of qsos .
for an equatorial outflow the maximum ratio between the equivalent width(ew ) of the scattering emission to the bal ew increases from 11% to 18.6% if the covering factor of balr increases from 20% to 40% .
the scattering emission of nv is even stronger because of the scattering of the strong ly@xmath3 emission line .
therefore , a strong nv may not indicate a high metal abundance but a large covering factor of the balr(wang et al .
2007 ) . , where @xmath4 is the radial terminal velocity of outflow . in the left panel
we only calculate the resonant scattering and in the right panel we consider both resonant and electron scattering.,title="fig : " ] , where @xmath4 is the radial terminal velocity of outflow . in the left panel
we only calculate the resonant scattering and in the right panel we consider both resonant and electron scattering.,title="fig : " ] elvis , m. 2000 , , 545 , 63 murray n. , chiang j. grossman s.a .
, voit g.m . , 1995 ,
apj , 451 , 498 - 509 ogle p.m. , cohen m.h . , miller j.s .
, tran h.d . , goodrich r.m . ,
martel a.r . , 1999 ,
125 , 1 - 34 proga , d. , stone , j. m. , & kallman , t. r. 2000 , , 543 , 686 wang , h .- y . ,
wang , t .-
, & wang , j .- x . 2005 , , 634 , 149 wang , h .- y . ,
wang , t .-
, & wang , j .- x . 2007 ,
apjs in press , astro - ph/0609469 | absorption lines bals ) in uv spectra 10 - 20% qsos important indicator for central engine qsos bal carries information region to continuum source global information in scattered polarized light bal qsos highly polarized population among radio quiet qsos ogle et al . 1999 studied polarization 36 bal qsos found position angle rotation bals blueshifted bals polarized flux deficit of polarized flux redward monte - carlo simulation resonant electron scattering for radial outflows equatorial funnel - shaped velocity wang et al.(2005 2007 ) details results in fig 1 for rotated outflow geometry similar elvis(2000 axisymmetric model pa rotation in bals when outflow carries angular momentum pa rotation increases with angular velocity subtending angle flow pa rotation subtending angle outflow larger than 25 red deficit of polarized flux in some bal qsos if larger than polarization plane back - scattered light perpendicular to continuum polarization vectors light redward bal cancel deficit of polarized flux(wang 2007 electron scattering region shielding gas2000 cospatial with outflow calculate effect electron scattering polarization bals rotation similar size scattering regions electron scattered light trough at @xmath1 @xmath2 rotational velocity outflow launching radius bal polarized flux blueshifted relative total flux shown fig . 1(right resonantly scattered light significant part nv line metal abundance qsos equatorial outflow maximum ratio scattering emission to bal ew increases 11% to 18.6% if covering factor balr increases 20% to 40% scattering emission nv stronger strong ly@xmath3 emission line strong nv indicate high metal abundance large covering factor balr(wang et al . 2007 ) @xmath4 radial terminal velocity outflow left panel calculate resonant scattering right panel both electron scattering @xmath4 radial terminal velocity outflow left resonant electron elvis ,. 2000 ,, 545 , 63 murray chiang j. grossman s voit g.m 1995 , apj , 451 , 498 - 509 ogle cohen miller tran goodrich martel.r1999 125 1 - 34 proga stone j. m. kallman 2000 543 686 wang h y t . j .- x 2005 634 149 wang h. y t j .- x 2007 apjs press astro - ph/0609469 | 0.184713 | 987 | 496 | , Saving $0.0 in GPT-4. |
55,341 | in this work , we provide two novel approaches to show that incompressible fluid flow in a finite domain contains at most a finite number vortices . we use a recently developed geometric theory of incompressible viscous flows along with an existing mathematical analysis concept to establish the finiteness .
we also offer a second proof of finiteness by roping in the kolmogorov s length scale criterion in conjunction with the notion of diametric disks . | the study of a general system of vortices generated due to flow separation in incompressible viscous flows in a finite domain has eluded proper investigation for quite a long period of time .
most of the available literatures are concerned about flow in some nice and standard geometry only , such as moffatt vortices ( sequence of vortices present in the solid corners of internal viscous incompressible flows ) @xcite , vortices in a semi - circular domain @xcite etc .
the finiteness of such system in the general scenario is still wide open . in the present work we propose a couple of rigorous proofs in the direction of finiteness of such an arbitrary system of vortices .
studies on the finiteness of critical points in a flow domain found through dynamical system approach can be found scantily in the existing literature @xcite .
however , the connection between these critical points and the finiteness of vortices in a bounded domain has yet to catch the attention of the scientific community .
to the best of our knowledge , no rigorous mathematical proof has been provided to establish this connection . in order to achieve this
, we utilize some ideas from the geometric theories of incompressible viscous flows along with the kolmogorov s length scale criterion @xcite .
kolmogorov proposed his famous 1st similarity hypothesis in the year 1941 @xcite .
he propounded that the only factors influencing the behavior of the small scale motions are the overall kinetic energy production rate ( which equals the dissipation rate ) and the viscosity @xcite .
while the dissipation rate is independent of viscosity , the scales at which this energy is dissipated depends both on the dissipation rate and viscosity .
if the dissipation rate per unit mass ( @xmath0 ) has dimensions ( @xmath1 ) and viscosity , @xmath2 has dimension ( @xmath3 ) then the length scale formed from these quantities is given by @xmath4 this length scale is the so called kolmogorov length scale , beyond which size eddies can not be formed either in laminar or turbulent flow @xcite . otherwise , the local value of power density ( @xmath0 ) would be so high that the kinetic energy would be fully dissipated as heat .
we provide two proofs which establish the finiteness of vortices in a bounded domain , one using the bolzano - weierstrass theorem @xcite and the other one using the diametric disk concept @xcite .
* @xmath5 is the set of real numbers . *
@xmath6 is a planar region . * on the boundary @xmath7 of the region @xmath6 , @xmath8 denotes the tangential vector and @xmath9 the normal vector .
* @xmath10 .
* @xmath11 is the tangent bundle of @xmath6 . *
assume @xmath12 is an integer .
let @xmath13 be the space of all @xmath14-th differentiable vector fields @xmath15 on @xmath6 .
* @xmath16 on @xmath17 * @xmath18 * @xmath19 on @xmath17 @xcite _ a point @xmath20 is called a singular point of @xmath21 if @xmath22 .
_ @xcite _ a singular point @xmath23 of @xmath21 is called non - degenerate if the jacobian matrix of @xmath15 at @xmath23 , @xmath24 is invertible . _
@xcite a vector field @xmath21 is called regular if all singular points of @xmath15 are non - degenerate .
@xcite [ lm_bd ] * @xcite * each non - degenerate @xmath25-singular point of @xmath26 is isolated .
in this section we will establish the finiteness of an arbitrary system of vortices ( in incompressible viscous flows ) through two different approaches .
the first proof has its origin in mathematical analysis @xcite and recently developed geometric theory of incompressible viscous flows @xcite .
the second proof leans more on the geometric aspects of the vortices which incorporates of the concept of diametric disk .
suppose @xmath27 is a closed subset of @xmath28 , representing an enclosed domain bounded by solid walls ( or combination of solid walls and free surfaces ) .
then for a steady incompressible viscous flow field defined by the velocity vector @xmath15 , for every point @xmath23 on the boundary , any neighbourhood of @xmath23 contains at most a finite number of vortices .
* it is well known that the process of flow separation on the solid boundaries of a domain leads to the creation of vortices in the flow field .
the points on the boundary where this flow separation takes place , are termed as separation points .
this class of critical points follow the definition of non - degenerate @xmath25-singular points @xcite .
+ if possible , suppose number of such critical points is infinite .
let us denote the set of separation points by @xmath29 .
but since the flow domain ( @xmath30 ) is always bounded , and all these critical points are points from the solid boundary of the flow domain , hence @xmath29 is an infinite bounded subset of @xmath28 . by bolzano - weierstrass theorem @xcite
, @xmath29 must have a limit point ( @xmath31 say ) in @xmath28 .
therefore , there exists a sequence @xmath32 such that @xmath33 as @xmath34 .
moreover , since flow separation takes place in presence of a solid boundary , this point @xmath31 has to lie in the flow domain on the solid boundary .
also , the flow field is continuously differentiable , hence a convergent sequence of @xmath25-singular points must converge to a @xmath25-singular point .
this contradicts the fact that @xmath25-critical points are isolated ( lemma [ lm_bd ] ) .
therefore , number of separation points must be finite .
+ since , separation points lead to the creation of vortices , the finiteness of separation points leads to the finiteness of vortices ( see the note at the end of proof 2 ) in the flow domain .
+ @xmath35 * let @xmath36 be the set representing a vortex
. define , @xmath37 consider the disk of radius @xmath38 centered at @xmath39 .
this is the largest disk inscribed in the vortex @xmath36 with center at @xmath39 .
we call them the diametric disk of @xmath36 @xcite .
refer to figure [ eq1 ] for a schematic of diametric disk in the lid - driven cavity flow @xcite . + + the total area of the flow domain covered by @xmath36 must be @xmath40 @xmath41 . moreover
since the vortices are non - overlapping , hence the diametric disks are mutually disjoint .
therefore , the total area covered by the vortices must be @xmath42 the sum of the areas of the diametric disks , i.e. @xmath43 where @xmath44 is an index set .
but since none of the diameters can drop below the kolmogorov length scale @xcite , therefore @xmath45 as the total area of the flow domain is itself finite , hence the result follows .
@xmath35 + + therefore , the finiteness of the system of vortices is indeed independent of the geometry of the flow domain .
the second proof suggests that even if the flow domain is not simply connected ( contains holes ) , still the finiteness of number of vortices is guaranteed .
* note : * + the separation points on the solid boundary paves the way for the formation of a new vortex .
but one separation point ( @xmath25-singular point ) does not necessarily generate only a single vortex in general over a period of time .
in fact , in the event of breakdown of vortices during vortex - shedding for flow past bluff bodies , at a particular instant of time the set of vortices generated from the same separation point is still finite . the instantaneous vortex will be attached to the solid boundary , whereas the earlier ones will be pushed into the downstream of the flow domain .
the second proof of finiteness of vortices ensures the finiteness of vortices corresponding to the same separation point .
so the set of all vortices in the flow domain being a finite union of finite sets , is finite .
the geometric theories of incompressible viscous flows as a discipline of topological fluid dynamics mainly revolve around the existence of critical points in a flow field . to the best of our knowledge
, no connection of these critical points to the number of vortices in a flow field has been established so far .
we provide two rigorous mathematical proofs to show that the number of vortices in an incompressible fluid flow in a finite domain is finite . for the first proof , a recently developed geometric theory of incompressible viscous flows along with
an existing mathematical analysis concept has been used .
the second proof adopts the kolmogorov s length scale criterion in conjunction with the notion of diametric disks .
b. andersson , r. andersson , l. hakansson , m. mortensen , r. sudiyo and b.v .
_ computational fluid dynamics for engineers
. _ cambridge university press , new york , 2012 .
s. biswas and j. c. kalita .
_ moffatt vortices in the lid - driven cavity flow .
_ doi : arxiv:1601.05186 .
m.d . deshpande and s. george milton . _ kolmogorov scales in a driven cavity flow . _ fluid dyn .
* 22 * ( 1998 ) , 359 - 381 .
u. frisch .
_ turbulence : the legacy of a. n. kolmogorov . _ cambridge university press , new york , 1995 .
r. glowinski , g. guidoboni and t .- w . pan .
_ wall - driven incompressible viscous flow in a two - dimensional semi - circular cavity .
_ j. comput
* 216 * ( 2006 ) , 76 - 91 .
j. jimnez . _ the contributions of a. n. kolmogorov to the theory of turbulence
. _ arbor clxxviii .
* 178 * ( 2004 ) , no .
704 , 589 - 606 .
j. c. kalita , s. biswas , s. panda and a. nal .
_ the finiteness of moffatt vortices .
_ submitted .
e. kirkinis and s. h. davis .
_ moffatt vortices induced by the motion of a contact line .
_ j. fluid mech .
* 746 * ( 2014 ) , r3 .
a. n. kolmogorov . _ the local structure of turbulence in incompressible viscous fluid for very large reynolds numbers .
_ c. r. acad .
30 * ( 1941 ) , 301 - 305 .
t. s. krasnopolskaya .
_ two - dimensional stokes flow near a corner in a right angle wedge and moffatt s eddies .
* 22 * ( 1995 ) , no . 1 , 9 - 14 . t. ma and s. wang .
_ geometric theory of incompressible flows with applications to fluid dynamics .
_ mathematical surveys and monographs , vol-119 , american mathematical society , united states of america , 2005 .
t. ma and s. wang .
_ structure of 2d incompressible flows with the dirichlet boundary conditions . _
s. * 1 * ( 2001 ) , no . 1 , 29 - 41 . v. s. malyuga .
_ viscous eddies in a circular cone
. _ j. fluid mech .
* 522 * ( 2005 ) , 101 - 116 .
h. k. moffatt .
_ viscous and resistive eddies near a sharp corner
. _ j. fluid mech . * 18 * ( 1964 ) , no . 1 , 1 - 18 . w. rudin .
_ principles of mathematical analysis . _ third edition , mcgraw - hill , inc . , new york , 1976 .
_ moffatt eddies in the cone
. _ j. fluid mech .
* 539 * ( 2005 ) , 113 - 135 . v. shtern . _ moffatt eddies at an interface .
_ theor . comp .
fluid dyn .
* 28 * ( 2014 ) , 651 - 656 .
p. walstra .
_ physical chemistry of foods .
_ marcel dekker , inc . , new york , 2003 . | study of system of vortices due to flow separation in incompressible viscous flows in finite domain eluded investigation for literatures concerned about flow in standard geometry moffatt vortices in semi - circular domain finiteness of system in general scenario wide open present work propose rigorous proofs finiteness of arbitrary system of vortices studies on finiteness of critical points in flow domain through dynamical system approach in existing literature connection between critical points and finiteness of vortices in bounded domain attention scientific community no rigorous mathematical proof to establish connection utilize ideas from geometric theories of incompressible viscous flows kolmogorov s length scale criterion kolmogorov proposed 1st similarity hypothesis in 1941 factors influencing behavior small scale motions are kinetic energy production rate dissipation rate and viscosity dissipation rate independent of viscosity energy depends on viscositydissipation rate per unit mass @xmath0 has dimensions @xmath1 viscosity @xmath2 dimension length scale by @xmath4 kolmogorov length scale beyond size eddies formed in laminar turbulent flow local power density @xmath0 high kinetic energy dissipated as heat two proofs finiteness of vortices in bounded domain bolzano - weierstrass theorem diametric disk concept @xmath5 real numbers @xmath6 planar region boundary @xmath8 tangential vector @xmath9 normal vector @xmath11 tangent bundle of @xmath6 @xmath12 integer @xmath13 space of @xmath14-th differentiable vector fields @xmath15 on @xmath6 @xmath16 @xmath17 @xmath19 point @xmath20 singular point of @xmath21 singular point @xmath23 of @xmath21 non - degenerate if jacobian matrix of invertible vector field @xmath21 regular if all singular points of @xmath15 non - degenerate non - degenerate @xmath25-singular point of @xmath26 isolatedestablish finiteness of arbitrary system of vortices in incompressible viscous flows through two approaches first proof in mathematical analysis geometric theory second proof on geometric aspects concept diametric disk @xmath27 closed subset of @xmath28 enclosed domain bounded by solid walls for steady incompressible viscous flow field defined by velocity vector @xmath15 every point @xmath23 on boundary neighbourhood contains finite number of vortices flow separation on solid boundaries leads to creation vortices points on boundary separation points critical points follow definition non - degenerate @xmath25-singular points number of critical points infinite denote set of separation points by @xmath29 flow domain @xmath30 always bounded critical points from solid boundary @xmath29 infinite bounded subset of @xmath28 bolzano - weierstrass theorem @xmath29 must have limit point @xmath31 in @xmath28 exists sequence @xmath32 @xmath33 as @xmath34 flow separation in solid boundary point @xmath31 in flow domain on solid boundaryflow field differentiable convergent sequence @xmath25-singular points converge to @xmath25-singular point contradicts @xmath25-critical points isolated number separation points finite separation points lead to creation vortices finiteness leads finiteness vortices 2 flow domain @xmath35 @xmath36 vortex @xmath37 disk of radius @xmath38 centered at @xmath39 largest disk in vortex center @xmath39 diametric disk of @xmath36 @xcite figure eq1 schematic diametric disk in lid - driven cavity flow total area flow domain covered by @xmath36 be @xmath40 @xmath41 vortices non - overlapping diametric disks mutually disjoint total area vortices must be @xmath42 sum of areas diametric disks @xmath43 @xmath44 index set diameters drop below kolmogorov length scale @xcite @xmath45 total area flow domain finite result follows @xmath35 finiteness system vortices independent of geometry flow domain second proof suggests flow domain not connected finiteness of number vortices guaranteedseparation points on solid boundary formation new vortex one separation point-singular point generate single vortex breakdown of vortices during vortex - shedding flow bluff bodies vortices from same separation point finite instantaneous vortex attached to solid boundary earlier pushed into downstream flow domain second proof of finiteness vortices ensures finiteness vortices same separation point all vortices in flow domain finite geometric theories of incompressible viscous flows revolve around critical points in flow field no connection critical points to number vortices flow established two mathematical proofs show number vortices in incompressible fluid flow finite domain finite first proof geometric theory existing mathematical analysis concept used second proof adopts kolmogorov s length scale criterion diametric disks b. andersson r. andersson l. hakansson m. mortensen r. sudiyo b.v computational fluid dynamics for engineers cambridge university press new 2012 s. biswas j. c. kalita moffatt vortices in lid - driven cavity flow arxiv:1601.05186 m deshpande s.george milton kolmogorov scales driven cavity flow fluid dyn 22 1998 359 - 381 frisch turbulence legacy kolmogorov cambridge university press new york 1995 glowinski guidoboni pan wall - driven incompressible viscous flow two - - circular cavity j comput 216 2006 76 - 91 jimnez contributions kolmogorov theory turbulence arbor clxxviii 178 2004 ) 704 589 - 606 kalita biswas panda a nal finiteness moffatt vortices kirkinis davis moffatt vortices motion contact line fluid mech 746 2014 n kolmogorov local structure turbulence incompressible viscous fluid large reynolds numbers c r. acad 30 1941 ) 301 - 305 krasnopolskaya two - stokes flow corner right angle wedge moffatt s eddies 22 1995 ) 1 9 - 14 ma wang geometric theory incompressible flows fluid dynamics mathematical surveys monographs vol-119 american mathematical society 2005 wang structure 2d incompressible flows dirichlet boundary conditions2001 ) 1 29 - 41 malyuga viscous eddies circular cone j fluid mech 522 2005 ) 101 - 116 moffatt viscous resistive eddies sharp corner fluid mech 18 1964 ) 1 1 - 18 rudin principles mathematical analysis third edition mcgraw - hill new york 1976 moffatt eddies cone fluid mech 2005 ) 113 - 135. shtern moffatt eddies interface fluid dyn 28 2014 ) 651 - 656 p. walstra physical chemistry foods marcel dekker new york 2003 | 0.43433 | 2,840 | 1,459 | , Saving $0.1 in GPT-4. |
55,344 | a planetary transit produces both a photometric signal and a spectroscopic signal .
precise observations of the transit light curve reveal the planetary radius and allow a search for timing anomalies caused by satellites or additional planets .
precise measurements of the stellar doppler shift throughout a transit ( the rossiter - mclaughlin effect ) place a lower bound on the stellar obliquity , which may be indicative of the planet s migration history .
i review recent results of the transit light curve project , and of a parallel effort to measure the rossiter effect for many of the known transiting planets . | i have great admiration for the people who discover transiting planets . identifying the candidate transit signals from among a hundred thousand light curves , and flushing out the numerous astrophysical false positives ,
are impressive feats .
this article , however , is not about transit discovery , but rather about the next step : performing high - precision photometry and spectroscopy of exoplanetary transits .
the goal of this step is to determine the planetary and stellar properties well enough to allow for meaningful comparisons with the familiar properties of the solar system , and to inform our theories of planet formation .
the most immediate result of transit photometry is a measurement of the planetary radius . in combination with the planetary mass , which can be inferred from the doppler orbit of the star , these data give the first clues about the composition , interior structure , and atmospheric energy balance of the planet .
an accurate radius is also needed to interpret the results of other observations , such as the detection of thermal emission or reflected light based on secondary - transit photometry .
the timings of the transits can be used to refine the measurement of the orbital period and search for additional bodies in the system . in
2 , i describe the transit light curve ( tlc ) project , an effort to gather high - precision photometry during exoplanetary transits .
the most prominent spectroscopic signal during a transit is the rossiter - mclaughlin ( rm ) effect .
this effect is an anomalous doppler shift that arises from stellar rotation . measuring this effect
allows one to assess the alignment between the planetary orbital axis and the stellar spin axis , a fundamental system property that provides clues about the process of planet migration .
i describe some recent measurements of the rm effect in 3 .
the most spectacular transit light curves have come from space - based observations with the _ hubble space telescope _ ( brown et al . 2001 ,
pont et al . 2007 ) and
the _ spitzer space telescope _
( knutson et al . 2007 ) . however , it is important to remember that the extreme precision with which the relative flux can be recorded ( @xmath0@xmath1 per 30 s sample ) does not guarantee an extremely precise value for the planetary radius .
rather , the transit depth @xmath2 is known precisely , but @xmath3 is known only as well as @xmath4 .
the stellar radius must be estimated using other observable properties of the star , such as its parallax , luminosity , angular diameter , or spectrum .
actually the situation is slightly more complex .
in addition to the transit depth , precise photometry provides a few other observables , among which are the photometric period @xmath5 , the total transit duration @xmath6 ( from first to fourth contact ) , and the full transit duration @xmath7 ( from second to third contact ) .
kepler s third law can be used to write an accurate expression for the mean stellar density in terms of observable quantities ( seager & mallen - ornelas 2003 ) : @xmath8 this approximation assumes @xmath9 , @xmath10 , and a circular orbit , but it is straightforward to generalize this expression .
thus , precise photometry reveals the mean stellar density .
this is helpful because it means that the light curve itself helps to estimate the stellar radius .
a traditional method for estimating the stellar radius is to compare the spectroscopic values of @xmath11 , @xmath12 , and metallicity with the output of theoretical models of stellar evolution .
with excellent transit photometry it is advantageous to use @xmath13 instead of @xmath12 ( see , e.g. , sozzetti et al . 2007 , holman et al .
a related point is that in some regions of the hr diagram , the stellar _ mass _ is constrained at least as well as the stellar radius . at fixed stellar density that is , with a perfect transit light curve the error in @xmath4 ( and hence @xmath3 )
varies only as the cube root of @xmath14 . in such cases the systematic error in @xmath3
would be reduced by a factor of three relative to a situation in which only the transit depth ( and not the durations ) were known .
another quantity that can be written purely in terms of observables is the planetary surface gravity ( southworth et al .
2007 ) . in this sense , what a transit observer _ really _ measures is the stellar mean density and the planetary surface gravity .
transforming these variables into @xmath3 and @xmath15 requires external information . in 2006 , matt holman and i wondered whether it would be possible to reach the limiting precision in @xmath3 by combining the results of repeated ground - based observations .
we began the transit light curve ( tlc ) project , with the aim of building a library of transit photometry of uniformly high quality , along with unified and rigorous methods for parameter determination .
our workhorse is the fred l. whipple 1.2 m telescope and keplercam detector .
we have also used the 6.5 m magellan telescopes for the ogle systems , which are both southerly and faint .
we have found that by employing these instruments and some straightforward observing protocols , it is possible in some cases to reach the limit in which the error in @xmath3 is dominated by the uncertainties in the stellar properties rather than the photometric uncertainties . among our observing protocols are the following .
when possible we center the field in such a way as to encompass at least 5 comparison stars of similar brightness to the target star .
we keep the image registration as consistent as possible throughout the night .
we defocus , if needed , to broaden the point - spread function to a consistent width of 5 - 7 pixels ; this often has the salutary effect of enhancing the duty cycle by lengthening the maxmimum exposure time .
we ensure that the flat - field calibration frame has enough counts to correct for @xmath00.1% pixel - to - pixel sensitivity variations .
we observe not only the entire transit but also at least 1 hr prior to ingress and 1 hr after egress .
we use the reddest available optical bandpass ( often sloan @xmath16 ) , for two reasons : to minimize the effect of differential extinction on the relative photometry , and to minimize the effect of limb darkening on the transit light curve . figure 1 is a gallery of some of our composite light curves , along with the standard deviation of the noise and citations to published results .
figure 2 shows the favorable noise characteristics that we have found for a subset of our target systems .
naturally , the best results are obtained when the field of view offers numerous good comparison stars and when the sky conditions are pristine .
( even relative photometry benefits from `` photometric '' nights . )
a photometric precision of 0.15% with 30-second sampling is typical .
we have written a comprehensive analysis code that simultaneously fits multiple transits observed in different bandpasses , secondary transits ( when such data are available ) , and radial velocities ( including the rossiter - mclaughlin effect ; see 3 ) .
we use a markov chain monte carlo algorithm to estimate the joint _ a posteriori _ probability distribution for all relevant parameters .
going forward , we intend to make some potentially important improvements to the code , including a more sophisticated treatment of `` red noise '' employing a model of the covariance matrix ( at the moment we use a crude approxmation ) , and a principal - component decomposition of the limb darkening function . over the longer term
, we hope to detect ( or rule out ) additional planets and satellites of the transiting planet .
these may manifest themselves through gradual changes in the transit duration caused by orbital precession ( miralda - escud 2002 , heyl & gladman 2007 ) and by short - term variations in the midtransit times due to multi - keplerian motions or mutual gravitational interactions .
we are gathering a large number of transit times with a precision of 0.25 - 1 min .
holman & murray ( 2005 ) and agol et al .
( 2005 ) have shown that transit - timing variations of that order could be produced by perturbers as small as the earth , as long as they are in mean - motion resonances with the transiting planet .
resonances might be a common outcome of planet formation and migration , and searching for transit - timing variations is one way to find out .
a striking pattern in the solar system is its coplanarity .
the orbital axes of the 8 planets line up to within a few degrees , and the sun s spin axis differs by only 7 degrees from the earth s orbital axis .
the observed coplanarity was the original inspiration for the theory that the sun and planets condensed from a single spinning disk
. it would be interesting to know whether this degree of alignment is typical of all planetary systems .
exoplanets have provided enough surprises that nothing should be taken for granted .
indeed there are reasons to expect at least occasional misalignments . among the surprises
to which i just alluded is that planetary orbital eccentricities are often large ; whatever mechanism perturbs orbital eccentricities may also perturb inclinations . for close - in planets , which are thought to have formed at large orbital distances and then migrated inward
, one may wonder whether the migration process disturbed the original alignment .
the various migration theories differ on this point .
migration via tidal interactions with the protoplanetary disk should not perturb the alignment and may even drive the system toward closer alignment ( ward & hahn 2003 ) . in contrast , migration via planet - planet scattering would magnify any initial misalignments ( chatterjee , ford & rasio 2007 ) , and migration via kozai cycles accompanied by tidal circularization results in a broad distribution of final inclination angles ( fabrycky & tremaine 2007 ) .
thus , measuring the alignment between the orbital axis and the stellar spin axis is a means for testing migration theories and for identifying planets that migrated in interesting ways .
transit observations can be used to assess the spin - orbit alignment , courtesy of a spectroscopic effect observed long ago by rossiter ( 1924 ) and mclaughlin ( 1924 ) in eclipsing binaries . during a transit ,
the planet hides part of the rotating stellar disk , causing an `` anomalous '' doppler shift .
when the planet transits the approaching ( blueshifted ) half of the star , the net starlight appears slightly redshifted , and vice versa , as shown in fig .
3 . for the configuration shown ,
in which the transit impact parameter is approximately 0.5 , the signal for a well - aligned planet is antisymmetric about the midtransit time , whereas a strongly misaligned planet that spends all its time covering the receding half of the star would produce only an anomalous blueshift .
the key parameter is @xmath17 , the angle between the sky projections of the orbital axis and the stellar rotation axis .
measurements of the projected spin - orbit angle @xmath17 have been published for 5 systems ( queloz et al .
2000 , winn et al .
2005 , 2006 , 2007 ; wolf et al . 2007 ;
narita et al .
2007 ) . in all cases
@xmath17 is consistent with zero at the 1 - 2@xmath18 level , with accuracies ranging from 130 degrees .
this has led to the working hypothesis that the migration of hot jupiters generally preserves spin - orbit alignment .
importantly , star - planet tidal interactions ( which are responsible for spin synchronization and orbital circularization of the planet ) do not confound the interpretation because the expected timescale for tidal reorientation is generally @xmath19 yr or longer , as estimated using the equilibrium - tide theory of hut ( 1981 ) .
observations of the rossiter - mclaughlin effect provide only a lower limit on the stellar obliquity , the angle between the orbital and rotation axes .
the rm waveform is sensitive the angle between the sky projections of those axes , and the inclinations with respect to the sky must be determined with other data .
the orbital inclination is often known to within @xmath201 deg from the transit light curve , but the stellar inclination is usually unknown .
the single exception thus far is hd 189733 .
that star is chromospherically active and exhibits quasiperiodic flux variations ( presumably due to star spots ) from which the rotation period can be measured ( henry & winn 2007 ) . the combination of a measured rotation period , stellar radius , and projected rotation rate ( which can be measured from either the amplitude of the rm effect or from the observed spectral - line broadening ) places a constraint on the stellar inclination , which is consistent with being edge - on .
more work remains to be done on confronting the specific predictions of planet - planet scattering and kozai cycles with the data .
the case of hd 147506 ( also known as hat - p-2 ) is especially interesting because the planet has a highly eccentric orbit ( @xmath21 ) , naturally raising the possibility of planet - planet scattering or kozai cycles , yet the system is well - aligned ( winn et al .
2007 ) . in this case , however , the achievable accuracy in @xmath17 was limited to about 14 degrees because the transit is nearly equatorial . in such cases
the rossiter waveform is always antisymmetric and the value of @xmath17 is encoded in the _ amplitude _ of the signal , which causes a strong degeneracy with the value of the stellar projected rotation velocity ( see gaudi & winn 2007 ) .
it is a pleasure to thank matt holman , my partner in the transit light curve project ; and john johnson , scott gaudi , geoff marcy , norio narita , yasushi suto , and ed turner , for their important contributions to the rossiter program .
i am very grateful to fred rasio , alex wolszczan , and the other conference organizers for the opportunity to share this work in a memorable setting on santorini .
winn , j. n. , et al .
2007a , , 665 , l167 winn , j. n. , et al .
2007b , arxiv e - prints , 707 , arxiv:0707.1908 winn , j. n. , et al . 2007c , , 133 , 1828 wolf , a. s. , laughlin , g. , henry , g. w. , fischer , d. a. , marcy , g. , butler , p. , & vogt , s. 2007 , , 667 , 549 | admiration for people transiting planets identifying candidate transit signals thousand light curves flushing out astrophysical false positives impressive feats this article not about transit discovery about next step high - precision photometry spectroscopy of exoplanetary transits goal to determine planetary stellar properties for comparisons with solar system inform theories planet formation immediate result transit photometry is measurement planetary radius with planetary mass inferred from doppler orbit data give clues about composition interior structure atmospheric energy balance planet accurate radius needed to interpret results other observations detection of thermal emission reflected light secondary transit photometry timings transits to refine measurement orbital period search for additional bodies transit light curve project high - precision photometry during exoplanetary transits prominent spectroscopic signal transit is rossiter - mclaughlin ( rm ) effect anomalous doppler shift from stellar rotation measuring effect alignment between planetary orbital axis stellar spin axis clues about planet migration describe recent measurements of rm effect spectacular transit light curves from space - observations with hubble space telescope spitzer space telescopeextreme precision relative flux recorded ( @xmath0@xmath1 per 30 s sample guarantee precise value planetary radius transit depth @xmath2 known precisely @xmath3 known only as @xmath4 stellar radius estimated using other properties star parallax luminosity angular diameter spectrum situation complex transit depth precise photometry provides other observables photometric period @xmath5 total transit duration @xmath6 first to fourth full transit duration @xmath7 second to third contact kepler s third law accurate expression mean stellar density ( 2003 : @xmath8 assumes @xmath9 @xmath10 circular orbit straightforward to generalize expression precise photometry reveals mean stellar density light curve helps estimate stellar radius traditional method radius spectroscopic values @xmath11 @xmath12 metallicity with theoretical models stellar evolution excellent transit photometry advantageous use @xmath13 instead of @xmath12 sozzetti et 2007 holman et al in some regions hr diagram stellar _ mass constrained stellar radiusfixed stellar density perfect transit light curve error in @xmath4 @xmath3 varies as cube root of @xmath14 systematic error in @xmath3 reduced by factor three transit depth not durations known quantity planetary surface gravity ( southworth et al . 2007 ) transit observer measures stellar mean density planetary surface gravity transforming variables into @xmath3 @xmath15 requires external information 2006 matt holman i reach limiting precision in @xmath3 combining results repeated ground - observations began transit light curve ) project aim building library of transit photometry high quality unified methods for parameter determination workhorse fred l. whipple 1.2 m telescope keplercam detector used 6.5 m magellan telescopes for ogle systems southerly employing instruments straightforward observing protocols reach limit error in @xmath3 dominated by uncertainties stellar properties photometric uncertainties observing protocols center field encompass 5 comparison stars similar brightness to target star keep image registration consistent defocus if needed broaden point - spread function to consistent width 5 - 7 pixels duty cycle maxmimum exposure timeflat - field calibration frame counts @xmath00.1% pixel - to - pixel sensitivity variations observe entire transit 1 hr prior ingress 1 hr after egress use reddest optical bandpass sloan @xmath16 ) minimize differential extinction relative photometry limb darkening transit light curve figure 1 composite light curves standard deviation noise citations published results figure 2 shows favorable noise characteristics for target systems best results when field of view good comparison stars sky conditions pristine relative photometry benefits from photometric nights photometric precision of 0.15% 30-second sampling typical written comprehensive analysis code fits multiple transits secondary transits radial velocities rossiter - mclaughlin effect use markov chain monte carlo algorithm estimate joint a posteriori probability distribution for parameters intend make improvements code sophisticated treatment of red noise model covariance matrix principal - component decomposition of limb darkening function longer term hope to detect or rule out additional planets satellites of transiting planetmanifest through changes transit duration orbital precession 2002 2007 short - term variations midtransit times due to multi - keplerian motions gravitational interactions gathering transit times precision 0.25 - 1 min holman murray 2005 agol et al 2005 shown transit - timing variations produced by perturbers small as earth in mean - motion resonances with transiting planet resonances outcome planet formation migration searching for transit - timing variations striking pattern in solar system coplanarity orbital axes 8 planets line up few degrees sun s spin axis differs 7 degrees from earth s orbital axis coplanarity original inspiration for theory sun planets from single spinning disk interesting know alignment typical of all planetary systems exoplanets surprises reasons to expect occasional misalignments planetary orbital eccentricities large inclinations close - in planets formed large distances migrated inward wonder migration disturbed original alignment migration theories differ migration via tidal interactions with protoplanetary disk perturb alignment may drive toward closer alignment ( ward hahn 2003 )migration via planet - scattering initial misalignments ( 2007 migration via kozai cycles tidal circularization broad final inclination angles ( fabrycky tremaine 2007 ) measuring alignment between orbital axis stellar spin axis testing migration theories identifying planets transit observations assess spin - orbit alignment spectroscopic effect observed by rossiter 1924 mclaughlin 1924 in eclipsing binaries during transit planet hides rotating stellar disk anomalous doppler shift planet transits approaching half star net starlight slightly redshifted shown in fig . 3 transit impact parameter 0.5 signal for well - aligned planet antisymmetric about midtransit time strongly misaligned planet anomalous blueshift key parameter is @xmath17 angle between sky projections orbital axis stellar rotation axis measurements of spin - orbit angle @xmath17 published for 5 systems ( queloz 2000 winn wolf @xmath17 consistent with zero at 1 - 2@xmath18 level accuracies from 130 degrees migration of hot jupiters preserves spin - orbit alignmentstar - planet tidal interactions responsible for spin synchronization orbital circularization confound interpretation expected timescale for tidal reorientation @xmath19 yr or longer estimated using equilibrium - tide theory of hut 1981 rossiter - mclaughlin effect provide lower limit on stellar obliquity angle between orbital rotation axes rm waveform angle between sky projections inclinations determined with other data orbital inclination known within @xmath201 deg from transit light curve stellar inclination unknown exception is hd 189733 star chromospherically active exhibits quasiperiodic flux variations due to star spots rotation period measured combination measured rotation period stellar radius projected rotation rate on stellar inclination consistent with edge - work confronting predictions of planet - planet scattering kozai cycles with data case of hd 147506 hat - p-2 ) eccentric orbit ( @xmath21 ) possibility of planet - planet scattering kozai cycles system well - aligned achievable accuracy in @xmath17 limited to 14 degrees transit nearly equatorialrossiter waveform antisymmetric value @xmath17 encoded in amplitude signal causes degeneracy stellar projected rotation velocity gaudi winn 2007 thank matt holman transit light curve project john johnson scott gaudi geoff marcy norio narita yasushi suto ed turner contributions rossiter program grateful to fred rasio alex wolszczan conference organizers work memorable setting santorini winn , j. n. et al . 2007a , , 665 , l167 2007b arxiv e - prints , 707 , arxiv:0707.1908 2007c, 133 , 1828 wolf laughlin henry fischer marcy butler p vogt 2007 , , 667 , 549 | 0.508846 | 3,194 | 1,592 | , Saving $0.1 in GPT-4. |
55,346 | the quantum mechanical wave - particle dualism is analyzed and criticized , in the framework of reichenbach s concepts of phenomenon and interphenomenon .
it is suggested that the dual pictures be de - emphasized in the study of quantum theory . in this connection
a view of the electron and other such particles as having discontinuous spatiality in time is presented and discussed . | textbooks on quantum mechanics typically begin by discussing the wave - particle dualism in its extreme aspects : such processes are best described in the particle picture , such other in the wave picture . some texts , like schiff s @xcite ,
for example , do then present a strictly correct , but too succinct quantum mechanical viewpoint .
after this the mathematical formalism is developed , and only occasionally the qualitative aspects of the theory are touched upon . it is left to the student the not - so - easy task of integrating the qualitative and quantitative fundamentals of quantum theory into a logically coherent , visualizable picture of the atomic world . in the present paper
i first make an analysis and criticism of the concept of dualism .
our purpose is to sharply expose the artificiality and non - uniqueness of both the wave and the particle pictures . to our knowledge
the best analysis of the quantum mechanical dualism is that of h. reichenbach , in his book _ philosophic foundations of quantum mechanics _
i am therefore guided by reichenbach s ideas in our study of the meaning of the wave - particle dualism .
second , and more important , a qualitative picture of microscopic reality is advocated , which is consistent with the more abstract interpretation of quantum mechanics .
as is well known , this interpretation assigns physical reality to observable facts only .
the key to our preferred language is to abandon thinking of micro - objects
i shall deal with _ electrons _ for definiteness as having permanent spatiality ; that is , as having spatial properties at all times .
for instance , the question where is the electron when it is not observable ? is meaningless , since the electron by definition does not have a place when it does not participate in an observable phenomenon .
this meaninglessness is not a mere consequence of unobservability .
it is rather a matter of adequate definition . in section 2
reichenbach s ideas of phenomenon and interphenomenon are introduced , as a basis for the discussion of the wave - particle dualism . in sec .
3 an alternative language is presented , which pre - empties micro - objects such as electrons or photons of classical - like spatial attributes , such as temporal continuity of position or shape , trajectory , or propagation .
i try to make it clear that there is no real need for such attributes in micro - objects .
next ( sec .
4 ) both viewpoints - i.e. , those of sections 2 and 3 - are applied to a physical example .
i call the reader s attention to the fact that physics progresses by increasing degrees of abstraction . concluding remarks are made in the last section .
reichenbach ( p. 21 of @xcite ) defines _ phenomena _ as occurrences which are so easily inferable from macroscopic data that they may be considered observable in a wider sense . in the same page
he rightly points out that the phenomena are ... determinate in the same sense as the unobserved objects of classical physics . as a matter of fact
, physicists are not troubled by the question of the reality of observable - but - unobserved events , no more than by the question of the reality of unobserved trees . as r. p. feynman puts it : nature does not know what you are looking at , and she behaves the way she is going to behave whether you bother to take down the data or not.@xcite in other words , the concept of phenomenon supposes berkeley s problem solved .
a satisfactory solution is given by a. j. ayer @xcite .
_ interphenomena _ are occurrences ... constructed in the form of an interpolation within the world of phenomena ( p. 21 of @xcite ) .
they are logical constructs similar to phenomena , usable to fill the gaps in the world of phenomena , so that the whole have an appearance of continuity in time .
notice that this is a purely mental process !
it is precisely this process that i intend to dispose of , as will be seen below .
next , reichenbach says that the distinction betien phenomena and interphenomena is the quantum mechanical analogue of the distinction between observed and unobserved things ( p. 21 of @xcite ) .
this analogy is somewhat misleading , because the relation between phenomena and interphenomena is not so simple as that between observed and unobserved macro - objects .
if one likes labels , we might call the question of interphenomena berkeley s problem of the second degree : formerly one had to construct objects out of actual or expected sense - experiences ; here one has to define entities related to experience only through its end points , the initial and final events .
after these preliminaries , it may be said that the particle and wave pictures in quantum theory represent different choices of interphenomena . in the former ,
the gap between , say , two consecutive electron appearances ( two happenings , or phenomena , involving the electron ) is filled by an imagined trajectory ( interphenomena ! ) traveled by the electron , imagined as a point - like object . with wave - like interphenomena , on the other hand , the same gap if filled by imagining the first phenomenon as a disturbance in a medium , then a wave - like propagation of the disturbance , and finally a quick coalescing of the wave into another point - like disturbance .
the considerations of the preceding section will be illustrated by an example in the next section . here
the dualistic view is countered with what i consider a more rational and satisfactory picture .
the point is this : there is no need for interphenomena . beween two consecutive , observable events nothing happens to the electron . during this time interval
the electron is nowhere , it is out of space .
one is not allowed to say that it has as indefinite position somewhere in the neighborhood of the events .
contrarily to the situation with macroscopic things , micro - objects like electrons do not constantly occupy a place , they have a place only when they appear .
electrons have a permanent potentiality to interact , mathematically expressed as its charge , and hence to appear in space , but no permanent spatiality .
( this is a qualitative content of the well known assertion that the state function @xmath0 gives a probability amplitude , not a wave amplitude .
see also section iv.3 in @xcite . ) in other words , its spatiality is limited , or discontinuous in time.i hope to make this point clearer in next section .
one might object that this picture has the disadvantage of being more abstract than the dual pictures .
my answer to this is : ( 1 ) abstract is a relative term : a moment s reflection may convince the reader that the classical concept of trajectory is also quite abstract , though less so than quantum mechanical concepts ; and ( 2 ) progress in physics is often reached by increasing degrees of abstraction , as for example the abandonment of a material ether as the medium of electromagnetic waves .
this is only natural , since concrete ideas are just those of our too limited , ordinary sense experience .
let us now illustrate the problem through an idealized experiment of diffraction of an electron by a slit in a diaphragm @xcite .
figure 1 is a sketch of the experimental situation : _ a _ is the electron source , _ s _ is the slit in the diaphragm , and _
b _ is the point on the screen where the electron is absorbed .
the source is supposed to be so weak that only one electron is emitted at a time . the emission of an electron at _ a _ and its absorption at _ b _ are point - like events , and are phenomena in the sense defined above .
let us call them process @xmath1 and process @xmath2 , respectively .
the space and time gap between these phenomena has been filled by reichenbach with particle - like interphenomena and with wave - like interphenomena , which i shall now discuss .
note that these constructs are not incorrect .
they are logically sound , indeed .
the point to be emphasized in this paper is rather their artificiality .
my claim is that it is more illuminating to discard interphenomena altogether , and to consider quantum mechanical objects as entities which do not always occupy space . in this picture
the electron is said to be a concentrated lump of matter , which after process @xmath1 at _ a _ travels through space along a straight path from _ a _ to _
then it proceeds along another straight line from _ s _ to _ b _ , where process @xmath2 occurs .
this is called a _ normal interpretation _ , which is another important concept in reichenbach s analysis
: it means that the constructed interphenomena obey physical laws which do not sharply differ from the laws of the observed world of phenomena @xcite .
( otherwise it is called an _
anomalous interpretation_. see below . )
note that here the direction the electron will take after passing _ s _ can only be predicted in probability terms : this does not make the interphenomena anomalous , since analogous situations exist in the world of phenomena - for example , quantum phenomena too can only be predicted in probability terms .
this picture is less trivial , and more relevant to the present discussion .
we say that the electron is a wave - like disturbance of a medium , initiated by phenomenon @xmath1 at _ a _ this is not an anomaly : the medium has been invented for the support of interphenomena , but its ability to oscillate is normal .
later the wave collapses in order to wholly pass through slit _ s _ , which is an anomaly : no such instantaneous collapse exists for real waves .
then the wave propagates from _ s _ to the screen , and finally collapses again at _ b _ for the production of phenomenon @xmath2 .
so the wave - like interpretation is anomalous .
but this does not make it wrong . in reichenbach
s theory , it is as true as the particle - like picture .
such anomalies are the price one pays for demanding filling the gaps between real events .
this picture says that between phenomena @xmath1 and @xmath2 nothing happens to the electron .
needless to say , it has been assumed all along that there is vacuum between the elements sketched in the figure .
the @xmath0-function and the rules of quantum theory let us calculate the probability of event @xmath2 after @xmath1 has happened , but no spatial picture of inter - events is allowed .
the electron has position _ a _ when @xmath1 occurs and position _ b _ when @xmath2 occurs . in between it
does not have either place or shape , it has no spatiality .
it is left as an exercise for the reader to make similar discussion for what is known in quantum mechanics as _ tunneling _ through a potential barrier @xcite . especially noteworthy
is the fact that its particle - like interphenomena have the anomaly of energy non - conservation @xcite .
reichenbach s analysis of the wave - particle dualism of quantum mechanics has the great merit of showing that these interpretations are logical constructs , neither unique nor unambiguous for a given physical situation .
while in a lower level - i. e. , in the case of unobserved facts or objects - similar constructs are very convenient and adequate , because unambiguous and normal , they are rather restrictive of the imagination in the case of inter - events .
i think it is more illuminating to discard the concept of dualism , and to recognize that there is no need to project all ideas of physical theory into space .
space , or more precisely position in space , is in a sense produced by physical phenomena .
the macrophysical , persistent continuity of spatial forms is an illusion analogous to many others , for example , the illusion of spatial continuity of macroscopic matter . the proper language of physics is mathematical , and in many cases mathematical statements do not allow for mapping on intuitive space .
thus an electron is a linguistic entity logically similar to common sense objects [ 4 ] , but having a much poorer spatial character .
of course , the language of the wave and particle pictures has heuristic value in problem solving and experiment setting .
it is also a practical jargon .
we may even say that these pictures have some reality .
however , the actual measurements and calculations are expressed in mathematical language , and physicists move so safely in this domain that they may not see how vague are qualitative statements made in terms of the dual pictures .
i can not propose that the idea of dualism be banned from physics .
but i do suggest that students pay more attention to the unrealism of considering an electron as sometimes a particle and sometimes a wave .
the electron is a mathematically defined thing , which has the physical attributes of form and position only intermittently : namely , when it participates in the production of a phenomenon .
* acknowledgements*the author wants to express his thanks to professor p. leal ferreira , of this institute , for his critical reading of the manuscript , and to professor c. e. bures , of the california institute of technology , for his encouragement .
this work was partly supported by finep . | textbooks on quantum mechanics wave - particle dualism extreme aspects processes described in particle wave picture some texts like schiff s @xcite present strictly correct succinct quantum mechanical viewpoint mathematical formalism developed occasionally qualitative aspects theory touched student qualitative quantitative fundamentals quantum theory into coherent visualizable picture atomic world present paper analysis criticism of concept dualism purpose expose artificiality non - uniqueness of wave particle pictures best analysis dualism h. reichenbach philosophic foundations of quantum mechanics guided by reichenbach s ideas in study wave - particle dualism qualitative picture of microscopic reality advocated consistent with abstract interpretation of quantum mechanics interpretation assigns physical reality to observable facts only key abandon thinking of micro - objects deal with electrons for as permanent spatiality properties question where is electron when not observable ? meaningless place when in observable phenomenon meaninglessness not consequence of unobservability matter of adequate definition in section 2 reichenbach s ideas of phenomenon interphenomenon introduced basis for discussion wave - particle dualism in sec alternative language presented pre - empties micro - objects electrons photons of classical spatial attributes temporal continuity position shape trajectory propagation no need for such attributes in micro - objects next sec . 4 ) both viewpoints sections 2 and 3 applied to physical example physics progresses by increasing degrees abstraction concluding remarks in last section reichenbach p. 21 defines phenomena _ as occurrences easily inferable from macroscopic data considered observable phenomena determinate same as unobserved objects of classical physics physicists not troubled by reality of observable - but - unobserved events reality unobserved trees feynman nature not know what you looking at behaves data or not concept of phenomenon supposes berkeley s problem solved satisfactory solution by a. j. ayer _ interphenomena _ are occurrences constructed in interpolation within world of phenomena 21 logical constructs similar to phenomena usable to fill gaps in world phenomena appearance of continuity in time this purely mental process this process i intend to dispose of belowreichenbach says distinction phenomena and interphenomena is quantum mechanical analogue of distinction between observed and unobserved things ( p. 21 @xcite ) analogy misleading relation between phenomena interphenomena not simple as observed macro - objects question of interphenomena berkeley s problem of second degree formerly objects out of sense - experiences here define entities related to experience through end points initial final events particle and wave pictures in quantum theory represent different choices of interphenomena gap between two consecutive electron appearances filled by imagined trajectory traveled by electron with wave - like interphenomena gap filled by imagining first phenomenon as disturbance medium wave - like propagation quick coalescing into point - like disturbance illustrated by example in next section dualistic view countered with rational picture no need for interphenomena beween two consecutive events nothing happens to electron during interval electron is nowhere out of space indefinite position in neighborhood events contrarily to micro - objects like electrons do not constantly occupy place only when they appearelectrons have permanent potentiality to interact expressed as charge appear in space no permanent spatiality qualitative content of state function @xmath0 gives probability amplitude not wave amplitude see section iv.3 in @xcite spatiality limited discontinuous in time clearer next section object picture disadvantage more abstract than dual pictures abstract relative term reflection classical concept trajectory abstract less than quantum mechanical concepts progress in physics reached by increasing degrees abstraction abandonment of material ether as medium of electromagnetic waves natural concrete ideas limited ordinary sense experience illustrate problem through idealized experiment of diffraction of electron by slit in diaphragm @xcite figure 1 experimental situation _ a _ electron source _ s _ slit in diaphragm _ b _ point electron absorbed source weak one electron emitted at time emission at _ a _ absorption at _ b _ are point - like events phenomena call process @xmath1 and process @xmath2 space and time gap between phenomena filled by with particle - wave - like interphenomena constructs not incorrect logically soundemphasized paper is artificiality illuminating to discard interphenomena consider quantum mechanical objects not always occupy space picture electron concentrated lump matter after process @xmath1 travels through space straight from a to proceeds from _ s _ to _ b _ process @xmath2 occurs called _ normal interpretation _ important in s analysis constructed interphenomena obey physical laws differ from observed phenomena @xcite otherwise called _ anomalous interpretation_ direction electron after passing _ s _ predicted in probability terms interphenomena anomalous analogous situations exist quantum phenomena probability picture less trivial relevant to discussion electron is wave - like disturbance of medium initiated by phenomenon @xmath1 at a not anomaly medium invented for interphenomena oscillate normal wave collapses pass through slit _ s _ anomaly wave propagates from _ s _ to screen collapses at _ b _ for production phenomenon @xmath2 wave - like interpretation is anomalous not make it wrong in reichenbach s theory as true as particle - like pictureanomalies are price for filling gaps between events picture says between phenomena @xmath1 and @xmath2 nothing happens to electron assumed vacuum between elements @xmath0-function and rules of quantum theory calculate probability of event @xmath2 after @xmath1 no spatial picture of inter - events allowed electron has position a _ when @xmath1 and b _ when @xmath2 place or shape no spatiality exercise for reader to discussion for quantum mechanics tunneling through potential barrier @xcite particle - like interphenomena have anomaly of energy non - conservation reichenbach s analysis of wave - particle dualism of quantum mechanics interpretations are logical constructs neither unique nor unambiguous for physical situation in lower level similar constructs convenient restrictive of imagination in inter - events to discard concept of dualism no need to project all ideas of physical theory into space space position produced by physical phenomena macrophysical continuity of spatial forms is illusion of matter language of physics is mathematical statements allow for mapping on intuitive spaceelectron is linguistic entity similar to objects poorer spatial character language of wave particle pictures has heuristic value in problem solving experiment setting practical jargon pictures have reality actual measurements calculations expressed in mathematical language physicists safely not see vague qualitative statements dual pictures propose idea dualism banned from physics suggest students pay attention to unrealism of considering electron as sometimes particle sometimes wave electron mathematically defined physical attributes of form position intermittently when participates in production of phenomenon author thanks to professor p. leal ferreira for critical reading professor c. e. bures california institute of technology for encouragement work supported by finep | 0.443059 | 2,747 | 1,310 | , Saving $0.1 in GPT-4. |
55,357 | perturbative calculations in quantum field theory often require the regularization of infrared divergences . in quantum electrodynamics ,
such a regularization can for example be accomplished by a photon mass introduced via the stueckelberg method .
the present work extends this method to the qed limit of the lorentz- and cpt - violating standard - model extension . | massive vector particles are of relevance for various subfields of physics including the weak interaction , test models for photon - mass searches , and the regularization of certain infrared divergences in quantum - field calculations . in this latter context
, the mass term should be introduced such that the symmetries of the original model remain unspoiled . for a u(1 ) lorentz - invariant gauge theory , this can be accomplished with the stueckelberg method.@xcite in recent years , lorentz - violating quantum field theories have become a focus of theoretical@xcite and experimental@xcite inquiry .
such theories also require the regularization of infrared divergences in some circumstances , and the question arises , as to whether the stueckelberg method can be adapted to such lorentz - breaking u(1 ) gauge theories . as in the lorentz - invariant case ,
the adapted stueckelberg method should be compatible with all symmetries of the theory to be regulated , so that the usual stueckelberg approach can be employed for the lorentz - violating extension of conventional qed .
however , the lorentz violation in this model allows a broader range of compatible stueckelberg terms that may be introduced , and this freedom can then be used to streamline calculations in lorentz - breaking qed .
the present work provides one possible class of extensions of the usual stueckelberg procedure to lorentz - violating qed .
section [ lagr ] gives a brief overview of the procedure at the lagrangian level .
some implications of the lorentz - breaking stueckelberg lagrangian are discussed in sec .
we begin with the usual free - photon lagrangian in the minimal standard - model extension coupled to a conserved source @xmath0 : @xmath1 here , lorentz and cpt breakdown is controlled by the spacetime - constant backgrounds @xmath2 and @xmath3 .
the absence of lorentz symmetry in the above lagrangian ( [ lagrangian ] ) allows us to drop the requirement of lorentz invariance for the mass - type term @xmath4 to be introduced for the photon .
however , the inclusion of arbitrary lorentz violation into @xmath4 may be problematic : consider the case in which @xmath2 and @xmath3 are such that a subgroup of the lorentz group remains unbroken . a regulator violating
this residual symmetry may be undesirable , so that the breakdown of the remaining invariant subgroup may have to be excluded from @xmath4 . for our present purposes , however
, we consider arbitrary lorentz violation in @xmath4 .
this yields more general results also relevant for purposes other than infrared regularization .
more specifically , we implement the stueckelberg procedure by introducing a scalar field @xmath5 as follows:@xcite @xmath6 where @xmath7 can later be identified with the photon mass and @xmath8 with @xmath9 a lorentz - violating operator that may contain derivatives but is otherwise spacetime constant . the small number of lorentz - symmetric pieces still contained in an arbitrary @xmath9 can be removed if necessary . as in the conventional stueckelberg case ,
the key feature of @xmath10 is its invariance ( up to total derivatives ) under a local gauge transformation @xmath11 with the addition of @xmath12-type gauge fixing @xmath13 and a faddeev
popov contribution @xmath14 , the model lagragian @xmath15 becomes@xcite @xmath16 note that the scalar @xmath5 and the ghosts @xmath17 and @xmath18 are now decoupled and can be integrated out .
we can therefore disregard these fields in what follows .
the lagrangian ( [ eq : totall ] ) yields the following equation of motion for the photon : @xmath19f_{\alpha\beta}&&\nonumber\\ { } + \left[m^2\hat{\eta}^{\mu\nu } + \frac{1}{\xi}\hat{\eta}^{\mu\alpha}\hat{\eta}^{\nu\beta}\partial_\alpha\partial_\beta\right]a_\mu & = & j^\nu \,.\end{aligned}\ ] ] taking the 4-divergence of this equation yields @xmath20 note that @xmath21 projects out one degree of freedom contained in @xmath22 .
equation ( [ longitudinal ] ) establishes that this degree of freedom is not excited by the source @xmath23 , so @xmath21 is an auxiliary mode .
this is consistent with the expectation of three physical degrees of freedom for a massive vector field . a plane - wave ansatz in eq .
( [ eom1 ] ) yields the model s dispersion relation ; it has the following structure:@xcite @xmath24 the first factor does not contain physical modes .
the @xmath25 piece corresponds to the auxiliary mode governed by eq .
( [ longitudinal ] ) .
the factor @xmath26 is associated with the physical degrees of freedom and has the structure @xmath27 here , the coefficients @xmath28 are coordinate scalars containing the lorentz - breaking tensors @xmath2 and @xmath3 .
they vanish in the limit @xmath29 .
the explicit expressions for the @xmath28 can be found in the literature.@xcite note that the physical dispersion relation ( [ physdr ] ) is consistent with the expectation of three conventional massive modes perturbed by small lorentz violation .
the exact expression for the corresponding propagator is somewhat unwieldy.@xcite but for most applications only leading - order lorentz - violating effects need to be taken into account .
the lorentz - breaking contributions can then be incorporated via this propagator insertion ( see also fig . [ fig]):@xcite @xmath30
this work has been supported in part by the indiana university center for spacetime symmetries , by universidad andrs bello under grant no .
unab di-27 - 11/r , as well as by the portuguese fundao para a cincia e a tecnologia . e.c.g .
stueckelberg , helv .
acta * 11 * , 225 ( 1938 ) ; helv .
acta * 11 * , 299 ( 1938 ) .
see , e.g. , d. colladay and v.a .
kosteleck , phys . rev .
d * 58 * , 116002 ( 1998 ) ; v.a .
kosteleck , phys .
d * 69 * , 105009 ( 2004 ) ; v.a .
kosteleck and r. lehnert , phys .
d * 63 * , 065008 ( 2001 ) ; a.j .
hariton and r. lehnert , phys .
lett . a * 367 * , 11 ( 2007 ) ; r. lehnert , phys . rev .
d * 74 * , 125001 ( 2006 ) ; m. cambiaso _ et al .
_ , phys .
d * 90 * , 065003 ( 2014 ) .
see , e.g. , v.a .
kosteleck and n. russell , arxiv:0801.0287v7 ; m.a .
hohensee _ et al .
_ , phys .
* 102 * , 170402 ( 2009 ) ; phys .
d * 80 * , 036010 ( 2009 ) ; j .-
bocquet _ et al . _
[ graal collaboration ] , phys .
lett . * 104 * , 241601 ( 2010 ) ; j.s .
daz _ et al .
_ , phys .
d * 88 * , 071902 ( 2013 ) .
m. cambiaso _ et al .
_ , phys .
d * 85 * , 085023 ( 2012 ) . | vector particles for physics weak interaction test models for photon - mass searches regularization infrared divergences in quantum - field calculations mass term symmetries original model unspoiled u(1 ) lorentz - invariant gauge theory with stueckelberg method lorentz - violating quantum field theories focus of inquiry require regularization infrared divergences stueckelberg method to lorentz - breaking gauge theories adapted stueckelberg method compatible with symmetries stueckelberg approach for lorentz - violating lorentz violation allows broader range of compatible stueckelberg terms streamline calculations in lorentz - breaking qed present work provides extensions stueckelberg procedure to lorentz - violating qed section overview procedure lagrangian level implications of lorentz - breaking stueckelberg lagrangian discussed in with usual free - photon lagrangian in minimal standard - model extension conserved source lorentz and cpt breakdown controlled by spacetime - constant backgrounds @xmath2 and @xmath3absence lorentz symmetry in lagrangian requirement lorentz invariance for mass - type term @xmath4 photon arbitrary lorentz violation @xmath4 problematic @xmath2 @xmath3 subgroup lorentz group unbroken regulator violating symmetry undesirable breakdown invariant subgroup from @xmath4 consider arbitrary lorentz violation in @xmath4 yields general results infrared regularization implement stueckelberg procedure scalar field @xmath5@xcite @xmath6 @xmath7 identified photon mass @xmath8 @xmath9 lorentz - violating operator derivatives spacetime constant lorentz - symmetric pieces in arbitrary @xmath9 removed if necessary key feature @xmath10 invariance to total derivatives under local gauge transformation @xmath11 addition @xmath12-type gauge fixing @xmath13 faddeev popov contribution @xmath14 model lagragian @xmath15 becomes@xcite @xmath16 scalar @xmath5 ghosts @xmath17 @xmath18 decoupled integrated disregard fieldslagrangian yields equation motion for photon @xmath19f\alpha\beta}\nonumber\\ +\left[m^2\hat{\eta}\mu\nu }\frac{1}\xi}\alpha\beta\partial\mu j^\nu 4-divergence equation yields @xmath20 @xmath21 projects one degree freedom in @xmath22 freedom not excited by source @xmath23 @xmath21 auxiliary mode consistent with expectation three physical degrees freedom for massive vector field plane - wave ansatz in eq . yields model s dispersion relation structure @xmath24 first factor physical modes @xmath25 auxiliary mode factor @xmath26 associated with physical degrees freedom structure @xmath27 coefficients @xmath28 are coordinate scalars lorentz - breaking tensors @xmath2 and @xmath3 vanish in limit @xmath29 explicit expressions for @xmath28 in literature physical dispersion relation consistent with three massive modes perturbed by small lorentz violation exact expression propagator unwieldy most applications leading - order lorentz - violating effectslorentz - breaking contributions incorporated propagator insertion fig . work supported indiana university center spacetime symmetries universidad andrs bello grant unab di-27 - 11/r portuguese fundao cincia tecnologia stueckelberg helv . acta 11 * 225 ( 1938 ) 299 ( 1938 ) d colladay kosteleck . rev. d 58 * 116002 ( 1998 ) 69 * 105009 ( 2004 ) lehnert . d 63 * 065008 ( 2001 ) hariton r lehnert . 367 * 11 ( 2007 ) lehnert d 74 * 125001 ( 2006 ) cambiaso phys . d * 90 * 065003 ( 2014 ) kosteleck n. russell arxiv:0801.0287v7 hohensee phys . 102 * 170402 ( 2009 ) d 80 * 036010 ( 2009 ) bocquet phys . lett. 104 * 241601 ( 2010 ) daz phys . d * 88 * 071902 ( 2013 ) cambiasophys * 85 085023 2012 ) | 0.310993 | 1,755 | 918 | , Saving $0.1 in GPT-4. |
55,360 | a decade of observations in geodetic mode with the vlba network gave valuable results .
approximately 1/5 of all geodetic observations are recorded in vlba mode and processed at the socorro correlator .
ten years of observations allowed us to reliably measure slow intra - plate motion of the vlba stations located on the north - american plate .
it also helped us to achieve scientific objectives of geodetic programs : verify models of harmonic and anharmonic site position variations caused by various loadings , improve models of core - mantle boundary , investigate mantle rheology and solve other tasks . | method of vlbi first proposed by matveenko , kardashev and sholomitsky ( 1965 ) is used for two major applications : making images of compact radio sources and measuring antenna s motion with respect to extragalactic source .
fringe amplitudes and phases measured with precision @xmath0@xmath1 are used as primary observables for imaging ; group delays measured with relative precision @xmath2@xmath3 are used as observables for geodynamics applications .
group delay is sensitive to the projection of the baseline vector to the source direction .
having observed many sources in different directions and different baselines we are able to solve for source positions , site positions , earth orientation parameters ( eop ) and the rate of their changes in a least square solution .
the paper of overs , fanselow and jacobs ( 1998 ) provides an overview of geodetic vlbi technique .
there are several measures of how precise results are . formal uncertainties , which are calculated according to a law of propagation of errors of an individual observation , are the first measure . this estimate gives us an upper limit of precision , since an estimation model usually does not account correlations correctly .
another measure of precision is repeatability the root mean square deviation of the series of estimates of the targeted parameter with respect to the average . and
the last measure is accuracy : the measure of how far our estimate of the targeted parameter is from the true value .
table [ t : acc ] summarizes the current level of precision and accuracy of geodetic vlbi .
|l|l| r@r @|@ r@r | & & + ' '' '' & & & + ' '' '' & & vert & horiz & vert & horiz + precision & 0.3 nrad & 2.0 mm & 1.5 mm & 0.7 mm & 0.2 mm ' '' '' + repeatability & 0.5 nrad & & & 8 mm & 2 mm ' '' '' + accuracy & 0.7 nrad & 10 mm & 3 mm & 10 mm & 3 mm ' '' '' + [ t : acc ] |l| r@r @|@ r@r | ' '' '' + ' '' '' & & + & vert & horiz & vert & horiz ' '' '' + precision & 14 mm & 0.20.7 mm & 0.71.5 mm & 0.20.4 mm ' '' '' + accuracy & 25 mm & 0.41.2 mm & 1.02.0 mm & 0.30.6 mm ' '' '' +
although site positions and source coordinates are determined in processing vlbi observations , and they can be useful for particular applications , such as source mapping , determination of _ positions _ is not the objective per se .
the objective of observations is to measure _ changes _ in site positions and source coordinates .
investigation of changes in positions of the sites allows us to make judgments about the forces which caused the motion and the response the earth to external forcing .
when the whole vlbi dataset is analyzed , the following parameters are estimated : * secular site velocities .
causes : 1 ) plate tectonic ; 2 ) po - seismic slips ; 3 ) post - glacial isostatical adjustment ; 4 ) environmental changes 5 ) ? ?
* amplitudes of harmonic site position variations .
causes : 1 ) solid earth tides ; 2 ) ocean tides ; 3 ) ? ?
* anharmonic site position variations .
causes : 1 ) atmospheric pressure loading ; 2 ) hydrology loading ; 3 ) non - tidal ocean loading ; 4 ) ? ?
* variations in the earth orientation parameters .
causes : 1 ) torques exerted by the sun and planets ; 2 ) exchange of angular momentum with the atmosphere ; 3 ) exchange of angular momentum with the ocean ; 4 ) earthquakes ?
5 ) ? ?
one of the examples of achieving these objectives are results of analysis of the amplitudes of the earth s nutation fitted to vlbi data .
frequency , rad / sec frequency , rad / sec in the case if the earth responses to external torques as an absolutely rigid body , the complex transfer function would be ( 1,0 ) . the presence of the liquid core causes the resonance at the retrograde free core nutation frequency .
dissipation in the ocean and in the liquid body causes appearance of the non - zero imaginary part of the transfer function .
fitting empirical transfer function to a model allowed buffett , mathews , & herring , ( 2002 ) to estimate so - called basic earth paramenters which includes the core mantle boundary flattening and the magnitude of the magnetic field at the core mantle boundary ( 0.7mt ) .
vlbi observations allowed us to detect a secondary tidal effect : site displacements caused by water mass redistribution due to ocean tides .
mass loading results in crust deformation even thousands kilometers away from the mass load .
it was found that observed amplitudes are in a rather good agreement with models ( fig .
[ f : oclo ] .
comparison of the observed amplitudes with theoretical amplitudes at many vlbi sites , including vlba antennas , allowed petrov and ma ( 2003 ) to discriminate between competing models of ocean tides .
it is relatively easy to validate ocean tides models close to the shore where tide gauges are located , but it is a challenge to validate ocean tides models at open ocean . [
f : oclo ] similarly to mass loading , changes of atmospheric mass above the surface cause crust deformation due to pressure loading . unlike tides ,
changes in atmospheric pressure are not predictable . however , global numerical weather models based on past observations of the world meteorological network allow us to compute displacements due to atmospheric pressure loading with latency several days . examples of the atmospheric pressure loading are presented in fig .
[ f : aplo ] .
comparison of the admittance factors of the theoretical time series for all sites , including vlba sites , showed surprisingly well agreement between the theoretical model and observations , even for horizontal loading which does not exceed 1 mm for most of the sites .
this allowed petrov and boy ( 2003 ) to set the upper limit ( 4% ) of possible errors of loading love numbers which describe elastic properties of the earth .
[ f : aplo ]
vlba stations started to participate in geodetic program even before construction of the vlba network has been completed . by june 2003 the total number of observations used in analysis was .
the number of observations with at least one vlba site was .
thus , approximately 1/5 of geodetic observations are done at the vlba network .
.vlba station velocities with respect to the rigid north american plate .
unit : @xmath4 ( 1 mm / yr @xmath5 v / c ) .
last column indicates whether the station was used as defining for horizontal or vertical motion of the plate .
[ cols="<,>,>,^,>,>,>,^,>,>,>,^,>,<",options="header " , ] results of estimation of secular station velocity are presented in table [ t : sec_vel ] .
the station of mk - vlba belongs to the pacific tectonic plate , sc - vlba belongs to the caribbean plate .
they move rapidly ( in geophysical meaning of this word ) with respect to the north american plate where 8 other stations are located .
the station of ov - vlba is located rather close to the plate boundary and this station , and in a lesser degree br - vlba , are affected by the plate deformation near the edge of the plate .
peculiar horizontal velocity of pietown is not yet completely understood .
as gordon ( 2003 ) indicated , it may be related to antenna s tilting .
six other stations do not exhibit relative horizontal motion with respect to each other at the level exceeding 1 mm during entire history of the vlba .
exceptionally high precision of the earth orientation parameters makes 20 station experiments correlated at the vlba correlator especially attractive for investigation of free core nutation a poorly understood phenomenon of variable amplitude with the period around 430 days . unlike forced nutation caused by torques exerted by moon , sun and planets , the mechanism of excitation of the free core nutation
is not known .
currently it can not be predicted and should be monitored like polar motion . | vlbi proposed by matveenko kardashev sholomitsky ( 1965 ) used for applications images compact radio sources measuring antenna s motion extragalactic source fringe amplitudes phases measured @xmath0@xmath1 for imaging group delays measured @xmath2@xmath3 for geodynamics group delay sensitive to projection baseline vector to source direction observed many sources different directions baselines source positions site positions earth orientation parameters rate changes in least square solution paper of overs fanselow jacobs ( 1998 ) overview of geodetic vlbi technique several measures results formal uncertainties calculated according to law of propagation errors first measure upper limit of precision another measure repeatability mean deviation of estimates last measure accuracy estimate true value table [ t : acc ] summarizes current precision accuracy of geodetic vlbi|l|l| r@r @| | & & + ' '' vert & horiz + precision 0.3 nrad 2.0 mm 1.5 mm 0.7 mm 0.2 mm repeatability 0.5 nrad 8 mm & 2 mm + accuracy 0.7 nrad 10 mm 3 mm |l| r@r vert & horiz + precision 14 mm 0.20.7 mm 0.71.5 mm 0.20.4 mm accuracy 25 mm 0.41.2 mm 1.02.0 mm 0.30.6 mm site positions source coordinates determined in vlbi observations for not objective objective measure changes in site positions source coordinates investigation of changes in positions judgments about forces motion response earth to external forcing vlbi dataset analyzed parameters estimated : secular site velocities causes plate tectonic po - seismic slips post - glacial isostatical adjustment environmental changes 5 ) amplitudes of harmonic site position variations causes solid earth tides ocean tides anharmonic site position variations causes atmospheric pressure loading hydrology loading non - tidal ocean loading variations in earth orientation parameters causes torques by sun planets exchange angular momentum with atmosphere ocean earthquakes examples objectives analysis amplitudes earth s nutation to vlbi data frequency rad / sec earth responses to torques as rigid body complex transfer function ( 1,0 ) liquid core causes resonance at retrograde free core nutation frequency dissipation in ocean liquid body causes non - zero part of transfer function fitting transfer function to model allowed buffett mathews herring estimate basic earth paramenters core mantle boundary flattening magnitude magnetic field at core mantle boundary ( 0.7mt ) vlbi observations secondary tidal effect site displacements water mass redistribution due to ocean tides mass loading results in crust deformation observed amplitudes in agreement with models fig comparison of observed amplitudes with theoretical amplitudes at vlbi sites allowed petrov ma 2003 to discriminate between models ocean tides easy to validate ocean tides close to shore challenge to at open ocean changes atmospheric mass above surface cause crust deformation unlike changes atmospheric pressure not predictableglobal weather models compute displacements due to atmospheric pressure loading with latency days examples pressure loading in fig . comparison of admittance factors theoretical time series for all sites including vlba sites showed agreement between theoretical model observations even for horizontal loading not 1 mm allowed petrov and boy 2003 set upper limit ( 4% ) of errors loading vlba stations geodetic program before construction network completed by june 2003 total number observations was observations with at least one vlba site was approximately 1/5 of geodetic observations at vlba network.vlba station velocities rigid north american plate unit : @xmath4 ( 1 mm / yr @xmath5 v / c ) column indicates station for horizontal or vertical motion plate results estimation secular station velocity in table station mk - vlba pacific tectonic plate sc - vlba caribbean plate move rapidly north american plate where 8 other stations located station ov - vlba close to plate boundary br - vlba affected by plate deformation near edgehorizontal velocity of pietown not understood gordon 2003 may related to antenna tilting six stations exhibit relative horizontal motion exceeding 1 mm during history vlba high precision of earth orientation parameters makes 20 station experiments vlba correlator attractive for investigation free core nutation variable amplitude period around 430 days forced nutation mechanism excitation free core nutation not known should monitored like polar motion | 0.316059 | 1,843 | 899 | , Saving $0.1 in GPT-4. |
55,376 | we discuss sauron absorption line strength maps of a sample of 24 early - type spirals , mostly sa . from the lick indices h@xmath0 , mg@xmath1 and fe 5015 we derive ssp - ages and metallicities . by comparing the scaling relations of mg@xmath1 and h@xmath0 and central velocity dispersion with the same relation for the edge - on sample of falcn - barroso et al .
( 2002 ) we derive a picture in which the central regions of sa galaxies contain at least 2 components : one ( or more ) thin , disc - like component , often containing recent star formation , and another , elliptical - like component , consisting of old stars and rotating more slowly , dominating the light above the plane .
if one defines a bulge to be the component responsible for the light in excess of the outer exponential disc , then many sa - bulges are dominated by a thin , disc - like component containing recent star formation . | the measurement of absorption line strengths in combination with stellar population models has been used for many years to probe the ages and metallicities from integrated stellar populations of galaxies .
although there exist many studies of elliptical galaxies and s0 s , absorption line studies of spiral galaxies are lagging behind .
they are , however , important to understand the origin of galactic bulges .
are bulges old , elliptical - like objects in the middle of a large , spiral disk , which formed first ? or are they mass and light concentrations that formed from internal processes in the disk ?
recent reviews about this topic are given by kormendy & kennicutt ( 2004 ) and athanassoula ( 2005 ) .
stellar population studies of early - type spirals ( jablonka et al . 1996 , proctor & sansom 2002 ) see very little difference between the central stellar populations of spirals and s0 s .
both studies , however , contain few fainter galaxies with a central velocity dispersion smaller than 120 km / s .
those objects show stellar populations with a variety of properties ( moorthy & holtzman 2006 ) . at present
it looks as if both types of bulges ( elliptical - like and disk - like ) exist .
it is not clear , however , why this is , and how the bulge - type is related to its stellar populations . in this work we present high quality
, two - dimensional absorption line maps of a sample of 24 early - type spirals . here
some highlights of the work are given .
more details are found in peletier et al .
we have obtained integral field spectroscopy in a field of 33@xmath2 @xmath3 41@xmath2 , with a spatial sampling of 0.94@xmath2 @xmath3 0.94@xmath2 using sauron at the wht in la palma .
the observations are part of the sauron survey , described in de zeeuw et al .
the sample consists of 24 early - type spiral galaxies , for which the kinematics of gas and stars have been presented in falcn - barroso et al .
the spectra , which have a wavelength range from 4790 to 5300 , were fitted with the stellar population models of vazdekis ( 1999 ) , allowing us to separate the emission lines from the absorption line spectrum ( for details about this procedure see sarzi et al .
2006 and falcn - barroso et al .
2006 ) . from the cleaned spectra we obtained the line indices
h@xmath0 , mg@xmath1 and fe 5015 . in the way we described in kuntschner et al .
( 2006 ) we determined ages , metallicities and abundance ratios at every position , assuming that the stellar populations there could be represented by a single - age , single metallicity stellar population . in practise , we determined the ssp for which the line strengths fe 5015 , h@xmath0 and mg@xmath1 fitted best in the @xmath4 sense ( fig . 1 ) .
although we know that it is a great over - simplification to represent the stellar populations ( even locally ) of a galaxy by a ssp , in some , especially elliptical galaxies this approach gives a good first - order approximation . , fe 5015 and mg@xmath1 .
second row : reconstructed intensity , logarithmic age , metallicity ( log z / z@xmath5 ) and [ @xmath6/fe ] .
the reconstructed intensity is overlaid in white contours on the maps . ]
early - type galaxies show a tight mg@xmath7 @xmath8 relation ( terlevich et al .
deviations from the relations correlate well with parameters indicating the presence of young stellar populations . in falcn - barroso et al .
( 2002 ) we used the relation to show that the stellar populations in a sample of inclined early - type spirals are generally old . in fig .
2 we show the central mg@xmath1 and h@xmath0 indices of our sample as a function of the central velocity dispersion @xmath9 . in the figure
are shown the galaxies of this sample , together with the ellipticals and lenticulars of kuntschner et al .
( 2006 ) ( at r@xmath10/8 ) , and a number of literature samples of early - type spirals ( see caption ) .
the black line is a best fit to the ellipticals and s0 galaxies in the coma cluster of jrgensen et al .
the mg@xmath1 - @xmath8 relation of elliptical galaxies and s0 s acts as an upper envelope for the sa galaxies .
although some sa galaxy centre measurements lie close to the relation , a significant fraction of the galaxies falls below it .
the same effect is seen for the h@xmath0 - @xmath8 relation . using the argumentation of schweizer et al .
, the line of galaxies in coma would correspond to old stellar populations , while deviations would be caused by younger stars .
the fact that our sa bulges mostly lie below the mg@xmath1 - @xmath8 relation or above the h@xmath0 - @xmath8 relation would indicate that the centres of sa bulges generally are significantly younger than early - type galaxies in the coma cluster .
this result appears to contradict the tight mg@xmath7
@xmath8 relation for bulges found by fb02 and also the relation by jablonka et al .
it confirms , however , the results of prugniel et al .
( 2001 ) , who find several early - type spiral galaxies lying considerably below the mg@xmath7 @xmath8 relation .
notice that there are several s0 galaxies that are far away from the relation defined by elliptical galaxies , in the same location as the spirals with the lowest mg@xmath1 values . .
the black line is the least - squares fit to the ellipticals and s0 galaxies in coma of jrgensen et al .
( 1996 ) . as a comparison
we also show a few literature samples in black : the highly - inclined bulges of fb02 ( filled triangles ) , the bulges of bender et al .
( 1993 , open triangles ) , and bulges of jablonka et al .
( 1996 , crosses ) . ]
in the region of interest ( @xmath8 @xmath11 120 km / s ) the galaxies of fb02 generally have higher mg@xmath1 than the galaxies of this sample .
why this difference ?
the only important difference between the two samples is the inclination distribution .
if sa galaxies would contain young stellar populations that would be situated in the plane , we would see them in the sauron sample . in fb02 , however , where we looked at 5@xmath2 from the center on the minor axis , we would not have seen them , if the young stellar populations were limited to the very central regions . since for these sa galaxies the light in the central regions is completely dominated by the bulge , it seems that bulges consist of 2 , maybe more components : one which is old , elliptical - like , and slowly rotating , and another more flattened , disk - like component containing often young stellar populations .
support to this idea is given by the fact that the sauron maps show that the young stellar populations can often be found in circumnuclear rings , features with a small vertical extent .
another supporting argument is the fact that many of the galaxies have local central velocity dispersion minima ( falcn - barroso et al .
2006 , see also ganda et al .
2006 , emsellem et al .
these are most likely caused by central discs , some of which contain young stellar populations . according to the current literature
there are several kinds of bulges .
bulges that photometrically ( r@xmath12 surface brightness law ) and kinematically ( still relatively high @xmath13 , emsellem et al .
2007 ) resemble elliptical galaxies are often called classical bulges ( kormendy & kennicutt 2004 ) .
a bulge consisting only of the fast - rotating component is called a pseudo - bulge in this reference .
athanassoula ( 2005 ) claims that there are three types of bulges : the classical bulges , which form by collapse or merging , disc - like bulges , which result from the inflow of ( mainly ) gas to the centre - most parts , and subsequent star formation , and boxy and peanut bulges , which are seen in near - to - edge - on galaxies and which are in fact just a part of the bar seen edge - on , and therefore not part of the bulge in the definition of this paper . here
we add another piece of the puzzle . from the stellar population distribution , by comparing a sample uniformly distributed in inclination with a sample biased towards high inclination we infer that galactic bulges have more than one physical component : generally they have a slowly - rotating , elliptical - like component , and one or more fast - rotating components in the plane of the galaxy .
this picture also nicely explains the fact that bulge populations in general are very similar to those in the disc ( e.g. peletier & balcells 1996 ) . | measurement absorption line strengths with stellar population models used ages metallicities populations studies elliptical galaxies absorption line studies spiral galaxies lagging behind important origin galactic bulges are bulges old elliptical objects spiral disk or mass light concentrations internal processes ? reviews by kormendy & kennicutt 2004 athanassoula 2005 stellar population studies early - type spirals proctor sansom 2002 see little difference between central stellar populations s0 studies contain few fainter galaxies with central velocity dispersion smaller than 120 km / s show stellar populations variety properties moorthy holtzman 2006 both types bulges elliptical disk - like exist why bulge - type to stellar populations quality two - absorption line maps of 24 early - type spirals highlights details in peletier et al obtained integral field spectroscopy in 33@xmath2 @xmath3 41@xmath2 spatial sampling of 0.94@xmath2 using sauron at la palma observations part of sauron survey de zeeuw et al .sample 24 early - type spiral galaxies kinematics gas presented in falcn - barroso et al spectra 4790 to 5300 fitted with stellar population models vazdekis ( 1999 emission lines from absorption line spectrum sarzi et al . 2006 falcn - barroso et al . 2006 cleaned spectra obtained line indices h@xmath0 mg@xmath1 fe 5015 determined ages metallicities abundance ratios at every position single - age single metallicity stellar population determined ssp line strengths fe 5015 h@xmath0 mg@xmath1 best in @xmath4 sense ( fig . 1 ) stellar populations ssp approximation fe 5015 mg@xmath1 reconstructed intensity logarithmic age metallicity [ @xmath6/fe reconstructed intensity overlaid in white contours on maps early - type galaxies show tight mg@xmath7 @xmath8 relation deviations correlate with young stellar populations falcn - barroso et al . 2002 relation stellar populations in inclined early - type spirals generally old fig .central mg@xmath1 h@xmath0 indices central velocity dispersion @xmath9 figure galaxies ellipticals lenticulars kuntschner et al . 2006 r@xmath10/8 literature samples early - type spirals black line best fit ellipticals s0 galaxies coma cluster jrgensen et al mg@xmath1 - @xmath8 relation upper envelope sa galaxies galaxy centre measurements close relation galaxies below same effect h@xmath0 - @xmath8 relation argumentation schweizer et al . line galaxies coma old stellar populations deviations caused younger stars sa bulges below mg@xmath1 - @xmath8 relation above h@xmath0 - @xmath8 relation centres younger than early - type galaxies tight mg@xmath7 @xmath8 relation for bulges fb02 relation jablonka et al confirms results prugniel et al . 2001 early - type spiral galaxies below mg@xmath7 @xmath8 relation several s0 galaxies far away from relation spirals lowest mg@xmath1 values black line least - squares fit ellipticals s0 galaxies coma jrgensen et al . 1996comparison show literature samples in black : highly - inclined bulges of fb02 bender et al . 1993 triangles jablonka et al . 1996 crosses region interest ( @xmath8 @xmath11 120 km / s ) galaxies of fb02 have higher mg@xmath1 than this sample difference inclination distribution if sa galaxies contain young stellar populations in plane see them in sauron sample in fb02 5@xmath2 them young stellar populations limited to central regions sa galaxies light in central regions dominated by bulge bulges consist of 2 components one old elliptical - like slowly rotating another flattened disk - like component containing young stellar populations support sauron maps show young stellar populations found in circumnuclear rings small vertical extent argument galaxies have local central velocity dispersion minima falcn - barroso et 2006 ganda et 2006 likely caused by central discs some contain young stellar populations current literature several kinds of bulges photometrically kinematically resemble elliptical galaxies called classical bulges ( kormendy & 2004 )bulge fast - rotating component pseudo - bulge athanassoula 2005 ) claims three types of bulges classical form by collapse merging disc - like bulges from inflow gas star formation boxy peanut bulges seen in near - to - edge - on galaxies part of bar edge - on not part of bulge definition stellar population distribution comparing sample uniformly distributed with high inclination galactic bulges have more than one physical component slowly - rotating elliptical - like component one or more fast - rotating components in plane galaxy picture explains bulge populations similar to disc ( peletier & balcells 1996 ) | 0.31639 | 2,015 | 1,034 | , Saving $0.1 in GPT-4. |
55,380 | we present a construction of an effective yang - mills action for qcd , from the expansion of the fermionic determinant in terms of powers of the chemical potential at high temperature for the case of massless quarks .
we analyze this expansion in the perturbative region and find that it gives extra spurious information .
we propose for the non - perturbative sector a simplified effective action which , in principle , contains only the relevant information . | there has been an increasing interest in the few last years in the _ sign problem _ or _ phase problem _ in qcd . for finite chemical potential ,
the fermion determinant matrix is non - positive definite , so it is not possible to perform monte - carlo simulations @xcite . the glasgow method @xcite and rewheighting techniques @xcite have made great advances in the description of phase transitions on the lattice , considering a set of parameters near the transition line .
however , the problem is still far from being solved . there is special interest in the region of high temperature and low chemical potential , since it corresponds to the high - energy heavy - ion collision scenario . in this regime
it is possible to expand the fermion determinant in powers of @xmath0 @xcite .
this kind of expansion is what we are interested in investigating .
as one performs this expansion , some questions arrise : what order is enough to cut the expansion ? which is its range of validity ? how to distinguish between real and complex terms ?
differently from what is done in the previously mentioned papers , we keep the expanded series as a functional of the gauge fields , giving rise to an effective non - local yang - mills action .
for simplicity , consider one - flavor massless quarks .
the qcd generating functional at finite chemical potential in euclidean space is defined as @xmath1},\ ] ] where @xmath2 are the gluon fields , also present in the covariant derivative @xmath3 , and @xmath4 is the yang - mills action .
we can expand the fermion determinant in powers of the chemical potential assuming that @xmath5 : @xmath6 where @xmath7 is the dressed fermion propagator , which can be expressed as a series in powers of the gauge field and the free fermion propagator .
the expansion , then , will contribute to additional terms in the yang - mills effective action @xmath8 where the vertices @xmath9 are series in powers of the chemical potential : @xmath10 , with @xmath11 . in this way we have a positive - definite fermion determinant and
the contribution from the chemical potential will be part of an effective yang - mills action .
the problem now is to find criteria to cut this infinite series in terms of the chemical potential and gauge fields .
to test the expansion , we study it firstly in the perturbative sector . setting @xmath12 ,
we find naturally the way to cut the gauge field series , which is given by the order of perturbative corrections . for one - loop corrections ,
we just need the effective action up to order @xmath13 , so @xmath14 we calculate , then , the pressure up to order @xmath13 . considering the expansion on chemical potential up to order @xmath15 ,
we surprisingly obtain the same result as in the usual pertubative qcd calculation with the chemical potential included in the quark propagator @xcite . however , the next terms in the chemical potential series are non - zero . in particular , the next vacuum contribution for the pressure is @xmath16 which is negligible only for @xmath17 .
more than establishing the range of validity , this result shows that this kind of expansion contains some spurious information , at least for the massless case , and in principle the only indication on where to cut the series is the dimension of the desired observable .
for the non - perturbative regime , we need another criterium to cut the series .
the most appropriate is the weinberg power counting @xcite , i.e. considering the external momentum as a small parameter , @xmath18 , and also assuming that the chemical potential and the expectation value of the fields are of the same order , @xmath19 .
this gives a low - energy theory where the effective lagrangian can be expanded in soft modes . in the case of yang - mills theories ,
this soft - mode expansions for high temperatures correspond to the hard thermal loop ( htl ) approximation @xcite .
the minimal action must be of order @xmath20 .
so , applying the power counting , the minimal effective action is @xmath21 = s_{{\mbox{\tiny{ym}}}}[g ] + s^{(0,2)}(\mu ) + s^{(0,4)}(\mu ) + s_{{\mbox{\tiny{lo}}}}^{(2,2)}[g,\mu]+ s_{{\mbox{\tiny{lo}}}}^{(3,1)}[g,\mu]~,\ ] ] where the indices @xmath22 denote powers of @xmath23 in the gauge field and @xmath24 in the chemical potential , and lo means leading order in the htl approximation . in this case
, the appearance of imaginary terms will happen for @xmath25 .
the whole series of gauge fields is gauge invariant in every order in the @xmath26 expansion , i.e. @xmath27 is gauge invariant for all @xmath24 . in the case of the minimal effective action , the contributions @xmath28 and @xmath29 are independently gauge invariant , and htl preserves this feature
. of course this effective action must be tested , but in principle it could provide rich information on the confinement - deconfinement transition line .
we thanks phillipe de forcrand for comments and suggestions .
we acknowledge financial support from capes , claf , cnpq , faperj and fujb - ufrj . | increasing interest in sign phase in qcd finite chemical potential fermion determinant matrix non - positive possible monte - carlo simulations glasgow method rewheighting techniques phase transitions lattice parameters transition line problem far from solved special interest in high temperature low chemical potential high - energy heavy - ion collision scenario possible to expand fermion determinant in powers @xmath0 @xcite expansion investigating questions order range of validity distinguish between real complex terms ? keep expanded series as functional of gauge fields effective non - local yang - mills action consider one - flavor massless quarks qcd generating functional finite chemical potential in euclidean space defined as @xmath1}, @xmath2 are gluon fields @xmath3 @xmath4 yang - mills action expand fermion determinant in powers chemical potential @xmath5 : @xmath6 @xmath7 is dressed fermion propagator series in powers of gauge field free fermion propagator expansion to additional terms in yang - mills effective action @xmath8 vertices @xmath9 are series in powers of chemical potential @xmath10 @xmath11positive - definite fermion determinant contribution chemical potential effective yang - mills action problem criteria to cut infinite series chemical potential gauge fields test expansion study perturbative sector setting @xmath12 cut gauge field series by order perturbative corrections for one - loop corrections need effective action up to @xmath13 calculate pressure up to @xmath13 expansion on chemical potential up to @xmath15 same result pertubative qcd calculation chemical potential quark propagator @xcite next terms in chemical potential series non - zero next vacuum contribution for pressure @xmath16 negligible for @xmath17 shows expansion contains spurious information only indication cut series dimension desired observable non - perturbative regime need another criterium to cut series weinberg power counting @xcite considering external momentum small parameter @xmath18 chemical potential expectation value fields same order @xmath19 low - energy theory effective lagrangian expanded in soft modes yang mills soft - mode expansions for high temperatures correspond to hard thermal loop ) approximation @xcite minimal action order @xmath20 applying power counting minimal effective action @xmath21 = s\mbox[g +(0,2)(\mu ) +(0,4)}( ) +{(2,2)}[g{(3,1)}[g indices @xmath22 powers @xmath23 @xmath24 chemical potential lo leading order htl approximation imaginary terms for @xmath25 gauge fields invariant @xmath26 expansion @xmath27 invariant for @xmath24 minimal effective action contributions @xmath28 @xmath29 gauge invariant htl preserves feature effective action tested provide rich information confinement - deconfinement transition line phillipe de forcrand comments suggestions acknowledge financial support from capes claf cnpq faperj fujb - ufrj | 0.186716 | 1,168 | 592 | , Saving $0.0 in GPT-4. |
55,381 | les cw - complexes globulaires et les flots sont deux modlisations gomtriques des automates parallles qui permettent de formaliser la notion de dihomotopie .
la dihomotopie est une relation dquivalence sur les automates parallles qui prserve des proprits informatiques comme la prsence ou non de deadlock .
on construit un plongement des cw - complexes globulaires dans les flots et on dmontre que deux cw - complexes globulaires sont dihomotopes si et seulement si les flots associs sont dihomotopes .
globular cw - complexes and flows are both geometric models of concurrent processes which allow to model in a precise way the notion of dihomotopy .
dihomotopy is an equivalence relation which preserves computer - scientific properties like the presence or not of deadlock .
one constructs an embedding from globular cw - complexes to flows and one proves that two globular cw - complexes are dihomotopic if and only if the corresponding flows are dihomotopic .
= 1 | cette note est la premire de deux notes prsentant quelques rsultats de @xcite .
tous les espaces topologiques sont supposs faiblement spars et compactement engendrs , cest - - dire dans ce cas homomorphes la limite inductive de leurs sous - espaces compacts ( cf .
lappendice de @xcite pour un survol des proprits de ces espaces ) . on travaille ainsi dans une catgorie despaces topologiques ( note @xmath0 ) qui est non seulement complte et cocomplte mais en plus cartsiennement ferme @xcite . en dautres termes , le foncteur @xmath1 a un adjoint droite @xmath2 .
dans la suite , pour @xmath3 , @xmath4 est le disque ferm de dimension @xmath5 et @xmath6 est le bord de @xmath4 , savoir la sphre de dimension @xmath7 . en particulier la sphre de dimension @xmath8 est la paire @xmath9 .
si @xmath10 est un espace topologique non vide , on note @xmath11 lespace topologique obtenu en partant de @xmath12 $ ] et en quotientant par les relations @xmath13 et @xmath14 pour tout @xmath15 .
cet espace est muni de lordre partiel suivant : @xmath16 si et seulement si @xmath17 et @xmath18 .
la classe dquivalence de @xmath19 ( resp .
@xmath20 ) est infrieure ( resp .
suprieure ) tous les points de @xmath11 .
le _ @xmath8-squelette _
@xmath21 de @xmath11 sera la paire constitue de la classe de @xmath19 et de celle de @xmath20 . on pose
@xmath22 qui a donc pour @xmath8-squelette @xmath23 .
un _ cw - complexe globulaire _
@xmath24 est , par dfinition , obtenu comme limite inductive @xmath25 despaces topologiques de la faon suivante . on part dun
espace discret @xmath26 appel le _ @xmath8-squelette _ et on commence par lui attacher des @xmath27 pour obtenir un premier espace topologique @xmath28 .
puis on suppose @xmath29 construit pour @xmath30 . on obtient alors
@xmath31 en attachant @xmath29 des @xmath32 le long de @xmath33 .
tout cela de telle faon que les attachements des @xmath27 et les morphismes dattachement @xmath34 soient des morphismes de cw - complexes globulaires ( cf .
ci - dessous ) .
width 5 cm exglob.png pour un espace topologique @xmath24,width=188 ] un _
chemin dexcution _
@xmath35 est une application continue localement croissante pour lordre partiel dfini sur chaque globe de @xmath24 et telle que @xmath36 .
lensemble de ces @xmath37 _ non - constants _ est not @xmath38 .
un morphisme de cw - complexes globulaires @xmath39 pour @xmath24 et @xmath40 quelconques est une application continue telle que @xmath41 et telle que @xmath42 induise une application de @xmath38 dans @xmath43 .
lhypothse de non - contractibilit des chemins dexcution par @xmath44 , qui peut paratre trange en premire lecture , est en fait essentielle pour pouvoir faire certaines constructions homologiques : cf .
par exemple le chapitre
`` why non - contracting maps ? '' de @xcite ou bien @xcite .
@xcite deux morphismes de cw - complexes globulaires @xmath44 et @xmath45 de @xmath24 dans @xmath40 sont s - homotopes sil existe une application continue @xmath46\rightarrow y$ ] telle que 1 .
@xmath47 est un morphisme de cw - complexes globulaires de @xmath24 dans @xmath40 pour tout @xmath48 $ ] 2 .
@xmath49 et @xmath50 on crit @xmath51 .
en particulier deux morphismes de cw - complexes globulaires s - homotopes concident sur le @xmath8-squelette . de l on dfinit la notion de cw - complexes globulaires s - homotopes : @xcite deux cw - complexes globulaires @xmath24 et @xmath40 sont s - homotopes sil existe un morphisme de cw - complexes globulaires @xmath39 et un morphisme de cw - complexes globulaires @xmath52 telle que @xmath53 et @xmath54 .
@xcite une t - homotopie @xmath39 est un morphisme de cw - complexes globulaires qui induit un homomorphisme entre les espaces topologiques sous - jacents . on dit
alors que @xmath24 et @xmath40 sont t - homotopes . on peut alors
dmontrer le @xcite la localisation @xmath55 de la catgorie @xmath56 des cw - complexes globulaires par rapport aux s - homotopies et aux t - homotopies existe .
la forme de globe suffit pour modliser les automates parallles @xcite . de plus la s - homotopie et la t - homotopie ne modifient pas des proprits informatiques comme la prsence ou non de deadlock @xcite .
la catgorie @xmath55 est donc un cadre possible pour ltude des automates parallles homotopie prs .
@xcite un flot @xmath24 consiste en la donne dun ensemble @xmath26 appel @xmath8-squelette , dun espace topologique @xmath57 appel espace des chemins dexcution ( non - constants ) , de deux applications continues @xmath58 et @xmath59 ( @xmath26 tant muni de la topologie discrte ) et dune application continue @xmath60 satisfaisant les axiomes @xmath61 , @xmath62 et enfin @xmath63 pour tout @xmath64 .
un morphisme de flots @xmath44 de @xmath24 vers @xmath40 est une application continue de @xmath65 vers @xmath66 tels que @xmath41 , @xmath67 , @xmath68 , @xmath69 , @xmath70 .
la catgorie correspondante est note @xmath71 .
si @xmath10 est un espace topologique , le flot @xmath72 est dfini comme suit : @xmath73 , @xmath74 , et enfin @xmath75 et @xmath76 ( il ny a pas de chemins dexcution composables ) .
@xcite deux morphismes de flots @xmath44 et @xmath45 de @xmath24 dans @xmath40 sont s - homotopes sil existe une application continue @xmath46\rightarrow y$ ] telle que 1 .
@xmath47 est un morphisme de flots de @xmath24 dans @xmath40 pour tout @xmath48 $ ] 2 .
@xmath49 et @xmath50 on crit @xmath51 .
on remarque que deux morphismes de flots s - homotopes concident sur le @xmath8-squelette .
@xcite deux flots @xmath24 et @xmath40 sont s - homotopes sil existe un morphisme de flots @xmath39 et un morphisme de flots @xmath52 tels que @xmath53 et @xmath54 .
@xcite soit @xmath24 un flot et soit @xmath40 un sous - ensemble de @xmath26 .
la restriction @xmath77 de @xmath24 @xmath40 est le flot dfini par 1 .
le @xmath8-squelette de @xmath77 est @xmath40 2 . @xmath78 avec une
dfinition vidente des source , but et loi de composition .
si @xmath24 est un flot , lensemble @xmath79 ( resp .
@xmath80 ) est , dans la dfinition qui suit , lensemble des @xmath81 tels que @xmath82 ( resp .
@xmath83 ) quotient par la relation dquivalence identifiant @xmath37 avec @xmath84 ( resp .
@xmath85 et @xmath37 ) .
cest - - dire que @xmath79 ( resp .
@xmath80 ) est lensemble des classes dquivalence de chemins dexcution non - constants commenant ( resp .
terminant ) de la mme faon en @xmath86 .
@xcite un morphisme de flots @xmath39 est une t - dihomotopie quand 1 .
le morphisme de flots @xmath87 est un isomorphisme de flots 2 .
pour @xmath88 , @xmath89 et @xmath90 sont des singletons 3 .
pour tout @xmath91 , il existe @xmath92 tels que @xmath93 soit dans limage @xmath94 ( avec les conventions @xmath95 ) .
on dit alors que les flots @xmath24 et @xmath40 sont t - dihomotopes .
@xcite il existe un et un seul foncteur @xmath96 ( appel ralisation catgorique ) de @xmath56 dans @xmath71 tel que @xmath97 , @xmath98 , et @xmath99 o @xmath10 parcourt la dcomposition cellulaire du cw - complexe globulaire @xmath24 .
si @xmath24 est un espace discret , on pose naturellement @xmath100 .
si @xmath10 est un compact , on pose naturellement @xmath101 .
puis on dfinit le plongement sur les objets en posant @xmath102 o @xmath10 parcourt la dcomposition cellulaire de @xmath24 .
le plongement sur les morphismes est obtenu comme suit . on construit
dabord explicitement pour tout morphisme de cw - complexes globulaires @xmath103 le morphisme de flots @xmath104 .
puis on pose pour @xmath105 un morphisme de cw - complexes globulaires quelconque @xmath106 o @xmath10 parcourt la dcomposition cellulaire de @xmath24 . on a donc par construction la bijection @xmath107 .
[ sur ] @xcite soit @xmath108 ( resp .
@xmath109 ) lespace des morphismes de cw - complexes globulaires de @xmath24 vers @xmath110 ( resp .
de flots de @xmath111 vers @xmath112 ) muni de la kelleyfication de la topologie induite par celle de @xmath113 ( resp .
@xmath114 ) . alors @xmath96 induit une homotopie faible @xmath115,\linebreak[0]cat(u)).\ ] ] de plus il existe une application continue @xmath116 de @xmath117 dans @xmath108 telle que @xmath118
. considrons dabord le cas @xmath119 .
la construction de @xmath116 se fait alors explicitement et on dmontre que cest un inverse homotopique de @xmath120 .
cela est essentiellement d au fait que par un point donn de @xmath11 autre que les deux points du @xmath8-squelette , il ne passe quun et un seul chemin dexcution non - constant ( reparamtrisation prs ) .
on montre ensuite que les deux tours despaces donnant les deux limites projectives homotopiques sont des tours fibrantes , savoir que toutes les applications apparaissant dans ces tours sont des fibrations despaces topologiques .
cela est essentiellement d au fait que les inclusions @xmath123 et donc @xmath124 sont des cofibrations @xcite . on en
dduit donc , encore grce @xcite ou @xcite , les quivalences dhomotopie faible @xmath125 et @xmath126x^1_n\linebreak[0])\linebreak[0],\linebreak[0]cat(\linebreak[0]u\linebreak[0])\linebreak[0]\linebreak[0 ] ) { \cong}\holimproj_n { { { \rm{\mathbf{flow}}}}}(cat(x^1_n),cat(\linebreak[0]u\linebreak[0])\linebreak[0])$ ] .
le rsultat dcoule alors des homomorphismes @xmath127,\linebreak[0]u\linebreak[0])\linebreak[0]{\cong}\linebreak[0]{\varprojlim}_n { { { \rm{\mathbf{gltop}}}}}(\linebreak[0]x^1_n , u)\ ] ] et @xmath128,\linebreak[0]cat(u)).\ ] ] les espaces topologiques @xmath108 et @xmath109 sont faiblement homotopes , et ont donc les mmes composantes connexes par arc .
comme on travaille dans une catgorie despaces topologiques qui est cartsiennement ferme , une composante connexe par arc de @xmath108 ( resp .
@xmath130cat(\linebreak[0]x)\linebreak[0],\linebreak[0]cat(u))$ ] ) est une classe de s - homotopie de morphismes de cw - complexes globulaires ( resp
. de flots ) .
le rsultat dcoule alors de la surjectivit de lapplication @xmath120 . | note premire deux notes rsultats @xcite espaces topologiques faiblement spars compactement engendrs homomorphes limite inductive sous - espaces compacts lappendice de @xcite travaille catgorie despaces topologiques note @xmath0 complte cocomplte ferme @xcite foncteur @xmath1 adjoint droite @xmath2 @xmath3 @xmath4 disque ferm dimension @xmath5 @xmath6 bord de @xmath4 sphre dimension @xmath7 sphre dimension @xmath8 paire @xmath9 @xmath10 espace non vide note @xmath11 lespace topologique obtenu partant de @xmath12 quotientant relations @xmath13 @xmath14 @xmath15 espace lordre partiel @xmath16 @xmath17 @xmath18 classe dquivalence de @xmath19 @xmath20 points de @xmath11 @xmath8-squelette @xmath21 de @xmath11 paire de classe @xmath19 @xmath20 pose @xmath22 @xmath8-squelette @xmath23 complexe globulaire @xmath24 obtenu limite inductive @xmath25 despaces topologiquespart espace discret @xmath26 @xmath8-squelette @xmath27 premier espace topologique @xmath28 suppose @xmath29 construit @xmath30 obtient @xmath31 attachant @xmath29 @xmath32 @xmath33 attachements @xmath27 morphismes @xmath34 cw - complexes globulaires width 5 cm exglob.png espace topologique @xmath24=188 chemin dexcution @xmath35 application continue localement croissante partiel chaque globe @xmath24 @xmath36 @xmath37 non - constants @xmath38 cw - complexes globulaires @xmath39 @xmath24 @xmath40 application continue @xmath41 @xmath42 application @xmath38 @xmath43 non - contractibilit chemins dexcution @xmath44 essentielle constructions homologiques chapitre non - contracting maps ? @xcite morphismes cw - complexes globulaires @xmath44 @xmath45 @xmath24 @xmath40 s - homotopes application continue @xmath46 @xmath47 morphisme cw - complexes globulaires @xmath24 @xmath40 @xmath48 @xmath49 @xmath50 crit @xmath51deux morphismes - complexes globulaires s - homotopes concident sur @xmath8-squelette dfinit notion s - homotopes @xmath24 et @xmath40 sont s - homotopes @xmath39 @xmath52 t - homotopie @xmath39 est morphisme induit homomorphisme entre espaces topologiques sous jacents @xmath24 et @xmath40 sont t - homotopes localisation @xmath55 de catgorie @xmath56 des - complexes globulaires s - t - homotopies forme de globe suffit pour modliser automates parallles s - t - homotopie ne modifient pas proprits informatiques deadlock catgorie @xmath55 cadre pour ltude des automates parallles flot @xmath24 ensemble @xmath26 appel @xmath8-squelette espace topologique @xmath57 chemins non deux applications continues @xmath58 et @xmath59 application continue @xmath60 satisfaisant axiomes @xmath61 @xmath62 @xmath63 @xmath64morphisme flots @xmath44 de @xmath24 vers application continue @xmath65 vers @xmath66 @xmath41 catgorie correspondante @xmath71 @xmath10 espace topologique flot @xmath72 dfini @xmath73 @xmath75 chemins composables flots @xmath44 @xmath45 de @xmath24 dans @xmath40 s - homotopes application continue @xmath46 @xmath47 flots @xmath24 dans @xmath40 @xmath48 @xmath49 @xmath50 @xmath51 deux morphismes flots s - homotopes concident sur @xmath8-squelette flots @xmath24 @xmath40 s - homotopes @xmath39 @xmath52 @xmath53 @xmath54 @xmath24 flot @xmath40 sous - ensemble de @xmath26 restriction @xmath77 de @xmath24 @xmath40 flot dfini @xmath8-squelette de @xmath77 est @xmath40 @xmath78 dfinition source but loi composition @xmath24 flot lensemble @xmath79 @xmath80 lensemble @xmath81 @xmath82@xmath83 quotient relation dquivalence identifiant @xmath37 @xmath84 @xmath85 @xmath37 @xmath79 @xmath80 classes dquivalence chemins dexcution non - constants @xmath86 morphisme flots @xmath39 t - dihomotopie @xmath87 isomorphisme flots @xmath88 @xmath89 @xmath90 singletons @xmath91 @xmath92 @xmath93 @xmath94 conventions @xmath95 flots @xmath24 @xmath40 t - dihomotopes foncteur @xmath96 @xmath56 dans @xmath71 @xmath97 @xmath98 @xmath99 @xmath10 parcourt dcomposition cellulaire - complexe globulaire @xmath24 @xmath24 espace discret pose @xmath100 @xmath10 compact pose @xmath101 dfinit plongement objets posant @xmath102 @xmath10 parcourt dcomposition cellulaire @xmath24 plongement sur morphismes construit morphisme - complexes globulaires @xmath103 morphisme flots @xmath104 pose @xmath105 - complexes globulaires @xmath106 @xmath10 parcourt dcomposition cellulaire @xmath24construction bijection @xmath107 sur @xcite @xmath108 @xmath109 ) lespace morphismes cw - complexes globulaires de @xmath24 vers @xmath110 flots @xmath111 vers @xmath112 ) kelleyfication topologie induite @xmath113 @xmath114 ) @xmath96 induit homotopie faible @xmath115 application continue @xmath116 de @xmath117 dans @xmath108 @xmath118 cas @xmath119 construction de @xmath116 explicitement inverse homotopique de @xmath120 par point de @xmath11 @xmath8-squelette passe chemin dexcution non - constant reparamtrisation deux tours despaces limites projectives homotopiques tours fibrantes applications tours fibrations despaces topologiques inclusions @xmath123 et @xmath124 cofibrations @xcite dduit @xcite quivalences dhomotopie faible @xmath125 @xmath126x^1_n\linebreak[0]\linebreak[0]]cat\cong}\holimproj_n{\mathbf{flow(cat(x^1\linebreak[0]u[0] rsultat homomorphismes @xmath127,\linebreak[0]u\varprojlim}_n\mathbf{gltop\linebreak[0]x^1_n @xmath128,\linebreak[0]cat(u)). espaces topologiques @xmath108 @xmath109 faiblement homotopes composantes connexes par arc catgorie topologiques cartsiennement ferme composante connexe par arc @xmath108 @xmath130cat(\linebreak[0]x[0][0]cat(u))$ ] classe s - homotopie morphismes cw - complexes globulaires flots rsultat surjectivit lapplication @xmath120 . | 0.430575 | 3,167 | 1,771 | , Saving $0.1 in GPT-4. |
55,383 | doppler - based planet surveys have shown that , besides metallicity , the planet occurrence is also correlated with stellar mass , increasing from m to f - a spectral types .
however , it has recently been argued that the subgiants ( which represent a stars after they evolve off the main sequence ) may not be as massive as suggested initially , which would significantly change the correlation found . to start investigating this claim ,
we have studied the subgiant star hd 185351 , which has precisely measured physical properties based on asteroseismology and interferometry .
an independent spectroscopic differential analysis based on excitation and ionization balance of iron lines yielded the atmospheric parameters @xmath0 = 5035 @xmath1 29 k , @xmath2 g = 3.30 @xmath1 0.08 and [ fe / h ] = 0.10 @xmath1 0.04 . these were used in conjunction with the parsec stellar evolutionary tracks to infer a mass m = 1.77 @xmath1 0.04 m@xmath3 , which agrees well with the previous estimates .
lithium abundance was also estimated from spectral synthesis ( a(li ) = 0.77 @xmath1 0.07 ) and , together with @xmath0 and [ fe / h ] , allowed to determine a mass m = 1.64 @xmath1 0.06 m@xmath3 , which is independent of the star s parallax and surface gravity .
our new measurements of the stellar mass support the notion that hd185351 is a retired a star with a mass in excess of 1.6 m@xmath3 . | it is well - known that giant planet occurrence is strongly dependent on stellar metallicity for main - sequence and subgiant stars ( e.g. , @xcite ) .
recent works suggest that planet occurrence is also correlated with stellar mass , with the probability of a star hosting a jovian planet rising linearly over the mass range 0.2 to 2.0 m@xmath3 ( e.g. , @xcite ) . however , it has recently been argued that the subgiants / giants studied at the massive end off the planet - host mass distribution ( representing a stars after they evolve off the main sequence ) may not be as massive as initially suggested @xcite .
if the stellar masses have been previously underestimated , the nature and interpretation of the correlation between planet occurrence and stellar mass would need to be revised .
an accurate assessment of the stellar masses of subgiants and giants is thus required to fully understand this apparent correlation . in this work ,
we start to address this question by performing a spectroscopic analysis of the giant / subgiant star hd 185351 .
three spectra were obtained for hd 185351 using the high resolution echelle spectrometer ( hires ; @xcite ) on the keck i 10-meter telescope ( mauna kea , hawaii ) .
they have a nearly complete wavelength coverage from 3650 to 7950 , resolution r = 77,000 and s / n @xmath4 250 per resolution element at @xmath46700 .
we measured the atmospheric parameters ( @xmath0 , @xmath2 g , [ fe / h ] and @xmath5 ) in lte using an iterative method based on the excitation and ionization equilibria of fe i and fe ii lines , ares , moog @xcite and kurucz odfnew model atmospheres @xcite .
the results were : @xmath0 = 5035 @xmath1 29 k , @xmath2 g = 3.30 @xmath1 0.08 , [ fe / h ] = 0.10 @xmath1 0.04 and @xmath5 = 1.08 @xmath1 0.03 km s@xmath6 .
we also measured the lithium abundance in lte using a spectral synthesis technique , the atmospheric parameters above and moog .
our iterative procedure ( see @xcite for details ) tested different combinations for a(li ) and a single gaussian broadening fhwm@xmath7 .
a reduced @xmath8 minimization was used to find the best fit between the observed and synthetic spectra , which occurred for the following parameters : a(li ) = 0.77 @xmath1 0.07 and fhwm@xmath7 = 0.152 ( see figure [ synth_li ] ) .
we measured the physical parameters of hd 185351 using the atmospheric parameters @xmath0 and [ fe / h ] , the v magnitude ( 5.18 @xmath1 0.01 ) from the hipparcos catalog @xcite , the revised hipparcos parallax ( 24.49 @xmath1 0.22 mas ; @xcite ) , and the param web interface with the parsec evolutionary tracks @xcite .
the results were : m = 1.77 @xmath1 0.04 m@xmath3 , r = 4.85 @xmath1 0.11 r@xmath3 , @xmath2 g = 3.29 @xmath1 0.02 , and age = 1.79 @xmath1 0.09 gyr .
following the analysis of , we also obtained an independent estimate of the stellar mass from the analysis of the non - standard mixing history of hd 185351 : m = 1.64 @xmath1 0.06 m@xmath3 ( see figure [ models_lithium ] ) .
the independently measured masses are not only consistent , but also lie within the interval ( @xmath91.6 @xmath10 2.0 m@xmath3 ) defined by the extensive analysis of @xcite , which includes results from interferometry , spectroscopy , and asteroseismology ( see their figure 7 ) .
the mass of hd 185351 was determined using different methods and all estimated values are higher than 1.5 m@xmath3 .
these measurements are consistent with hd 185351 being the evolved counterpart of an a dwarf , which supports the results from @xcite .
we plan to extend the spectroscopic analysis described here to the entire sample of @xmath9300 evolved stars .
bressan , a. , marigo , p. , girardi , l. , et al .
2012 , , 427 , 127 castelli , f. , & kurucz , r. l. 2004 , arxiv : astro - ph/0405087 do nascimento , j. d. , jr . , castro , m. , melndez , j. , et al .
2009 , , 501 , 687 esa 1997 , the hipparcos and tycho catalogues , sp 1200 ( noordwijk : esa ) fischer , d. a. , & valenti , j. 2005 , , 622 , 1102 ghezzi , l. , cunha , k. , smith , v. v. , & de la reza , r. 2010 , , 724 , 154 hui - bon - hoa , a. 2008 , , 316 , 55 johnson , j. a. , aller , k. m. , howard , a. w. , & crepp , j. r. 2010 , , 122 , 905 johnson , j. a. , huber , d. , boyajian , t. , et al . 2014 , arxiv:1407.2329 lloyd , j. p. 2011
, , 739 , l49 lloyd , j. p. 2013
, , 774 , l2 pietrinferni , a. , cassisi , s. , salaris , m. , & castelli , f. 2004 , , 612 , 168 schlaufman , k. c. , & winn , j. n. 2013 , , 772 , 143 sneden , c. a. 1973 , ph.d . thesis , univ . texas , austin sousa , s. g. , santos , n. c. , israelian , g. , mayor , m. , & monteiro , m. j. p. f. g. 2007 , , 469 , 783 van leeuwen , f. 2007 , hipparcos : the new reduction of the raw data ( astrophysics and space science library , vol .
350 ; dordrecht : springer ) vogt , s. s. , allen , s. l. , bigelow , b. c. , et al .
1994 , proc .
spie , 2198 , 362 | giant planet occurrence on stellar metallicity for sequence subgiant stars works suggest planet occurrence with stellar mass probability star hosting jovian planet rising over mass range 0.2 to 2.0 m@xmath3 argued subgiants giants at massive end off planet - host mass distribution not as massive as suggested if stellar masses underestimated correlation accurate assessment of stellar masses of subgiants giants required correlation spectroscopic analysis of giant / subgiant star hd 185351 three spectra obtained for using high resolution echelle spectrometer keck i 10-meter telescope nearly complete wavelength coverage from 3650 to 7950 resolution r = 77,000 s / n @xmath4 250 per resolution element at @xmath46700 measured atmospheric parameters ( @xmath0 @xmath2 g fe / h ] @xmath5 ) using iterative method ionization equilibria results @xmath0 = 5035 @xmath1 29 k @xmath2 g = 3.30 @xmath1 0.08 [ fe / h ] = 0.@xmath1 0.04 @xmath5 1.08 0.03 measured lithium abundance lte spectral synthesis atmospheric parameters procedure tested combinations a(li ) single gaussian broadening fhwm@xmath7 reduced @xmath8 minimization fit observed synthetic spectra a(li ) = 0.77 @xmath1 0.07 fhwm@xmath7 = 0.152 measured physical parameters hd 185351 atmospheric parameters @xmath0 fe h v magnitude 5.18 @xmath1 0.01 hipparcos catalog revised hipparcos parallax 24.49 @xmath1 0.22 web interface parsec evolutionary tracks results m = 1.77 @xmath1 0.04 r = 4.85 0.11 @xmath2 g = 3.29 0.02 age = 1.79 @xmath1 0.09 independent estimate stellar mass non - standard mixing history hd 185351 m = 1.64 @xmath1 0.06 m@xmath3 measured masses consistent within interval @xmath91.6 @xmath10 2.0 m@xmath3 defined analysis @xcite interferometry spectroscopy asteroseismologymass hd 185351 determined methods estimated values higher than 1.5 m@xmath3 measurements consistent evolved counterpart dwarf results @xcite plan extend spectroscopic analysis sample @xmath9300 evolved stars bressan marigo girardi 2012 , 427 , 127 castelli kurucz 2004 arxiv astro - ph/0405087 nascimento castro melndez 2009 , 501 , 687 esa 1997 hipparcos tycho catalogues sp 1200 noordwijk esa ) fischer valenti 2005 622 , 1102 ghezzi cunha smith de la reza 2010 724 , 154 hui - bon - hoa 2008 316 , 55 johnson j aller howard crepp 2010 122 , 905 johnson huber boyajian 2014 arxiv:1407.2329 lloyd . 2011 739 , l49 2013 774 , l2 pietrinferni cassisi salaris castelli 2004 612 , 168 schlaufman winn j2013 772 , 143 sneden 1973 ph.d thesis univ. texas austin sousa santos israelian mayor monteiro 2007 469 , 783 van leeuwen 2007 hipparcos new reduction raw data astrophysics space science library vol . 350 dordrecht springer vogt s allen. bigelow b. c. 1994 proc spie 2198 , 362 | 0.264012 | 1,578 | 839 | , Saving $0.0 in GPT-4. |
55,391 | we report the updated results of direct @xmath0 violation measurements from belle and babar for the following modes : @xmath1 , @xmath2 and @xmath3 , where @xmath4 stands for a charged kaon or pion .
the data used in these studies are up to 386 million @xmath5 pairs for belle and 232 million @xmath5 pairs for babar .
the current significant measurements include : @xmath6 , @xmath7 , and @xmath8 . | in the standard model ( sm ) @xmath0 violation arises via the interference of at least two diagrams with comparable amplitudes but different @xmath0 conserving and violating phases .
mixing induced @xmath0 violation in the @xmath9 sector has been established in @xmath10 transitions @xcite . in the sm
, direct @xmath0 violation is expected to be sizable in the @xmath9 meson system @xcite .
charmless @xmath9 decays provide a rich sample to understand @xmath9 decay dynamics and to search for @xmath0 violation .
the first experimental evidence for direct @xmath0 violation was shown by belle for the decay mode @xmath11 @xcite .
this result suggests large interference between tree and penguin diagrams and the existence of final state interactions @xcite .
the three - body charmless hadronic @xmath9 mesons decays can provide new possibilities for @xmath0 violation searches . in spite of measuring the difference between @xmath9 and @xmath12 decay rates
, one can also measure the difference in relative phase between two quasi - two - body amplitudes , which are often dominated in three - body @xmath9 decays .
this can be achieved via amplitude ( dalitz ) analysis . as currently direct @xmath0 violation has been observed only in decays of neutral @xmath13 mesons @xcite and recently in neutral @xmath9 meson decays @xcite .
large direct @xmath0 violation is expected in charged @xmath9 decays to some quasi - two - body charmless hadronic modes @xcite .
the partial rate @xmath0 violating asymmetry is defined as : @xmath14 where @xmath15 is the yield for the @xmath16 decays and @xmath17 denotes that of the charge - conjugate mode .
we report the updated direct @xmath0 violating measurements from both belle and babar .
the data used in these study are up to 386 milion @xmath5 pairs collected at belle @xcite detector and 274 milion @xmath5 pairs at babar .
the belle detector is located at the kekb @xmath18 asymmetric - energy ( 3.5 on 8 gev ) collider @xcite and babar detector @xcite is at the slac pep - ii asymmetric - energy ( 3.1 on 9 gev ) @xmath18 storage ring @xcite .
both of them are operating at the @xmath19 resonance .
charged tracks are required to have momenta transverse to the beam greater than 0.1 gev/@xmath20 and to be consistent with originating from the interaction point ( ip ) . for particle identification of charged tracks , a combined likelihood of information from @xmath21 of drift chamber ,
time - of - flight and aerogel cherenkov counter is used by belle while an associated cherenkov - angle from the ring image counter ( dirc ) is used by babar .
for some of the modes described below , charged tracks that are positively identified as electrons or protons are excluded .
candidate @xmath22 mesons are selected by requiring the two - photon invariant mass to be in the @xmath23 mass window . the momentum vector of each photon is then readjusted to constrain the mass of the photon pair to the nominal @xmath22 mass .
candidate @xmath24 mesons are reconstructed by combining a @xmath22 with at least 250 mev/@xmath20 laboratory momentum with a pair of oppositely charged tracks that originate from the interaction point ( ip ) . for belle ,
the following requirements are made on the invariant mass of the @xmath24 candidates : 516 mev/@xmath25 mev/@xmath26 for @xmath27 and 539 mev/@xmath28 mev/@xmath26 for @xmath29 s . as to babar ,
the following cuts are applied : 490 mev/@xmath30 mev/@xmath26 for @xmath31 and 520 mev/@xmath32 mev/@xmath26 for @xmath33 . after the selection of each candidate , the @xmath24 mass constraint is implemented by readjusting the momentum vectors of the daughter particles .
the @xmath34 mesons are reconstructed via two decay chains : @xmath35 ( with @xmath36 and @xmath37 . in addition , we require the following . all photons are required to have an energy of at least 50 mev , photons from @xmath34 in @xmath37 of at least 100 mev .
the transverse momenta of @xmath38 for @xmath39 candidates have to be greater than 200 mev/@xmath20 .
@xmath40 candidates are reconstructed from pairs of oppositely - charged tracks with an invariant mass ( @xmath41 ) between 480 and 516 mev/@xmath26 .
each candidate must have a displaced vertex with a flight direction consistent with that of a @xmath40 originating from the ip .
@xmath9 meson candidates are identified using two kinematic variables : the energy difference @xmath42 and the beam constrained mass @xmath43 @xmath44 is the run - dependent beam energy in the @xmath19 rest frame and is determined from @xmath45 events , and @xmath46 and @xmath47 are the momentum and energy of the @xmath9 candidate in the @xmath19 rest frame .
the resolutions on @xmath48 and @xmath49 are about 3 mev/@xmath26 and 2030 mev , respectively .
events with @xmath50 gev/@xmath26 and @xmath51 gev are selected for the analysis .
the dominant background comes from the @xmath52 continuum , where @xmath53 or @xmath20 . to distinguish signal from the jet - like continuum background ,
event shape variables and @xmath9 flavor tagging information are employed .
we combine information of correlated shape variables into a fisher discriminant @xcite and compute the likelihood as a product of probabilities of this discriminant and @xmath54 , where @xmath55 is the angle between the @xmath9 flight direction and the beam direction in the @xmath19 rest frame .
a likelihood ratio , @xmath56 , is formed from signal ( @xmath57 ) and background ( @xmath58 ) likelihoods , obtained using events from the signal monte carlo ( mc ) and from data with @xmath59 gev/@xmath26 , respectively .
additional background discrimination is provided by @xmath9 flavor tagging .
an event that contains a lepton ( high quality tagging ) is more likely to be a @xmath60 event so a looser @xmath61 requirement can be applied .
we divide the data into six sub - samples based on the quality of flavor tagging @xcite .
continuum suppression is achieved by applying a mode dependent requirement on @xmath61 for events in each sub - sample according to @xmath62 , where @xmath63 is the expected signal from mc and @xmath64 denotes the number of background events estimated from data .
the signal yields and partial rate asymmetries are obtained using an extended unbinned maximum - likelihood ( ml ) fit with input variables @xmath48 and @xmath49 .
the likelihood is defined as : @xmath65 p_j(m_{{\rm bc}i } , \delta e_i),\end{aligned}\ ] ] where @xmath66 is the identifier of the @xmath66-th event , @xmath67 is the two - dimensional probability density function ( pdf ) in @xmath48 and @xmath49 , @xmath68 indicates the @xmath9 meson flavor , @xmath69 or @xmath70 , @xmath71 is the number of events for the category @xmath72 , which corresponds to either signal , @xmath73 continuum , a reflection due to @xmath13-@xmath74 misidentification , or background from other charmless @xmath9 decays . for the neutral @xmath9 mode , @xmath75 in the equation above
is simply @xmath76 and there is no reflection component . as to the three - body direct @xmath0 violation study , an unbinned maximum likelihood fit method is also used .
for the study performed by belle , the distribution of background events is parametrized by an empirical function with 11 parameters@xcite . as found in the previous paper by belle @xcite ,
the three - body @xmath77 amplitude is well - described by a coherent sum of @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 and @xmath83 quasi - two - body channels and a non - resonant amplitude . in order to describe an excess of signal events at @xmath84 gev/@xmath26
a @xmath82 channel was introduced .
the best fit is achieved assuming @xmath85 is a scalar state ; the mass and width determined from the fit ( see below ) are consistent with those for @xmath86 @xcite .
each quasi - two - body amplitude includes a breit - wigner function , a @xmath9 decay form - factor parametrized in a single - pole approximation , a blatt - weisskopf factor @xcite for the intermediate resonance decay , and a function that describes angular correlations between final state particles .
this is multiplied by a factor of @xmath87 that describes the relative magnitude and phase of the contribution .
the non - resonant amplitude is parametrized by an empirical function @xmath88 where @xmath89 , @xmath90 and @xmath91 are fit parameters , @xmath92 , and @xmath93 . in the analysis performed by belle , a modified model that changing the parameterization of the @xmath94 line - shape from a breit - wigner function to a flatt parameterization@xcite and adding two more channels : @xmath95 and @xmath96 is used . for @xmath0 violation studies
the amplitude for each quasi - two - body channel is modified from @xmath87 to @xmath97 , where the plus ( minus ) sign corresponds to the @xmath98 ( @xmath99 ) decay .
with such a parameterization the charge asymmetry , @xmath100 , for a particular quasi - two - body @xmath9@xmath101@xmath102 channel is given by @xmath103 it is worth noting that in this parametrization zero relative phase between @xmath99 and @xmath98 amplitudes is assumed .
the first evidence of direct @xmath0 violation in two - body @xmath9 decays in @xmath104 was claimed by belle @xcite .
however , this is not confirmed by the babar experiment @xcite .
there are both tree diagram and penguin diagram contributing to @xmath105 .
therefore , large @xmath0 violation is expected in this decay channel . in the summer of 2005 ,
both belle and babar reported their updated measurements for the @xmath106 , @xmath107 and @xmath108 modes @xcite . in spite of the results from b factories ,
cdf also report their measurements on the @xmath106 mode and is included in the world average shown in table [ tab_b2kpi ] .
note that the deviation between @xmath109 and @xmath110 is of @xmath111 for belle only and the world average is now about @xmath112 .
.summary table of @xmath113 .
[ cols="<,^,^,^",options="header " , ] [ tab_3body ]
there are lots of fruitful experimental results since last fpcp meeting and many improvements on the theoretical calculations .
we now have observation of direct @xmath0 violation in @xmath105 mode and its sign and amplitude can be better understood with next to leading order ( nlo ) calculations @xcite .
however , the deviation between @xmath114 and @xmath115 may need further clarifying theoretically @xcite .
other than @xmath105 , there are some hints or evidence of direct @xmath0 asymmetry seen by belle , but those are not yet confirmed by babar .
therefore , more statistics are still needed for both b - factories to clarify the situation .
this proceeding for fpcp 2006 includes experimental results on charmless @xmath9 meson decays measured by both belle and babar .
the support for belle experiment includes mext and jsps ( japan ) ; arc and dest ( australia ) ; nsfc ( contract no . 10175071 , china ) ; dst ( india ) ; the bk21 program of moehrd and the chep src program of kosef ( korea ) ; kbn ( contract no .
2p03b 01324 , poland ) ; mist ( russia ) ; mess ( slovenia ) ; nsc and moe ( taiwan ) ; and doe ( usa ) .
the work of babar are supported by doe and nsf ( usa ) , nserc ( canada ) , ihep ( china ) , cea and cnrs - in2p3 ( france ) , bmbf and dfg ( germany ) , infn ( italy ) , fom ( the netherlands ) , nfr ( norway ) , mist ( russia ) , and pparc ( united kingdom ) .
na31 collaboration , g. barr _ et al .
b * 317 * , 233 ( 1993 ) ; na48 collaboration , j.r .
batley _ et al .
b * 544 * , 97 ( 2002 ) ; ktev collaboration , a. alavi - harati _ et al .
d * 67 * , 012005 ( 2003 ) .
the fox - wolfram moments were introduced in g. c. fox and s. wolfram , phys .
* 41 * ( 1978 ) 1581 .
the modified moments used in this paper are described in belle collaboration , s. h. lee _
et al . _ ,
phys . rev .
lett . * 91 * , 261801 ( 2003 ) . | standard model @xmath0 violation arises interference two diagrams comparable amplitudes different conserving violating phases mixing violation established @xmath10 transitions direct @xmath0 violation sizable @xmath9 meson system charmless @xmath9 decays provide rich sample decay dynamics search violation first evidence direct @xmath0 violation belle decay mode @xmath11 suggests interference between tree penguin diagrams final state interactions three - body @xmath9 decays possibilities violation searches @xmath12 decay rates measure difference phase between quasi - two - body amplitudes amplitude dalitz analysis direct @xmath0 violation observed in decays neutral @xmath13 mesons neutral @xmath9 meson decays large direct @xmath0 violation expected in charged @xmath9 decays quasi - two - body charmless hadronic modes partial rate @xmath0 violating asymmetry defined as @xmath14 yield decays charge - conjugate mode updated direct @xmath0 violating measurements from belle babardata study 386 milion @xmath5 pairs at belle 274 milion pairs at babar belle detector at kekb @xmath18 asymmetric - energy 3.5 8 gev collider babar detector at slac pep - ii asymmetric - energy 3.1 9 gev @xmath18 storage ring operating at @xmath19 resonance charged tracks momenta transverse greater than 0.1 gev/@xmath20 consistent with interaction point ip particle identification @xmath21 by belle by babar charged tracks as electrons or protons excluded candidate @xmath22 mesons selected two - photon invariant mass in @xmath23 mass window momentum vector readjusted mass to nominal @xmath22 mass @xmath24 mesons @xmath22 with 250 mev/@xmath20 laboratory momentum oppositely charged tracks from interaction point ip for belle requirements on invariant mass @xmath24 candidates : 516 mev/@xmath25 mev/@xmath26 for @xmath27 539 mev/@xmath28/@xmath26 for @xmath29babar cuts applied 490 mev/@xmath30@xmath26 @xmath31 520 mev/@xmath32 @xmath33 selection @xmath24 mass constraint momentum vectors daughter particles @xmath34 mesons reconstructed via decay chains @xmath35 photons energy 50 mev 100 mev transverse momenta @xmath38 @xmath39 candidates greater than 200 mev/@xmath20 @xmath40 candidates reconstructed from oppositely - charged tracks invariant mass between 480 and 516 mev/@xmath26 displaced vertex flight direction consistent with @xmath40 @xmath9 meson candidates identified variables energy difference @xmath42 beam constrained mass @xmath43 @xmath44 run - dependent beam energy @xmath45 events @xmath46 @xmath47 momentum energy @xmath9 candidate resolutions on @xmath48 @xmath49 3 mev/@xmath26 2030 mev events with @xmath50/@xmath26 @xmath51 gev selected analysis dominant background from @xmath52 continuum @xmath53 @xmath20 distinguish signal event shape variables @xmath9 flavor tagging information employedcombine shape variables fisher discriminant @xcite compute likelihood probabilities @xmath54 @xmath55 angle @xmath9 flight direction beam direction @xmath19 rest frame likelihood ratio @xmath56 signal @xmath57 background @xmath58 likelihoods signal monte carlo data @xmath59@xmath26 background discrimination @xmath9 flavor tagging event lepton high quality tagging likely @xmath60 event looser @xmath61 requirement divide data six sub - samples quality flavor tagging @xcite continuum suppression mode dependent requirement @xmath61 @xmath62 @xmath63 expected signal mc @xmath64 background events data signal yields partial rate asymmetries obtained extended unbinned maximum - likelihood ml ) input variables @xmath48 @xmath49likelihood defined as @xmath65 @xmath66-th event @xmath67 two - dimensional probability density function @xmath48 @xmath49 @xmath68 @xmath9 meson flavor @xmath69 @xmath71 number events category @xmath72 corresponds signal @xmath73 continuum reflection @xmath13-@xmath74 misidentification or background @xmath9 decays neutral @xmath9 mode @xmath75 @xmath76 no reflection component three - body direct @xmath0 violation study unbinned maximum likelihood fit method used study distribution background events parametrized by empirical function 11 parameters@xcite three - body @xmath77 amplitude described by sum @xmath78 @xmath79 @xmath80 @xmath82 @xmath83 quasi - two - body channels non - resonant amplitude excess signal events at @xmath84@xmath26 @xmath82 channel introduced best fit @xmath85 scalar state mass width consistent with @xmath86 @xcitequasi - two - body amplitude includes breit - wigner function @xmath9 decay form - factor blatt - weisskopf factor resonance function angular correlations between final state particles multiplied by factor @xmath87 magnitude phase non - resonant amplitude parametrized by function @xmath88 @xmath89 parameters belle modified model @xmath94 line to channels @xmath95 and @xmath96 used for @xmath0 violation studies amplitude quasi - two - body channel modified from @xmath87 to @xmath97 plus sign corresponds to @xmath98 ( ) decay charge asymmetry @xmath100 given by @xmath103 zero relative phase between @xmath99 and @xmath98 amplitudes assumed direct @xmath0 violation in two - body @xmath9 decays in @xmath104 claimed by belle not confirmed by babar experiment tree diagram penguin diagram to @xmath105 large @xmath0 violation expected in decay channel summer 2005 belle babar reported updated measurements for @xmath106 @xmath107 @xmath108 modesresults b factories cdf report measurements on @xmath106 mode included in world average table deviation between @xmath109 and @xmath110 @xmath111 for belle only world average about @xmath112.summary table of @xmath113 fruitful experimental results since last fpcp meeting improvements on theoretical calculations observation of direct @xmath0 violation in @xmath105 mode sign amplitude understood with leading order calculations @xcite deviation between @xmath114 and @xmath115 may need clarifying hints direct @xmath0 asymmetry by belle not confirmed by babar more statistics needed for b - factories clarify proceeding fpcp 2006 includes results on charmless @xmath9 meson decays measured by belle and babar support for belle experiment includes mext jsps japan arc dest australia nsfc dst bk21 program of moehrd program kosef korea kbn mist russia mess slovenia nsc moe taiwan doe usawork babar supported doe nsf usa nserc canada ihep china cea cnrs in2p3 france bmbf dfg germany infn italy fom netherlands nfr norway mist russia pparc united kingdom na31 collaboration g. barr al 317 233 1993 ) na48 collaboration j batley 544 97 2002 ) ktev collaboration a. alavi - harati al 67 012005 2003 ) fox - wolfram moments introduced c. fox s. wolfram phys. 41 1978 ) 1581 modified moments paper described belle collaboration lee lett 91 261801 2003 ) | 0.494135 | 3,094 | 1,614 | , Saving $0.1 in GPT-4. |
55,398 | observations made using _ cobe _ , scuba , _ iso _ and mambo have provided a reasonable working knowledge of both the intensity of the submm and far - infrared background radiation and the source counts of luminous high - redshift dusty galaxies . however , because there are uncertainties in the background intensity determinations , the samples of detected galaxies are small , and most importantly , their redshift distributions are very incomplete , details of the evolution of dusty galaxies remain unresolved . the next steps forward in the field will be the launches of _ sirtf _ and _ astro - f _ , the commissioning of sofia and new , more capable ground - based mm / submm - wave cameras bolocam , sharc - ii and scuba - ii the use of ultra - long duration balloon experiments , such as blast , the construction of alma and the arrival of _ first _ , and ultimately the advent of space - borne far - infrared interferometers , such as _ specs_. there are also exciting prospects for direct mm / submm - wave co - line redshift surveys using wide - band spectrographs . using these new facilities ,
the number of high - redshift dusty galaxies known will be increased dramatically .
spectroscopy using _ sirtf _ , sofia and _ first _ will probe the astrophysical processes within these sources in detail , hopefully addressing the open question of the fraction of the counts and background radiation that is generated by the formation of high - mass stars and by active galactic nuclei ( agns ) .
the spatial and spectral structure of distant dusty galaxies will finally be resolved in detail using alma and _
specs_. # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | the determination of the intensity of extragalactic background radiation in the submm and far - infrared wavebands has been a very important development in observational cosmology ( puget et al .
1996 ; hauser et al . 1998 ; schlegel , finkbeiner & davis 1998 ; finkbeiner , davis , & schlegel 2000 ) . over the same period ,
the first observations that were sufficiently deep to detect high - redshift galaxies in the mm , submm and far - infrared wavebands were made using the 1.25-mm mambo bolometer camera at the iram 30-m telescope ( bertoldi et al .
2000 ) , the 450/850-@xmath0 m scuba camera at the jcmt ( smail , ivison , & blain 1997 ; for a current summary see blain et al .
2000a ) and the phot instrument aboard _
iso _ ( kawara et al . 1998 ; puget et al .
1999 ; juvela , mattila , & lemke 2000 ) .
most of the results for both counts and background have been summarized elsewhere , see for example figs9 and 12 in blain et al .
( 1999b ) .
progress has been rapid : not only has the background radiation spectrum been determined for the first time , but most of the 850-@xmath0 m background radiation , and about 30% and 10% of the background at wavelengths of 450 and 175@xmath0 m respectively has been accounted for as coming from individual detected galaxies .
this progress has been possible only because of tremendous improvements in the sensitivity and field of view of instruments in these wavebands .
many instrumental references can be found in blain ( 1999a ) : to save space in this brief article they are generally not repeated here .
the important pieces of missing information and the prospects for future developments of observations in the mm , submm and far - infrared wavebands are discussed below .
the mm far - infrared background radiation spectrum is generated by thermal emission from interstellar dust in all the galaxies that lie along the line of sight , right back to the epochs when the first dust was formed during the early process of galaxy formation .
it potentially includes emission from redshifts beyond recombination , as a neutral universe is transparent to long - wavelength radiation .
the intensity and spectral shape of the background radiation provides information about the volume emissivity , intrinsic spectral energy distribution and redshift of dusty galaxies ; however , this information is not unambiguous .
the counts of galaxies that make up the brighter fraction of the population contributing to the background can be used to refine and confirm the results of analyses of the background spectrum , but unfortunately neither the values of these quantities nor different scenarios of galaxy evolution can be distinguished strongly .
indeed , from the background and count data alone , the median redshift of galaxies detected in deep scuba images at 850-@xmath0 m could plausibly lie between 1 and 5 . in order to discriminate between different well fitting models , measurements of redshift distributions are necessary ; determinations of accurate counts at other mm / submm wavebands are useful , but less important . for more details
see blain et al .
( 1999a , b ) unfortunately , the redshift distribution of scuba galaxies is hard to determine for two reasons .
first , it is difficult to identify counterparts for submm - selected galaxies in other wavebands because their positions are relatively uncertain .
the jcmt beam is 15arcsec wide , and so the centroids of even the most significant detections are uncertain to within several arcseconds .
secondly , counterparts are reasonably expected to be extremely faint , perhaps so faint that optical spectroscopy will remain challenging until _
_ is available ( smail et al .
2000 ) .
deep radio observations are very useful for finding the positions of counterparts ( ivison et al .
1998 ; 2000 ) .
the flux density ratio between the submm and radio wavebands can also be used to provide an indication of redshift ( carilli & yun 1999 , 2000 ; smail et al .
2000 ; barger , cowie , & richards 2000 ) .
note however that the result is offset to lower values by the presence of any radio emission from a buried agn , and depends on the dust temperature in the galaxy ( blain 1999b ) , with hotter galaxies at higher redshifts being difficult to distinguish from cooler ones closer by .
observations of high - resolution dust emission using mm - wave interferometer arrays can also be used to obtain better positions ( downes et al .
1999 ; frayer et al .
2000 ; gear et al .
2000 ) ; however , accurate relative optical mm astrometry may be difficult in the absence of deep radio data because of the small fields of both the optical and mm - array images . to confirm an identification beyond reasonable doubt , it is necessary to obtain an optical / near - infrared redshift for the candidate and then detect co line emission at the same redshift and position .
however , this is a slow process and an accurate redshift is essential .
so far only three submm - selected galaxies have been observed in this way ( frayer 1998 , 1999 ; kneib et al . in prep ) , less than 10% of those described in the literature .
future wide - band mm and submm spectrographs will improve this situation .
at present , the scuba camera detects sources at the rate of about one per 8-hour shift at the jcmt . within the next few years , new large - format mm submm cameras on ground - based telescopes including bolocam for the cso/50-m lmt , sharc - ii ( dowell , moseley , & phillips 2000 ) for the cso and scuba - ii for the jcmt will offer more rapid detection rates and access new wavebands . at shorter wavelengths , the mips instrument onboard _ sirtf _
, the hawc camera aboard sofia and balloon - borne instruments such as blast ( devlin et al . 2000 ) will provide similar improvements over existing surveys .
the simple increase in the detection rate of galaxies will assist follow - up studies by providing larger samples of objects to sift .
in addition , the higher spatial resolution of images from scuba - ii at 450@xmath0 m , sharc - ii at 350@xmath0 m , and bolocam at 1.1 mm on the lmt will make the identification process easier .
wider survey fields and greater numbers of detections will allow large - scale statistical studies of the distribution and clustering of high - redshift dusty galaxies to be carried out ( haiman & knox 2000 ) .
there are also excellent prospects for more rapid detection of redshifted co emission from candidate galaxies , mainly by increasing the bandwidth of instruments from the current maximum of 4ghz to several 10 s of ghz
. this will be sufficient to yield a realistic chance of detecting simultaneously adjacent co lines separated by @xmath1ghz in the observer s frame for a galaxy at redshift @xmath2 from a submm - selected galaxy , and so determine a redshift without requiring optical observations ( see blain et al .
2000b ; blain 2000 ) . in the longer term , the ground - based alma interferometer array and space - borne mid- and far - ir interferometers such as _ specs _ will provide wide spectral coverage , sub - arcsec spatial resolution and great sensitivity . using these instruments the detailed astrophysics within distant dusty galaxies will be resolved in detail .
however , interferometers have small fields of view .
wide - field surveys made using the _ first _ and _ planck surveyor _
space missions will be required provide extensive lists of targets , in addition to the legacy of observations left by surveys made using _
sirtf _ and _ astro - f _ ( takeuchi et al .
a wide range of new instruments for studying the evolution of galaxies in the mm , submm and far - infrared are proposed and under construction .
observations using these facilities will enhance significantly the size and spatial resolution of catalogs of galaxies selected in these wavebands , and the amount of spectroscopic information available .
the goal of understanding the history of dust - enshrouded galaxy formation and evolution in detail by determining the redshift distributions of galaxies selected in these wavebands will be realized .
barger , a. j. , cowie l. l. , & richards , e. a. , 2000 , aj , 119 , 2092 bertoldi , f. , et al 2000 , a&a , 360 , 92 blain , a. w. 1999a , asp conf .
191 , 255 ( astro - ph/9906141 ) blain , a. w. 1999b , mnras , 309 , 955 blain , a. w. , smail i. , ivison r. j. , & kneib j .-
1999a , mnras , 302 , 632 blain , a. w. , et al .
1999b , mnras , 309 , 715 blain , a. w. 2000 , asp conf .
, in press ( astro - ph/9911449 ) blain , a. w. , et al .
2000a , asp conf .
ser . vol .
193 , 246 ( astro - ph/9908111 ) blain , a. w. , frayer , d. t. , bock , j. , & scoville , n. 2000b , mnras , 313 , 559 carilli , c. l. , & yun , m. s. 1999 , apj , 513 , l13 carilli , c. l. , & yun , m. s. 2000 , apj , 530 , 618 devlin , m. , et al .
2000 , proc .
umass conference , in press dowell , c. d. , moseley , s. h. , & phillips , t. g. 2000 , asp conf .
, in press downes , d. , et al . 1999 ,
a&a , 347 , 809 finkbeiner , d. p. , davis , m. , et al .
2000 , apj , in press ( astro - ph/0004175 ) frayer , d. t. , et al .
1998 , apj , 506 , l7 frayer , d. t. , et al . 1999 ,
apj , 514 , l13 frayer , d. t. , et al .
2000 , aj , in press ( astro - ph/0005239 ) gear , w. k. , et al .
2000 , mnras , in press ( astro - ph/0007054 ) haiman , z. , & knox , l. 2000 , apj , 530 , 124 hauser , m. g. , et al .
1998 , apj , 508 , 106 ivison , r. j. , et al .
1998 , mnras , 298 , 583 ivison , r. j. , et al .
2000 , mnras , 315 , 209 juvela , m. , mattila , k. , & lemke , d. 2000 , a&a , in press ( astro - ph/0005564 ) kawara , k. , et al .
1998 , a&a , 336 , l9 puget , j .- l . , et al .
1996 , a&a , 308 , l5 puget , j .- l . , et al . 1999 , a&a , 345 , 29 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 smail , i. , ivison , r. j. , and blain , a. w. 1997 , apj , 490 , l5 smail , i. , ivison , r. j. , owen , f. n. , et al .
2000 , apj , 528 , 612 smail , i. , et al .
2000 , proc .
umass conference ( astro - ph/0008237 ) takeuchi , t. t. , et al . 1999 ,
pasp , 111 , 288 | determination intensity extragalactic background radiation in submm far - infrared wavebands important development observational cosmology ( puget 1996 hauser 1998 schlegel finkbeiner davis 1998 2000 first observations deep high - redshift galaxies in - infrared wavebands using 1.25-mm mambo bolometer camera iram 30-m telescope bertoldi 2000 ) 450/850-@xmath0 m scuba camera at jcmt smail ivison blain 1997 blain 2000a phot instrument aboard iso _ ( kawara 1998 puget 1999 juvela mattila lemke 2000 ) results for counts background summarized elsewhere figs9 12 blain et al . 1999b ) progress rapid background radiation spectrum determined first most 850-@xmath0 m background radiation 30% 10% background at wavelengths 450 175@xmath0 m accounted from detected galaxies progress improvements in field of view of instruments wavebands instrumental references in blain ( 1999a ) not repeated missing information prospects for future developments observations - infrared wavebands discussed belowmm far - infrared background radiation spectrum generated by thermal emission from interstellar dust in galaxies along line of sight galaxy formation potentially includes emission from redshifts beyond recombination neutral universe transparent to long - wavelength radiation intensity spectral shape background radiation provides information about volume emissivity spectral energy distribution redshift of dusty galaxies not unambiguous counts of galaxies brighter fraction results analyses values scenarios galaxy evolution distinguished median redshift of galaxies in deep scuba images at 850-@xmath0 m between 1 and 5 measurements of redshift distributions necessary accurate counts at other mm / submm wavebands useful less important see blain et al . 1999a redshift distribution of scuba galaxies hard to determine for difficult to identify counterparts for submm - galaxies in other wavebands positions uncertain jcmt beam 15arcsec wide centroids detections uncertain counterparts expected to be extremely faint optical spectroscopy challenging until _ available smail 2000 deep radio observations useful for finding positions of counterparts ivison1998 ; 2000 flux density ratio between submm radio wavebands redshift ( carilli & yun 1999 2000 smail et 2000 barger cowie richards 2000 result offset by radio emission from buried agn depends on dust temperature ( blain 1999b hotter galaxies at higher redshifts difficult to distinguish from cooler ones observations of high - resolution dust emission using mm - wave interferometer arrays better positions ( downes 1999 frayer 2000 gear et al . 2000 accurate relative optical mm astrometry difficult deep radio data small fields of optical mm - array images confirm identification optical / near - infrared redshift for detect co line emission at same redshift position slow process accurate redshift essential three submm - selected galaxies observed ( frayer 1998 1999 kneib less than 10% future wide - band mm submm spectrographs improve scuba camera detects sources one per 8-hour shift at jcmt new large - format mm submm cameras including bolocam sharc - ii offer rapid detection rates access new wavebandsshorter wavelengths mips instrument sirtf hawc camera sofia balloon - borne instruments blast devlin 2000 provide improvements surveys increase detection rate studies larger samples higher spatial resolution images from scuba - ii 450 sharc - ii 350 bolocam 1.1 mm identification easier wider survey fields greater numbers detections allow large - scale statistical studies distribution clustering high - redshift dusty galaxies haiman knox 2000 ) prospects rapid detection redshifted co emission from galaxies increasing bandwidth instruments from 4ghz to 10 s ghz chance detecting adjacent co lines separated by @xmath1ghz galaxy redshift @xmath2 without optical observations blain et al . 2000b ground - based alma interferometer array space - borne mid- far - ir interferometers specs provide wide spectral coverage sub - arcsec spatial resolution great detailed astrophysics distant dusty galaxies resolved interferometers have small fields of viewwide - field surveys first planck surveyor missions provide extensive lists targets legacy sirtf astro - f takeuchi new instruments evolution mm submm far - infrared proposed under construction observations enhance size spatial resolution catalogs galaxies spectroscopic information goal understanding dust - enshrouded galaxy formation evolution determining redshift distributions galaxies wavebands realized barger a. j. cowie l. l. richards , e. a. 2000 ,, 119 , 2092 bertoldi , f. al 2000 a&a , 360 , 92 blain 1999a asp conf . 191 , 255 ( astro - ph/9906141 ) 1999b mnras , 309 , 955 smail ivison r kneib 1999a mnras 302 , 632 1999b mnras 309 , 715 2000 asp conf. press astro - ph/9911449 ) 2000a asp conf . 193 , 246 ( astro - ph/9908111 ) frayer bock j scoville n2000b mnras 313 559 carilli l. yun m.. 1999 apj 513 l13 carilli yun 2000 apj 530 , 618 devlin m. umass conference press dowell moseley s t. 2000 asp conf press downes d.. 1999 a&a 347 , 809 finkbeiner d. davis m. 2000 apj press astro - ph/0004175 ) frayer d. t. 1998 apj 506 , l7 frayer 1999 apj 514 l13 frayer 2000 press astro - ph/0005239 ) gear w k. mnras press astro - ph/0007054 ) haiman z knox 2000 apj 530 , 124 hauser m. g. 1998 apj 508 , 106 ivison r. j. mnras 298 , 583 ivison 2000 mnras 315 , 209 juvela mattila k lemke 2000 a&a press astro - ph/0005564 ) kawara k.1998 a&a 336 l9 puget 1996 a&a 308 l5 puget 1999 a&a 345 29 schlegel d. j finkbeiner davis 1998 apj 500 , 525 smail ivison r. j. blain 1997 apj 490 l5 smail ivison. owen f. n. 2000 apj 528 612 smail 2000 umass conference astro - ph/0008237 ) takeuchi t. t. al 1999 pasp 111 , 288 | 0.433875 | 2,768 | 1,454 | , Saving $0.1 in GPT-4. |
55,403 | we study the energy dependence of global polarization of @xmath0 hyperons in peripheral @xmath1 collisions .
we combine the calculation of vorticity and strange chemical potential in the framework of kinetic quark - gluon string model with the anomalous mechanism related to axial vortical effect .
we pay special attention to the temperature dependent contribution related to gravitational anomaly and found that the preliminary rhic data are compatible with its suppression discovered earlier in lattice calculations . | the experimental evidences for polarization of hyperons in heavy - ion collisions found by star collaboration @xcite attracted recently much attention @xcite . the studies of polarization are often performed @xcite in the framework of approach exploring local equilibrium thermodynamics @xcite and hydrodynamical calculations of vorticity @xcite .
there is another approach to polarization first proposed in @xcite and independently in @xcite .
the so - called axial vortical effect ( see e.g. @xcite and references therein ) being the macroscopic manifestation of axial anomaly @xciteleads to induced axial current of strange quarks which may be converted to polarization of @xmath0-hyperons @xcite .
the effect is proportional to vorticity and helicity of the strong interacting medium which was calculated @xcite in the kinetic quark - gluon - string model(qgsm)@xcite . in this way , in fact ,
some particular case of fundamental relation between kinetic and hydrodynamic description was considered .
this calculation were later made more detailed@xcite , including the spatial and temporal dependence of strange chemical potential which is also the ingredient of anomalous approach to polarization .
anomalous mechanism of polarization makes this effect qualitatively similar to the local violation @xcite of discrete symmetries in strongly interacting qcd matter .
the most well known here is the chiral magnetic effect ( cme ) which uses the ( c)p - violating ( electro)magnetic field emerging in heavy ion collisions in order to probe the ( c)p - odd effects in qcd matter .
the polarization , in turn , is similar to chiral vortical effect ( cve)@xcite due to coupling to p - odd medium vorticity leading to the induced electromagnetic and all conserved - charge currents @xcite , in particular the baryonic one .
recently the pioneering preliminary results on global polarization of @xmath0 and @xmath2 hyperons in @xmath1 collisions in beam energy scan at rhic were released @xcite and the qualitative tendency of polarization decrease with energy in agreement with the prediction @xcite was revealed . the recent theoretical analysis @xcite suggested that decrease of polarization with energy is related to the suppression of axial magnetic effect contribution in strongly correlated qcd matter found in lattice simulations .
indeed , the chiral vorticity coefficient describing the axial vortical effect @xmath3 contains temperature dependent term related to gravitational anomaly @xcite , and naively it can be quite substantial and increase with energy . however , lattice simulations @xcite lead to the zero result in the confined phase and to the suppression by one order of magnitude at high temperatures .
neglecting axial chemical potential , the coefficient @xmath4 takes the form @xmath5 as soon as for free fermion gas the @xmath6 term is recovered @xcite for large lattice volume at fixed temperature , the above - mentioned suppression should be attributed to the correlation effects .
it was suggested@xcite , that the accurate measurements of polarization energy dependence may serve a sensitive probe of strongly correlated qcd matter . in the current paper
we perform numerical simulations to implement this suggestion .
we performed the numerical simulations in qgsm model @xcite .
we decomposed the space - time to the cells , allowing to define velocity and vorticity in the kinetic model , as described in detail in @xcite . to define the strange chemical potential ( assuming that @xmath0 polarization is carried by strange quark ) we used the matching procedure @xcite of distribution functions to its ( local ) equilibrium values
. in this paper
, we also determine in this way the values of temperature .
in general , let us stress that we realized in our particular case the relation between kinetics , hydrodynamics and thermodynamics .
we first neglect the gravitational anomaly contribution and start by considering the energy dependence of polarization ( described in detail in @xcite ) for three values of impact parameter .
the results are presented at fig.1 .
the curves correspond to @xmath7 .
we continue by the inclusion of contribution related to gravitational anomaly , which is the central issue of this paper .
the results are presented at fig.2 .
we consider as a starting point the original value of anomaly coefficient@xcite @xmath6 which is reproduced for large lattice volume at fixed temperature@xcite .
we present the curves following from the coefficients suppressed by factor @xmath8 ( [ cvl ] ) resulting from the lattice calculations @xcite .
we compare values of @xmath9 with @xmath10 .
as one can see , the lattice - supported value @xmath11 is most close to the behavior of preliminary data which may be considered as a signal of strongly correlated matter formation .
the closeness of @xmath12 curve to the experimental points may be related to the contribution of confinement phase , where lattice calculations @xcite lead to zero temperature - dependent effect . at the same time ,
already @xmath13 leads to the curve growing with energy , while @xmath9 leads to extremely strong growth .
we numerically studied the generation of polarization by the anomalous mechanism ( axial vortical effect ) and compared it with the preliminary observed data .
first we neglected the gravitational anomaly related temperature dependent contribution when the decrease of chemical potential with energy leads in turn to the decrease of polarization .
we considered this effect for three impact parameters .
we also included the contribution related to gravitational anomaly proportional to @xmath14 and studied its possible suppression in strongly correlated matter .
we found that the preliminary data are in accordance with suppression effect found on the lattice .
b. betz , m. gyulassy and g. torrieri , phys .
c * 76 * ( 2007 ) 044901 doi:10.1103/physrevc.76.044901 [ arxiv:0708.0035 [ nucl - th ] ]
. l. p. csernai , v. k. magas and d. j. wang , phys . rev .
c * 87 * ( 2013 ) no.3 , 034906 doi:10.1103/physrevc.87.034906 [ arxiv:1302.5310 [ nucl - th ] ] .
o. rogachevsky , a. sorin and o. teryaev , `` chiral vortaic effect and neutron asymmetries in heavy - ion collisions , '' phys .
c * 82 * , 054910 ( 2010 ) [ arxiv:1006.1331 [ hep - ph ] ] .
t. kalaydzhyan , `` temperature dependence of the chiral vortical effects , '' phys .
d * 89 * , no .
10 , 105012 ( 2014 ) [ arxiv:1403.1256 [ hep - th ] ] .
d.t . son and p.
surowka , phys . rev
. lett . * 103 * , 191601 ( 2009 ) m. baznat , k. gudima , a. sorin and o. teryaev , `` helicity separation in heavy - ion collisions , '' phys .
c * 88 * , 061901 ( 2013 ) [ arxiv:1301.7003 [ nucl - th ] ] .
j. -h .
gao , z. -t .
liang , s. pu , q. wang and x. -n .
wang , `` chiral anomaly and local polarization effect from quantum kinetic approach , '' phys .
lett . * 109 * , 232301 ( 2012 ) [ arxiv:1203.0725 [ hep - ph ] ] .
k. fukushima , d.e .
kharzeev and h.j .
warringa , `` the chiral magnetic effect , '' phys .
d * 78 * , 074033 ( 2008 ) [ arxiv:0808.3382 [ hep - ph ] ] .
d. kharzeev and a. zhitnitsky , `` charge separation induced by p - odd bubbles in qcd matter , '' nucl .
a * 797 * , 67 ( 2007 ) [ arxiv:0706.1026 [ hep - ph ] ] .
a. sorin and o. teryaev , arxiv:1606.08398 [ nucl - th ] ; phys .
c , in print .
k. landsteiner , e. megias , l. melgar and f. pena - benitez , `` holographic gravitational anomaly and chiral vortical effect , '' jhep * 1109 * ( 2011 ) 121 v. braguta , m.n .
chernodub , k. landsteiner , m.i .
polikarpov , m.v .
ulybyshev phys.rev .
d88 ( 2013 ) 071501 doi : 10.1103/physrevd.88.071501 e - print : arxiv:1303.6266 [ hep - lat ] ; v. braguta , m.n .
chernodub , v.a .
goy , k. landsteiner , a.v .
molochkov , m.i .
polikarpov , phys.rev .
d89 ( 2014 ) no.7 , 074510 doi : 10.1103/physrevd.89.074510 e - print : arxiv:1401.8095 [ hep - lat ] ; v. braguta , m. n. chernodub , v. a. goy , k. landsteiner , a. v. molochkov and m. ulybyshev , `` study of axial magnetic effect , '' aip conf . proc . * 1701 * , 030002 ( 2016 ) ; doi:10.1063/1.4938608 ; v. goy , `` investigation of su(2 ) gluodynamics in the framework of the latice approach '' , phd thesis ( in russian ) , vladivostok , 2015 .
p. v. buividovich , `` axial magnetic effect and chiral vortical effect with free lattice chiral fermions , '' j. phys .
ser . * 607 * , no . 1 , 012018 ( 2015 ) doi:10.1088/1742 - 6596/607/1/012018 [ arxiv:1309.4966 [ hep - lat ] ] . | experimental evidences for polarization of hyperons in heavy - ion collisions by star collaboration @xcite attention studies polarization performed exploring local equilibrium thermodynamics hydrodynamical calculations of vorticity another approach polarization proposed in @xcite axial vortical effect manifestation of axial anomaly to induced axial current of quarks converted to polarization of-hyperons effect proportional to vorticity helicity of interacting medium calculated in kinetic quark - gluon - string model relation between kinetic hydrodynamic description considered calculation detailed including spatial temporal dependence of chemical potential anomalous approach to polarization mechanism polarization effect similar to local violation of symmetries in interacting qcd matter known chiral magnetic effect violating field in heavy ion collisions - effects in qcd matter polarization similar to chiral vortical effect due to p - odd medium vorticity to induced electromagnetic conserved - charge currents baryonicpreliminary results on global polarization of @xmath0 @xmath2 hyperons in @xmath1 collisions energy released polarization decrease with energy with prediction analysis suggested decrease polarization with energy related to suppression axial magnetic effect in qcd matter in lattice simulations chiral vorticity coefficient axial vortical effect @xmath3 contains temperature dependent term related to gravitational anomaly can increase with energy lattice simulations lead to zero result in confined phase suppression at high temperatures coefficient @xmath4 takes form @xmath5 free fermion gas @xmath6 term recovered for large lattice volume at fixed temperature suppression attributed to correlation effects accurate measurements of polarization energy dependence may serve probe of qcd matter current paper numerical simulations to suggestion in qgsm model decomposed space - time to cells velocity vorticity in kinetic model define strange chemical potential @xmath0 polarization carried by quark used matching procedure of distribution functions to local equilibrium values values of temperature realized relation between kinetics hydrodynamics thermodynamicsneglect gravitational anomaly contribution energy dependence polarization @xcite for three values impact results presented fig.1 curves correspond to @xmath7 continue contribution gravitational anomaly central issue results at fig.2 consider starting point original value anomaly coefficient@xcite @xmath6 reproduced large lattice volume fixed temperature present curves from coefficients suppressed by factor @xmath8 lattice calculations compare values @xmath9 with @xmath10 lattice - supported value @xmath11 close to preliminary data signal of strongly correlated matter formation closeness @xmath12 curve to experimental points confinement phase lattice calculations lead zero temperature - dependent effect @xmath13 leads to curve growing energy @xmath9 strong growth studied generation polarization by anomalous mechanism compared with preliminary data neglected gravitational anomaly temperature dependent contribution decrease chemical potential leads decrease polarization considered effect for three impact parameters included contribution gravitational anomaly proportional to @xmath14 studied suppression in strongly correlated matter preliminary data with suppression effect lattice betz gyulassy torrieri phys 2007 044901:0708.0035 nucl - th csernai magas j wang phys . rev 87 2013 ).3 034906.1103/physrevc.87.034906 arxiv:1302.5310 nucl - th rogachevsky sorin teryaev chiral vortaic effect neutron asymmetries heavy - ion collisions . c 82 054910 2010 arxiv:1006.1331 hep - ph kalaydzhyan temperature dependence chiral vortical effects . d 89 105012 2014 arxiv:1403.1256 hep - th son surowka phys . rev. 103 191601 2009 baznat gudima sorin teryaev helicity separation heavy - ion collisions . c 88 061901 2013:1301.7003 nucl - th gao liang pu wang wang chiral anomaly local polarization effect quantum kinetic approach phys . 109 232301 2012 ) arxiv:1203.0725 hep - ph fukushima kharzeev warringa chiral magnetic effect phys . d * 78 074033 2008 arxiv:0808.3382 hep - ph kharzeevzhitnitsky charge separation p - bubbles qcd matter nucl 797 67 2007 arxiv:0706.1026 hep - ph sorin teryaev arxiv:1606.08398 nucl - th phys. landsteiner megias melgar. pena - benitez holographic gravitational anomaly chiral vortical effect jhep 1109 2011 braguta chernodub landsteiner polikarpov ulybyshev phys.rev d88 2013 ) 071501 10.1103/physrevd arxiv:1303.6266 hep - lat braguta chernodub goy landsteiner molochkov polikarpov.rev d89 2014 ) no.7 074510 10.1103/physrevd.89.074510 arxiv:1401.8095 hep - lat braguta chernodub goy landsteiner molochkov ulybyshev study axial magnetic effect conf proc 1701 030002 2016 doi:10.1063/1.4938608 v. goy investigation su(2 ) gluodynamics latice approach phd thesis russian vladivostok 2015buividovich axial magnetic chiral vortical lattice fermions phys. ser. 607 1 012018 2015.1088/1742 6596/607/1/012018 arxiv:1309.4966 hep | 0.353579 | 2,302 | 1,152 | , Saving $0.1 in GPT-4. |
55,409 | we present a new , semi - phenomenological method to extract the condensate fraction @xmath0 from two sets of measured responses of @xmath1he with identical kinematics at @xmath2 and @xmath3 . | over the past decades attempts have been made to determine the condensate fraction @xmath4 of liquid @xmath1he , which is defined as the asymptotic limit of the single - atom density matrix ( sadm ) @xmath5 .
@xmath6 @xmath7 is the number density .
we shall use the notation @xmath8 in spite of the increasing quality of data , the various experimental results show an unsatisfactory spread in @xmath9 .
the reason may well be that there is no unequivocal , direct link between the above definition of @xmath4 and experimental information .
the quality of an extracted condensate fraction thus depends on the accuracy of the approximate isolation of the relevant sadm , which contains the condensate fraction . without striving for completeness
we briefly review : \i ) computations from first principles ; \ii ) results from the comparison of specific theories and data ; \iii ) data analysis with minimal theoretical input ; we then propose and evaluate a novel extraction method which , except for minor assumptions , is largely model - independent .
@xmath10 : in principle stochastic methods for a representative sample of a finite number of atoms determine the ground state wave functions and thus @xmath11 .
those wave functions , and thus the sadm , can be computed up to large values of @xmath12 and ( [ a1 ] ) thus directly provide @xmath9 ( see for instance [ ] ) . in practice
the method is accurate only for @xmath13 k , and from sufficiently large @xmath12 one extracts @xmath14 .
nevertheless , even for @xmath13 , different stochastic methods produce a spread in @xmath9 of , as much as 25% @xcite . for finite @xmath15 a calculation of @xmath11 requires a canonical average @xmath16|{{\mbox{\boldmath $ r$}}}_1-s\hat{{\mbox{\boldmath $ i$}}}{_z},{{\mbox{\boldmath $ r$}}}_k ) \rangle \label{a2}\end{aligned}\ ] ] with @xmath17 , the partition function .
results for @xmath18 ) known to us , are for @xmath19 and for a number of discrete @xmath20 around @xmath21 @xcite , but beyond @xmath22 the computed sadm apparently have sizable inaccuracies .
this reflects on condensate fractions , which have been extracted from @xmath12 in the range @xmath23 where a constant value of the sadm is only reached on the average @xcite .
@xmath24 : efforts to obtain the condensate fraction naturally focuss on the extraction of the sadm from an observable . a prime example is the cross section for inclusive neutron scattering on he , which is proportional to the response @xmath25 . the latter is a function of two kinematic variables , e.g. the momentum transfer @xmath26 and the grs - west scaling variable , @xmath27 , with @xmath28 , the mass of an atom . for fixed @xmath26
the variable @xmath29 is a measure for the energy loss @xmath30 @xcite . in order to compare with experiment
the above response has to be folded into the experimental resolution function @xmath31 .
one may then formally write the response and its fourier transform ( ft ) in a closed form ( see for instance ref . [ ] ) [ a3 ] @xmath32 the first factor @xmath33 is the asymptotic limit of the response , and depends on the single - atom momentum distribution @xmath34 .
its fourier transform is just the sadm .
final state interactions ( fsi ) are formally contained in @xmath35 , which can be expanded in powers of @xmath36 and which depends on higher order density matrices @xmath37 . in practice
it suffices to retain only the dominant fsi @xmath38 . finally , @xmath31 is experimental resolution ( er ) of the measuring devise . in the standard approach one
computes from dynamics the fsi factor @xmath35 , which modifies the asymptotic response @xmath39 for finite @xmath26 ( see for instance ref . [ ] ) .
in addition many - body density matrices which weigh all components are required .
we focus on @xmath39 , which is related to the momentum distribution @xmath40 and which for @xmath41 requires modeling @xmath42+a(t_<)n^{no}(p;t _ < )
f(0;t _ < ) ] \label{a4}\end{aligned}\ ] ] in the above parametrization ( [ a4 ] ) one finds in addition to the macroscopic fraction @xmath4 of atoms with momentum @xmath43 , @xmath44 , the fraction of atoms relative to @xmath9 with momenta @xmath45 in the immediate neighborhood of @xmath46 @xcite . finally , @xmath47 above is the normal part of the momentum distribution of atoms with @xmath48 ; @xmath49 cares for proper normalization .
albeit dominant for @xmath50 is not directly measurable and is for low @xmath15 frequently assumed to be @xmath15-independent : @xmath51 .
for @xmath52 it is the measurable @xmath40 .
equivalently for the sadm [ a5 ] @xmath53 \rho_1^{no}(s;t_<)/\rho(t _ < ) \label{a5b}\end{aligned}\ ] ] above , @xmath54 is the difference between the sadm and its asymptotic limit @xmath9 , which will be called the deficit function . by definition @xmath55 .
we summarize : except for @xmath13 , there is no accurate information on the sadm for large @xmath12 . for finite @xmath12 an expression for @xmath54 , e.g. ( [ a5b ] ) ,
is model - dependent and without knowledge of both @xmath56 and @xmath4 there is no accurate way to reach the deficit function @xmath54 .
@xmath57 : one method used in the past relates the integral of the difference between responses for @xmath58 in the region @xmath59 to the fraction of atoms having momenta in that region .
the method is not accurate because of imprecise isolation of fsi and er broadening in ( [ a3 ] ) ( see for instance ref .
[ ] ) . of similar nature
are cumulant parametrizations of reponses for @xmath41 @xcite .
even if the above assumption on @xmath60 is made , the data do not permit the extraction of an accurate value of @xmath4 and @xmath61 .
instead one studies fits for varying @xmath4 @xcite .
we now describe a new method for the extraction of the condensate fraction as it appears in the sadm for @xmath41 .
we try to avoid the described insufficient information on the latter by exploiting data on inclusive scattering on @xmath62 at precisely the same momentum transfer @xmath26 and energy loss @xmath30 , or the related scaling variable @xmath29 .
fairly recent data of that type exist for @xmath63 k @xcite ( mari data ) .
( [ a6a ] ) expresses the @xmath15-independence of the er for the mari data and as a consequence the ratio of actual convoluted data are replaced by deconvoluted ones .
those @xmath65 depend on @xmath26 and are complex , because the ft of the data are complex .
( [ a6b ] ) is of a different nature .
it assumes that the fsi function @xmath35 is at most weakly @xmath15- dependent .
although verified within the accuracy of the data @xcite , we cite weak @xmath15-dependence on the measured pair - distribution function @xmath66 @xcite which , at least in some theories , enters the fsi function @xmath67 @xcite . the price for @xmath68removing@xmath69 and
the er is the appearance of two sadm , instead of the desired @xmath5 . since the precision with which one ultimately extracts the condensate fraction depends on the quality of the input , we discuss the latter .
we note that @xmath65 in ( [ a6 ] ) uses not actual data , but instead their ft . with substantial noise , in particular in the larger @xmath70 tails of the data for @xmath25 , one can not avoid smoothing , or even cutting out those low - intensity data , which affect the accuracy of @xmath65 .
we first apply the method to the above - mentioned mari data at @xmath71 k @xcite from which we chose a representative sample of highest quality for which @xmath72 . from those we extract the ratio @xmath73/{\rm re}[\eta^{exp}(q , s)]$ ] , which is well determined in the range @xmath74 . for growing @xmath12 , the imaginary part increases from 0 but is , even at medium @xmath75 , only 2% of the real part , and can thus for all purposes be neglected .
this confirms that at least the ratio of the involved fsi functions @xmath35 is virtually @xmath15-independent .
next we investigate the @xmath26-dependence of @xmath76 $ ] and show in fig . 1 data for some individual @xmath26 .
before judging the quality we mention that , contrary to predictions , some data for @xmath52 and varying @xmath26 , and more for @xmath41 , are not uniformly smooth in @xmath26 .
this may cause @xmath77-dependence in @xmath78 and justifies the use of an average for @xmath79 @xmath80\bigg\rangle_q \label{a7}\end{aligned}\ ] ] + + as a first step we test ( [ a6b ] ) with the only available computed results on sadm for @xmath84 @xcite . without any dynamical cause , the outcome for @xmath15=1.54 k closest to @xmath15=1.6 k of the mari data , is not very close to values from interpolation between neighboring @xmath15 .
we suspect the mentioned inaccuracies which are also apparent in non - vanishing condensate fractions for @xmath52 and the non - smooth dependence of @xmath4 as function of @xmath15 outside their estimated theoretical error bars .
nevertheless we use interpolated computed results and define for @xmath85 k @xmath86 the above as function of @xmath87 is also entered in fig . 1 and is seen to represent the data rather poorly .
we shall thus exclusively use @xmath88 and return to eq .
( [ a5a ] ) , which by means of ( [ a7 ] ) and ( [ a6b ] ) becomes @xmath89 we note that , contrary to ( [ a1 ] ) , eq . ( [ a9 ] ) is not applicable in the asymptotic region , because in order to produce a constant , numerator and denominator in the ratio in ( [ a9 ] ) have to tend to zero in exactly the same fashion .
this is an impossibly stringent demand .
although eq .
( [ a9 ] ) provides in principle a value of the condensate fraction for @xmath90 , it is clearly of greater interest to consider a range of @xmath12 .
in fact , the right hand side of ( [ a9 ] ) is a function of @xmath12 , whereas the left hand side is a constant , implying a consistency test , to be passed as a condition for a succesful extraction of @xmath4 .
the size of the @xmath12-range depends on the available information on the various functions in ( [ a9 ] ) .
the most delicate source of information is the deficit function @xmath54 .
after the warnings above , one clearly does not want to use ( [ a5b ] ) and has thus to rely on the model - independent , not too accurate results of ref .
[ ] , which for @xmath91 , provide @xmath0 with uncertainties of the order of 15% .
the crucial point is that those uncertainties decrease relatively to @xmath92 which increases for decreasing @xmath12 .
we thus conclude that , up to medium @xmath79 , one can trust and use the computed @xmath54 .
the above happens to also be the range for which the available data determine @xmath78 sufficiently well .
we thus perform the consistency test , implied by ( [ a8 ] ) and it comes as a surprise that the right hand side in the considered @xmath12-range is only weakly @xmath12-dependent , leading to @xmath93 the latter is the result for a standard extrapolation to @xmath13 .
the small error limits are due to averaging results for different @xmath12 ; we could not estimate the same for uncertainties in @xmath54 .
we note that the errors on the three functions in ( [ a9 ] ) are uncorrelated , which underscores the feducity of the extracted value of @xmath4 .
the above numbers use @xmath94 , interpolated between values for @xmath15= 1.54,1.82k as reported by ceperley .
we mentioned that the former set , does not interpolate smoothly between values for @xmath15=1.18 , 1.82 k. smoothing leads to a lower condensate fraction .
obviously the accuracy with which one can extract the condensate fraction depends on the same for the input .
above we also listed older argonne data for fixed @xmath95 and some 10 temperatures around @xmath96 .
unfortunately we do not have available the actual data and the er functions , and have to analyze instead parametrizations of the above . as given by the authors .
those are given by the authors in the form of double gaussians for data deconvoluted from er , and in addition also deconvoluted from fsi .
the information no doubt reduces the required accuracy . yet , in view of scarcity of information we performed the above analysis for the argonne data . in order to compare with the mari data
and we fix @xmath52 at 2.3 k. one first observes that the computed ft of the above parametrized data are essentially real .
next , ratios of those ft for and @xmath41 and the above @xmath52 are identical within 1% , in support of ( [ a6 ] ) .
nevertheless , only a limited part of the information can be used for an analysis of the above type .
thus for @xmath97 k , @xmath4 is the difference of nearly equal terms in ( [ a9 ] ) .
a reliable determination requires a precision on @xmath98 and @xmath92 , which is beyond the ceperley results .
a different difficulty occurs for @xmath99 k , where one has to make an uncertain extrapolation , using the lowest @xmath15 results of ceperley .
the following outcome @xmath100 carries some 20 - 25% uncertainties , which appreciably exceed the one in ( [ a10 ] ) based on the mari data .
one notes that @xmath101 k ) from the mari and argonne data approximately agree .
however , in view of the fact that we had to use parametrizations of the argonne data , we do not attach too much significance to the correct trend of @xmath0 as function of @xmath15 and the somewhat low , but not unreasonable , average @xmath102=0.079 . in conclusion ,
we have suggested and worked out a new method to extract the condensate fraction in @xmath1he from the fourier transforms of data sets on structure functions at the same momentum and energy transform , but for two @xmath15 below and above @xmath96 .
in addition the method requires some previously computed dynamical information . the expression ( [ a9 ] ) for @xmath4 appears in principle as a function of @xmath12 , but is in practice a well - determined and thus meaningful constant , the condensate fraction . | past decades attempts determine condensate fraction @xmath4 of liquid @xmath1he defined as asymptotic limit of single - atom density matrix @xmath5 @xmath6 @xmath7 is number density use notation @xmath8 increasing quality data experimental results show unsatisfactory spread in @xmath9 no direct link between definition @xmath4 experimental information quality of extracted condensate fraction depends on accuracy isolation relevant sadm review computations from first principles results comparison of theories data data analysis minimal theoretical input propose evaluate novel extraction method model - independent @xmath10 stochastic methods for finite atoms determine ground state wave functions @xmath11 functions sadm computed up to large values of @xmath12 provide @xmath9 method accurate only for @xmath13 k from large @xmath12 extracts @xmath14 for @xmath13 stochastic methods produce spread in @xmath9 25% @xcite finite @xmath15 calculation @xmath11 requires canonical average @xmath16\mbox\boldmath r$-s_z}_k\rangle{aligned @xmath17 partition function results @xmath18 @xmath19 discrete @xmath20 @xmath21 @xcite beyond @xmath22 computed sadm inaccuracies reflects condensate fractions extracted from @xmath12 range @xmath23 constant value sadm reached average @xcite @xmath24 efforts obtain condensate fraction focuss extraction sadm observable example cross section inclusive neutron scattering he proportional to response @xmath25 function two kinematic variables momentum transfer @xmath26 grs - west scaling variable @xmath27 @xmath28 mass atom fixed @xmath26 variable @xmath29 measure energy loss @xmath30 @xcite compare experiment response folded into experimental resolution function @xmath31 write response fourier transform closed form @xmath32 first factor @xmath33 asymptotic limit response depends on single - atom momentum distribution @xmath34 fourier transform sadmstate interactions ( fsi in @xmath35 expanded in powers @xmath36 depends on higher density matrices @xmath37 retain dominant fsi @xmath38 @xmath31 is experimental resolution of measuring devise standard approach computes from dynamics fsi factor @xmath35 modifies asymptotic response @xmath39 for finite @xmath26 many - body density matrices weigh all components required focus on @xmath39 related to momentum distribution @xmath40 @xmath41 requires modeling @xmath42+a parametrization finds macroscopic fraction of atoms with momentum @xmath43 @xmath44 fraction of atoms relative to @xmath9 with momenta @xmath45 neighborhood of @xmath46 @xmath47 normal part of momentum distribution of atoms with @xmath48 @xmath49 cares for normalization dominant @xmath50 not directly measurable @xmath15-independent for @xmath52 measurable @xmath40 @xmath53 @xmath54 is difference between sadm and asymptotic limit @xmath9 called deficit functiondefinition @xmath55 except @xmath13 no accurate information sadm large @xmath12 finite @xmath12 expression @xmath54 model - dependent without knowledge @xmath56 @xmath4 no accurate way reach deficit function @xmath54 @xmath57 method relates difference responses @xmath58 region @xmath59 to fraction atoms momenta method not accurate imprecise isolation fsi er broadening a3 ] ) similar cumulant parametrizations reponses @xmath41 @xcite assumption @xmath60 data permit extraction accurate value @xmath4 @xmath61 studies varying @xmath4 @xcite new method extraction condensate fraction sadm @xmath41 avoid insufficient information exploiting data inclusive scattering @xmath62 same momentum transfer @xmath26 energy loss @xmath30 scaling variable @xmath29 recent data @xmath63 @xcite mari data [ a6a ] ) expresses @xmath15-independence er mari data ratio convoluted data replaced by deconvoluted ones @xmath65 depend on @xmath26 complex data complex [ a6b ] different nature assumes fsi function @xmath35 weakly @xmath15- dependentverified accuracy data @xcite cite weak @xmath15-dependence on measured pair - distribution function @xmath66 enters fsi function @xmath67 price @xmath68removing@xmath69 appearance two sadm instead desired @xmath5 precision condensate fraction depends quality input discuss latter @xmath65 uses not actual data ft substantial noise larger @xmath70 tails data @xmath25 avoid smoothing cutting low - intensity data affect accuracy @xmath65 apply method to data @xmath71 k @xcite chose representative sample highest quality @xmath72 extract ratio @xmath73/{\rm re}[\eta^{exp}(q , s)]$ ] determined range @xmath74 growing @xmath12 imaginary part increases from 0 medium @xmath75 only 2% real part neglected confirms ratio involved fsi functions @xmath35 virtually @xmath15-independent investigate @xmath26-dependence of @xmath76 show fig . 1 data individual @xmath26 data @xmath52 @xmath26 @xmath41 not uniformly smooth in @xmath26cause @xmath77-dependence in @xmath78 justifies average for @xmath79 @xmath80 test ( [ a6b ] ) computed results sadm @xmath84 @xcite outcome @xmath15=1.54 k closest to @xmath15=1.6 k data not close to values interpolation between @xmath15 suspect inaccuracies in non - vanishing condensate fractions for @xmath52 non - smooth dependence of @xmath4 function @xmath15 outside estimated theoretical error bars use interpolated results define @xmath85 @xmath86 function of @xmath87 entered fig . 1 data poorly use @xmath88 return to eq . ( [ a5a ] ) a6b becomes @xmath89 eq . ( [ a9 ] ) not applicable in asymptotic region constant numerator denominator ratio a9 tend to zero stringent demand eq . [ a9 ] ) provides value condensate fraction for @xmath90 interest consider range of @xmath12 right side of [ a9 ] ) function of @xmath12 left side constant consistency test for succesful extraction @xmath4size @xmath12-range depends on information functions in ( [ a9 ] ) delicate is deficit function @xmath54 warnings use ( [ a5b ] ) on model - independent results of ref . [ ] @xmath0 uncertainties 15% uncertainties decrease to @xmath92 increases for @xmath12 up to medium @xmath79 trust computed @xmath54 range data determine @xmath78 well consistency test ( [ a8 ] right hand side @xmath12-range weakly @xmath12-dependent leading to @xmath93 result standard extrapolation to @xmath13 small error limits due to averaging results for different @xmath12 estimate uncertainties in @xmath54 errors on three functions in ( [ a9 ] ) uncorrelated underscores feducity of extracted value @xmath4 numbers use @xmath94 interpolated between values @xmath15= 1.54,1.82k interpolate smoothly between values @xmath15=1.18, 1.82 k smoothing leads to lower condensate fraction accuracy depends on input older argonne data for @xmath95 temperatures around @xmath96actual data er functions analyze parametrizations given by authors double gaussians for data deconvoluted from er fsi reduces accuracy scarcity information performed analysis for argonne data compare with mari data fix @xmath52 at 2.3 k computed ft of parametrized data real ratios of ft for @xmath41 @xmath52 identical within 1% support ( [ a6 ] ) limited part of information used for analysis @xmath97 k @xmath4 nearly equal terms in ( [ a9 ] ) reliable determination requires precision on @xmath98 and @xmath92 beyond ceperley results difficulty for @xmath99 k uncertain extrapolation using lowest @xmath15 results outcome @xmath100 20 - 25% uncertainties exceed in ( [ a10 ] ) mari data @xmath101 k ) from mari argonne data agree parametrizations of argonne data attach significance to correct trend of @xmath0 as function of @xmath15 low average @xmath102=0.079 .suggested new method extract condensate fraction @xmath1he from fourier transforms data sets same momentum two @xmath15 below above @xmath96 method requires previously computed dynamical information expression ( [ a9 ] ) @xmath4 function @xmath12 - determined meaningful constant condensate fraction | 0.565415 | 3,511 | 1,875 | , Saving $0.1 in GPT-4. |
55,417 | the question of how the scattering cross section changes when the spectra of the colliding nuclei have low - excitation particle - emitting resonances is explored using a multi - channel algebraic scattering ( mcas ) method . as a test case , the light - mass nuclear target @xmath0be , being particle - unstable , has been considered .
nucleon - nucleus scattering cross sections , as well as the spectra of the compound nuclei formed , have been determined from calculations that do , and do not , consider particle emission widths of the target nuclear states .
the resonant character of the unstable excited states introduces a problem because the low - energy tails of these resonances can intrude into the sub - threshold , bound - state region .
this unphysical behaviour needs to be corrected by modifying , in an energy - dependent way , the shape of the target resonances from the usual lorentzian one .
the resonance function must smoothly reach zero at the elastic threshold .
ways of achieving this condition are explored in this paper . | the advent of radioactive ion beam ( rib ) physics prompts the consideration of new theoretical challenges involving weakly - bound systems .
radioactive nuclei , especially those close to the drip lines , can have quite low particle emission thresholds and consequently have low - lying resonance states in their spectra .
low - energy scattering of rib is a coupled - channel problem that involves such low - lying resonant states of the scattered ion . to address this , a multi - channel algebraic scattering ( mcas ) formalism @xcite
is used , in which momentum space solutions of coupled lippmann - schwinger equations are found , including also all negative - energy bound states . a finite - rank separable representation using an `` optimal '' set of sturmian functions @xcite serves to construct an input matrix of nucleon - nucleus interactions .
the mcas method has the ability to locate all compound system resonance centroids and widths , regardless of how narrow . for full details
see reference @xcite .
also , by use of orthogonalizing pseudo - potentials ( opp ) in generating sturmians , it ensures the pauli principle is not violated @xcite , despite the collective model formulation of nucleon - nucleus interactions used .
otherwise , some compound nucleus wave functions possess spurious components . in the following sections
, we first briefly summarize the mcas method @xcite , which has been extended to unstable target states in ref .
@xcite . in section
ii , we display and discuss our results for neutron scattering from @xmath0be , with and without the excited target states having non - zero widths . in section iii
, we look briefly at the implications of letting the target - state widths have energy dependence , to ensure that these resonances do not extend to negative energies , where they would produce unphysical behaviour for bound , sub - threshold , states . finally , section iv gives our brief concluding remarks .
@xmath1-matrix equations in the mcas methodology take the form : @xmath2^{-1 } \right)_{nn ' } \hat{\chi}_{c'n ' } \sqrt{k_{c'}}\ ; . \label{smx}\ ] ] traditionally , all target states are taken to have eigenvalues of zero width .
then the integrals in the ( complex ) green s functions are evaluated using the method of principal parts , where , in the limit @xmath3 , the green s functions take the form @xmath4_{nn ' } = \mu \left [ \sum^{\text{open}}_{c = 1 } \int^{\infty}_0 \hat{\chi}_{cn}(x ) \frac{x^2 } { k^2_c - x^2 + \color{red}\textrm{i}\epsilon\color{black } } \hat{\chi}_{cn'}(x ) \ , dx \right . - \left .
\sum^{\text{closed}}_{c = 1 } \int^{\infty}_{0 } \hat{\chi}_{cn}(x ) \frac { x^2 } { h^2_c + x^2 } \hat{\chi}_{cn'}(x ) \ , dx \right ] .
\label{s - g}\ ] ] is a column vector of sturmian eigenvalues and @xmath5 are form factors determined from the chosen sturmian functions .
the factor @xmath6 in eqs .
[ smx ] and [ s - g ] is @xmath7 , with @xmath8 being the reduced mass .
this method assumes time evolution of target states is given by @xmath9 however , if states decay , they evolve as @xmath10 thus , in the green s function , channel energies become complex , as do the squared channel wave numbers , @xmath11 where @xmath12 is half the width of the target state associated with channel @xmath13 .
the green s function matrix elements are then @xmath4_{nn ' } = \mu \left [ \sum^{\text{open}}_{c = 1 } \int^{\infty}_0 \hat{\chi}_{cn}(x ) \frac{x^2 } { k_c^2 - x^2 + \color{red}\textrm{i } \mu { \textstyle\frac{\gamma_c } { 2 } } \color{black } } \ ; \hat{\chi}_{cn'}(x ) \ , dx \right
. - \left .
\sum^{\text{closed}}_{c = 1 } \int^{\infty}_{0 } \hat{\chi}_{cn}(x ) \frac { x^2 } { h_c^2 + x^2 - \color{red}\textrm{i } \mu { \textstyle\frac{\gamma_c } { 2 } } \color{black } } \ ; \hat{\chi}_{cn'}(x ) \ , dx \right ] .
\label{s - ggamma2}\ ] ] when poles are moved significantly off the real momentum axis , integration along this axis is feasible .
the low excitation spectrum of @xmath0be has a @xmath15 ground state that decays into two @xmath16-particles ( width : @xmath17 mev ) , a @xmath18 resonance state ( centroid : 3.03 mev ; width : 1.5 mev ) and a @xmath19 resonance state ( centroid : 11.35 mev ; width : @xmath20 mev ) . in table
[ be8-param ] are given the parameters in the channel - coupling potentials , as well as the @xmath21 values giving the strengths of pauli hindrance .
[ ht ] > p28 mm p27 mm < p27 mm < & odd parity & even parity + @xmath22 ( mev ) & -31.5 & -42.2 + @xmath23 ( mev ) & 2.0 & 0.0 + @xmath24 ( mev ) & 12.0 & 11.0 + @xmath25 ( mev ) & -2.0 & 0.0 + >
p19.5 mm p21 mm < >
p21 mm p19 mm < geometry & @xmath26 fm & @xmath27 fm & @xmath28 + > p28 mm p27 mm < p27 mm < & @xmath29 & @xmath30 + @xmath31 & 1000 & 0.0 + @xmath32 & 2.0 & 0.2 + @xmath33 & 0.0 & 0.0 + two evaluations of the @xmath34be cross section are obtained using this spectrum ; one with target - state widths set to zero , and the other taking into account the widths of the excited levels .
the two results will be identified by the terms ` no - width ' and ` width ' , respectively .
both cases are calculated with the same nuclear interaction , taken from a rotor model @xcite .
table [ tab2 ] displays @xmath14be spectrum results from these calculations .
.[tab2 ] widths of resonances in @xmath35 scattering for the no - width and width cases , compared to experimental values .
boldface entries highlight the matches we find between theory and experiment to within a factor of 3 . [
cols="^,^,^,^,^,^",options="header " , ] taking the excited states of @xmath0be to be resonances gives the same spectral list as when they are treated as zero - width , but the evaluated widths of the compound nuclear states significantly increase , as reflected in the cross sections .
these increases bring the theoretical @xmath14be state widths closer , often significantly , to experimental values , as evidenced by comparison of the final columns of the two tables .
only the @xmath36 , also poorly recreated in centroid , has a worse match with widths applied .
the resultant elastic and reaction cross sections are shown in figure [ fig1 ] .
the blue line depicts the results without widths , the red line with constant widths multiplied by the heaviside ( step ) function : @xmath37 for @xmath38 ; @xmath39 for @xmath40 ( for the green line , see section [ scaling ] . ) introducing target state widths , the resonances in @xmath14be are suppressed but still present ; their widths increasing and magnitudes decreasing so as all but the @xmath41 , and arguably @xmath42 , can not be discerned from the background .
the reaction cross section found in the no - width case is zero until 3.4 mev ( 3.03 mev in the centre of mass frame ) , the first inelastic threshold .
when the target state has width different from zero the elastic cross section has flux loss from zero projectile energy upwards .
there is an asymptotic behaviour as energy approaches zero . in the reaction cross section from the width calculation
, there is a broad peak at 1.5 mev , corresponding to the @xmath43 resonance .
using widths not varying with energy ( see the next section ) , the bound states , @xmath44 , of the compound @xmath14be system are unstable due to the couplings to the decaying excited states of the @xmath0be core .
for this reason the widths have been multiplied by a heaviside function to ensure that the compound system is stable at negative energies .
an improvement of this technique is discussed in the next section .
a smooth description of the energy dependence of the decay width is needed to avoid any pathological behaviour caused by the heaviside function .
this function allows maintaining the stability of the compound system in the sub - threshold region , by setting to zero all widths for @xmath45 , but can generate `` ghost '' levels where the application of widths could move a sub - threshold state above @xmath46 , thus creating two copies of the same state .
also , a dramatic overestimation of the reaction cross - section occurs in the region close to threshold .
we therefore need to improve the model with a suitable parametrization of the energy - dependence of the decaying width for the excited levels , with the constraint that the width must be zero at the scattering threshold , and equal to the experimental width at the resonance centroid energy .
we are presently exploring various forms of energy dependences for the widths of these excited states ( @xmath47 ) .
examples of such expressions are @xmath48 where @xmath49 , @xmath50 is the heaviside function defined above and @xmath51 is a parameter , and the gaussian form @xmath52 in this paper we study only the first form . the second , and others , will be explored in future work .
also yet to be taken into account is causality , as needed when dealing with energy - dependent widths @xcite . in figure [ fig3 ]
we give the induced energy dependence for the function @xmath53 with respect to the renormalized energy @xmath54 .
the curves differ for different values of the parameter @xmath51 .
the green line in figure [ fig1 ] shows the effect of this energy - dependent width on elastic and reaction cross sections , respectively , when @xmath55 .
clearly , for this value of @xmath51 , the asymptotic behaviour in the reaction cross section at @xmath56 is suppressed , but not resolved .
the use of an energy dependence in the widths of the excited states of the target introduces a new aspect that has to be taken into consideration . in principle , an energy dependence in the widths induces a corresponding energy - dependence in the positioning of the resonance centroids .
this additional effect is a direct consequence of the principle of causality which demands that the real and imaginary part of any green s function matrix elements are related by a dispersion integral .
such an aspect has been extensively discussed in the study of the properties of the nuclear optical potential @xcite .
eq . [ rational ] is particularly convenient in this regard , because the corresponding dispersion integral leads to an analytical energy - dependent shift of the resonance centroids . such analytical forms
have been derived in ref .
@xcite for a class of functions which includes eq .
[ rational ] .
the resulting energy shift function @xmath57 is given by the analytical expression : @xmath58 where @xmath59 and @xmath60
an extension to the multi - channel algebraic scattering ( mcas ) formalism that considers particle - decay widths of target nucleus eigenstates has been applied to a range of light mass nuclear targets @xcite , in addition to the results for @xmath34be shown in this paper . to ensure that any resonance aspect of target states does not have influence below the nucleon - nucleus threshold in our formalism ,
we have used a heaviside function to cut off tails of the positive - energy resonances below zero energy of the compound system .
this procedure , however , is too simplistic since it introduces non - physical singularities in cross sections at the approach to zero energy , and may also incorrectly overestimate reaction cross just above the threshold . a better energy dependent scaling factor is needed : one with value one at the centroid energy and approaching zero for lower and higher energies .
we propose here possible means of achieving such modified lorentzians , and show , for one example , their effect on the elastic and reaction cross sections in the scattering of neutrons from @xmath0be .
work is in progress to apply these and other forms to the light - nuclear systems considered in our previous work .
work is in progress , also , for preserving causality in the presence of energy - dependent widths by the use of appropriate dispersion relations .
this work is supported by the national research foundation , south africa , the natural sciences and engineering research , council , canada , infn , italy and the australian academy of science .
99 amos k , canton l , pisent g , svenne j p and van der knijff d 2003 _ nucl .
phys . _ * a728 * 6595 rawitscher g h and canton l 1991 _ phys .
_ c * 44 * 6066 canton l , pisent g , svenne j p , van der knijff d , amos k and karataglidis s 2005 _ phys . rev .
lett . _ * 94 * 122503 fraser p , amos k , canton l , pisent g , karataglidis s , svenne j p and van der knijff d 2008 _ phys .
lett . _ * 101 * 242501 fraser p , amos k , canton l , karataglidis s , svenne j p and van der knijff d 2010 submitted to _
revista mexicana de fisica _ in review ; arxiv:1005.5351 [ nucl - th ] ( 2010 ) mahaux c , ng h and satchler g r 1986 _ nuclear physics _ * a449 * 354394 cattapan g , canton l , pisent g 1991 _ phys . rev .
_ c * 45 * 1395 - 1407 | radioactive ion beam physics prompts new theoretical challenges weakly - bound systems radioactive nuclei close to drip lines low particle emission thresholds low - lying resonance states low - energy scattering of rib coupled - channel problem low - lying resonant states multi - channel algebraic scattering ( mcas ) formalism @xcite used momentum space solutions of lippmann - schwinger equations including negative - energy bound states finite - rank separable representation optimal sturmian functions input matrix of nucleon - nucleus interactions mcas method all compound system resonance centroids widths @xcite orthogonalizing pseudo - potentials in generating sturmians ensures pauli principle not violated some compound nucleus wave functions possess spurious components summarize mcas method @xcite extended to unstable target states section ii results for neutron scattering from @xmath0be with without target states non - zero widths implications of target - state widths energy dependence resonances extend to negative energies unphysical behavioursection iv concluding remarks @xmath1-matrix equations mcas methodology form @xmath2^{-1 }_{nn {\chi}_{c'n\sqrt{k_{c'}}{smx} target states eigenvalues zero width integrals green s functions evaluated method principal parts limit @xmath3 green s functions form @xmath4_{nn ' } {\text{open}}_{c = 1 }{\chi}_{cn}(x ) \frac{x^2 } k^2_c - x^2 +{red}{black_{cn'}(x ){\text{closed}}_{c = 1{\infty}{}_{cn}(x ) \frac { x^2 } h^2_c + x^2 }{s - g} column vector of sturmian eigenvalues @xmath5 form factors chosen sturmian functions factor @xmath6 s - g is @xmath7 @xmath8 reduced massmethod assumes time evolution of target states given by @xmath9 if states decay evolve as @xmath10 green s function channel energies become complex squared channel wave numbers @xmath11 @xmath12 half width of target state channel @xmath13 green s function matrix elements are @xmath4_{nn = \mu \left\text{open}}{c = 1{\chi}_{cn}(x ) \frac{x^2 } k_c^2 - x^2 +{red}\textrm{i\gamma_c 2{black(x\right\left{closed}}{c = 1\chi}_{cn}(x ) \frac { x^2 } h_c^2 + x^2 -{red}\textrm{i\gamma_c 2{black{cn'}(x\right when poles moved off momentum axis integration along axis feasiblelow excitation spectrum @xmath0be @xmath15 ground state @xmath16-particles width mev @xmath18 resonance state centroid 3.03 mev width 1.5 mev @xmath19 resonance state centroid 11.35 mev width mev table be8-param channel - coupling potentials @xmath21 values strengths pauli hindrance p28 p27 mm odd parity even parity @xmath22 -31.5 -42.2 @xmath23 2.0 0.0 @xmath24 12.0 11.0 @xmath25 -2.0 0.0 p19.5 mm p21 mm p19 mm @xmath26 @xmath27 fm @xmath28 p28 p27 mm @xmath29 @xmath30 @xmath31 1000 0.0 @xmath32 2.0 0.2 @xmath33 0.0 two evaluations @xmath34be cross section target - state widths zero widths excited levels results identified no - width width ' cases calculated same nuclear interaction rotor model @xcite table tab2 displays @xmath14be spectrum results widths resonances @xmath35 scattering no - width width cases compared experimental values boldface entries highlight matches theory experiment factor 3 cols="^ excited states @xmath0be resonances same spectral list zero - width evaluated widths compound nuclear states increase cross sections increases bring theoretical @xmath14be state widths closer to experimental values evidenced comparison final columns tables @xmath36 worse match with widths elastic reaction cross sections shown in figure blue line results without widths red line constant widths multiplied by heaviside function @xmath37 @xmath38 @xmath39 @xmath40 green line section scaling introducing target state widths resonances in @xmath14be suppressed present widths increasing magnitudes decreasing @xmath41 @xmath42 discerned background reaction cross section no - width case zero until 3.4 mev ( 3.03 mev first inelastic threshold target state width different from zero elastic cross section flux loss from energy asymptotic behaviour as energy approaches zero peak at 1.5 mev @xmath43 resonance widths not varying with energy bound states @xmath44 @xmath14be system unstable due to couplings decaying excited states @xmath0be corewidths multiplied by heaviside function compound system at negative energies improvement technique discussed next section smooth description of energy dependence decay width needed avoid pathological behaviour heaviside function stability in sub - threshold region zero widths for @xmath45 generate ghost levels sub - threshold state above @xmath46 creating two copies state overestimation of reaction cross - section occurs in close to threshold need improve model with parametrization energy - dependence decaying width for excited levels width zero at scattering threshold equal to experimental width at resonance centroid energy exploring energy dependences for widths of excited states ( @xmath47 ) examples expressions are @xmath48 @xmath49 @xmath50 @xmath51 gaussian form @xmath52 first form second explored in future work causality energy - dependent widths figure fig3 induced energy dependence for function @xmath53 renormalized energy @xmath54 curves differ for values of parameter @xmath51 green line shows effect of energy - dependent width on elastic reaction cross sections when @xmath55value @xmath51 asymptotic behaviour reaction cross section @xmath56 suppressed not resolved energy dependence in widths excited states introduces new aspect energy dependence widths induces energy - dependence positioning resonance centroids principle causality real imaginary green s function matrix elements related by dispersion integral discussed nuclear optical potential @xcite eq . [ rational ] convenient dispersion integral leads to energy - dependent shift resonance centroids analytical forms derived in @xcite functions eq . [ rational ] resulting energy shift function @xmath57 given expression @xmath58 @xmath59 @xmath60 multi - channel algebraic scattering formalism particle - decay widths target nucleus eigenstates applied to light mass nuclear targets @xcite results @xmath34be resonance target below nucleon - nucleus threshold used heaviside function to cut off positive - energy resonances below zero energy compound procedure simplistic introduces non - physical singularities in cross sections at zero energy overestimate reaction cross above threshold better energy dependent scaling factor needed value one at centroid energy approaching zero for lower higher energiespropose means achieving modified lorentzians show effect elastic reaction cross sections scattering neutrons @xmath0be work progress apply forms light - nuclear systems work preserving causality energy - dependent widths dispersion relations supported national research foundation south africa natural sciences engineering research council canada infn italy australian academy science amos k canton l pisent g svenne j p van der knijff d 2003 nucl . . a728 * 6595 rawitscher g canton l 1991 * 44 * 6066 karataglidis s 2005 . rev . 94 * 122503 fraser p amos canton l van der knijff d 2008 phys 101 * 242501 2010 revista mexicana de fisica review arxiv:1005.5351 nucl - th 2010 mahaux c ng h satchler g r 1986 nuclear physics a449 * 354394 cattapan g canton l pisent g 1991 phys . rev . * 45 * 1395 - 1407 | 0.505715 | 3,257 | 1,689 | , Saving $0.1 in GPT-4. |
55,424 | the structural and electronic properties in common - anion + ( gasb)@xmath0/(insb)@xmath0 and common - cation ( inas)@xmath0/(insb)@xmath0 [ 111 ] ordered superlattices have been determined using the local density total energy full potential linearized augmented plane wave method .
the influence of the ordering direction , strain conditions and atomic substitution on the electronic properties of technological and experimental interest ( such as energy band - gaps and charge carrier localization in the different sublattices ) were determined .
the results show an appreciable energy band - gap narrowing compared to the band - gap averaged over the constituent binaries , either in [ 001 ] ordered structures or ( more markedly ) in the [ 111 ] systems ; moreover energy band - gaps show an increasing trend as the substrate lattice parameter is decreased .
finally , the systems examined offer interesting opportunities for band - gap tuning as a function of the growth condition ( about 0.7 ev in ( gasb)@xmath0/(insb)@xmath0 and 0.3 ev in ( inas)@xmath0/(insb)@xmath0 ) . | ternary systems based on iii - v semiconductors ( such as disordered alloys @xcite , heterostructures @xcite or quantum well systems @xcite ) have been the subject of wide scientific interest and of accurate theoretical studies , since they could be used as fundamental components in a large class of important devices ( laser diodes or infrared detectors , to name just a few ) @xcite . in the present work ,
we focus our attention on superlattices ( sl ) , whose structural , electronic and transport properties can be opportunely tuned by varying the constituent materials , the strain , the ordering direction or the layers thickness . to this end
, we have examined the properties of interest in ultrathin [ 111 ] ordered sl , specifically in common - anion ( gasb)@xmath0/(insb)@xmath0 systems and in common - cation ( inas)@xmath0/(insb)@xmath0 systems , using the self - consistent all - electron flapw method within the density functional formalism .
the systems considered here are now under experimental investigation ( results obtained for strained - layer insb / gasb quantum well @xcite , in@xmath1ga@xmath2sb / gasb heterostructures @xcite and inas@xmath2sb@xmath3 alloys @xcite have already been published ) ; at the present time , however , we are not yet able to compare our predicted results with experimental values regarding the sl .
the appreciable mismatch between the lattice parameters of the binary constituents ( 5.7 @xmath4 in ( gasb)@xmath0/(insb)@xmath0 and 6.4 @xmath4 in ( inas)@xmath0/(insb)@xmath0 ) gives the opportunity to study the effects of the strain conditions on the sl electronic properties .
in analogy with the common experimental approach @xcite we have considered various growth conditions for the sl , leading to different strain modes in the structure : ( i ) pseudomorphic growth on a substrate usually constituted by one of the binary constituents ( in which the lattice constant parallel to the growth plane is taken equal to that of the bulk semiconductor composing the substrate ) and ( ii ) a `` free standing mode '' ( fsm ) , in which no constraints are imposed on the bond - lengths , leading to relaxed lattice constants for the binary constituents , both different from their bulk values .
the structural parameters for all of the structures considered have been chosen through total energy minimization or according to the macroscopic theory of elasticity ( mte ) @xcite .
the choice made towards the [ 111 ] direction is encouraged by recent experimental observations @xcite that suggested the spontaneous ordering along this particular axis ( the so called cupt structure ) shown by some iii - v alloys during vapour - phase growth .
the strong influence of the ordering direction on the sl electronic properties is immediately clear if we consider the ternary sl brillouin zones ( bz ) as obtained from the binary zincblende bz through folding operations , which are obviously different for [ 111 ] or [ 001 ] ordered systems ; the immediate consequence of this is a noticeable difference in the electronic properties . in order to study the dependence of the electronic properties on the ordering direction
, we have studied the common - anion and the common - cation systems in the three different growth conditions both in the cuau - like ( having [ 001 ] direction as growth axis ) and cupt - like structure . following the model proposed by wei and zunger @xcite
, we consider the sl as obtained first from an ideal virtual crystal ( a common - anion ( common - cation ) system having the cation ( anion ) with intermediate properties between the two cations relative to the binary constituents ) .
we then introduce a perturbative potential , having a structural part ( due to atomic displacements and relaxation due to epitaxial strain ) and a chemical one ( due to the electronegativity difference between the constituent atoms ) . in order to separate the effects due to the two different terms in the expression of the potential ,
we have studied strained ( ac)@xmath0/(ac)@xmath0-type systems ideally obtained by monolayer deposition of the same binary constituent ( ac ) along the [ 111 ] direction , in which the two different ac bond - lengths are equal to those in the equivalent ( ac)@xmath0/(bc)@xmath0 sl .
we will discuss the results obtained in this work as follows : first of all , we will briefly expose the computational details and parameters used in the calculations ( section ii ) ; in section iii our structural results will be reported for all the different systems considered . in section iv we will discuss the electronic properties of the sl , with particular attention to the quantities of technological interest ( such as energy band - gaps and crystal field splittings ) and their dependence on growth direction , atomic substitution and strain conditions .
we will also discuss the charge density distribution and in particular the localization of the charge carriers in the different constituents sublattices .
section v summarizes our main results and draws some conclusions .
we have determined the properties of the structures considered using the density functional formalism , within the local density approximation ( lda ) @xcite exchange and correlation potential as parametrized by hedin - lundqvist @xcite .
the calculations were performed using the _ ab initio _ all - electron full - potential linearized augmented plane wave method ( flapw ) @xcite .
core electrons as well as valence ones are treated using a self - consistent procedure ; the shallow ga @xmath5 and in @xmath6 states are considered as valence states , for which scalar - relativistic effects are included in the self - consistent calculation , whereas spin - orbit effects are treated in a perturbative approach . for
[ 111 ] ordered systems , angular momenta up to @xmath7 = 6 in the muffin tin spheres ( with radius @xmath8 = 2.4 a.u . for all the constituents atoms ) and plane waves with wave vector up to @xmath9 = 3.3 a.u . are used , leading to about 600 basis functions . to perform integrations in reciprocal space , a set of four special * k
* points is chosen in the trigonal brillouin zone ( bz ) , following the monkhorst - pack scheme @xcite .
similar values for these computational parameters have been used for [ 001 ] ordered systems , with the only exceptions represented by @xmath7 = 8 and a set of three special * k * points used for the integration over the tetragonal brillouin zone .
finally , the broyden @xcite method is used to accelerate the convergence in the self - consistent iterations .
the atomic ordering along the [ 111 ] direction of a sl grown on a ( 111 ) substrate gives origin to a trigonal bravais lattice with @xmath10 ( schenflies notation ) space group @xcite .
the unit cell in real space contains 4 atoms and the origin is taken on a cation site @xcite .
the free structural parameters are determined following the macroscopic theory of elasticity ( mte ) @xcite , taking into account the elastic properties of the constituent materials , and then compared with those obtained through total energy minimization .
we observe that in each cell , there are two atoms belonging to the same chemical species ( the two sb anions in the ( gasb)@xmath0/(insb)@xmath0 systems and the two in cations in the ( inas)@xmath0/(insb)@xmath0 systems ) , which are not equivalent from the coordination point of view . as an example
, we consider the particular case of ( gasb)@xmath0/(insb)@xmath0 sl , having the first sb@xmath11 bound with three ga and one in , and the second sb@xmath12 showing a complementary situation .
the total energy minimization procedure that considers all the free parameters in the unit cell as different degrees of freedom is a very onerous computational problem ; this encouraged some simplifications , such as considering equal bond lengths between equal atomic species ( _ i.e. , _ in the common - anion system , we have chosen @xmath13 ) , thus reducing to three the number of degrees of freedom ( in - plane lattice parameter and two different bond - lengths ) .
although this simplification is frequently used in total energy minimization @xcite , one should be aware that this approximation results in considering an average over the two different local environments and therefore that the real elastic structures may slightly differ from our optimized sls .
according to mte , using the same notation as in ref.@xcite , the structural parameters for the epilayer are determined as follows : @xmath14 } \:\left(\frac{a_{\parallel}}{a^{epi}}-1 \right)\right]\\ \label{d111 } d^{[111 ] } & = & 2\ : \left ( \frac{c_{11}+2\:c_{12}-2\:c_{44}}{c_{11}+2\ : c_{12}+4\:c_{44 } } \right)\\ \nonumber \epsilon^{epi}_{\parallel } & = & \frac{a_{\parallel}}{a^{epi}}-1 \\
\nonumber \epsilon^{epi}_{\perp } & = & \frac{a_{\perp}^{epi}}{a^{epi}}-1\end{aligned}\ ] ] where @xmath15 are the elastic constants for the bulk epilayer ( we have used the experimental values reported in ref.@xcite ) . through total energy minimization of the `` ideal '' ( ac)@xmath0/(bc)@xmath0 unrelaxed structures ( in which all the atoms are arranged in a zincblende structure with lattice constant @xmath16 - our free parameter - and with bond - lengths @xmath17 ) , we have found an in - plane lattice constant very close to the average of the bulk constituents , according to vegard s rule .
we have thus examined a free standing mode structure ( indicated in the following as elastically relaxed or simply er ) , that has this value for the in - plane lattice constant . in table
[ parstruttan ] and table [ parstruttcat ] we report the calculated structural parameters for the ternary common - anion and common - cation systems : the s1 ( s2 ) system is a common - anion sl grown on a gasb ( insb ) substrate , while the s3 ( s4 ) system is a common - cation sl grown on an inas ( insb ) substrate . in the case of pseudomorphic growth on a substrate ,
we have found general agreement between the structures obtained through total energy minimization and those given by mte ; this fully justifies our having considered this approximation to determine the five unknown parameters in the free standing mode ( fsm ) structure .
note that deviations from the results predicted by mte occur in the case of insb strained to gasb or to inas ; in both these structures , total energy minimization gives an insb bond - length that is systematically larger ( within 0.6 @xmath4 ) than the one expected according to mte , even though the difference between the total energies for the elastic sl and for the total energy minimal structures is very small ( barely larger than the numerical uncertainty of 1 mry / unit cell ) .
however , this can be justified considering that , due to its elastic properties , this material could easily be out of the linear elastic region .
in fact , the elastic constants for insb are quite smaller than those for gasb and inas @xcite , resulting in a larger effective strain - due to the mismatch - in the former case ; this is also confirmed by the non - linear behaviour of the band - gap as a function of the strain , as will be discussed later .
as expected from the similarity of the gasb and inas elastic constants @xcite and bulk moduli ( the experimental values are @xmath18 = 0.579 mb and @xmath19 = 0.578 mb @xcite ) , we obtain similar deformations for these two constituents respectively in the common - anion and common - cation sls .
we also notice that strains ( either parallel or perpendicular ) and percentage deviations from bulk bond - lengths are more pronounced in the common - cation systems , compared to the common - anion systems : this is obviously a consequence of the greater mismatch between the constituent lattice parameters in the ( inas)@xmath0/(insb)@xmath0 structures . in the case of [ 001 ] ordered systems ( whose
structural parameters are reported in table [ parstruttan ] and table [ parstruttcat ] respectively for common - anion and common - cation sl ) , we obtain a tetragonal bravais lattice with @xmath20 space group @xcite and a unit cell in real space with 4 atoms ( two of which are equivalent ) ; the origin is taken on a cation ( anion ) site for the common - anion ( common - cation ) system .
the mte relations reported in eqs.([d111 ] ) are still valid for [ 001 ] ordered sl , if the parameter @xmath21 is redefined as : @xmath22 } = 2\ : \left ( \frac{c_{12}}{c_{11 } } \right ) .
\nonumber\ ] ] a comparison between the [ 111 ] and [ 001 ] ordered structures ( having the same chemical composition ) shows that in the same growth conditions ( fsm or growth on a substrate ) the structural parameters are not equal .
in particular we notice larger perpendicular strains in the [ 001 ] compared to the [ 111 ] ordered structures , while the parallel strains are obviously equal in considering the same growth conditions ; what we find is thus a smaller deviation from bulk bond - lengths , due to a more effective relaxation .
the determination of the sl electronic energy levels is a fundamental point for most of the properties of interest in the systems considered . in table
[ sllivris ] we report the calculated electronic levels ( with a numerical uncertainty of @xmath23 0.04 ev , equal for all the energies reported in the present work , unless otherwise specified ) at the bz center ( @xmath24 ) , for the different [ 111 ] ordered systems considered ( free standing mode and pseudomorphic growth on the two substrates ) , both for the common - anion and for the common - cation systems .
we also report the zincblende state from which the sl state derives through folding the f.c.c .
brillouin zone back into the smaller ternary trigonal zone .
the splitting , @xmath25 , of the triply degenerate @xmath26 zincblende state is due to the non - cubic crystal field and is conventionally taken positive if the doubly degenerate state @xmath27 has a higher energy compared to the state @xmath28 . as can be seen from table [ sllivris ]
, we obtain a negative @xmath25 in the case of gasb ( inas ) strained to insb for common - anion ( cation ) systems - corresponding to an in - plane extensive strain @xmath29 - whereas the complementary case ( pseudomorphic growth on a gasb - substrate ( inas - substrate ) ) and the free standing mode produce a positive @xmath25 .
the introduction of spin - orbit coupling removes the double degeneracy of the @xmath27 state and yields the electronic energy levels illustrated in fig.[elenlev ] as a function of the substrate lattice constant .
the topmost valence bands ( e@xmath0 , e@xmath30 , e@xmath31 ) have been labeled according to the quasi - cubic " model @xcite ( taking the centre of gravity of the sl valence bands as zero ) ; | systems on iii - v semiconductors disordered alloys heterostructures quantum well systems subject of scientific interest theoretical studies fundamental components in important devices laser diodes infrared detectors present work on superlattices sl structural electronic transport properties tuned by varying constituent materials strain ordering direction layers thickness examined properties in ultrathin [ 111 ] ordered sl common - anion gasb)@xmath0/(insb) common - cation inas)@xmath0/ systems using self - consistent all - electron flapw method density functional formalism systems under experimental investigation results for strained - layer insb / gasb quantum well@xmath2sb / heterostructures inas@xmath2sb@xmath3 alloys published not compare predicted results with experimental values mismatch between lattice parameters binary constituents ( 5.7 @xmath4 6.4 @xmath4 in/ opportunity to study effects strain conditions on sl electronic propertiesanalogy with experimental approach @xcite considered growth conditions for sl different strain modes pseudomorphic growth on substrate by binary constituents lattice constant parallel growth plane equal to bulk semiconductor free standing mode no constraints on bond - lengths relaxed lattice constants for binary constituents different from bulk values structural parameters for structures chosen through energy minimization or macroscopic theory of elasticity choice towards [ 111 ] direction encouraged by observations spontaneous ordering along axis cupt structure by iii - v alloys during vapour - phase growth influence of ordering direction on sl electronic properties clear ternary sl brillouin zones obtained from binary zincblende bz folding different for [ 111 ] or [ 001 ] ordered systems noticeable difference in electronic properties study dependence properties ordering direction studied common - anion common - cation systems in three growth conditions cuau - like direction cupt - like structure following model by wei and zunger @xcite consider sl as obtained from ideal virtual crystal common - anion system intermediate properties binary constituentsintroduce perturbative potential structural part to atomic displacements relaxation epitaxial strain chemical electronegativity difference atoms separate effects studied strained ac)@xmath0/(ac)@xmath0-type systems monolayer deposition same binary constituent ac ) 111 ] direction two ac bond - lengths equal to equivalent ( ac)@xmath0/(bc)@xmath0 sl discuss results expose computational details parameters calculations section ii ) section iii structural results reported systems section iv discuss electronic properties sl attention technological interest energy band - gaps crystal field splittings on growth direction atomic substitution strain conditions discuss charge density distribution localization charge carriers constituents sublattices section v summarizes results conclusions determined properties structures using density functional formalism local density approximation ) @xcite exchange correlation potential parametrized hedin - lundqvist @xcite calculations all - electron full - potential linearized augmented plane wave method core electrons valence ones treated self - consistent procedure shallow ga @xmath5 @xmath6 states valence states scalar - relativistic effects included self - consistent calculation spin - orbit effects treated perturbative approach[ 111 ] ordered systems angular momenta @xmath7 = 6 muffin tin spheres radius @xmath8 = 2.4 a.u constituents atoms plane waves vector @xmath9 = 3.3 a.u . 600 basis functions integrations reciprocal space four special * k * points trigonal brillouin zone monkhorst - pack scheme similar values [ 001 ] ordered systems exceptions @xmath7 = 8 three special * k * points integration tetragonal brillouin zone broyden @xcite method convergence self - consistent iterations atomic ordering [ 111 ] direction 111 ) substrate origin trigonal bravais lattice @xmath10 space group @xcite unit cell real space contains 4 atoms origin cation site @xcite free structural parameters determined macroscopic theory elasticity ) @xcite elastic properties constituent materials compared total energy minimization each cell two atoms same chemical species sb anions gasb)@xmath0/(insb) in cations not equivalent coordinationexample consider case gasb)@xmath0/(insb)@xmath0 sl first sb@xmath11 bound three ga one in second sb@xmath12 complementary situation total energy minimization procedure considers free parameters different degrees freedom onerous encouraged simplifications considering equal bond lengths between atomic species common - anion system chosen @xmath13 ) reducing to three degrees of freedom plane lattice parameter two bond - lengths simplification used energy minimization @xcite average over two local environments real elastic structures may differ from optimized sls mte same notation ref.@xcite structural parameters epilayer determined @xmath14\left(\frac{a{\parallel}}{a{epi}}-1\right\label{d111{[111 2\\left\frac{c_{11}+2_{12}-2_{44}}{11}+2 c_{12}+4_{44 }\right\nonumber\epsilon^{epi}_{\parallel }\frac{a_{\parallel}}{a{epi}}-1\nonumber\epsilon{epi}_{\perp }\frac{a\perp}^{epi}}{a^{epi}}-1{aligned} @xmath15 elastic constants bulk epilayer used experimental values ref.@xcite energy minimization ideal unrelaxed structures atoms zincblende structure lattice constant @xmath16 bond - lengths @xmath17 found in - plane lattice constant close average bulk constituents vegard s rule examined free standing mode structure elastically relaxed value in - plane lattice constanttable [ parstruttan parstruttcat ] report calculated structural parameters for ternary common - anion common - cation systems s1 ( s2 ) system common - anion sl on gasb ( insb ) substrate s3 s4 ) system common - cation sl inas ( insb ) substrate pseudomorphic growth agreement between structures total energy minimization mte justifies approximation five unknown parameters in free standing mode fsm ) structure deviations from results mte occur insb strained to gasb or inas total energy minimization gives insb bond - length larger ( 0.6 @xmath4 ) than mte difference between total energies elastic sl total energy minimal structures small barely than uncertainty 1 mry / unit cell ) elastic properties material could be out of linear elastic region elastic constants for insb smaller than gasb and inas @xcite larger effective strain confirmed by non - linear behaviour of band - gap strain similarity gasb inas elastic constants bulk moduli experimental values similar deformations for constituents in common - anion and common - cation slsstrains percentage deviations from bulk bond - lengths pronounced in common - cation systems anion mismatch between constituent lattice parameters in)@xmath0/(insb)@xmath0 structures [ 001 ] ordered systems structural parameters reported in table parstruttan parstruttcat for common - anion cation tetragonal bravais lattice with @xmath20 space group @xcite unit cell in space with 4 atoms two equivalent origin on cation ( anion ) site for common - system mte relations in eqs.([d111 ] valid for [ 001 ] ordered sl if parameter @xmath21 redefined as @xmath22 } = 2\ : \left ( \frac{c_{12}}{c_{11 \right comparison between [ 111 ] [ 001 ] ordered structures shows same growth conditions structural parameters not equal larger perpendicular strains in [ 001 ] parallel strains equal smaller deviation from bulk bond - lengths due to effective relaxation determination of sl electronic energy levels fundamental for propertiestable [ sllivris ] calculated electronic levels uncertainty @xmath23 0.04 ev equal energies bz center @xmath24 ) different 111 ordered systems free standing mode pseudomorphic growth substrates common - anion common - cation systems report zincblende state derives folding f brillouin zone ternary trigonal zone splitting @xmath25 triply degenerate @xmath26 zincblende state due non - cubic crystal field positive if doubly degenerate state @xmath27 higher energy @xmath28 negative @xmath25 gasb ( inas ) strained insb common - anion cation systems - plane extensive strain @xmath29 complementary case pseudomorphic growth gasb - free standing mode produce positive @xmath25 spin - orbit coupling removes double degeneracy @xmath27 state yields electronic energy levels fig. function substrate lattice constant topmost valence bands@xmath0@xmath30 ) labeled quasi - cubic model @xcite centre of gravity valence bands zero | 0.569895 | 3,415 | 1,775 | , Saving $0.1 in GPT-4. |
55,446 | we report the outbursts from a newly discovered x - ray transient in the magellanic - type , sb(s)m galaxy ngc 55 .
the transient source , xmmu j001446.81 - 391123.48 , was undetectable in the 2001 _ xmm - newton _ and 2004 _ chandra _ observations , but detected in a 2010 _ xmm - newton _ observation at a significance level of @xmath0 in the 0.38 kev energy band .
the _ xmm - newton _ spectrum is consistent with a power law with photon index @xmath1 , but is better fit with a @xmath2kev disk blackbody .
the luminosity was @xmath3ergs@xmath4 , and the source displayed strong short - term x - ray variability .
these results , combined with the hardness ratios of its emission , strongly suggest an x - ray binary nature for the source . the follow - up studies with _ swift _ xrt observations revealed that the source exhibited recurrent outbursts with period about a month .
the xrt spectra can be described by a power law ( @xmath52.9 ) or a disk blackbody ( @xmath61.0kev ) , and the luminosity was in a range of 10@xmath710@xmath8ergs@xmath4 , with no evidence showing any significant changes of the spectral parameters in the observations .
based on the x - ray spectral and temporal properties , we conclude that xmmu j001446.81 - 391123.48 is a new transient x - ray binary in ngc 55 , which possibly contains a black hole primary .
[ firstpage ] x - rays : general x - rays : binaries x - rays : bursts x - rays : galaxies x - rays : individual(xmmu j001446.81 - 391123.48 ) | transient x - ray sources have been extensively studied from the beginning of x - ray astronomy .
most of them are binary systems with a black hole ( bh ) or a neutron star ( ns ) as the primary .
these systems have been primarily discovered when they entered outbursts characterised by an episode of high accretion rates and abrupt increases of x - ray luminosity by several orders of magnitude . during an outburst
, the source goes through different spectral states : _ low / hard state _ ( lhs ) and _ high / soft state _ ( hss ) are the two principal states of black hole x - ray binaries ( bhxbs ) . in the lhs , the spectrum is described by a hard power law with photon index in the range of 1.5 - 2.0 , and in the latter , thermal emission ( @xmath9 1 kev ) from an optically thick and geometrically thin accretion disc dominates .
there is also a _ steep power law state _ ( spl ) , described by a disk blackbody component plus power law with photon index @xmath10 ( see * ? ? ?
* ; * ? ? ?
* for extensive reviews on spectral states ) .
the outbursts typically last a few months with recurrence period of many years ( @xmath11 ; * ? ? ?
* ; * ? ? ?
ns x - ray binaries also have such outbursts with much shorter recurrence period ( @xmath12 d ; * ? ? ?
* ; * ? ? ?
* ) compared to the bh systems .
however , bh and ns systems both show many remarkable similarities in their spectral states @xcite .
ngc 55 is a magellanic - type , sb(s)m galaxy and its x - ray properties has been well studied by two _ xmm - newton _ observations conducted in 2001 @xcite .
they identified 137 x - ray sources in the field of view and classified 42 x - ray sources in the @xmath13 region of the galaxy as x - ray binaries ( xrbs ) , supernova remnants , and very soft sources .
moreover , this galaxy hosts a very bright bhxb candidate , which showed a marked upward drift , significant chaotic variability and pronounced dips in the _ xmm - newton _
observations @xcite .
this particular source belongs to the ultraluminous x - ray source class with x - ray luminosity @xmath14 . in this paper , we report the outbursts from a new x - ray transient in ngc 55 .
the transient source was discovered serendipitously from the inspection of archival _ xmm - newton _ observations , which were primarily selected for studying the variability of super soft x - ray sources . while a more detailed report on the variability of super soft x - ray sources will be presented later , in this paper
, we concentrate on this transient which exhibited recurrent outbursts in the _ swift _ x - ray telescope ( xrt ) observations . in the following ,
2 describes the observations and data reduction techniques used .
we explain the analysis and results in 3 and discuss the results and possible nature of the transient in 4 .
we used archival _ xmm - newton _ observations of ngc 55 performed in 2001 november and 2010 may .
the data from the european photon imaging camera ( epic ) pn detector were reduced and analysed using standard tools of the _ xmm - newton _ science analysis software ( sas ) version 13.5.0 .
the full - field background light curve was extracted from the pn camera to remove the particle flaring background and create the good time intervals file .
a count rate of @xmath15 in the 1012 kev light curve was used for the rejection .
we used the pn events with the best quality data ( flag = 0 ) and pattern @xmath16 , and removed the hot pixels in the data by using the flag expression @xmath17 .
the source detection routine ( edetect_chain ) was carried out using the standard parameters for epic - pn data over the entire energy bands and obtaining the final source list from 2001 and 2010 observations .
we corrected the x - ray source positions by correlating the final source list with the usno a2.0 optical catalogue @xcite , using the sas task eposcorr .
we also analysed two archival _ chandra _ observations conducted in 2001 and 2004 .
chandra _ data were reduced and reprocessed using the science threads of _ chandra _ interactive analysis of observations ( ciao ) version 4.6 and heasoft version 6.15.1 .
in addition , we used _ swift _
xrt observations of the region carried out during 2013 april to 2014 october .
we selected the observations with an exposure time of @xmath18 from the _ swift _ program , which resulted 24 observations ( see table [ log ] ) .
we reduced the _ swift _ data using heasoft and the calibration database ( caldb ) files as of 2014 june 10 .
we processed the photon counting mode of _ swift _ observations using the _ swift _ specific ftool xrtpipeline , by following the standard procedures .
.observation log [ cols="^,^,^,^ " , ] [ fit ]
we serendipitously discovered a new transient source in ngc 55 using the archival _ xmm - newton _ data obtained in 2010 .
the source was undetected in the _ xmm - newton _ 2001 observation , and 2001 and 2004 _ chandra _ observations , whose deep sensitivities indicate a flux change of at least two orders of magnitude from @xmath19 to @xmath3ergs@xmath4 .
thus , these observations establish that xmmu j001446.81 - 391123.48 is a new x - ray transient source in ngc 55 . in considering the nature of xmmu j001446.81 - 391123.48
, the x - ray colour classification scheme of @xcite identifies this source as an x - ray binary system .
the high luminosities ( @xmath20 ) during the outbursts are consistent with the luminosity range of bright x - ray binaries , and also help to rule out the low - luminosity source classes , such as magnetic and non - magnetic cataclysmic variables @xcite and magnetars @xcite , as a possible candidate for this transient . apart from the high luminosity , the source displayed a strong short - term x - ray variability in the 2010 observation , further supporting the x - ray binary nature .
there is no statistically significant difference between the power law and disk blackbody fits for the _ xmm - newton _ spectrum .
the best fit photon index ( @xmath21 3.2 ) is too high for the hard state @xcite ; furthermore , the photon index is likely too high for the steep power law state also since the disk blackbody contribution makes the spl appear hard in the observed energy range @xcite . however , the best fit disk blackbody model is consistent with the thermally dominated spectrum @xcite , and we find this to be the most likely spectral state for the transient .
moreover , such a thermally dominated spectrum for an x - ray transient has never been seen in ns x - ray binaries , suggesting that the accretor in the transient is a black hole @xcite .
the follow - up studies with _ swift _ xrt revealed the source s outburst activity and it is possibly a fred phenomenon . during the 2013 may _ swift _ observations , the source had an outburst and reached the peak intensity , by having a factor @xmath22 flux increase , within @xmath23d .
after the outburst , the flux decayed exponentially with an e - folding time of @xmath24d .
this time scale is consistent with the e - folding time of black hole candidates in outburst @xcite .
there are secondary peaks occurred prior ( 2013 april observations ) and after the outburst ( 2013 august observations ) , but the peaks only reached a level of @xmath25 of the main outburst peak .
in addition , the peak prior to the outburst decayed very rapidly compared to the main outburst decay .
thus , the source possibly showed a repeated outburst . the outburst recurrence period would be about a month , which is much smaller than the reported recurrence periods for the ns and bh systems @xcite .
our co - added _ swift _ spectra only provide approximate spectral properties for the source . in the fainter state ( _ swift1 _ ) ,
the spectrum is described by power law with index @xmath26 or disk blackbody with @xmath27kev .
the source changed its luminosity by an order of magnitude and reached @xmath28 in the _ swift5 _ compared to _
swift1 _ , but not much change in the power law index or disk blackbody temperature .
although the large uncertainties on the parameters do not allow us to draw a clear conclusion , the results suggest that the source changed its luminosity by an order of magnitude without change in its state . in the observations ,
the x - ray luminosity of xmmu j001446.81 - 391123.48 is @xmath29 of the eddington limit for a canonical @xmath30 ns , again suggesting that the primary is more likely a bh .
finally , the transient source is located in the bar region which is displaced @xmath21 3 arcmin from the geometrical centre of ngc 55 @xcite .
the very large array observations at 6 cm and 21 cm revealed that the radio continuum emission is concentrated on the bar region @xcite .
moreover , the 6 cm radio emission is dominated by a triple source and this triple coincides with discrete @xmath31 and @xmath32 emission .
also , the on - going star formation rate ( sfr ) in the bar region is consistent with the global sfr of ngc 55 ( @xmath33 ; * ? ? ?
* ) and suggests that the bar region is a young star formation complex with an age of @xmath34 .
thus , the transient source could be a young stellar system in the bar region of ngc 55 . in summary , xmmu j001446.81 - 391123.48 is a new x - ray transient in the young stellar region of ngc 55 , possibly being a black hole x - ray binary .
the repeated outbursts and prominent radio emission from the young stellar region made this source as a good candidate for the further follow - up studies at x - ray and radio wavelengths .
such multi - wavelength studies can shed further light on the nature of xmmu j001446.81 - 391123.48 .
we thank referee for constructive suggestions , and thank ranjeev misra and fuguo xie for useful comments on the manuscript . this work has made use of data obtained from the high energy astrophysics science archive research center ( heasarc ) , provided by nasa s goddard space flight center .
this research was funded by chinese academy of sciences president s international fellowship initiative ( grant no .
2015pm059 ) , and was supported in part by the strategic priority research program the emergence of cosmological structures " of the chinese academy of sciences ( grant no .
xdb09000000 ) .
zw is a research fellow of the one - hundred - talents project of chinese academy of sciences . | transient x - ray sources studied from beginning astronomy binary systems with black hole bh or neutron star ns primary systems discovered outbursts high accretion rates increases x - ray luminosity during outburst source goes through spectral states low / hard state and high / soft state hss principal states of black hole x - ray binaries in lhs spectrum described by hard power law photon index 1.5 - 2.0 latter thermal emission 1 kev from thick thin accretion disc dominates steep power law state spl ) by disk blackbody component plus power law with photon index @xmath10 outbursts last few months recurrence many years ns x - ray binaries have outbursts shorter recurrence period compared to bh systems bh and ns systems show similarities in spectral states ngc 55 magellanic - type sb(s)m galaxy x - ray properties studied by two xmm - newton observations in 2001 identified 137 x - ray sources classified 42 in @xmath13 region galaxy as x - ray binaries ( xrbs supernova remnants very soft sourcesgalaxy hosts bright bhxb candidate showed upward drift chaotic variability dips in xmm - newton observations @xcite source ultraluminous x - ray class luminosity @xmath14 outbursts from new x - ray transient in ngc 55 discovered from archival xmm - newton observations variability super soft x - ray sources detailed report later concentrate on transient recurrent outbursts in swift _ x - ray telescope observations describes observations data reduction techniques analysis results nature transient used archival xmm - newton observations of ngc 55 2001 november 2010 may data from european photon imaging camera pn detector reduced analysed using xmm - newton science analysis software version 13.5.0 full - field background light curve extracted remove particle flaring background good time intervals file count rate of @xmath15 1012 kev light curve used for rejection used pn events with best quality data flag = 0 pattern @xmath16 removed hot pixels flag expression @xmath17 source detection routine standard parameters epic - pn data final source list from 2001 2010 observationscorrected x - ray source positions correlating final source list with usno a2.0 optical catalogue @xcite using sas task eposcorr analysed two archival chandra _ observations 2001 2004 data reduced reprocessed using chandra interactive analysis version 4.6 heasoft version 6.15.1 used swift _ xrt observations 2013 april to 2014 october selected observations exposure time @xmath18 program resulted 24 observations reduced swift data using heasoft calibration database files 2014 june 10 processed photon counting mode swift observations specific ftool xrtpipeline standard procedures discovered new transient source in ngc 55 archival xmm - newton data 2010 source undetected in 2001 2004 chandra observations flux change two orders of magnitude from @xmath19 to @xmath3ergs@xmath4 establish xmmu j001446.81 - 391123.48 new x - ray transient source in ngc 55 x - ray classification scheme identifies source as x - ray binary systemhigh luminosities @xmath20 ) during outbursts consistent with bright x - ray binaries out low - luminosity source classes for transient source displayed strong short - term x - ray variability in 2010 supporting x - ray binary nature no significant difference between power law disk blackbody fits for xmm - newton spectrum best fit photon index ( @xmath21 3.2 ) too high for hard state @xcite too high for steep power law state disk blackbody hard best fit disk blackbody model consistent with thermally dominated spectrum @xcite most likely state for transient thermally dominated spectrum for x - transient never seen in ns x - ray binaries accretor is black hole @xcite follow - up studies with swift _ xrt source s outburst activity possibly fred phenomenon 2013 may source outburst reached peak intensity @xmath22 flux increase after flux decayed with e - folding time of @xmath24d consistent with of black hole candidates in outburst @xcite secondary peaks occurred prior after reached level of @xmath25 of main outburst peakpeak outburst decayed rapidly source possibly repeated outburst outburst recurrence period about a month smaller than ns bh systems @xcite co - added swift _ spectra provide approximate spectral properties source fainter state swift1 spectrum described by power law index @xmath26 disk blackbody @xmath27kev source changed luminosity reached @xmath28 in swift5 not change in power law index disk blackbody temperature large uncertainties parameters results suggest source changed luminosity without change state x - ray luminosity of xmmu j001446.81 - 391123.48 @xmath29 of eddington limit for @xmath30 ns primary likely bh transient source in bar region displaced @xmath21 3 arcmin from centre ngc 55 @xcite large array observations at 6 cm 21 cm radio continuum emission concentrated on bar region 6 cm radio emission dominated by triple source coincides with @xmath31 and @xmath32 emission on - star formation rate bar region consistent with global sfr of ngc 55 young star formation complex age of @xmath34 transient source could be young stellar system region ngc 55 xmmu j001446.81 - 391123.48 new x - ray transient possibly black hole x - ray binary repeated outbursts radio emission good candidate for studies x - ray radio wavelengths multi - studies light nature xmmu j001446.81 - 391123.48 thank referee suggestions ranjeev misra fuguo xie comments work data high energy astrophysics science archive research center nasa s goddard space flight center funded by chinese academy sciences president s international fellowship initiative grant 2015pm059 supported strategic priority research program cosmological structures academy grant xdb09000000 research fellow one - hundred - talents project academy | 0.362592 | 2,451 | 1,239 | , Saving $0.1 in GPT-4. |
55,450 | we have updated our phoenix model atmospheres and theoretical spectra for ultracool dwarfs with new opacity data for methane based on line strength predictions with the stds software . by extending the line list to rotational levels of @xmath0
we can significantly improve the shape of the near - ir absorption features of ch@xmath1 , and in addition find an enhanced blanketing effect , resulting in up to 50% more flux emerging in the @xmath2 band than seen in previous models , which may thus contribute to the brightening in @xmath2 and blue ir colors observed in t dwarfs .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in | absorption bands of h@xmath3o and ch@xmath1 dominate the ir spectra of dwarfs cooler than @xmath41400k and are a backbone of the t dwarf classification schemes defined by geballe et al .
( 2002 ) and burgasser et al .
accurate and complete opacity data for these molecules are thus essential for modelling the atmospheres of t dwarfs . we have updated the phoenix model atmosphere code with a new methane list based on theoretical calculations with the spherical top data system ( stds ) ( wenger & champion 1998 ) . compared to our ames - cond models ( allard et al .
2001 ) , with @xmath437000 ch@xmath1 lines from the hitran and geisa databases , this has added 1.2@xmath5 lines ( homeier et al .
we have computed opacity sampling spectra in the limiting case of fully settled dust clouds as described in allard et al .
the main difference in the line lists is a much larger set of faint lines , mostly due to a coverage in rotational quantum number extending up to @xmath2=40 , as compared to typically @xmath610 in the hitran and geisa data .
we discuss the changes in these models on the case of the t4.5 dwarf sdss0207 + 00 observed by geballe et al .
( 2002 ) . as can be seen in the absorption feature around 2.2@xmath7 m ,
rotational levels up to about @xmath2=30 need to be included to reproduce the band shape over its full width .
less evident in the spectra is the dense background of faint lines at higher @xmath2 , which effectively closes any windows of small opacity throughout all of the @xmath8 and @xmath9 bands .
this results in a stronger redistribution of flux towards shorter wavelengths , and thus a brightening of the @xmath2 band of about 05 compared to an ames - cond model with the same parameters .
we note that the colors can be `` fixed '' with lower log@xmath10 models in the ames - cond grid , but at the cost of an even worse spectral fit in the @xmath8 and @xmath9 bands .
we could significantly improve our synthetic ir spectra for t dwarfs with the new stds line list for methane .
in addition , the new models produce a shift of emergent flux towards the @xmath2 band due to stronger blanketing , reproducing well the blue ir colors of t dwarfs . to clarify the exact role of the onset of ch@xmath1 in the rapid brightening in @xmath2 observed at the l - t transition ,
a new set of models will be necessary that also provides a consistent treatment of the settling of the cloud layer below the photosphere expected to occur in about the same temperature range ( allard et al .
, these proceedings ) . * acknowledgments . *
we thank d. alexander , v.boudon , j .-
champion and u. jrgensen for helpful discussions and s. leggett for access to observational data .
this work is supported by nfs grant n - stars rr185 - 258 .
allard , f. , hauschildt , p. h. , alexander , d. r. et al .
2001 , apj 556 , 357 burgasser , a. j. , kirkpatrick , j. d. , brown , m. e. et al .
2002 , apj 564 , 421 geballe , t. r. , knapp , g. r. and leggett , s. k. et al .
2002 , apj 564 , 466 homeier , d. , hauschildt , p. h. and allard , f. 2002 , in : http://adsabs.harvard.edu/cgi-bin/nph-abs_connect?bibcode=2002sam..worke[stellar atmosphere modeling , eds .
k. werner , i. hubeny & d. mihalas , pasp , _ in press _ ] , http://arxiv.org/abs/astro-ph/0209163[astro-ph/0209163 ] wenger , c. & champion , j. p. 1998 , jqsrt 59 , 471 | absorption bands h@xmath3o ch@xmath1 dominate ir spectra dwarfs cooler than @xmath41400k backbone t dwarf classification schemes by geballe 2002 burgasser et accurate opacity data essential for modelling atmospheres t dwarfs updated phoenix model atmosphere code with new methane list calculations spherical top data system compared to ames - cond models added 1.2@xmath5 lines computed opacity sampling spectra in fully settled dust clouds main difference line larger set faint lines due to coverage rotational quantum number up to @xmath2=40 compared to @xmath610 hitran geisa discuss changes t4.5 dwarf sdss0207 + 00 observed by geballe et 2002 2.2@xmath7 m rotational levels up to @xmath2=30 need to reproduce band shape dense background faint lines at higher @xmath2 closes windows small opacity @xmath8 @xmath9 bands results in stronger redistribution flux towards shorter wavelengths brightening of @xmath2 band 05 compared to ames - cond modelcolors fixed with lower log@xmath10 models ames - cond grid worse spectral fit in @xmath8 @xmath9 bands improve synthetic ir spectra for t dwarfs with new stds line list methane new models shift flux towards @xmath2 band stronger blanketing reproducing blue ir colors t dwarfs role ch@xmath1 in rapid brightening in @xmath2 l - t transition new models necessary consistent treatment settling cloud layer below photosphere same temperature range ( allard et al . proceedings ) thank d. alexander v.boudon j. champion u. jrgensen for discussions s. leggett access observational data work supported by nfs grant n - stars rr185 - 258 allard hauschildt alexander 2001 apj 556 , 357 burgasser kirkpatrick brown 2002 564 421 geballe knapp leggett 2002 564 466 homeier 2002 http://adsabs.harvard.edu?bibcode=2002sam..worke[stellar atmosphere modeling eds k. werner i. hubenymihalas pasp press http://arxiv.org/astro-ph/0209163 wenger champion j p. 1998 jqsrt 59 | 0.183073 | 943 | 478 | , Saving $0.0 in GPT-4. |
55,483 | we introduced a new approach to describe java software as graph , where nodes represent a java file - called compilation unit ( cu ) - and an edges represent a relations between them .
the software system is characterized by the degree distribution of the graph properties , like in - or - out links , as well as by the distribution of chidamber and kemerer metrics computed on its cus .
every cu can be related to one or more bugs during its life .
we find a relationship among the software system and the bugs hitting its nodes .
we found that the distribution of some metrics , and the number of bugs per cu , exhibit a power - law behavior in their tails , as well as the number of cus influenced by a specific bug .
we examine the evolution of software metrics across different releases to understand how relationships among cus metrics and cus faultness change with time .
an analysis of bug distribution in object oriented systems | large software systems can be analysed as graphs so huge and intricate that can be studied using complex network theory .
+ in the case of object oriented ( oo ) software systems nodes are the classes or the interfaces , and oriented edges are the various kinds of relationships between them , inheritance , composition , dependence
. for oo systems there exist also some consolidated software metrics , also associated to the graph , usually computed at class level , the most used being the chidamber and kemerer ( ck ) suite of metrics @xcite .
the relationship between metrics and software quality is fuzzy , and is still the subject of ongoing research .
related to software quality are software bugs .
several researchers analysed software evolution in order to understand the relationship between software management and bug issues .
purushothaman et al .
@xcite analyzed software development process to identify what are the relationships between small changes to the code and bug growth .
kim et al . @xcite analyzed micro - pattern evolution in java classes to identify which of them is more bug - prone .
liwerski et al .
@xcite analyzed the fix - inducing changes , i.e. software updates that trigger the appearance of bugs . in their work , the revision history
associated to compilation units ( cus ) was examined to understand where bugs issues are introduced during cu evolution .
compilation units , the basic blocks examined in this paper , are files containing one or more classes , for which it is possible to compute software metrics similar to those used for classes . a complete analysis of the relationships between graph properties of large software systems , statistic of software metrics , and the introduction and distribution of bugs in such graphs is , to our knowledge , completely missing .
zimmerman et al . considered a network analysis on dependences graphs , built on binary files @xcite , and how dependencies correlate with , and predict , defects .
andersson et al .
@xcite discussed the pareto distribution of bugs in classes , without entering into the details of the statistical properties of software which determine such distribution .
zhang found that the bug distribution across compilation packages in eclipse java system seems to follow a weibull distribution @xcite .
+ the aim of this paper is study oo systems using complex network theory , to improve the knowledge of bugs causes and to statistically determine their distribution into the system . we extend the definitions of ck software metrics to cus to understand the evolution of faultness , i.e. how a metric variation affects the number of bugs hitting a cu . a deeper understanding of the dynamics of software development could be useful for software engineers to identify which system components will be more prone to bugs , thus focusing testing and code reviews on these components .
+ we also study the time evolution of software systems and of the related graphs and metrics , analysing both the source code and the bugs of various releases of two large java systems , eclipse @xcite and netbeans @xcite . for each release we computed the associated software graph and the ck metrics for each class .
furthermore , we study the number of defects associated to cus , as found in the bug - tracking system used for development .
we computed the correlation between oo metrics and bugs and analyzed the evolution of these metrics between one release and the next , correlating metrics changes with the number of defects .
we present a scheme of classification of cus into categories which allows us to identify which parts of the software are the most fault - prone , and how these are correlated to ck software metrics
. we support our findings with significance tests .
we analyze the source code of object - oriented systems written in java . both use cvs as version control system .
eclipse uses bugzilla as issue tracker system , while netbeans uses issuezilla .
the cvs keeps track of the source code history , bugzilla and issuezilla keep track of the bugs history .
an oriented graph is associated to oo software systems , where the nodes are the classes and the interfaces , and the edges are the relationships between classes , namely inheritance , composition and dependence .
+ the number and orientation of edges allow to study the coupling between nodes . in this graph
the in - degree of a class is the number of edges directed toward the class , and measures how much this class is used by other classes of the system .
the out - degree of a class is the number of edges leaving the class , and represents the level of usage the class makes of other classes in the system . in this context ck suite
is a common metrics employed in classes analysis .
we calculated for each node the values of the four most relevant ck metrics of the associated class : * weighted methods per class ( wmc ) . a weighted sum of all the methods defined in a class .
we set the weighting factor to one to simplify our analysis . * coupling between objects ( cbo ) .
the counting of the number of classes which a given class is coupled to .
* response for a class ( rfc ) .
the sum of the number of methods defined in the class , and the cardinality of the set of methods called by them and belonging to external classes . * lack of cohesion of methods ( lcom ) .
the difference between the number of non cohesive method pairs and the number of cohesive pairs .
we also computed the lines of code of the class ( loc ) , excluding blanks and comment lines .
this is useful to keep track of cu dimension because it is known that a `` long '' class is more difficult to menage than a short class .
every system class resides inside a java file , called cu .
while most files include just one class , there are files including more than one class . in eclipse 10% of cus host more than one class , whereas in netbeans this percentage is 30% . in commit messages issues and issue
fixing always refer to cus . to make consistent issue tracking with source code
, we decided to extend ck metrics from classes to cus .
cus represent therefore the main element of our study .
so , we defined a cu graph whose nodes are the cus of the system .
two nodes are connected with a directed edge if at least one class inside the cu associated with the first node has a dependency relationship with one class inside the cu associated with the second node .
we refer to this graph for computing in - links and out - links of a cu - node .
we reinterpreted ck metrics onto this cu - graph : * cu locs is the sum of the locs of classes contained in the cu ; * cu cbo is the number of out - links of each node , excluding those representing inheritance .
this definition is consistent with that of cbo metrics for classes ; * cu lcom and cu wmc are the sum of lcom and wmc metrics of the classes contained in the cu , respectively ; * cu rfc is the sum of weighted out - links of each node , each out - link being multiplied by the number of specific distinct relationships between classes belonging to the cus connected to the related edge . for each cu
we have thus a set of 6 metrics : in - links , out - links , cu - locs , cu - lcom , cu - wmc , cu - rfc and cu - cbo .
this was made for all versions of eclipse and netbeans . onto the cu graph
we look for nodes hit by issues .
to obtain this information it is necessary to check the cvs log file , and the data contained in the its .
+ we consider a cu as affected by an issue when it is modified for issue fixing .
developers record on the cvs log all fixing activities .
all commit operations are tracked in the cvs log as single entries .
each entry contains various data , among which the date , the developer who made the changes , an annotation referring to the reasons of the commit , and the list of cus interested by the commit . in case of commits associated to an issue fixing activity ,
this is written in the annotation , though not in a standardized way .
it is not simple to obtain a correct mapping between issue(s ) and the related cu(s ) @xcite @xcite . in our approach
, we first analyzed the cvs log , to locate commit messages associated to fixing activities .
then , the extracted data are matched with information found in the its .
each issue is identified by a whole positive number ( i d ) . in commit messages it can appear a string such as
`` fixed 141181 '' or `` bug # 141181 '' , but sometimes only the i d is reported .
every positive integer number is a potential issue . to discern among issues and simple numbers we applied the following strategies : 1 .
we considered only positive integer numbers present in the issue tracker as valid issue ids related to the same release ; 2 .
we did not consider some numeric intervals particularly prone to be a false positive issue i d .
the latter condition is not particularly restrictive in our study , because we do not consider the first releases of the studied projects , where issues with `` low '' i d appear .
+ all ids not filtered out are considered issues and associated to the addition or modification of one ore more cus , as reported in the commit logs .
the total number of issues hitting a cu in each release constitutes the issue metric we consider in this study .
note that an issue reported in an issue management system has a broad sense .
it may denote an error in the code , but also an enhancement of the system , or a features request , or fixing a requirement error .
moreover , when many cus are affected by a single bug , it is possible that some of them are modified not because they have the issue , but as a side - effect of modifications made in other cus .
the subjects of our study were eclipse and netbeans projects , both open source , object oriented , java based systems .
table [ tabella3.0.1 ] and [ tabella3.0.2 ] show the number of cus involved in the main releases of eclipse and netbeans , respectively .
+ [ tabella3.0.1 ] .number of cus of eclipse for each main release [ cols="<,^,^,^,^,^",options="header " , ] all data in tables [ tabella3.2.8 ] and [ tabella3.2.9 ] , show positive correlations .
correlation values are quite similar for the same pair of subsequent releases , whereas they show larger fluctuations for different metrics .
a comparison of tables [ tabella3.2.8 ] and [ tabella3.2.9 ] shows that correlation values in netbeans are often lower than in eclipse .
this result can be partially due to the less clear subdivision between main and patching releases in netbeans project . + according to tables [ tabella3.2.3 ] , [ tabella3.2.6 ] , [ tabella3.2.8 ] and [ tabella3.2.9 ] , bug introduction is mainly due to updating and adding cus .
this is valid for each metric considered .
a statistical description of large software systems as directed graphs can provide much additional information on the system features with respect to more traditional approaches , from the software engineering perspective . adopting a graph as a model for the software system , we used the compilation units as the basic software module in order to build a software graph , and redefined the ck suite of metrics to cope with cus .
these metrics were then used to investigate , with a statistical analysis , how and where bugs were introduced into two big , oo software projects like eclipse and netbeans .
we wrote two different parsers to analyze the cvs log file and the issue tracker repositories in order to automatically associate bugs and cus . in this paper , we introduced the concept of compilation unit graph , and of oo metrics related to compilation units , with the purpose of analyzing software projects managed using a configuration management system and a corresponding bug tracking system . + the picture of the software system as a graph allowed us to detect fat - tail distributions , well described by power - laws , for different features of the system , suggesting the same general underlying framework of many other complex networks .
in particular , we found that bugs distribution among cus , number of cus affected by bugs , metrics distributions ( namely locs , number of in - links and out - links of the class graph , ck metrics wmc , cbo , rfc and lcom ) , all exhibit power - laws fat - tails .
+ inside this framework it is possible to identify strong correlations among bugs and those metrics related to the number of external dependencies which , in the graph representation , are easily described as directed links . all these findings together indicate a possible strategy to optimize resources and efforts in software engineering for finding , forecasting , and fixing software defects . once the software graph reveals the fat - tail in the relationships between bug and cus , one may identify which parts of the software are the most fault - prone and focus fixing efforts on them .
following @xcite , if one ranks cus according to these power - laws , the review of a small fraction among the highest ranked may have an exponential impact on the overall amount of software defects detectable and fixable .
+ our analysis goes one step further , examining software evolution across many releases .
this study identifies how metric evolution is related to bug introduction .
the change of some particular metrics may result in a higher probability of introducing a bug .
in particular , we identify different families of cus , related to ck metrics , showing different robustness with respect to bug affection .
our categorization into families is related to the software evolution and it is useful to investigate correlations among bugs and software metrics changes .
this classification may be particularly useful to software engineers in order to decide onto which parts of a big software project it is better to concentrate efforts and resources .
99 chidamber s. r , darcy d. p , kemerer c.f , _ managerial use of metrics for object oriented software : an exploratory analysis _ , ieee trans .
software eng . ,
vol 24 , no . 8 , pp .
629 - 639 , 1998 .
purushothaman r , dewayne e.p .
_ toward understanding the rhetoric of small source code changes _ , ieee trans.software eng . , vol . 31 , no . 6 , june 2005 kim s. , pan k. , whitehead e.j.jr . _ micro pattern evolution _
( msr06 ) , may 22 - 23 , 2006 , shanghai , china .
liwerski j , zimmermann t , zeller a. _ when do changes induce fixes?_. proc .
international workshop on mining software repositories ( msr05 ) , st .
louis , missouri , u.s .
, may 2005 .
zimmermann t , nagappan n. _ predicting defects using network analysis on dependency graphs _ , icse08 , may 10 - 18 , 2008,leipzig , germany .
andersson c , runeson p. _ a replicated quantitative analysis of fault distributions in complex software systems _ , ieee trans.software eng .
5 , may 2007 , pp . 273 - 286 .
zhang h. _ on the distribution of software faults _ , ieee trans.software eng .
2 , march / april 2008 , pp .
301 - 302 .
fischer m , pinzger m , gall h. _ analyzing and relating bug report data for feature tracking_. in proc .
10th working conference on reverse engineering ( wcre03 ) , victoria , british columbia , canada , nov .
concas g , marchesi m , pinna s , serra n. _ power - laws in a large object - oriented software system_. ieee trans.software eng .
33 , no . 10
687 - 708 , 2007 .
juran j.m , gryna f.m .
jr . _ quality control handbook _
, fourth ed .
mcgraw - hill , 1988 .
barabasi a , albert r. _ emergence of scaling in random networks_. science , vol .
509 - 512 , 1999 .
louridas p , spinellis d , vlachos v. _ power laws in software _
, acm transactions on software engineering and methodology , vol . 18 , no.1 , september 2008 .
newman m. e. j. _ power laws , pareto distributions and zipf s law _ , contemporary physics , vol .
323 - 351 , 2005 . | software systems analysed as graphs studied using complex network theory object oriented ) software systems nodes are classes interfaces oriented edges are relationships inheritance composition dependence for systems consolidated software metrics associated to graph computed at class level used chidamber and kemerer suite of metrics @xcite relationship between metrics and software quality fuzzy subject of ongoing research related to software quality are software bugs researchers analysed software evolution software management bug issues purushothaman et al analyzed software development process relationships between small changes code bug growth kim et analyzed micro - pattern evolution in java classes bug - liwerski et analyzed fix - inducing changes software updates bugs revision history compilation units examined bugs issues compilation units are files containing classes compute software metrics similar to classes complete analysis of relationships between graph properties of large software systems software metrics introduction distribution of bugs missing zimmerman et al considered network analysis on dependences graphs on binary files @xcite dependencies correlate with predict defects andersson et al@xcite discussed pareto distribution bugs in classes without statistical properties software zhang found bug distribution across compilation packages in eclipse java system weibull distribution @xcite aim paper oo systems complex network theory improve knowledge bugs causes statistically determine distribution extend definitions ck software metrics to cus understand evolution of faultness variation bugs deeper understanding dynamics software development for engineers identify system components prone to bugs focusing testing code reviews study time evolution of software systems related graphs metrics analysing source code bugs of releases java systems eclipse @xcite netbeans @xcite each release computed software graph ck metrics class study number defects associated cus bug - tracking system computed correlation between oo metrics bugs analyzed evolution metrics release correlating defects present scheme classification of cus into categories identify parts software most fault - prone ck software metrics support findings with significance tests analyze source code of object - oriented systems java both use cvs version control system eclipse uses bugzilla netbeans issuezilla cvs source code history bugzilla issuezilla bugs historyoriented graph to software systems nodes are classes interfaces edges relationships between classes inheritance composition dependence number orientation of edges study coupling between nodes in - degree of class number edges directed toward class measures used by other classes out - degree number edges leaving class represents level usage other classes ck suite common metrics in classes analysis calculated for each node values four relevant ck metrics class weighted methods per class ) sum weighting factor to one coupling between objects ( cbo ) number of classes coupled to response for class ( rfc ) sum of number of methods defined cardinality of set methods external classes lack of cohesion of methods ( lcom ) difference between non cohesive pairs cohesive pairs computed lines of code of class ( loc ) excluding blanks comment lines useful cu dimension long '' class difficult to menage than short class every system class inside java file cu most files include one class files more than one class in eclipse 10% of cus host more than one class in netbeans 30% commit messages issues issue fixing refer to cus ck metrics from classes to cus cus represent main element study defined cu graph nodes are cus system two nodes connected with edge if one class first has dependency with second node refer to graph for computing in - links out - links of cu - node reinterpreted ck metrics cu - graph : cu locs sum of locs classes cu cu cbo number out - links each node excluding inheritance definition consistent with cbo metrics for classes cu lcom and cu wmc sum of lcom wmc metrics cu rfc sum of weighted out - links each node multiplied by relationships between classes cus for each cu 6 metrics : in - links out - links cu - locs cu - lcom cu - wmc cu - rfc cu - cbo made for all versions eclipse and netbeans cu graph for nodes hit by issues check cvs log file data cu affected by issue when modified for issue fixing developers record cvs log fixing activities commit operations tracked as single entries each entry contains data date developer changes annotation reasons commit list of cus interested commits associated to issue fixing activity written in annotationnot simple correct mapping between issue related cu @xcite analyzed cvs log commit messages fixing activities extracted data matched with information in each issue identified by positive number i d commit messages fixed 141181 or bug # 141181 sometimes only i d reported every positive integer number is potential issue discern issues simple numbers applied strategies considered only positive integer numbers issue tracker as valid issue ids release consider numeric intervals prone false positive issue i d not restrictive consider first releases issues with low i d appear all ids not filtered out considered issues associated to addition or modification of one more cus reported logs total number of issues hitting cu in each release constitutes issue metric issue reported in issue management system broad sense may denote error code enhancement system features request fixing requirement error many cus affected by single bug possible some modified side - effect of modifications subjects study were eclipse netbeans projects open source object oriented java based systems tabella3.0.1 and tabella3.0.2 show number of cus involved in main releases of eclipse and netbeans tabella3.0.1number cus eclipse each main release cols="<^,^^",options="header " data in tables tabella3.2.8 tabella3.2.9 ] show positive correlations correlation values similar for same subsequent releases show larger fluctuations for different metrics comparison shows correlation values in netbeans often lower than eclipse due to less clear subdivision between main patching releases tables tabella3.2.3.2.6.2.8.9 bug introduction due to updating adding cus valid for each metric statistical description large software systems graphs provide additional information system features adopting graph model software system used compilation units basic software module build software graph redefined ck suite metrics cope with cus metrics used investigate bugs introduced into big software projects eclipse netbeans wrote two parsers to analyze cvs log file issue tracker repositories automatically associate bugs cus introduced concept of compilation unit graph oo metrics related to compilation units analyzing software projects managed using configuration management system bug tracking systemsoftware system as graph fat - tail distributions described by power - laws for different features suggesting framework complex networks found bugs distribution among cus number cus affected metrics distributions in - - links ck metrics rfc lcom exhibit power - laws fat - tails strong correlations among bugs metrics related to external dependencies directed links findings indicate strategy to optimize resources efforts software engineering for finding forecasting fixing software defects software graph reveals fat - tail relationships between bug cus identify parts software most fault - prone focus fixing efforts ranks cus according to power - laws review small fraction highest ranked impact on software defects detectable fixable analysis software evolution across releases identifies metric evolution related to bug introduction change of metrics higher probability introducing bug different families of cus related to ck metrics showing different robustness bug affection categorization into families related to software evolution investigate correlations among bugs software metrics changes classification useful to software engineers decide parts software project concentrate efforts resources chidamber s. r darcy d. p kemerer c.managerial metrics object oriented software exploratory analysis software eng vol 24 8 629 - 639 1998 purushothaman dewayne e understanding small source code changes.software vol 31 6 june 2005 kim s pan whitehead micro pattern evolution may 22 - 23 2006 shanghai liwerski zeller changes induce fixes international workshop mining software repositories st. louis missouri may 2005 nagappan n predicting defects network analysis dependency graphs icse08 may 10 - 18 2008,leipzig germany andersson runeson p analysis fault distributions complex software systems.software eng 5 may 2007 273 - 286 zhang h distribution software faults.software 2 march april 2008 301 - 302 fischer pinzger gall h analyzing bug report data feature tracking 10th working conference reverse engineering victoria british columbia canada concas marchesi pinna serra n power - laws large object - oriented software system trans.software eng 33 10 687 - 708 2007 juran jgryna f quality control handbook fourth ed mcgraw - hill 1988 barabasi albert scaling random networks science 509 - 512 1999 louridas spinellis vlachos power laws software transactions software engineering methodology 18 no september 2008 newman power laws pareto distributions zipf s law contemporary physics 323 - 2005 | 0.57646 | 3,446 | 1,661 | , Saving $0.1 in GPT-4. |
55,486 | it is suggested by chen
_ et al .
_ that the earth s surface ultraviolet irradiance ( @xmath0 nm ) could influence the earth s surface temperature variation by highly excited water vapor " ( hewv ) effect . in this manuscript ,
we reconstruct the developing history of the ozone layer depth variation from 1860 to 2011 based on the hewv effect .
it is shown that the reconstructed ozone layer depth variation correlates with the observational variation from 1958 to 2005 very well ( @xmath1 , @xmath2 ) . from this reconstruction
, we may limit the spectra band of the surface ultraviolet irradiance referred in hewv effect to ultraviolet b ( @xmath3 nm ) . | it is suggested that the surface solar ultraviolet ( uv ) irradiance ( @xmath0 nm , similarly hereinafter ) may have the ability to influence the earth s surface temperature by
highly excited water vapor " ( hewv ) effect @xcite . based on the hewv effect
, it is schematically shown in fig .
[ fig : phys - rel ] that the physical way of solar uv irradiance influences the earth s surface temperature : part of the uv irradiance emitted by the sun is absorbed by the earth s ozone layer and cloud , while the rest reaches to the lower troposphere and influences the surface temperature via hewv effect . in this manuscript
, we suggest a model to obtain the uv irradiance absorbed by ozone layer and the cloud by employing the earth s surface temperature data and solar uv irradiance data at the top of the earth s atmosphere .
the uv irradiance absorbed by ozone layer and the cloud could describe the variation of ozone layer or cloud .
in our model , we assume a signal @xmath4 goes through a linear system box , which is in a steady state . it is schematically shown in fig .
[ fig : model ] .
then , we can obtain a new signal @xmath5 , and @xmath6 has a linear relation with @xmath7 , namely @xmath8 [ h ] changes to @xmath7 after getting through the linear system , title="fig : " ] based on hewv effect , we suggest the surface uv irradiance influences the surface temperature linearly . equation ( [ eq : model ] ) can be employed to describe the relation between surface temperature and surface uv irradiance in spectra band @xmath0 nm . in our model
, the sun emits @xmath9 solar uv irradiance in spectra band 280 - 400 nm to the top of the earth s atmosphere at the time @xmath10 , and @xmath11 is absorbed by ozone layer and cloud .
the rest part @xmath12 goes to the lower troposphere and influences the surface temperature @xmath13 in a linear relation : @xmath14 or , it can be rewritten as follow : @xmath15 the solar uv irradiance data in spectra band @xmath0 nm @xmath9 and surface temperature data @xmath13 can be obtained .
the amounts absorbed by ozone layer and cloud @xmath11 can be obtained from eq .
( [ eq : alpha ] ) for given @xmath16 and @xmath17 , which can describe the variation of the ozone layer or cloud .
an appropriate way to obtain @xmath16 and @xmath17 is to find the years when the ozone layer and cloud did not change ( i.e. @xmath18 ) .
if this happened , eq .
( [ eq : alp ] ) can be rewritten as @xmath19 where @xmath20 .
( [ eq : y - c ] ) can be employed to obtain the coefficients @xmath16 and @xmath21 using the data of @xmath6 and @xmath7 during these years .
once we know the values of @xmath16 and @xmath21 , the uv irradiance anomaly absorbed by the ozone layer and the cloud , @xmath22 , can be obtained as follows @xmath23 where @xmath24 .
it is clearly that @xmath25 can also describe the variation of ozone layer or cloud .
however , we could only find the years when the cloud and ozone layer are keeping the same roughly .
this means we could just obtain @xmath26 . in this case
we can search for the years when the @xmath9 correlated with @xmath13 well , and employ the linear fitting to obtain the values of @xmath16 and @xmath21 .
based on eq .
( [ eq : moment ] ) , hewv effect can be shown via comparing @xmath27 with the observed ozone layer depth data or cloud data .
in this manuscript , the detail data of the surface temperature and solar uv irradiance can be taken from the public data source , they are listed as below : global yearly mean surface temperature anomaly from 1860 to 2011 is obtained from hadcrut3 in climate research unit ( cru ) @xcite .
solar spectra irradiance ( ssi ) variation in spectra band @xmath0 nm from 1860 to 2005 is reconstructed by krivova _
_ @xcite .
also , the total solar irradiance ( tsi ) can also be employed to reconstruct @xmath28 , because the variation of the tsi can describe the property of the ssi in spectra band @xmath0 nm very well .
we test the relation between reconstructed ssi and tsi from 1860 to 2005 and find they correlate with each other very well ( @xmath29 , @xmath30% ) .
moreover , we can obtain satellite observational data after 1979 , which could make the value closer to the real tsi variation and may reconstruct the @xmath28 more accurate .
the tsi data from 1850 to 1978 was reconstructed by judith lean @xcite .
satellite data from 1979 to 2003 is from the virgo experiment on the cooperative esa / nasa mission soho @xcite .
satellite data from 2002 to 2011 is from sorce tim total solar irradiance @xcite .
the annual average of global ( @xmath31 ) column total ozone variation from 1958 to 2005 was compiled by johnston @xcite , which was synthesized the ground - based observations data and the satellite observations data .
the ground - based observations data from 1958 to 1977 was analyzed by angell and korshover in carbon dioxide information analysis center ( cdiac ) @xcite .
the 1978 to 1992 satellite data was presented by total ozone mapping spectrometer ( toms ) , nasa , measured by nimbus-7 satellite @xcite .
the ozone layer data from 1993 to 1994 was obtained from meteor-3 satellite by toms , nasa @xcite .
the 1995 data was analyzed by weber @xcite .
the ozone layer data from 1996 to 2005 was presented by toms @xcite .
figure [ fig : ssifit](a ) shows the comparison of st ( @xmath7 ) and ssi ( @xmath6 ) variation from 1910 to 1970 in 11-yr moving average .
good correlation is found between them ( @xmath32 , @xmath30% ) , so we suppose ozone layer or cloud did not change much in this period ( we deduct the time series during 1937 and 1945 , in the global war ii , when there were not enough observational data @xcite ) and employ them to calculate the parameters @xmath16 and @xmath21 using linear fitting method . the linear fitting result is shown in fig . [
fig : ssifit](b ) .
the parameters : @xmath33 and @xmath34 .
nm and surface temperature ( st ) from 1910 to 1970 ( both data are in @xmath35-yr moving average ) , data from 1937 to 1945 being deducted ( in red color ) ; ( b ) linear fit of the two series . ]
we can use the parameters @xmath16 and @xmath21 to calculate the variation of @xmath25 based on eq .
( [ eq : moment ] ) .
the result of @xmath25 is shown in fig .
[ fig : dalpha_ssi ] . fig .
[ fig : dalpha_ssi](a ) shows the 11-yr moving average variation from 1860 to 2005 , compared with observed 11-yr moving average ozone layer depth variation from 1958 to 2005 .
we find a good correlation between them ( @xmath36 , @xmath30% ) .
[ fig : dalpha_ssi](b ) shows the yearly variation of @xmath25 , compared with the yearly ozone layer depth observed variation .
good correlation between them is also found : @xmath37 , @xmath30% . from 1860 to 2005 ( by employing ssi ) .
( a ) : the 11-yr moving average variation of @xmath25 , compared with the observed 11-yr moving average ozone layer depth variation from 1958 to 2005 .
( b ) : yearly variation of @xmath25 , compared with the yearly variation of observational ozone layer depth from 1958 to 2005 ] based on the good correlation above , we can reconstruct the ozone layer depth variation from 1860 to 2005 .
firstly , we fit the linear relation between @xmath25 and ozone layer depth by applying @xmath25 and observational ozone layer depth data from 1958 to 2005 , and it can be written as follows @xmath38 where @xmath39 is the ozone layer depth .
we expand the relation described by eq .
( [ eq : re_ssi ] ) to the whole @xmath25 and get the reconstructed ozone layer variation from 1860 to 2005 .
this is shown in fig .
[ fig : re_ssi ] . the tsi @xmath40 and st @xmath41 data from 1910 to 1970 are applied ( time series during 1937 and 1945 is deducted ) ( see fig . [
fig : dalpha_tsi](a ) ) .
we employ linear fitting and get @xmath42 and @xmath43 ( see fig .
[ fig : dalpha_tsi](b ) ) .
@xmath16 and @xmath21 are employed to calculate the variation of @xmath25 based on eq .
( [ eq : moment ] ) .
the result of @xmath25 is shown in fig .
[ fig : re_tsi ] . fig .
[ fig : re_tsi](a ) shows the 11-yr moving average variation from 1860 to 2011 , compared with observed 11-yr moving average ozone layer depth variation from 1958 to 2005 .
we find a good correlation between them ( @xmath44 , @xmath30% ) .
[ fig : re_tsi](b ) shows the yearly variation of @xmath25 , compared with the yearly ozone layer depth observed variation .
good correlation between them is also found ( @xmath1 , @xmath30% ) .
this result of @xmath25 is much better than that in ssi occasion . from 1860 to 2011
( by employing tsi ) .
( a ) : the 11-yr moving average variation of @xmath25 , compared with the observed 11-yr moving average ozone layer depth variation from 1958 to 2005 .
( b ) : yearly variation of @xmath25 , compared with the yearly variation of observational ozone layer depth from 1958 to 2005 ] we could also reconstruct the ozone layer depth variation from 1860 to 2011 based on the linear fitting relation between @xmath25 and observed ozone layer depth from 1958 to 2005 : @xmath45 we expand the relation described by eq .
( [ eq : re_tsi ] ) to the whole @xmath25 and get the reconstructed ozone layer variation from 1860 to 2011 as shown in fig .
[ fig : tsi ] .
we could see from fig . [
fig : tsi ] that the reconstructed ozone layer depth data describes the property of the ozone layer depth variation very well , but there are still some differences , such as the opposite variation during 1993 and 2000 .
the difference may led by the volcano eruption in 1991 and el nio and la nio in 1997 - 1999 .
researchers have pointed the cooling effect of the volcanic eruption to the surface temperature , such as robock and mao shows that for two years following great volcanic eruptions , the surface of the earth could be cooled significantly by @xmath46 in the global mean scale @xcite . in 1991 ,
the volcanic eruption of mount pinatubo occurred , cooling the earth s surface in the following @xmath47 years , which will lead to a lower value of the reconstructed surface uv irradiance , then a higher uv irradiance absorbed by ozone layer in the reconstruction , and eventually cause a higher reconstructed ozone layer depth near 1993
. this suggestion could also interpret the higher reconstructed ozone layer depth data in 1985 , when the volcanic eruption in 1983 .
el nio could lead a warmer phenomenon in the earth s
surface @xcite .
the el nio appearance in 1997 - 1998 could lead a warmer surface temperature , which will lead to a lower value of the reconstructed ozone layer depth variation near 1998 .
we can see the same phenomenon in the year of 1987 - 1988 , when the el nio occurred .
the signal of the cloud cover does nt be recognized obviously .
cloud cover is compared with the @xmath25 in fig .
[ fig : cloud ] .
the global mean cloud coverage data is obtained from international satellite cloud climatology project ( isccp ) , nasa @xcite .
they do nt show an apparent correlation ( @xmath48 , @xmath49% ) , as obviously as the @xmath25 with ozone layer depth ( @xmath1 , @xmath30% ) .
the reason why the cloud cover signal does nt been recognized obviously need to do more researches . but based on the results , we may suggest that the ozone layer plays more important role in the hewv effect .
the spectra band absorbed by ozone layer is uvb ( @xmath3 nm ) , and this result may limit the spectra that works in hewv effect to uvb irradiance . more deeper researches should be done before we get a more clear conclusion .
15ifxundefined [ 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 ( ) , link:\doibase 10.1029/2005jd006548 [ * * , ( ) ] @noop * * ( ) @noop * * , ( ) @noop `` , '' ( ) @noop `` , '' ( ) @noop `` , '' ( ) @noop `` , '' ( ) @noop `` , '' ( ) @noop `` , '' ( ) @noop `` , '' @noop `` , '' @noop * * , ( ) @noop * * , ( ) @noop `` , '' | suggested surface solar ultraviolet irradiance @xmath0 nm may influence earth s surface temperature by water vapor hewv effect hewv effect shown in fig . solar uv irradiance influences surface temperature part uv irradiance absorbed by ozone layer cloud rest lower troposphere influences surface temperature suggest model to obtain uv irradiance absorbed by ozone layer cloud surface temperature data solar uv irradiance data top atmosphere uv irradiance absorbed by ozone layer cloud could describe variation of ozone layer cloud model assume signal @xmath4 goes through linear system box steady state shown in fig . new signal @xmath5 @xmath6 linear relation with @xmath7 @xmath8 changes to @xmath7 after linear system hewv effect surface uv irradiance influences surface temperature linearly equation relation between surface temperature uv irradiance in spectra band @xmath0 nm sun emits @xmath9 solar uv irradiance in spectra band 280 - 400 nm earth atmosphere @xmath10 @xmath11 absorbed by ozone layer cloud@xmath12 lower troposphere influences surface temperature @xmath13 linear relation @xmath14 rewritten @xmath15 solar uv irradiance data spectra band @xmath0 nm @xmath9 surface temperature data @xmath13 obtained amounts absorbed by ozone layer cloud @xmath11 obtained from eq . alpha @xmath16 @xmath17 variation ozone layer cloud obtain @xmath16 find years ozone layer cloud change @xmath18 ) eq alp rewritten as @xmath19 @xmath20 . eq : y - c ] ) obtain coefficients @xmath16 @xmath21 data @xmath6 @xmath7 values @xmath16 uv irradiance anomaly absorbed by ozone layer cloud @xmath22 obtained @xmath23 @xmath24 @xmath25 describe variation ozone layer cloud find years cloud ozone layer same obtain @xmath26 search for years @xmath9 correlated with @xmath13 employ linear fitting obtain values @xmath16 @xmath21 eq . moment hewv effect comparing @xmath27 with ozone layer depth data cloud datamanuscript detail data surface temperature solar uv irradiance from public data source global yearly surface temperature anomaly 1860 to 2011 from hadcrut3 climate research unit solar spectra irradiance ssi variation band @xmath0 nm 1860 to 2005 reconstructed by krivova @xcite total solar irradiance tsi reconstruct @xmath28 ssi relation between reconstructed ssi tsi 1860 to 2005 correlate satellite observational data after 1979 value closer to real tsi variation reconstruct @xmath28 tsi data 1850 to 1978 reconstructed by judith lean satellite data 1979 to 2003 from virgo experiment data 2002 to 2011 from sorce tim total solar irradiance annual average global column total ozone variation 1958 to 2005 compiled by johnston @xcite synthesized ground - based satellite observations data ground - based observations data 1958 to 1977 analyzed by angell korshover 1978 to 1992 satellite data by total ozone mapping spectrometer nimbus-7 satellite ozone layer data 1993 to 1994 from meteor-3 satellite 1995 data analyzed by weber ozone layer data 1996 to 2005 by toms @xcitefigure shows comparison st ssi variation 1910 to 1970 11-yr moving average good correlation @xmath32 @xmath30% suppose ozone layer cloud change much deduct time series during 1937 1945 global war ii enough data employ calculate parameters @xmath16 @xmath21 linear fitting method fitting result shown in fig parameters @xmath33 @xmath34 nm surface temperature st 1910 to 1970 data @xmath35-yr moving average data 1937 to 1945 deducted red color linear fit series use parameters @xmath16 @xmath21 calculate variation @xmath25 based eq result shown in fig . dalpha_ssi shows 11-yr moving average variation 1860 to 2005 compared with observed ozone layer depth variation 1958 to 2005 good correlation ( @xmath36 @xmath30% shows yearly variation of @xmath25 compared with yearly ozone layer depth observed variation good correlation found @xmath37 @xmath30% 1860 to 2005 11-yr moving average variation of @xmath25 compared with observed ozone layer depth variation 1958 to 2005yearly variation @xmath25 compared with observational ozone layer depth 1958 to 2005 based correlation reconstruct ozone layer depth variation from 1860 to 2005 fit linear relation between @xmath25 ozone layer depth applying observational data from 1958 to 2005 written @xmath38 where @xmath39 is ozone layer depth expand relation to @xmath25 reconstructed ozone layer variation from 1860 to 2005 shown in fig . @xmath40 and @xmath41 data from 1910 to 1970 applied time series 1937 and 1945 deducted employ linear fitting get @xmath42 and @xmath43 @xmath16 and @xmath21 employed calculate variation @xmath25 result shown in fig . shows 11-yr moving average variation from 1860 to 2011 compared with observed 11-yr moving average ozone layer depth variation 1958 to 2005 good correlation ( @xmath44 , @xmath30% ) shows yearly variation of @xmath25 compared with yearly ozone layer depth observed variation good correlation found result of @xmath25 better than from 1860 to 201111-yr moving average variation @xmath25 compared with observed 11-yr ozone layer depth variation 1958 to 2005 yearly variation of @xmath25 compared with observational ozone layer depth 1958 to 2005 reconstruct ozone layer depth variation from 1860 to 2011 based linear fitting relation between @xmath25 observed ozone layer depth expand relation to @xmath25 reconstructed ozone layer variation 1860 to 2011 in fig . reconstructed ozone layer depth data describes property well differences opposite variation during 1993 and 2000 led by volcano eruption 1991 el nio and la nio 1997 - 1999 cooling effect volcanic eruption to surface temperature surface earth cooled significantly by @xmath46 global mean scale 1991 volcanic eruption of mount pinatubo cooling surface following years to lower value reconstructed surface uv irradiance higher uv irradiance ozone layer higher reconstructed ozone layer depth near 1993 interpret higher reconstructed ozone layer depth data in 1985 volcanic eruption in 1983 el nio could lead warmer phenomenon surfaceel nio 1997 - 1998 warmer surface temperature lower reconstructed ozone layer depth variation near 1998 same phenomenon 1987 - 1988 el nio occurred signal cloud cover recognized cloud cover compared with @xmath25 fig . global mean cloud coverage data from international satellite cloud climatology project nasa @xcite apparent correlation @xmath48 , @xmath49% @xmath25 with ozone layer depth @xmath1 , @xmath30% ) cloud cover signal more researches ozone layer important in hewv effect spectra band absorbed ozone layer uvb ( @xmath3 nm ) may limit spectra hewv effect uvb irradiance deeper researches before clear conclusion15ifxundefined [ 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 link\doibase 10.1029/2005jd006548 [ * * @noop * * @noop @noop @noop @noop @noop @noop * * | 0.513541 | 3,227 | 1,668 | , Saving $0.1 in GPT-4. |
55,499 | starting with a thermal squeezed state defined as a conventional thermal state based on an appropriate hamiltonian , we show how an important physical property , the signal - to - noise ratio , is degraded , and propose a simple model of thermalization ( kraus thermalization ) . | we first of all propose a simple definition of a thermal squeezed state .
previous definitions of thermal coherent states and squeezed states@xcite have used the thermofield approach of takahashi and umezawa@xcite .
an early paper by vourdas@xcite discusses the superposition of squeezed states with thermal photons ; a more recent paper@xcite considers only the case of thermal coherent states .
an early study of canonically transformed thermal states similar to the approach of the present work was made in @xcite , and also in @xcite .
two further papers adopt essentially the same definition as the present note s@xcite using a slightly different approach . the method that we take here
is the straightforward one of defining the hamiltonian whose eigenstates are squeezed states ( sometimes known as squeezed - coherent states to distinguish them from squeezed vacuum states ) .
the complete set of eigenstates are in fact squeezed displaced number states .
these are the states @xmath0 , where @xmath1 is the displacement operator , and @xmath2 the squeezing operator .
the thermal squeezed state we define is simply a mixed state given as a thermal sum of these states .
the simplicity of this approach in , for example , calculating thermal averages is due to the group - theoretical structure of these states , which essentially enables such calculations to be reduced to free hamiltonian averages .
this structure is a consequence of the definition of the squeezed - coherent states as transformed vacua under the automorphism group of the canonical commutation relations , which we shall describe in the next section . to exemplify how the physical properties are degraded by the thermalization process
, we calculate the important signal - to - quantum noise ratio for these thermal squeezed states .
finally , we use the kraus formalism to propose a model for the thermalization .
the canonical commutation relations for the boson ( photon ) creation and annihilation operators @xmath3 and @xmath4 are @xmath5=i.$ ] we shall consider canonical transformations of the form @xmath6 the approach of this note is to consider the relevant quantum optics states as vacuum states for the transformed operator @xmath7 . the _ vacuum state _ is defined as the normalized state satisfying @xmath8 we note that the vacuum state corresponding to @xmath7 , defined by @xmath9 is given by @xmath10 in the spirit of our approach , the ordinary vacuum state corresponds to the trivial unitary transformation @xmath11 , @xmath12 with @xmath13 less trivially , we shall now consider some important states which may be considered in the same light . *
* coherent states * these states correspond to the ( next ) most simple canonical transformation , that corresponding to @xmath14 where @xmath15 is the unit operator .
this canonical transformation is generated by the unitary displacement operator @xmath16 which operates on @xmath4 by @xmath17 the vacuum state of @xmath7 is the coherent state @xmath18 and the vacuum equation for @xmath7 is @xmath19 which is equivalent to @xmath20 this is the usual coherent state@xcite @xmath21 where @xmath22 is the normalization , and has all the well - known properties , such as minimizing the heisenberg uncertainty principle . *
* squeezed vacuum * this state corresponds to the canonical transformation @xmath23 , @xmath24 where @xmath25 and @xmath26 satisfy @xmath27 here @xmath28 is the squeezing operator , given by @xmath29 where we have put @xmath30 the squeezed vacuum satisfying @xmath19 is given by @xmath31 ; that is , it satisfies @xmath32 + the 2-parameter operator @xmath28 is not the most general element of the 3-parameter vacuum - squeezing group @xmath33 an additional operator @xmath34 , where @xmath35 supplies the third parameter and completes the generators of the group . * * squeezed states * standard squeezed states correspond to the canonical transformation @xmath36 where @xmath37 the squeezed state @xmath38 thus defined may be written explicitly as @xmath39 and satisfies the usual equation@xcite @xmath40 we tabulate the above results in table 1 .
.canonical transformations and corresponding states [ cols="^,^,^,^",options="header " , ] the motivation for the foregoing observations is the following . since we have an explicit unitary transformation @xmath41 from the vacuum to a squeezed state , we have analogously the transformation from a free thermal state to our thermal squeezed state .
this enables any calculations involving the thermal squeezed state to be performed very easily . and
this may be expedited further by using the following 3-dimensional representation of the automorphism group of the canonical commutation relations generated by @xmath42 , which we may call the _ squeezing group _ @xmath43 .
this is a 5-parameter group corresponding to the action of the semi - direct sum of the 3-dimensional heisenberg - weyl algebra ( the central element of this algebra acts trivially ) and the 3-dimensional algebra @xmath44 on @xmath45 and is a subgroup of the inhomogeneous pseudounitary group @xmath46 with fundamental representation @xmath47\ ; \ ; \ ; \ ; \ ; ( \lambda,\mu,\alpha \in { \mathcal c } ; \ ; \ ; |\lambda|^2-|\mu|^2=1).\ ] ] the value of this representation is due to the fact that for any operator @xmath48 the inverse transformed operator @xmath49 may be simply obtained by use of the inverse matrix @xmath50 @xmath51\ ] ] where @xmath52 we shall make extensive use of this inverse transformation in the calculations which follow .
we define a thermal squeezed state quite conventionally as that obtained in the usual way from a hamiltonian .
our starting point is the single - mode free boson hamiltonian ( harmonic oscillator ) @xmath53=1\ ] ] with complete eigenspectrum @xmath54 .
we shall use the unitary transformation defined in the preceding section to produce the following hamiltonian with which to define our thermal states , writing the @xmath55 dependence explicitly .
@xmath56 where @xmath57 explicitly @xmath58 the eigenstates of @xmath59 are @xmath60 which are squeezed ( displaced ) number states .
the partition function @xmath61 for the hamiltonian eq.([ham1 ] ) is just the standard @xmath62 since the trace is invariant under the unitary transformation eq.([trans ] ) .
we define the thermal squeezed state @xmath63 by means of the hamiltonian eq.([ham1 ] ) in the usual way @xmath64 thermal averages in the state @xmath63 are immediate due to the unitary equivalence with the free thermal state @xmath65 defined in terms of the free boson hamiltonian . generally , for the operator @xmath48 , the thermal average is @xmath66 where @xmath67 is the free thermal average .
as mentioned above , such thermal averages are simply evaluated using the inverse transformation @xmath50 of eq.([inverse ] ) .
for example , the thermal averages of the quadratures are given by @xmath68 where @xmath69 is the average number of photons in the _ free thermal state_. we obtain here the same expressions as for the @xmath70-added squeezed state@xcite with @xmath71 substituted for @xmath70 .
in an important yet elementary paper , yuen@xcite showed that the absolute optimal value for the signal - to - noise ratio @xmath72 in a quantum system is given by @xmath73 , where @xmath74 is the photon intensity ( number of photons per unit time ) .
this limiting ratio is in fact attained by a standard squeezed state .
it requires only a straightforward calculation to evaluate the corresponding optimal ratio @xmath75 in a thermal squeezed state , following the methods of @xcite and @xcite .
we assume that the maximum photon intensity @xmath74 is used , that all the signal is in one quadrature @xmath76 , so that @xmath77 we then optimize @xmath78 with respect to @xmath79 subject to the constraint given by eqs([avs1 ] , [ avs2 ] ) @xmath80 eq.([sig2 ] ) is minimal for real @xmath81 , so that @xmath82 , giving @xmath83 the optimal value is given by @xmath84 which value is attained for @xmath85 as noted , @xmath71 is the average number of photons in the _ free thermal state _
@xmath86 in the figure we show how the yuen limit is degraded by increasing temperature @xmath87 for our thermal squeezed states .
( we have normalized @xmath88 to the optimal @xmath89 value @xmath90 , and measure @xmath87 in units of @xmath91 . )
another advantage of the unitary transformation approach described above is that it enables us to relate the thermalization of squeezed states to that of the free boson vacuum state . since we are not interested here in details of the dissipative evolution of our system , we eschew the more usual lindblad differential equation approach and use the global , kraus formulation@xcite .
this gives the dissipative transformation of the state @xmath92 as @xmath93 where @xmath94 and the operators ( matrices ) @xmath95 are otherwise arbitrary .
it is clear that this preserves both the positivity of @xmath92 and the tr @xmath96 property .
( perhaps less obvious is the fact that the set @xmath2 of transformation sets @xmath97 gives rise to a semi - group , since for any two elements @xmath98 and @xmath99 the set product @xmath100 also satisfies the kraus conditions , @xmath101 .
the semi - group @xmath2 has an identity , but the only elements with inverses are the single - element sets @xmath102 , where @xmath41 is unitary . ) as a simple 2-level illustration of the kraus procedure , we may describe the dissipation of the pure state @xmath103 $ ] to the mixed state @xmath104 $ ] by @xmath105 with @xmath106 \ ; \ ; \ ; w_1=\sqrt{1-p } \left [ \begin{array}{cc } 0 & \sqrt{1-p^2 } \\ 1 & -p \end{array } \right]\ ] ] of course the kraus transformation set @xmath107 is not unique .
perhaps the general pattern of this kraus set will emerge more easily if we write down the @xmath108 case .
the set of matrices @xmath97 which transform @xmath109 \longrightarrow \rho = \left[\begin{array}{ccc}p_0&0&0\\0&p_1&0\\0&0&p_2 \end{array } \right]\ ] ] is given by : @xmath110 \ ; \ ; w_1= \sqrt{p_1}\left [ \begin{array}{ccc } \begin{array}{c}0\\1\\0 \end{array } \left|\begin{array}{c } \mbox { } \underline{p^1_1 } \\
\mbox { } \end{array}\right| \begin{array}{c } \mbox { } \underline{p^1_2 } \\\mbox { } \end{array } \end{array } \right ] \ ; \ ; w_2= \sqrt{p_2}\left [ \begin{array}{ccc } \begin{array}{c}0\\0\\1 \end{array } \left|\begin{array}{c } \mbox { } \underline{p^2_1 } \\ \mbox { } \end{array}\right| \begin{array}{c } \mbox { } \underline{p^2_2 } \\\mbox { } \end{array } \end{array } \right]\ ] ] where the vectors @xmath111 satisfy the orthonormality relations @xmath112 as well as the probability conditions @xmath113 note that we are numbering our matrix elements from @xmath114 . in the general case
the conditions on @xmath115 are 1 .
orthonormality : @xmath116 2 .
probability matching : @xmath117 3 .
probability orthogonality : @xmath118 the foregoing permits an immediate description of the thermalization process for a squeezed state . consider a kraus set @xmath97 which thermalizes the pure state @xmath119 giving the free thermal state @xmath120 @xmath121 then the pure squeezed state @xmath122 dissipates to the thermal squeezed state @xmath123 using the kraus set @xmath124 , @xmath125 where @xmath126 .
we have shown that using a group theoretic definition of squeezed states , it is a straightforward matter to define a thermal squeezed state , which lends itself simply to evaluation of physical quantities .
we exemplified this by evaluating the yuen limit for the signal - to - noise ratio in such a thermal squeezed state as a function of the temperature @xmath87 , showing that it is reduced below the optimal @xmath89 value .
finally , we have outlined a general model for the thermalization of the standard squeezed state using the kraus formalism .
99 barnett s m and knight p l 1985 _ j opt .
b _ * 2 * 467 mann a , revzen m , nakamura k.,umezawa h and yamanaka y , 1989 _ j. math . phys._*30 * 2883 takahashi y and umezawa h 1975 _ collect . phenom._*25 * 55 vourdas a 1986 _ phys .
a _ * 34 * 3466 vourdas a and bishop r f 1994 _ physical review a _ * 50 * 3331 dodonov v v , manko v i and semjonov v v 1984 _
nuovo cimento b _ * 83 * , 145 vourdas a and weiner r m 1987 _ phys .
a _ * 36 * 5866 musslimani z h , braunstein s l , mann a and revzen m 1995 _ phys .
a _ * 51 * 4967 marian p and marian t a 1993 _ phys .
* 4474 marian p and marian t a 1993 _ phys . rev .
_ * 47 * 4487 glauber r j 1963 _ phys .
rev . _ * 130 * 2529 yuen h p 1976 _ phys .
* 13 * 2226 feng y and solomon a i 1998 _ optics communications_*152 * 299 yuen h p 1976 _ phys . lett._*56a * 105 solomon a i 1994 _ phys . lett.a
_ * 188 * 215 kraus k 1971 _ ann.phys . _ * 64 * 311 | propose simple definition thermal squeezed state previous definitions used thermofield approach takahashi umezawa early paper vourdas discusses superposition squeezed states with thermal photons recent paper considers only thermal coherent states early study of canonically transformed thermal states similar made in @xcite two further papers adopt same definition different approach method defining hamiltonian eigenstates squeezed states squeezed - coherent states complete set eigenstates are squeezed displaced number states states @xmath0 @xmath1 displacement operator @xmath2 squeezing operator thermal squeezed state mixed state thermal sum of states simplicity approach calculating thermal averages due to group - theoretical structure of states enables calculations to free hamiltonian averages structure consequence of definition squeezed - coherent states as transformed vacua under automorphism group canonical commutation relations physical properties thermalization calculate signal - to - quantum noise ratio for thermal squeezed states use kraus formalism propose model for thermalizationcanonical commutation relations boson photon creation annihilation operators @xmath3 @xmath4 @xmath5=i. consider canonical transformations @xmath6 quantum optics states vacuum states transformed operator @xmath7 vacuum state defined normalized state satisfying @xmath8 vacuum state @xmath7 defined @xmath9 @xmath10 ordinary vacuum state corresponds trivial unitary transformation @xmath11 @xmath12 @xmath13 consider important states coherent states correspond simple canonical transformation @xmath14 @xmath15 unit operator generated by unitary displacement operator @xmath16 operates @xmath4 @xmath17 vacuum state @xmath7 coherent state @xmath18 vacuum equation @xmath7 @xmath19 equivalent @xmath20 usual coherent state @xmath21 @xmath22 normalization properties minimizing heisenberg uncertainty principlesqueezed vacuum corresponds transformation @xmath23 @xmath24 @xmath25 @xmath26 satisfy @xmath27 @xmath28 squeezing operator by @xmath29 squeezed vacuum @xmath19 by @xmath31 satisfies @xmath32 2-parameter operator @xmath28 not general element 3-parameter vacuum - squeezing group @xmath33 additional operator @xmath34 @xmath35 supplies third parameter completes generators squeezed states correspond transformation @xmath36 @xmath37 squeezed state @xmath38 @xmath39 satisfies equation @xmath40 tabulate results in table 1 transformations corresponding states motivation explicit unitary transformation @xmath41 from vacuum to squeezed state transformation free thermal state to thermal squeezed state enables calculations thermal squeezed state 3-dimensional representation automorphism group canonical commutation relations by @xmath42 squeezing group @xmath435-parameter group semi - direct sum 3-dimensional heisenberg - weyl algebra central element acts trivially 3-dimensional algebra @xmath44 @xmath45 subgroup inhomogeneous pseudounitary group @xmath46 fundamental representation @xmath47\lambda\mu\alpha\mu value representation due operator @xmath48 inverse transformed operator @xmath49 obtained inverse matrix @xmath50 @xmath51 @xmath52 extensive use inverse transformation calculations define thermal squeezed state conventionally from hamiltonian starting point single - mode free boson hamiltonian harmonic oscillator @xmath53=1 complete eigenspectrum @xmath54 unitary transformation hamiltonian define thermal states @xmath55 dependence @xmath56 @xmath57 @xmath58 eigenstates @xmath59 @xmath60 squeezed displaced ) number states partition function @xmath61 hamiltonian eq.([ham1 ] standard @xmath62 trace invariant under unitary transformation eq.([trans ] define thermal squeezed state @xmath63 hamiltonian eq.@xmath64 thermal averages state @xmath63 immediate equivalence with free thermal state @xmath65 boson hamiltonian operator @xmath48 thermal average @xmath66 @xmath67 free thermal average thermal averages evaluated using inverse transformation @xmath50 of eq.([inverse ] ) thermal averages quadratures by @xmath68 @xmath69 average photons free thermal state same expressions @xmath70-added squeezed state@xcite @xmath71 substituted for @xmath70 optimal value signal - to - noise ratio @xmath72 quantum system @xmath73 @xmath74 photon intensity limiting ratio attained by standard squeezed state straightforward calculation evaluate optimal ratio @xmath75 thermal squeezed state maximum photon intensity @xmath74 used signal in one quadrature @xmath76 @xmath77 optimize @xmath78 @xmath79 constraint by eqs([avs1 ] avs2 ] ) @xmath80 eq.[sig2 minimal @xmath81 @xmath82 @xmath83 optimal value by @xmath84 attained for @xmath85 @xmath71 average photons free thermal state @xmath86 yuen limit degraded by increasing temperature @xmath87 thermal squeezed states normalized @xmath88 to optimal @xmath89 value @xmath90 measure @xmath87 in units @xmath91 unitary transformation approach thermalization squeezed states free boson vacuum state eschew lindblad differential equation approach use global kraus formulation@xcite dissipative transformation state @xmath92 as @xmath93 @xmath94 operators @xmath95 arbitrary preserves positivity of @xmath92 @xmath96 property set @xmath2 transformation sets @xmath97 semi - group @xmath98 @xmath99 set product @xmath100 satisfies kraus conditions @xmath101 semi - group @xmath2 identity only elements with inverses single - element sets @xmath102 @xmath41 unitary2-level illustration kraus procedure describe dissipation pure state @xmath103 mixed state @xmath104 @xmath105 @xmath106 w_1=\sqrt{1-p }\left\begin{array}{cc 0\sqrt{1-p^2 } 1 -p\end{array } kraus transformation set @xmath107 not unique general pattern kraus set write @xmath108 casematrices @xmath97 transform @xmath109\begin{array}{ccc}p_0&0&0\\0&p_1&p_2 \end{array given by @xmath110 w_1= \sqrt{p_1}\left \begin{array}{ccc }}0\\1\\0\end{array \underline{p^1_1 }\end{array}\right|\begin\underline{p^1_2 }\end{array w_2= \sqrt{p_2}\left\begin{array}{ccc }\begin{array{c}0\\0\\1 \end{array }|\begin{array{c\mbox\underline{p^2_1 } \end{array}\right|\underline{p^2_2 }\end{array } vectors @xmath111 satisfy orthonormality relations @xmath112 probability conditions @xmath113 numbering matrix elements from @xmath114 conditions on @xmath115 orthonormality : @xmath116 probability matching : @xmath117 3probability orthogonality @xmath118 permits description thermalization process squeezed state kraus set @xmath97 thermalizes pure state @xmath119 free thermal state @xmath120 pure state @xmath122 dissipates thermal @xmath123 kraus set @xmath124 @xmath126 group theoretic definition squeezed states straightforward define thermal squeezed state evaluation physical quantities yuen limit signal - to - noise ratio thermal state function temperature @xmath87 reduced below optimal @xmath89 value outlined general model thermalization standard squeezed state kraus formalism barnett s m knight p l 1985 467 mann revzen nakamura yamanaka y 1989 phys. * 2883 takahashi y umezawa h 1975 55 vourdas a 1986 phys 34 3466 bishop r f 1994 physical review 50 * 3331 dodonov manko semjonov 1984 145 vourdas weiner r 1987 phys 36 * 5866 musslimani z braunstein mann revzen m 1995 phys 51 * 4967 marian p marian t a 1993 phys4474 marian marian t 1993 phys rev 4487 glauber r 1963 phys 130 2529 yuen 1976 2226 feng y solomon 1998 optics communications 299 yuen 1976 . 105 solomon 1994 phys 188 215 kraus k 1971 ann.phys . 64 311 | 0.60711 | 3,296 | 1,788 | , Saving $0.1 in GPT-4. |
55,508 | in this paper , we study a realistic model of quintessential inflation with radiation and matter . by the analysis of the dynamical system and numerical work about the evolution of the equation of state and cosmic density parameter
, we show that this model is a good match for the current astronomical observation .
the conclusion we obtain is in favor of the model where the modular part of complex field plays the role of the inflaton whereas the argument part is the quintessence field .
the numerical calculation shows that a heteroclinic orbit ( solution of dynamical system ) interpolates between early - time de sitter phase ( an unstable critical point ) and a late - time de sitter attractor . | astronomical observation on the cosmic microwave background(cmb ) anisotropy @xcite , supernova type ia(snia ) @xcite and sloan digital sky survey(sdss ) @xcite converge on that our universe is spatially flat , with about seventy percent of the total density resulting from dark energy that has an equation of state @xmath0 and drives the accelerating expansion of the universe which began at a redshift of order one - half . the origin of the dark energy remains elusive from the point of view of general relativity and standard particle physics .
several candidates to represent dark energy have been suggested and confronted with observation : cosmological constant , quintessence with a single field @xcite or with n coupled field @xcite , phantom field with canonical @xcite or born - infield type lagrangian @xcite , k - essence @xcite and generalized chaplygin gas(gcg ) @xcite . among these models ,
the most typical ones are cosmological constant and quintessence which has caught much attention ever since its invention .
the idea of inflation is legitimately regarded as an great advancement of modern cosmology : it solves the horizon , flatness and monopole problem , and it provides a mechanism for the generation of density perturbations needed to seed the formation of structures in the universe . in standard inflationary models ,
the physics lies in the inflation potential .
the underlying dynamics is simply that of a single scalar field rolling in its potential .
this scenario is generically referred to as chaotic inflation in reference to its choice of initial conditions .
this picture is widely favored because of its simplicity and has received by far the most attention to date .
the properties of inflationary models are also tightly constraint by the recent result from the observation .
the standard inflationary @xmath1cdm model provides a good fit to the observed cosmic microwave background ( cmb ) anisotropy .
peeples and vilenkin proposed and quantitatively analyzed the intriguing idea @xcite that a substantial fraction of the present cosmic energy density could reside in the vacuum potential energy of the scalar field responsible for inflation ( quintessential inflation ) . after that , there were some models to be presented in succession .
however , most of these models involve non - renormalizable , special potentials and these potentials usually do not have a local minimum , making the conventional reheating process unoperative .
on the other hand , the peccei - quinn ( pq ) symmetry is the most elegant solution to the strong cp problem of qcd .
the global @xmath2 is a spontaneously symmetry breaking ( ssb ) .
weinberg @xcite and wilczek @xcite pointed out that there is a nambu - goldstone boson , the axion , associated with ssb .
such a global symmetry often arises in supersymmetric and superstring - inspired models .
a priori pq symmetry breaking scale is arbitrary , which can be taken a value between @xmath3 gev and @xmath4 gev . in most axion models pq symmetry
breaking happens at a temperature @xmath5 , when a complex scalar field @xmath6 , which carries @xmath2 charge , develops a vacuum expectation value . at temperatures below @xmath5
the renormalizable potential for @xmath6 is @xmath7 where @xmath8 .
however , the argument of @xmath9 is left undecided when the @xmath2 has been broken because of @xmath10 is independent of @xmath11 .
the massless @xmath12 degree of freedom is the axion : @xmath13 , where @xmath14 is the color anomaly of the pq symmetry @xcite .
the axion is massless at @xmath15 , but the axion develops a mass due to instanton effects at low temperatures . in the @xmath16 case ,
the axion mass is easy to concretize by using a potential @xmath17,\ ] ] where @xmath18 and the axion develops a mass @xmath19 .
the conception that the modular part of a complex field with pq symmetry may drive inflation is also not new @xcite .
another version of a complex field was proposed as a realistic candidate for inflation @xcite and , separately , for quintessence @xcite in a natural fashion .
the two - axion model has also been suggested for quintessence @xcite or inflation @xcite .
the axion model can also be used for quintessence @xcite and , separately , for inflation @xcite from extra dimensions .
recently , a fascinating model for quintessential inflation has been proposed by rosenfeld and frieman ( rf ) with aid of the axion theory . in their model ,
a renormalizable complex scalar field @xmath6 described by a lagrangian with a @xmath20 symmetry spontaneously broken at a high energy scale @xmath21 gev and explicitly broken by instanton effects at a much lower energy can account for both the early inflationary phase and the recent accelerated expansion of the universe .
the modular and argument parts of the field @xmath6 were identified with the inflaton and the quintessence fields respectively . in this paper
, we put the emphasis on the study of the dynamics of rf model with radiation and matter composition .
according to the phase space analysis and numerical calculation , we show that how the modular part of complex field plays the role of the inflaton whereas the argument part is the quintessence field .
phase space methods are particularly useful when the equations of motion are hard to solve analytically for the presence of radiation and dust matter .
in fact , the numerical solutions with random initial conditions are not a satisfying alternative because of these may not reveal all the important properties . therefore , combining the information from the critical points analysis with numerical solutions , one is able to give the complete classification of solutions according to their early - time and late - time behavior . there are stable and unstable critical points for rf model with radiation and dust matter composition .
critical points are always exact constant solution in the context of autonomous dynamical systems .
these points are often the extreme points of the orbits and therefore describe the asymptotic behavior . in our case
, the unstable critical points are corresponding to a de sitter behavior at very first moments of the universe .
the stable critical points are corresponding to the second stage of accelerated expansion began at a redshift @xmath22 and is still operative , which alleviates the fine tuning problem .
the equation of state @xmath23 varies with the cosmic evolution and approaches towards -1 asymptotically showing the existence of a cosmological constant at late times . if the solutions interpolate between critical points they can be divided into heteroclinic orbit and homoclinic orbit . in our case ,
the heteroclinic orbit connects unstable and stable critical points .
the numerical calculation shows that brought the universe back to the usual friedman - robertson - walker expansion , then the second stage of accelerated expansion began at @xmath24 .
for convenience , we investigate firstly the global structure of the dynamical system without radiation and matter .
the equations of motion for rf model are @xmath25 and the einstein equation is @xmath26 where @xmath27 equals @xmath28 is hubble parameter and @xmath29 . here
we define the energy densities : @xmath30 to gain more insights into the dynamical system , we introduce the new variables @xmath31 then the equations of motion could be reduced to @xmath32 and @xmath27 could be rewritten as @xmath33^{1/2}\ ] ] here we have @xmath34 and @xmath35 $ ] .
before we carry out the numerical study , we would like to analyze the system of equations qualitatively .
the critical point of the autonomous system eqs.([eq : auto1 ] ) is @xmath36 , where @xmath37 are defined by @xmath38 respectively .
so one can obtain the critical points : @xmath39 where @xmath40 is any integer . in order to investigate the stability of these critical points , we can linearize the eqs.([eq : auto1 ] ) about the critical points @xmath41 here @xmath42 and @xmath43 is short for the right hand sides of eqs.([eq : auto1 ] ) .
then we have the stability matrix @xmath44 @xmath45 the eigenvalues of the stability matrix are @xmath46 . \end{split}\end{aligned}\ ] ] it is clear that the critical points are always stable when @xmath40 is even because the real part of all the eigenvalues is negative .
so these critical points correspond to late - time attractor solutions .
when @xmath40 is odd , the fourth eigenvalue in eq.([eigenvalue ] ) is always larger than zero and the corresponding critical points are unstable , which are exact constant solution with de sitter behavior at early - time of the universe .
this is to say that the dynamical system has stable critical points at the minimum of the potential @xmath10 and the minimum of the potential @xmath47 .
next , let s read out the physical implications when the system is at the critical point regime .
the cosmic parameters are @xmath48 and the equation of state of the complex field is @xmath49 clearly , from eqs.([omega ] ) and ( [ w ] ) , one can find that @xmath50 , @xmath51 and @xmath52 at the late time attractor .
if we are to search for a realistic model , we must know exactly how the complex scalar field @xmath6 couples to ordinary matter . for convenience , we assume that coupling is very weak following traditional ideas .
in fact , the most significant feature of all the axion couplings is that they are proportional to @xmath53 : the larger the pq ssb scale , the weaker the axion couples .
furthermore , we have taken the @xmath54 gev so that couples can be neglected .
certainly , the ordinary matter still affect the evolution of @xmath6 via their contribution to the general expansion of the universe .
now we solve the equations of motion via a numerical calculation at the presence of radiation and matter and obtain the results that will confirm our qualitative analysis . to do this ,
we introduce the new variable @xmath55 therefore , the eqs.([eq : auto1 ] ) are reduced to @xmath56 and @xmath57 where @xmath58 denote the hubble parameter at an initial time .
@xmath59 is defined as @xmath60^{1/2}. \end{split}\ ] ] @xmath61 and @xmath62 are the cosmic density parameters for radiation and dust matter at the initial time .
note that when n goes to be very large , or at late - time , the contribution to @xmath27 from matter and radiation will become negligible .
solving these equations will give us some insights into the evolution of the fields and the quantities of interest .
the corresponding conclusions of numerical calculation are shown in fig.[fig1 ] and fig.[fig2 ] .
this numerical solution describes a heteroclinic orbit , which interpolate between an unstable de sitter critical point and a late - time de sitter attractor .
therefore , this model is more natural to explain the two stages of acceleration .
a point worth emphasizing is that the ordinary matter ( radiation and dust ) affect the evolution of the scalar field via their contribution to the general expansion of the universe because of the couples can be neglected in the model . in fig.[fig1 ] , the behavior of the equation of state parameter is shown . at initial time , @xmath50 corresponds the de sitter expansion ( inflationary phase ) , then it increases and becomes positive . after arriving at the value @xmath63
, the universe comes to the radiation dominated epoch and @xmath23 stays on a broad platform .
next , @xmath23 drops to zero and stays on a narrow platform .
finally , @xmath23 drops below zero and approaches to @xmath64 , which corresponds second stage of accelerated expansion .
the evolution of cosmic density parameters are shown in fig.[fig2 ] .
the argument part contribution @xmath65 stays at @xmath66 at the very first moments of the universe and becomes @xmath67 in late - time , which plays the role of quintessence field . on the other hand
, the modular part contribution @xmath68 plays the role of inflaton . during the whole evolution ,
the radiation energy density and the dust matter energy density become dominate respectively . therefore , we see that the constraints arising from cosmological nucleosynthesis and structure formation are satisfied .
in this paper , we have discussed the cosmological implication of a complex scalar model of quintessence inflation with radiation and matter . using the numerical calculation for this model ,
we show that the argument contribution is negligible at the early epoch of the universe while it becomes dominate with the time evolving and the evolution of the modular part is the opposite , which is a viable way to unify the description of the inflationary stage and the current accelerated expansion .
analysis to the dynamical evolution of the complex scalar model indicates that it admits a late - time attractor solution , at which the field behaves as a cosmological constant .
obviously , attractor and heteroclinic orbit are both alleviate the fine tuning problem .
we can also extend our discussion to the lagrangian with non - abelian symmetry , which is an interesting extended work and we will discuss in another preparing work .
this work is supported by national natural science foundation of china under grant no .
10473007 and by shanghai rising - star program under grant no . 02qa14033 .
bennett c _ et al _ 2003 _ astrophys j suppl _ * 148 * 1 tonry j l _ et al _ 2003 _ astrophys j _ * 594 * 1 tegmark m _
et al _ 2004 _ astrophys j _ * 606 * 702 peebles p j e and ratra b 2003 _ rev mod phys _ * 75 * 599 ; padmanabhan t 2003 _ phys rep _ 2003 * 380
* 235 boyle l a , caldwell r r and kamionkowski m 2002 _ phys lett _ * b545 * 17 ; li x z , hao j g and liu d j 2002 _ class quantum grav _ * 19 * 6049 ; gu j a and huang w y p 2001 _ phys lett _ * b517 * 1 ; hao j g and li x z 2004 _ class quantum grav _ * 21 * 4771 ; li x z and hao j g 2004 _ phys rev _
* d69 * 107303 hao j g and li x z 2003 _ phys rev _ * d67 * 107303 ; hao j g and li x z 2003 _ phys rev _ * d68 * 043501 ; liu d j and li x z 2003 _ phys rev _ * d68 * 067301 ; li x z and zhai x h 2003 _ phys rev _ * d67 * 067501 ; chen c y and shen y g 2004 _ chin phys lett _ * 21 * 2320 | astronomical observation on cosmic microwave background ) anisotropy @xcite supernova type ia(snia ) @xcite sloan digital sky survey ) @xcite converge on universe spatially flat seventy percent total density from dark energy equation state @xmath0 drives accelerating expansion began redshift one - half origin dark energy elusive general relativity standard particle physics candidates energy suggested : cosmological constant quintessence phantom field k - essence @xcite generalized chaplygin gas(gcg ) @xcite typical cosmological constant quintessence caught attention since invention idea of inflation great advancement of modern cosmology solves horizon flatness monopole problem provides mechanism for generation density perturbations formation structures standard inflationary models physics in inflation potential dynamics single scalar field rolling in potential scenario referred as chaotic inflation picture favored simplicity received most attention properties of inflationary models constraint by recent result observation standard inflationary @xmath1cdm model good fit to cosmic microwave background anisotropypeeples vilenkin proposed analyzed idea cosmic energy density in vacuum potential energy scalar field inflation models presented involve non - renormalizable special potentials local minimum conventional reheating unoperative peccei - quinn symmetry solution to cp problem qcd global @xmath2 symmetry breaking weinberg wilczek nambu - goldstone boson associated with ssb global symmetry arises in supersymmetric superstring - inspired models pq symmetry breaking scale arbitrary between @xmath3 @xmath4 gev models pq symmetry breaking at temperature @xmath5 complex scalar field @xmath6 develops vacuum expectation value temperatures below @xmath5 renormalizable potential for @xmath6 is @xmath7 @xmath8 argument of @xmath9 undecided when @xmath2 broken @xmath10 independent of @xmath11 massless @xmath12 degree freedom axion : @xmath13 @xmath14 color anomaly of pq symmetry axion massless at @xmath15 develops mass due to effects at low temperatures@xmath16 case axion mass potential @xmath17 @xmath18 axion develops mass @xmath19 modular part complex field pq symmetry drive inflation not new another version complex field proposed candidate for inflation @xcite quintessence @xcite two - axion model suggested for quintessence inflation axion model used for quintessence @xcite inflation @xcite from extra dimensions model quintessential inflation proposed by rosenfeld and frieman axion theory renormalizable complex scalar field @xmath6 lagrangian @xmath20 symmetry broken high energy scale @xmath21 gev broken effects lower energy account for early inflationary phase recent accelerated expansion universe modular argument parts field @xmath6 identified with inflaton quintessence fields emphasis study dynamics rf model with radiation matter composition phase space analysis numerical calculation modular part complex field inflaton argument part quintessence field phase space methods useful equations of motion hard solve presence radiation dust matter numerical solutions with random initial conditions not satisfying not reveal important propertiescombining critical points analysis with numerical solutions classification early late time behavior stable unstable critical points for rf model with radiation dust matter composition constant autonomous dynamical systems extreme orbits describe asymptotic behavior unstable points de sitter behavior first moments universe stable points second stage accelerated expansion @xmath22 operative fine tuning problem equation of state @xmath23 varies with cosmic evolution approaches towards -1 cosmological constant at late times solutions interpolate between points divided into heteroclinic homoclinic orbit heteroclinic orbit connects unstable stable points numerical calculation shows universe to friedman - robertson - walker expansion second stage accelerated expansion began at @xmath24 investigate global structure dynamical system without radiation matter equations of motion for are @xmath25 einstein equation @xmath26 @xmath27 @xmath28 hubble parameter @xmath29 define energy densities @xmath30 introduce new variables @xmath31 equations of motion reduced to @xmath32 @xmath27 rewritten as @xmath33 @xmath34 @xmath35 numerical study analyze system of equations qualitativelycritical point autonomous system eqs. auto1 is @xmath36 @xmath37 defined by @xmath38 obtain critical points @xmath39 @xmath40 is integer investigate stability linearize eqs. auto1 about critical points @xmath41 @xmath42 @xmath43 for right hand sides of eqs stability matrix @xmath44 @xmath45 eigenvalues are @xmath46 critical points stable when @xmath40 even part eigenvalues negative critical points correspond to late - time attractor solutions when @xmath40 odd fourth eigenvalue larger than zero critical points unstable constant solution with behavior at early - time universe dynamical system has stable critical points at minimum potential @xmath10 minimum @xmath47 physical implications when system at critical point regime cosmic parameters are @xmath48 equation of state complex field is @xmath49 eqs @xmath50 @xmath51 @xmath52 at late time attractor realistic model know how complex scalar field @xmath6 couples to ordinary matter assume coupling is weak traditional ideassignificant feature axion couplings proportional to @xmath53 larger pq ssb scale weaker axion couples taken @xmath54 gev couples neglected ordinary matter evolution of @xmath6 expansion universe solve equations of motion presence radiation matter results confirm qualitative analysis introduce new variable @xmath55 eqs. reduced to @xmath56 @xmath57 @xmath58 hubble parameter initial time @xmath59 defined as @xmath60^{1/2} @xmath61 @xmath62 cosmic density parameters for radiation dust matter initial time when n large late - time contribution to @xmath27 from matter radiation negligible solving equations insights into evolution of fields quantities of interest conclusions calculation shown in fig.[fig1 ] and fig.[fig2 ] solution describes heteroclinic orbit between unstable de sitter critical point late - time de sitter attractor model explain two stages of acceleration ordinary matter ( radiation dust ) affect evolution scalar field expansion couples neglected fig.[fig1 ] behavior of equation of state parameter showninitial @xmath50 corresponds de sitter expansion inflationary phase increases becomes positive value @xmath63 universe radiation dominated epoch @xmath23 stays broad platform @xmath23 drops to zero narrow platform finally drops below zero approaches @xmath64 second stage accelerated expansion evolution cosmic density parameters shown in fig.[fig2 ] argument part contribution @xmath65 stays at @xmath66 becomes @xmath67 late - time quintessence field modular part contribution @xmath68 role inflaton radiation energy density dust matter energy density dominate constraints cosmological nucleosynthesis structure formation satisfied discussed cosmological implication complex scalar model of quintessence inflation with radiation matter numerical calculation argument contribution negligible early epoch dominate time evolving evolution modular part opposite unify inflationary stage current accelerated expansion analysis dynamical evolution complex scalar model admits late - time attractor solution field behaves as cosmological constant attractor heteroclinic orbit alleviate fine tuning problem extend discussion lagrangian non - abelian symmetry interesting another preparing worksupported national science foundation china grant 10473007 shanghai rising - star program grant 02qa14033 bennett 2003 astrophys 148 1 tonry j 594 1 tegmark m 2004 606 702 peebles ratra b 2003 phys 75 599 padmanabhan t 2003 phys rep 380 235 boyle caldwell kamionkowski 2002 phys lett b545 17 li x z hao j g liu d j 2002 quantum 19 6049 gu j a huang 2001 b517 hao j g li x z 2004 21 4771 li x z hao j g 2004 phys rev d69 107303 hao j li x z 2003 d67 107303 hao j li x z 2003 d68 043501 liu d j li x z 2003 067301 li x z zhai 2003 d67 067501 chen c y shen y g 2004 phys 21 2320 | 0.50679 | 3,278 | 1,683 | , Saving $0.1 in GPT-4. |
55,511 | in a few years astrometry with the venerable combination of hubble space telescope and fine guidance sensor will be replaced by sim , gaia , and long - baseline interferometry .
until then we remain a resource of choice for sub - millisecond of arc precision optical astrometry . as examples we discuss 1 ) the uses which can be made of our parallaxes of galactic cepheids , and 2 ) the determination of perturbation orbital elements for several exoplanet host stars , yielding true companion masses | in the early 1970 s , bob odell , then director of yerkes observatory , twisted the arm of bill van altena , encouraging bill to propose to study the feasibility of doing astrometry with what was then called the space telescope ( st ) .
bill was joined by larry frederick and otto franz .
their efforts insured that astrometry with a fine guidance sensor ( fgs ) would be an integral science component for st . in 1977 bill
jefferys formed a team to respond to the st announcement of opportunity .
the first author was a member of that team , which then included raynor duncombe , paul hemenway , and pete shelus .
we were eventually chosen as a component ( along with van altena , franz , and fredrick ) of the final stat ( space telescope astrometry team ) .
barbara mcarthur joined us in 1985 .
st was launched in 1990 , becoming the hubble space telescope ( hst ) .
because of seismic disturbances caused by day / night cycle temperature effects on the original solar arrays and slow , but fortunately steady , changes in the fgs , we were unable to prove that precision astrometry could be carried out with the fgs until after the first refurbishment mission in late 1992 . with new , re - designed solar arrays , hst and
the two fgs units we have used over the years , originally predicted to produce single - measurement astrometry with 3 millisecond of arc ( mas ) precision , have both routinely produced 1 mas or better per - observation precision astrometry .
calibration issues , many of which are non - trivial , are discussed in mcarthur ( 2002 , 2006 ) , benedict ( 1994 , 2002a ) , and for binary star fringe morphology analysis , franz ( 1998 ) .
the many results from the original stat include parallaxes of astrophysically interesting stars ( benedict 2000 , 2002a , 2002b , 2003 , mcarthur 1999 , 2001 ) , a parallax for the hyades ( van altena 1997 ) , a link between quasars and the hipparcos reference frame ( hemenway 1997 ) , determination of low - mass binary star masses ( franz 1998 , benedict 2001 ) , and searches for jupiter - mass companions to proxima cen and barnard s star ( benedict 1999 ) .
since 1999 , barbara and i have carried on as general observers , contributing to the study of the lower main sequence mass - luminosity relationship ( henry 1999 ) , the intercomparison of dwarf novae ( harrison 2004 ) , and parallaxes of cataclysmic variables ( beuermann 2003 , 2004 ; roelofs 2007 ) and the pleiades ( soderblom 2005 ) .
jacob bean , now a postdoc , joined us as a graduate student in 2002 .
two major themes of our investigations over the last few years have included the cosmic distance scale and extrasolar planetary systems .
the remainder of this paper will outline recent results , specifically drawing from our work ( benedict 2007 ) on the galactic cepheid period - luminosity relationship ( plr ) , and the determination of extrasolar planetary masses ( benedict 2002c , mcarthur 2004 , benedict 2006 , bean 2007 ) .
our goal was to determine trigonometric parallaxes for nearby fundamental mode galactic cepheid variable stars .
our target selection consisted in choosing the nearest cepheids ( using _ hipparcos _
parallaxes , perryman 1997 ) , covering as wide a period range as possible .
these stars were the brightest known cepheids at their respective periods .
our new parallaxes provided distances and ultimately absolute magnitudes , @xmath0 , in several bandpasses . additionally , our investigation of the astrometric reference stars provided an independent estimation of the line of sight extinction to each of these stars , a contributor to the uncertainty in the absolute magnitudes of our prime targets .
these cepheids , all with near solar metallicity , should be immune to variations in absolute magnitude due to metallicity variations , _
e.g. _ @xcite .
adding our previously determined absolute magnitude for @xmath1 cep , @xcite , we established v , i , k , and w@xmath2 period - luminosity relationships using ten galactic cepheids with average metallicity , @xmath3[fe / h]@xmath4=0.02 , a calibration that can be directly applied to external galaxies whose cepheids exhibit solar metallicity .
details are given in @xcite .
one of our interesting results is an independent determination of the distance modulus of the large magellanic cloud ( lmc ) , particularly given the recent exploration of possible socialogical influences on such determinations ( schaefer , 2007 ) .
ogle ( udalski 1999 ) has produced the largest amount of lmc cepheid photometry .
we worked with an apparent w@xmath2 plr for 581 cepheids in the lmc .
these data were carefully preened , selecting only cepheids with normal light curves and amplitudes from @xcite .
they provide a zero - point of 12.65 @xmath5 0.01 .
direct comparison of that w@xmath2 zero - point with the zero - point of the left - hand m@xmath6 plr in figure [ fig1 ] yields an lmc distance modulus 18.51 @xmath50.03 with no metallicity corrections .
@xcite demonstrate that a metallicity correction is necessary by comparing metal - rich cepheids with metal - poor cepheids in ngc 4258 . with a previously measured [ o / h ] metallicity gradient from @xcite , macri
find a cepheid metallicity correction in w@xmath2 , @xmath7 magnitude for 1 dex in metallicity , where r and s subscripts signify random and systematic .
this value is similar to an earlier w@xmath2 metallicity correction ( kennicutt et al .
1998 ) derived from cepheids in m101 ( -0.24@xmath5 0.16 ) . taking the weighted mean of the kennicutt and macri values and using the difference in metallicity of lmc and galactic cepheids ( -0.36 dex from means of the data in groenewegen 2004 tables 3 and 4 ) we find a metallicity correction of -0.10@xmath5 0.03 magnitude with the galactic cepheids being brighter . with this estimated metallicity correction
we obtain a corrected lmc modulus of 18.41 @xmath5 0.05 .
one other recent determination is noteworthy for its lack of dependence on any metallicity corrections . @xcite
derive 18.42@xmath5 0.04 , from eclipsing binaries , a modulus in close agreement with our new value . as an additional check on our adopted lmc modulus , m - m= 18.41 @xmath5 0.05
, a macri ( 2006 ) differential modulus ( lmc - ngc 4258 ) leads to a modulus of 29.29 @xmath5 0.08 for ngc 4258 , a value in excellent agreement with the maser - based distance for ngc 4258 , m - m = 29.29 @xmath5 0.15 from @xcite .
one of the first uses of our new cepheid parallaxes was the calibration of various techniques used to provide distances for many more long period ( and all far more distant ) cepheids as described in @xcite .
these techniques include infrared surface brightness and interferometric badde - wesselink , which compare measured physical diameters with radial velocities .
this calibration results in a densification and extension of the galactic cepheid plr , and is illustrated in figure [ fig1 ] .
the immediate value of the densification is a greater confidence in the absolute magnitudes of the longer - period , brighter cepheids , the ones critical for extragalactic distance determination .
currently , fewer than 10% of the more than 200 candidate exoplanets + ( http://exoplanet.eu ) orbiting nearby stars have precisely determined masses . because the most successful technique for detecting candidate exoplanets , the radial velocity method , suffers from a degeneracy between the mass and orbital inclination for most of the known exoplanet candidates , only a minimum mass , @xmath8 , is known . in principle
, radial velocities alone can be used to determine the masses of exoplanets in multiplanet systems , if two or more planets are experiencing significant mutual gravitational interactions on short timescales . however , only one such system has been investigated , results from different groups vary significantly , and the effects of non - coplanarity remain to be considered .
thus , establishing precise masses , rather than arguing statistically ( using @xmath8 , which everyone does , e.g. the otherwise excellent review of udry & santos 2007 ) , for the majority of exoplanet candidates requires additional observations .
the primary techniques that have been employed to break the mass - inclination degeneracy in radial velocity data are astrometry ( benedict 2002c , mcarthur 2004 , benedict 2006 , bean 2007 ) and transit ( e.g. , bakos 2007 ) observations . ideally , all candidates should have their masses determined .
obviously , not all systems are edge - on to us . and
, unfortunately , the size of the perturbation due to the companion decreases linearly with distance , placing many of them beyond the reach of hst fgs astrometry .
to partially rectify this lack of knowledge of the true masses , the hst allocation process awarded us over 100 orbits in cycle 14 and 15 to be used to determine actual masses for hd 47536 b , hd 136118 b , hd 168443 c , hd 145675 b , hd 38529 c , and hd 33636 b. the object chosen for discussion in this contribution , hd 33636 ( bean 2007 ) presents a cautionary tale . as we shall see , at least in some cases , @xmath8 is not a very good estimator of the actual companion mass .
we chose to work first on hd 33636 because we had finished collecting all the hst data for this object .
radial velocity data ( figure 2 ) were obtained from past investigators ( lick and keck data from butler 2006 and marcy ( 2007 , private communication ) , elodie data from perrier ( 2003 ) ) and from a high - cadence campaign with the hobby - eberly telescope high - resolution spectrograph ( hrs ; tull 1998 ) .
the latter data were used to search for shorter - period companions , the unknown and unmodeled presence of which would have introduced noise into our velocity data .
no such companions were found with detection limits detailed in bean ( 2007 ) .
limits depend on period and assumed eccentricity of the tertiary .
for example these velocity data would have permitted detection of a tertiary with @xmath14 for @xmath15 and eccentricity @xmath16 .
our techniques for combining radial velocity and hst fgs astrometry have been detailed in benedict ( 2001 , 2006 ) and bean ( 2007 ) . for hd 33636 we found an inclination , @xmath17 and , as shown in figure 3 , a perturbation semimajor axis , @xmath18 mas . assuming a mass for the primary , hd 33636 , @xmath19 @xmath20 ( takeda 2007 )
, we obtain a mass for hd 33636 b , @xmath21 @xmath20 .
hd 33636 b is actually hd 33636 b , an m6v star .
including it in any statistical studies depending on @xmath8 , such as metallicity , eccentricity , and semimajor axis , would corrupt those relationships . .
the inclination , @xmath22 , and size of the semi - major axis of the perturbation , @xmath23 mas , identify the companion as an m dwarf star .
the box marks the date of the next periastron passage.,width=384 ] we have made some progress .
table 1 collects the mass estimates that have already resulted from our combination of hst astrometry and high - precision radial velocities . over the next year
we will finish our analysis and determine actual masses for hd 47536 b , hd 136118 b , hd 168443 c , hd 145675 b , and hd 38529 c. but , before those results , we will have finished our analysis of the @xmath24 andromedae system ( mcarthur 2008 ) .
we have a preliminary indication that for the first time the degree of coplanarity of an extrasolar planetary system associated with a normal , main sequence star will be established .
beyond that , we are in the process of acquiring hst fgs data for similar coplanarity tests on the multiplanet systems associated with hd 128311 , hd 202206 , @xmath25 ara , and @xmath26 cep , paving the way for future studies of hundreds of such systems with ground - based long - baseline interferometry and the space - based projects , sim and gaia . | early 1970 s bob odell director yerkes observatory twisted arm bill van altena study feasibility astrometry with space telescope st ) bill joined by larry frederick otto franz efforts insured astrometry with fine guidance sensor fgs integral science component for st 1977 bill jefferys formed team respond st announcement first author member team included raynor duncombe paul hemenway pete shelus chosen van altena franz fredrick final stat space telescope astrometry team barbara mcarthur joined 1985 st launched 1990 hubble space telescope hst ) seismic disturbances temperature effects changes in fgs prove precision astrometry fgs until after first refurbishment mission late 1992 new - solar arrays hst two fgs units predicted 3 millisecond arc precision produced 1 mas or better per - observation precision astrometry calibration issues discussed in mcarthur ( 2002 2006 ) benedict ( 1994 binary star fringe morphology analysis franz ( 1998 )results original stat include parallaxes interesting stars benedict 2000 2003 mcarthur 1999 2001 parallax hyades 1997 link between quasars hipparcos reference frame 1997 low - mass binary star masses franz 1998 benedict 2001 searches jupiter - mass companions proxima barnard s star 1999 since 1999 barbara and i general observers lower main sequence mass - luminosity relationship henry 1999 intercomparison dwarf novae harrison 2004 parallaxes cataclysmic variables beuermann 2003 roelofs 2007 pleiades soderblom 2005 jacob bean postdoc joined graduate student 2002 major themes investigations cosmic distance scale extrasolar planetary systems remainder paper recent results work 2007 galactic cepheid period - luminosity relationship determination extrasolar planetary masses 2004 2006 2007 goal determine trigonometric parallaxes for nearby fundamental mode galactic cepheid variable stars target selection nearest cepheids hipparcos parallaxes 1997 wide period range brightest cepheids periods new parallaxes provided distances absolute magnitudes several bandpassesinvestigation astrometric reference stars provided estimation line of sight extinction uncertainty absolute magnitudes prime targets cepheids near solar metallicity immune to variations magnitude metallicity variations @xcite adding absolute magnitude for @xmath1 cep @xcite established v k w@xmath2 period - luminosity relationships using ten galactic cepheids with average metallicity @xmath3@xmath4=0.02 calibration applied to external galaxies solar metallicity details in @xcite independent determination distance modulus of large magellanic cloud socialogical influences ogle produced largest lmc cepheid photometry worked with w@xmath2 plr for 581 cepheids data preened normal light curves amplitudes from @xcite provide zero - point of 12.65 @xmath5 0.01 comparison with m@xmath6 plr yields lmc distance modulus 18.51 @xmath50.03 no metallicity corrections metallicity correction necessary comparing metal - rich cepheids with metal - poor in ngc 4258measured metallicity gradient from @xcite macri find cepheid metallicity correction w@xmath2 @xmath7 magnitude 1 dex metallicity random systematic similar w@xmath2 metallicity correction kennicutt 1998 from cepheids m101 -0.24@xmath5 0.16 weighted mean kennicutt macri values difference metallicity lmc galactic cepheids -0.36 dex groenewegen 2004 tables 3 4 metallicity correction -0.10@xmath5 0.03 magnitude galactic cepheids brighter correction corrected lmc modulus 18.41 @xmath5 0.05 dependence metallicity corrections @xcite 18.42@xmath5 0.04 from eclipsing binaries modulus new value check lmc modulus m - m= 18.41 @xmath5 0.05 macri 2006 differential modulus 4258 leads to modulus 29.29 @xmath5 0.08 for ngc 4258 maser - based distance ngc 4258 = 29.29 @xmath5 0.15 @xcite cepheid parallaxes techniques distances long cepheids infrared surface brightness interferometric badde - wesselink diameters radial velocitiescalibration results in densification extension of galactic cepheid plr illustrated in figure immediate value densification is greater confidence in magnitudes of longer - period brighter cepheids critical for extragalactic distance determination fewer than 10% of 200 candidate exoplanets orbiting nearby stars have precisely determined masses successful technique radial velocity method suffers from degeneracy between mass orbital inclination for candidates only minimum mass , @xmath8 known radial velocities can be used to determine masses of exoplanets in multiplanet systems if planets experiencing significant mutual gravitational interactions short only one system investigated results from groups vary effects of non - coplanarity remain to be considered establishing precise masses using @xmath8 requires additional observations primary techniques mass - inclination degeneracy are astrometry transit observations all candidates should have masses determined not all systems are edge - on size of perturbation due to companion decreases with distance many beyond reach of hst fgs astrometrylack knowledge masses hst allocation process awarded 100 orbits cycle 14 15 masses for hd 47536 b hd 136118 b hd 168443 c hd 145675 b hd 38529 c hd 33636 b hd 33636 ( bean 2007 ) presents cautionary tale @xmath8 not good estimator companion mass chose first on hd 33636 finished collecting hst data radial velocity data obtained from past investigators elodie high - cadence campaign with hobby - eberly telescope high - resolution spectrograph shorter - period companions noise no companions found detection limits bean ( 2007 ) limits depend on period assumed eccentricity tertiary data detection tertiary with @xmath14 @xmath15 eccentricity @xmath16 techniques combining radial velocity hst fgs astrometry detailed in benedict 2001 2006 bean ( 2007 ) for hd 33636 found inclination @xmath17 perturbation semimajor axis , @xmath18 mas assuming mass for primary hd 33636 @xmath19 @xmath20 obtain mass for hd 33636 b @xmath21 @xmath20hd 33636 b is m6v star in statistical studies @xmath8 corrupt relationships inclination @xmath22 size of semi - major axis perturbation @xmath23 identify companion as m dwarf star box marks date next periastron passage progress table 1 collects mass estimates from hst astrometry high - precision radial velocities next year finish analysis determine masses for hd 47536 b hd 136118 b hd 168443 c hd 145675 b hd 38529 c before finished analysis of @xmath24 andromedae system 2008 preliminary indication degree coplanarity of extrasolar planetary system with normal main sequence star established acquiring hst fgs data for coplanarity tests on multiplanet systems hd 128311 hd 202206 @xmath25 ara @xmath26 cep for future studies interferometry space - based projects | 0.438407 | 2,981 | 1,554 | , Saving $0.1 in GPT-4. |
55,517 | we consider the period of a simple pendulum in the gravitational field of the spherical earth . effectively , gravity is enhanced compared with the often used flat earth approximation , such that the period of the pendulum is shortened .
we discuss the flat earth approximation , and show when the corrections due to the spherical earth may be of interest . | the experimental fact that all objects fall with the same acceleration is known as the weak equivalence principle @xcite .
the first systematic tests of the universality of free fall were done by galileo @xcite , who measured the acceleration of freely falling objects , or of objects rolling down an inclined plane .
such an approach suffers from great inaccuracies , which are related to the short time scales involved .
indeed , galileo was unable to accurately determine @xmath0 , the gravitational acceleration at the surface of the earth , due to the lack of an accurate clock .
galileo concluded that a much better way to check the universality of free fall was to use a pendulum .
newton , and following him others , improved on galileo s experiments , and were able to determine @xmath0 quite accurately .
( notice that it is much harder to measure @xmath0 than it is to test the weak equivalence principle . ) because the period is cumulative , by measuring the time over many cycles of the pendulum one can increase the accuracy in the measurement of @xmath0 significantly .
is the value of the gravitational acceleration as determined by free - fall experiments identical to the value of the gravitational acceleration as determined by the period of a pendulum ? in this paper we shall study this question , and show that the spherical geometry of the earth affects the _ effective _ gravitational acceleration differently in the two types of experiments : the effective gravitational acceleration is greater than @xmath0 for pendulum experiments , and smaller than @xmath0 for free - fall experiments .
the approximation that both galileo and newton used ( and which is also used in virtually all physics textbooks from the high school level to the advanced undergraduate level @xcite ) , is to make the following two assumptions on the earth s gravitational field .
first , one assumes that the gravitational acceleration is independent of the altitude above the surface of the earth .
that is , one assumes that during its motion , a freely falling object or the pendulum s bob experiences a constant acceleration due to gravity , such that the variation of the gravitational acceleration with altitude ( `` vertical inhomogeneities '' of the gravitational field ) is neglected .
second , for the case of a pendulum , one assumes that the gravitational field lines are parallel .
that is , one neglects the variations in the direction of the gravitational field ( `` lateral inhomogeneities '' of the gravitational field ) which come about because the gravitational field lines direct towards the center of the earth . effectively , this latter approximation amount to adopting a model for the earth wherein the earth is flat .
we shall dubb this latter approximation henceforth as the `` flat earth '' ( fe ) approximation .
the former approximation , when taken in tandem with the fe approximation , is equivalent to assuming that the flat earth extends indefinitely is all directions . as in usual pendulum experiments both assumptions
are regularly made , we shall refer by the fe approximation to both assumptions taken together .
it turns out that the values of the gravitational acceleration as determined by free fall or by using a pendulum are not identical if one uses the flat earth approximation ( even without neglecting the vertical inhomogeneities ) . specifically , the finite size of the spherical earth acts in opposite ways : it effectively _ decreases _ the gravitational acceleration in free fall experiments , and effectively _ increases _ the gravitational acceleration in pendulum experiments . under most circumstances ,
the deviations of the effective gravitational acceleration from @xmath0 due to the spherical earth are tiny : the relative change in @xmath0 is of order of the ratio of the length of the arm of the pendulum to the radius of the earth . for short pendula ,
the systematic error involved in neglecting this effect is smaller than the systematic errors in the neglections of other effects , notably the finite amplitude effect , the mass of the wire which suspends the bob , the finite size of the bob , decay of finite amplitudes , or the buoyancy effect @xcite .
nelson and olsson @xcite listed many systematics , and included many sources for error which affect the measurement of @xmath0 with a pendulum only minutely .
even with a wire of moderate length ( nelson and olsson used a wire whose length was 3 m ) , the fe approximation is responsible for a systematic error which is considerably larger than some of the systematics which nelson and olsson did discuss . with longer arm ,
the spherical earth effect becomes even more important .
as we shall see , for pendula of lengths comparable with the usual length of the foucault pendulum , the spherical earth effect is comparable to the finite amplitude effect . and
yet , the effect of the spherical earth is usually neglected in most treatments .
the spherical earth effect was treated briefly by gough @xcite , who assumed an infinitely small amplitude .
it turns out , however , that the spherical earth effect couples with finite amplitudes in a rather complicated manner . in this paper
we shall consider the effect of a spherical earth on the period of a simple pendulum , and show that this effect acts to shorten the period .
that is , when one uses the fe model and fits experimental data to the parameters of that model , one gets a value for the gravitational acceleration which is slightly higher than the actual one . as the fe approximation is used so frequently at all level of instruction ( from the high school level to the advanced undergraduate level ) , it is instructive to understand the nature of the fe approximation .
in addition , the analysis in the paper is appropriate for classroom instruction or as a homework assignment at the intermediate or advanced undergraduate levels .
consider a simple pendulum of arm length @xmath1 , which moves under the earth s gravity , in the absence of any non - gravitational forces ( such as friction ) .
we assume that the pendulum moves in a perfect vacuum , is suspended from a perfectly rigid support by an inextensible massless string , and that the bob is a point .
we further assume that the earth is perfectly spherical .
the radius of the earth is @xmath2 , and its mass is @xmath3 . at the surface of the earth
the gravitational acceleration then is @xmath4 , where @xmath5 is newton s constant .
the opening angle of the pendulum is @xmath6 .
( see figure [ fig1 ] . )
we assume that the arm of the pendulum is extensionless and weightless , such that the entire mass of the pendulum @xmath7 is in the bob . denoting by @xmath8 the distance of the bob from the center of the earth , we find that @xmath9 .
the energy per unit mass @xmath10 of the bob is @xmath11 where an overdot denotes differentiation with respect to time . denoting @xmath12 ,
@xmath13 conservation of energy implies that @xmath10 equals at all times its value at the turning point of the motion , where @xmath6 equals its maximum value @xmath14 . at the turning point
@xmath15 , such that @xmath16 epsf 7.0 cm define next @xmath17 , and find that @xmath18 where @xmath19 .
notice that for small values of @xmath20 , @xmath21 is at order @xmath20 .
the period of the motion @xmath22 is given by @xmath23 this integral can be evaluated numerically .
however , we can gain much insight into it by approximating the integrand .
as @xmath24 , we can expand the expression inside the brackets as @xmath25 neglecting terms which are at order @xmath26 .
then , to that order , we find that @xmath27 the integral is the usual complete elliptic integral ( see appendix [ appa ] ) @xcite , such that @xmath28 recall that this result is accurate only to @xmath29 .
we write in eq .
( [ e8 ] ) the result in terms of the elliptic integral just for easy notation and comparison with the usual result of the fe model . in the limit of @xmath30
( radius of the earth goes to infinity , flat earth model " ) , we recover the regular result for the period of a simple pendulum , i.e. , @xmath31 . for finite values of @xmath20 , however , we find that the result can be expressed through an effective gravitational acceleration @xmath32 , which means that effectively gravity for a pendulum is enhanced in the spherical earth compared with the flat earth model .
consequently , we find that the period of a pendulum when the spherical earth is taken into consideration is shorter than in the flat earth model .
specifically @xmath33 we can also write the solution including higher - order corrections in @xmath20 . in this case
one can no longer write the solution easily in closed form . instead
, one can write it as a series expansion as @xmath34 \ , . \label{e9}\ ] ] notice that the period in eq .
( [ e8 ] ) separated into a product of a function of @xmath20 and a function of @xmath35 .
this happened just because of the small @xmath36 approximation we made .
in general , the period is not separable , as is clear from eq .
( [ e9 ] ) . as the elliptic function @xmath37 , in the limit of @xmath30 eq .
( [ e9 ] ) indeed agrees with eq .
( [ e8 ] ) .
notice that the usual effect of finite amplitude is to lengthen the period of the pendulum .
the effect of the spherical earth is in the opposite direction : it shortens the period .
this shortening of the period comes about because effectively the gravitational acceleration is enhanced compared with its value at the surface of the earth . to understand this heuristically ,
consider the forces which act on the bob .
because the bob is at height @xmath38 above the surface of the earth , the gravitational acceleration there is given by @xmath39 . because the weight of the bob directs towards the center of the earth ,
its angle with the string of the pendulum is not @xmath6 , but rather @xmath40 ( see fig .
[ fig1 ] for the definitions of the geometrical quantities . )
consequently , the projection of the bob s weight perpendicularly to the pendulum s arm , which is the force @xmath41 which determines the torque which drives the oscillations , is given by @xmath42 simple geometrical considerations show that @xmath43 , and @xmath44 .
expanding @xmath41 in small @xmath45 , and keeping only the leading terms in @xmath46 and @xmath45 , we find that @xmath47 this has the same form as the force which acts on the bob in the flat earth model , but with a slightly enhanced gravitational acceleration .
gravity being effectively stronger , it is clear why the period of the pendulum s oscillations shortens .
as the period of the pendulum is inversely proportional to the square root of the gravitational acceleration , we immediately recover eq .
( [ period ] ) . in the flat earth model
we effectively take the limit as @xmath48 .
this implies that when we compute the gravitational acceleration using a pendulum , we in fact measure @xmath49 , not @xmath0 , that is , we measure a slightly larger acceleration .
of course , with reasonable length for the pendulum , this effect is smaller than typical experimental errors . compare this with the gravitational acceleration which we measure using a free - fall experiment ( or an inclined surface ) .
the gravitational acceleration decreases with height , such that at all times it is slightly smaller than @xmath0 at the earth s surface .
specifically , the gravitational acceleration is given by @xmath50 .
when we use the flat earth ( with infinite extension ) model , we in fact neglect this small variation in the gravitational acceleration .
that is , when measuring the gravitational acceleration using a free fall experiment , we are in effect measuring @xmath50 , not @xmath0 .
notice that @xmath51 , whereas @xmath52 .
that is , when we use the flat earth model we measure different quantities for the gravitational acceleration when we do a pendulum experiment or a free fall experiment .
the fractional change in the pendulum s period being at order @xmath45 implies that this effect is very small under most circumstances .
as @xmath53 m , even if we take a long pendulum , of length @xmath54 m , the change in the period would be just one part in @xmath55 .
this effect , which is typically ignored in textbooks , however , can be at the same order of magnitude as other effects , which many standard textbooks do discuss @xcite .
specifically , one can choose reasonable parameters for which this effect is at the same order as the usual lengthening of the period due to finite amplitude effects .
in fact , because these two effects are competing , one can easily choose parameters for which the spherical earth effect compensates for the finite amplitude effect at leading order , such that dependence on amplitude would be at order @xmath56 ( rather than @xmath57 ) . comparing the two effects , we find that they are comparable if @xmath58 .
take , say , @xmath59rad
. the two effects are comparable if @xmath60 , or @xmath61 m .
this length is the length of the foucault pendulum in the science museum of virginia , and about the length of many foucault s pendula which are used in many science museums for demonstration of the earth s rotation .
the length of the arc of the pendulum s motion then is about @xmath62 m , which is large enough to be set conveniently .
i thank richard price for stimulating discussions .
this research was supported by the national science foundation through grant no .
in this appendix we show how to bring the integral in eq .
( [ per ] ) to the familiar form of the complete elliptic integral of the first kind .
the strategy is to change the variables such that the interval of integration would be from @xmath63 to @xmath64 .
we choose then a new variable @xmath65 , defined by @xmath66 .
the measure of the integral is @xmath67 substituting , we find that @xmath68 which is nothing but the complete elliptic integral @xcite @xmath69 .
i. ciufolini and j. a. wheeler , _ gravitation and inertia _
( princeton university press , princeton , nj , 1995 ) g. galilei _ dialogues concerning two new sciences _ ( dover , new york , 1954 ) h. goldstein , _ classical mechanics _ , 2nd ed ( addison - wesley , reading , ma , 1980 ) j. b. marion and s. t. thornton , _ classical dynamics of particles and systems _ , 4th ed ( saunders , fort worth , 1995 ) r. a. nelson and m. g. olsson , ` the pendulum rich physics from a simple system ' , am .
j. phys .
* 54 * ( 2 ) , 112 - 121 ( 1986 ) .
w. gough , ` the period of a simple pendulum is not @xmath70 ' , eur .
j. phys .
* 4 * , 53 ( 1983 ) . g. b. arfken and h. j. weber , _ mathematical methods for physicists _ , 5th ed ( hartcourt , san diego , 2001 ) i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series , and products _ , 6th ed ( academic press , san diego , 2000 ) | experimental fact all objects fall with same acceleration known as weak equivalence principle @xcite first tests of universality free fall done by galileo measured acceleration of freely falling objects approach suffers inaccuracies related to short time scales galileo unable to accurately determine @xmath0 gravitational acceleration at surface earth due to lack of accurate clock concluded better check free fall use pendulum newton others improved galileo s experiments @xmath0 accurately harder to measure @xmath0 than weak equivalence principle period cumulative measuring time over many cycles of pendulum increase accuracy @xmath0 is gravitational acceleration by free - fall experiments identical to period pendulum ? paper study spherical geometry of earth affects effective gravitational acceleration differently in two types experiments acceleration greater than @xmath0 for pendulum experiments smaller than @xmath0 for free - fall experiments approximation galileo and newton used used in all physics textbooks two assumptions on earth s gravitational field gravitational acceleration independent of altitude above surface earthone assumes during motion freely falling object or pendulum s bob experiences constant acceleration due to gravity variation gravitational acceleration with altitude vertical inhomogeneities neglected for pendulum assumes gravitational field lines parallel neglects variations in direction field ( lateral inhomogeneities lines direct towards center earth approximation adopting model for earth flat dubb approximation as `` flat earth '' ( fe ) approximation approximation with fe equivalent to assuming flat earth extends indefinitely all directions pendulum both assumptions regularly made refer by fe approximation to both assumptions together values of gravitational acceleration determined by free fall or pendulum not identical if flat earth approximation even without neglecting vertical inhomogeneities finite size of spherical earth acts opposite decreases acceleration in free fall increases in pendulum experiments deviations of effective gravitational acceleration from @xmath0 due to spherical earth tiny relative change in @xmath0 of ratio of length of arm pendulum to radius earthshort pendula systematic error neglecting this effect smaller than other effects finite amplitude effect mass of wire bob finite size bob decay of finite amplitudes buoyancy effect @xcite nelson and olsson listed many systematics included sources for error affect measurement of @xmath0 with pendulum minutely even wire moderate length used 3 m fe approximation responsible for systematic error larger than systematics nelson olsson longer arm spherical earth effect more important for pendula lengths comparable with length foucault pendulum spherical earth effect comparable to finite amplitude effect effect spherical earth usually neglected in most treatments spherical earth effect treated briefly by gough @xcite assumed small amplitude spherical earth effect couples with finite amplitudes complicated paper consider effect of spherical earth on period of simple pendulum period fe model fits experimental data value for gravitational acceleration slightly higher than actual fe approximation used frequently at all level instruction high school to level instructive to understand nature analysis appropriate for classroom instruction or homework assignment at intermediate advanced undergraduate levelspendulum arm length @xmath1 moves under earth s gravity non - gravitational forces friction pendulum moves in perfect vacuum suspended from rigid support by inextensible massless string bob point earth spherical radius @xmath2 mass @xmath3 gravitational acceleration @xmath4 @xmath5 newton s constant opening angle pendulum @xmath6 figure arm pendulum extensionless weightless entire mass @xmath7 in bob @xmath8 distance bob from center earth @xmath9 energy per unit mass @xmath10 bob @xmath11 overdot denotes differentiation time @xmath12 @xmath13 conservation of energy @xmath10 equals value turning point motion @xmath6 equals maximum value @xmath14 turning point @xmath15 @xmath16 7.0 cm define @xmath17 @xmath18 @xmath19 small values of @xmath20 @xmath21 order @xmath20 period of motion @xmath22 given by @xmath23 integral evaluated numerically insight approximating integrand @xmath24 expand expression inside brackets as @xmath25 neglecting terms order @xmath26 @xmath27 integral is usual complete elliptic integral appendix @xcite @xmath28 result accurate to @xmath29 write in eq . ( [ e8 ] ) result elliptic integral for notation comparison with result fe model limit of @xmath30 radius earth infinity flat earth model recover regular result for period simple pendulum @xmath31 finite values of @xmath20 result expressed through gravitational acceleration @xmath32 gravity for pendulum enhanced in spherical earth flat earth model period of pendulum spherical earth shorter than flat earth model @xmath33 write solution including higher - order corrections in @xmath20 closed form as series expansion as @xmath34 period in eq . ( [ e8 ] ) separated into product function @xmath20 and @xmath35 small @xmath36 approximation period not separable from eq . ( [ e9 ] ) elliptic function @xmath37 limit of @xmath30 eq e9 agrees with eq . ( [ e8 ] ) usual effect of finite amplitude period pendulum effect spherical earth opposite shortens periodshortening period gravitational acceleration enhanced surface earth consider forces on bob bob at height @xmath38 above surface earth gravitational acceleration given by @xmath39 weight bob towards center earth angle with string pendulum not @xmath6 but @xmath40 see fig for definitions geometrical quantities projection bob s weight perpendicularly to pendulum s arm force @xmath41 determines torque oscillations given by @xmath42 geometrical considerations show @xmath43 and @xmath44 expanding @xmath41 in @xmath45 keeping leading terms @xmath46 and @xmath45 @xmath47 same form as force on bob flat earth model slightly enhanced gravitational acceleration gravity stronger period of pendulum s oscillations shortens period inversely proportional to gravitational acceleration recover eq . ( period ] ) in flat earth model limit as @xmath48 gravitational acceleration pendulum measure @xmath49 not @xmath0 slightly larger acceleration with reasonable length pendulum effect smaller than typical experimental errors compare with gravitational acceleration free - fall experiment decreases with height slightly smaller than @xmath0 at earth s surfacegravitational acceleration by @xmath50 flat earth model neglect small variation measuring acceleration free fall experiment @xmath50 not @xmath0 @xmath51 @xmath52 flat earth model different quantities for gravitational acceleration pendulum or free fall experiment fractional change in pendulum s period at @xmath45 implies effect small @xmath53 m long pendulum @xmath54 m change in period one part in @xmath55 effect ignored in textbooks can be same order magnitude as other effects textbooks discuss choose parameters for effect same order as usual lengthening period due to finite amplitude effects parameters spherical earth effect compensates for finite amplitude effect at on amplitude at order @xmath56 @xmath57 comparing two effects comparable if @xmath58 @xmath59rad comparable if @xmath60 or @xmath61 m length is length of foucault pendulum in science museum of virginia many foucault s pendula in science museums for demonstration earth s rotation length of arc of pendulum s motion is about @xmath62 m large enough thank richard price for stimulating discussionsresearch supported national science foundation grant appendix integral complete elliptic integral strategy change variables interval integration @xmath63 to @xmath64 choose new variable @xmath65 defined @xmath66 measure integral @xmath67 @xmath68 complete elliptic integral @xmath69. ciufolini j a. wheeler gravitation inertia princeton university press 1995 g. galilei dialogues new sciences dover new york 1954 h. goldstein classical mechanics 2nd ed addison - wesley 1980 j. b. marion s t thornton classical dynamics particles systems 4th ed saunders fort worth 1995 r. nelson m g. olsson pendulum physics simple system j. phys . 54 112 - 121 1986 w. gough period simple pendulum not @xmath70 j . 4 53 1983. b. arfken h. j. weber mathematical methods physicists 5th ed hartcourt san diego 2001 i s. gradshteyn m. ryzhik table integrals series products 6th ed academic press san diego 2000 | 0.516661 | 3,392 | 1,726 | , Saving $0.1 in GPT-4. |
55,524 | we present high - resolution , high - sensitivity radio images of the ultra - luminous infrared galaxy ( ulirg ) iras23365@xmath03604 .
we performed contemporaneous observations at 1.7 and 5.0ghz , in three epochs separated by one year from each other , with the european very long baseline interferometry network ( evn ) .
we also present complementary multi - element radio linked interferometry network ( merlin ) at 1.6 and 5.0ghz , and archival very large array ( vla ) data , taken at 1.4 and 4.9ghz .
we find that the emission at @xmath15.0ghz remains quite compact as seen at different resolutions , whereas at @xmath11.7ghz , high resolution imaging reveals some extended structure .
the nuclear region has an approximate linear size of 200pc and shows the presence of two main emission components : i ) one with a composite spectrum due to ongoing non - thermal activity ( probably due to recently exploded supernovae and agn activity ) , ii ) another one with a steep spectrum , likely dominated by an old population of radio emitters , such as supernova remnants ( snrs ) .
radiative losses
are important , so re - acceleration or replenishment of new electrons is necessary .
we estimate a magnetic field strength of @xmath2 g at galactic , and @xmath3 g at nuclear scales , which are typical for galaxies in advanced mergers .
[ firstpage ] galaxies : individual : iras23365@xmath03604 galaxies : starburst radio continuum : general | galaxies with very high infrared ( ir ) luminosities ( @xmath4@xmath5>10^{12}$]l@xmath6 ) known as ultra luminous ir galaxies ( ulirgs ) , dominate the ir background and the star formation rate ( sfr ) density at z@xmath12 @xcite .
although uncommon at lower redshifts , the presence of ulirgs in the local universe offers the opportunity of investigating their parsec scale structure , while profiting from the high angular resolution provided by current instrumentation .
studying ulirgs in the local universe is of great importance since it can aid to understand their high - redshift analogues which dominate the sub - mm sky ( see e.g. , * ? ? ?
it is thought that ulirgs represent a key stage in the formation of optical quasi - stellar objects ( qsos ) and powerful radio galaxies ( e.g. , * ? ? ?
a study based on _ hst _ observations and _ n_-body simulations point to diverse evolutionary paths , not necessarily including a qso phase @xcite .
there is however a general agreement on gas - rich galaxy merging as the origin of ulirgs @xcite , and on the ubiquity of enhanced star - formation , which can be found in combination with different flavours of active galactic nucleus ( agn ) activity ( e.g. , * ? ? ?
* ) . which of these two contributions dominates and is primarily responsible for the overall dust heating , is still an open question . [
cols="<,<,<,^,^,^,^,^ " , ] from section [ sec : obs_str ] , we know that the radio emission at l- and c - bands has a different extent and peaks at different positions within the nuclear region , which strongly suggests the presence of different populations of particles .
this is more evident in the nuclear region mapped with the evn : in the innermost region , where there is an overlap between the emission at the two different frequencies , there would be a concentration of very energetic , short - lived particles , whereas the outer region would be populated by less energetic , long - lived particles , which had had time to diffuse from the inner regions into the outer ones . in table
[ tab : combovalues ] we show the values for @xmath7 , assuming that the critical frequency is either that of the l - band or the c - band . in all cases
we see that the l - band emission is tracing the emission from an older population of electrons ( regardless of the resolution ) than the one emitting at c - band frequencies .
the putative agn together with an ensemble of sne , for which evidence has been found in other studies ( see section [ sec : iras2336 ] ) , must be located within the c - band emission region as seen with the evn , where the magnetic field strength is larger , and where a composite spectrum ( which varies with time ) has been found ( section [ subsec : nucspec ] ) .
we note that the radio lifetime of the emitting source is not only determined by @xmath7 .
the lifetime of relativistic electrons might also be affected by compton losses given by @xmath8 ( following * ? ? ?
* ) since the electrons are immersed in a radiation field , @xmath9 for which we take @xmath10 as a good approximation to the bolometric luminosity @xmath11 .
@xmath12 varies from @xmath13erg@xmath14 at galactic scales , up to @xmath15erg@xmath14 at nuclear scales . to compare the different losses ,
we calculate their ratio , @xmath16 we note that at all scales ( nuclear , circumncuclear and galactic ) and at both l- and c - band , we obtain @xmath17 , with a ratio ranging between @xmath18 ( nuclear scales ) and @xmath19 ( galactic scales ) ; i.e. , the energy density of the radiation field , greatly exceeds the magnetic energy density .
radiative losses are important and we argue that there is need for injection of new electrons or a continuous acceleration to halt the energy depletion , otherwise radio emission would not be visible .
the re - acceleration or injection of new electrons in a ( u)lirg environment , is very likely provided in sn - shells and snrs by first order fermi acceleration .
the presence of sne , snrs and a strong magnetic field in iras23365 , agrees with this scenario . among the evn observing epochs , the third one benefited from a better coverage , and thus resulted in a smaller natural beam ( 19@xmath2025mas@xmath21 at 30 at l - band ) .
the l - band map ( figure [ fig : combo_resol ] , bottom - left ) shows the presence of two compact sources ( el1 and el2 ) within the nuclear region , without counterparts at c - band . on the other hand , the compact source that dominates the emission at c - band , labelled as ec1 ( figure [ fig : combo_resol ] , bottom - right ) , has no compact counterpart at l - band ,
although extended emission is present . to obtain ec1 , el1 and el2 peak intensities , we first estimated the background emission where these compact sources lay .
we solved for the zero level emission ( @xmath22 at c - band and @xmath23 at l - band ) using the task imfit within aips .
we then subtracted this value from the maximum intensity found at the positions of each compact source , in order to obtain their @xmath24 . in table
[ tab : cc_l3 ] we give the positions for ec1 , el1 and el2 , their peak intensities , as well as their estimated @xmath25 and @xmath26 , which are indicative of a non - thermal origin .
ec1 lays on a region where @xmath27 changes with time , suggesting variability within this region .
we note that ec1 is confined to a small area in the first epoch , and then appears to increase in size , as we have mentioned in section [ sec : obs_str ] .
el1 lays in a region with basically no c - band emission , unlike the region where el2 lays where more extended emission is being traced from the first epoch to the third one at c - band . as a consequence , el1 lays in a region which maintains a very steep @xmath27 through time , whilst el2 is found in a region with varying @xmath27 ( figure [ fig : alphapeak ] )
. however , we can not rule out that the @xmath27 variations at el2 are intrinsic . in the case of ec1 , both the variability of the radio emission ( see section [ subsec : nucspec ] ) and of the spectral index distribution ( see section [ subsec : specmas ] and figure [ fig : alphapeak ] ) are indicative of recent non - thermal activity ( probably due to sne and/or agn activity ) .
el1 and el2 display brightness temperatures similar to those expected from either type ii sne or snrs .
we note that the maximum linear size for el1 and el2 is set by the beam size to @xmath28pc , which is too large for characterising either an individual sn or a snr .
a scenario in which el1 and el2 are clusters of sne is difficult to reconcile with the absence of peaks of emission at c - band in all the evn observing epochs , and with the behaviour of @xmath27 at both el1 and el2 .
these facts suggest that there is no recent activity from young sne in those regions , and favour an scenario in which el1 and el2 are dominated by an old population of radio emitters .
we have presented state - of - the - art radio interferometric images of iras23365 , one of the brightest and farthest ulirgs in the local universe ( @xmath29 ) .
our images reveal the presence of a nuclear region , possibly a starburst - agn composite , with an approximate size of 200pc in l - band , and about 100pc in c - band .
we find that the l- and c - band radio emission peak at different positions , thus suggesting that the nuclear region is composed of at least two zones , dominated by distinct populations of radio emitters . in the region where the l- and c - band emission overlap
, there is evidence for ongoing non - thermal activity , characterised by very energetic , short - lived particles . during our evn monitoring of iras23365
, we have found flux density variability in the overlapping region , thus resulting in a variation of the spectral index .
this can be explained by the flux density variations of sources therein ( sne , agn , etc . ) and/or by the appearance of new sources ( e.g. , sne ) which would be seen first at higher frequencies and later at lower frequencies @xcite .
the edges of the overlapping region characterised by less energetic , long - lived particles , would be dominated by an old population of radio emitters , probably clumps of snrs , for which we have found two candidates in the third l - band evn epoch .
these facts agree with the classification of iras23365 as a composite system , made by @xcite .
the radio source lifetime at different scales ( as seen with the vla , merlin and the evn arrays ) and at both l- and c - bands , is limited by compton losses .
the sne and snrs , for which we have found evidence , are likely providing the mechanism of re - acceleration , or replenishment of new electrons that is needed to halt the radio energy depletion .
we have found that the equivalent magnetic field strength at galactic ( mapped with the vla ) and nuclear scales ( mapped with the evn ) , 18 and 175@xmath30 g , respectively , correspond to that of a galaxy in an advanced stage of interaction @xcite . the magnetic field in both nuclear and circumnuclear regions is stronger than at galactic scales , thus implying that the lifetime of the electrons undergoing synchrotron losses is shorter ( @xmath31myr ) in the innermost nuclear regions ( with linear size @xmath32kpc ) of iras23365 , and larger ( @xmath33myr ) in the outer regions ( @xmath34kpc ) .
our study of iras23365 ( at @xmath35 ) has shown that high - resolution , high - sensitivity observations are needed if we are to make significant improvement in the detailed understanding of nuclear and circumnuclear starbursts in the local universe .
the resolution we attained using a maximum baseline length of approximately 7,000 km , is not enough to resolve individual compact sources ( e.g. , sne , snrs , agn ) from each other , within the nuclear region of iras23365 ; yet , it could be possible to infer the activity of such compact sources , by carefully monitoring variations of total flux density and spectral index distribution .
for instance , @xcite were able to directly detect sn activity in the b1-nucleus of arp299 , by carefully monitoring the variations in @xmath36 , over several years of vla observations . without spatially resolving each individual sne , they estimated a lower limit for the @xmath37 in that lirg . in the case of iras23365 , where a larger number of sne are expected each year ,
several observations per year would be needed to perform such an indirect study of the sn population in its nuclear region , provided that we are able to distinguish between agn outbursts and sn explosions .
iras23365 is a good example of the situation to be faced when observing galaxies at higher redshifts .
it is expected that the square kilometer array ( ska ) , with a maximum baseline length @xmath38 km , will allow the detection of sources as faint as 50njy , e.g. ccsne , exploding at @xmath39 ( see e.g. , * ? ? ?
* ) . however , the angular resolution will be a strong limiting factor . in those cases where the nuclear and
even the circumnuclear regions ( i.e. , where we expect most of the sn activity to occur ) of the host galaxy can not be resolved out into their different components , ska s high sensitivity might be of great use to indirectly detect sn activity through the monitoring of flux density variations . ) , given pixel by pixel ( as obtained from aips ) in the nuclear zone , which we expect to be composed by non - thermal sources ( e.g. sne and agn ) .
note that while the peak of the spectral index distribution varies in the very central region , it becomes steeper towards the edges , thus representing older radio emission .
the sources ec1 , el1 and el2 are marked with crosses ( see section [ sec : compactel ] ) . ]
we thank our referee robert beswick for constructive comments and suggestions that have improved this manuscript .
we are grateful to the editor for useful suggestions on improving the presentation of our results .
we acknowledge financial support from the spanish micinn through grant aya2009 - 13036-c02 - 01 , co - funded with feder funds .
we also acknowledge support from the autonomic government of andalusia under grants p08-tic-4075 and tic-126 . our work has also benefited from research funding from the european community framework programme 7 , advanced radio astronomy in europe , grant agreement no . : 227290 , and sixth framework programme under radionet r113ct 2003 5058187 .
the authors are grateful to jive and especially to zsolt paragi and bob campbell for their assistance in this project . _
the european vlbi network is a joint facility of european , chinese , south african and other radio astronomy institutes funded by their national research councils_. this article is also based on observations made with merlin , a national facility operated by the university of manchester at jodrell bank observatory on behalf of pparc , and observations made with the very large array ( vla ) of the national radio astronomy observatory ( nrao ) ; the nrao is a facility of the national science foundation operated under cooperative agreement by associated universities , inc . | galaxies with high infrared ir luminosities ( @xmath4@xmath5>10^{12}$]l@xmath6 ) known as ultra luminous ir galaxies ( ulirgs dominate ir background star formation rate sfr density at z@xmath12 @xcite uncommon at lower redshifts presence ulirgs local universe opportunity parsec scale structure high angular resolution current instrumentation studying ulirgs understand high - redshift analogues sub - mm sky ulirgs key stage in formation of optical quasi - stellar objects powerful radio galaxies observations simulations point diverse evolutionary paths not including qso phase @xcite general agreement on gas - rich galaxy merging origin of ulirgs @xcite ubiquity of enhanced star - formation with active galactic nucleus activity which dominates responsible for dust heating open question radio emission at l- and c - bands different extent peaks at different positions nuclear region suggests presence of different populations of particlesevident in nuclear region mapped with evn innermost region overlap between emission at two frequencies concentration of energetic short - lived particles outer region populated by less energetic long - lived particles table combovalues values for @xmath7 critical frequency l - band or c - band l - band emission emission from older population electrons than at c - band frequencies putative agn ensemble of sne evidence found in other studies must be located within c - band emission region magnetic field strength larger composite spectrum varies with time found radio lifetime of emitting source not determined by @xmath7 relativistic electrons affected by compton losses by @xmath8 electrons immersed in radiation field @xmath9 take @xmath10 approximation to bolometric luminosity @xmath11 @xmath12 varies from @xmath13erg@xmath14 at galactic scales to @xmath15erg at nuclear scales compare losses calculate ratio , @xmath16 at all scales l- and c - band obtain @xmath17 ratio ranging between @xmath18 ( nuclear and @xmath19 ( galactic scales )energy density radiation field exceeds magnetic energy density radiative losses important need for injection new electrons or continuous acceleration to halt energy depletion radio emission visible re - acceleration injection new electrons environment likely provided in sn - shells snrs by first order fermi acceleration presence sne snrs strong magnetic field in iras23365 agrees scenario epochs third one benefited better coverage resulted in smaller natural beam ( 19@xmath2025mas@xmath21 at 30 l - band ) l - band map shows two compact sources ( el1 and el2 ) nuclear region without counterparts at c - band source emission at c - band ec1 no counterpart at l - band extended emission present obtain ec1 el1 el2 peak intensities estimated background emission solved for zero level emission subtracted value from maximum intensity each compact source obtain @xmath24 table positions for ec1 el1 el2 peak intensities estimated @xmath25 and @xmath26 indicative of non - thermal originec1 lays region where @xmath27 changes with time suggesting variability ec1 confined to small area in first epoch then in size mentioned el1 region no c - band emission unlike el2 more extended emission from first to third c el1 maintains steep @xmath27 through time el2 region varying @xmath27 @xmath27 variations at el2 intrinsic ec1 variability of radio emission spectral index distribution indicative of recent non - thermal activity probably due to sne agn activity el1 and el2 display brightness temperatures similar to type ii sne or snrs maximum linear size for el1 and el2 set by beam size to @xmath28pc too large for characterising individual sn or snr scenario el1 and el2 clusters of sne difficult to reconcile with absence of peaks of emission at c - band in epochs behaviour of @xmath27 at both el1 and el2 suggest no recent activity from young sne scenario el1 and el2 dominated by old population radio emitterspresented radio interferometric images of iras23365 brightest farthest ulirgs in local universe images reveal nuclear region possibly starburst - agn composite size 200pc in l - band 100pc in c - band l- c - band radio emission peak at different positions suggesting nuclear region two zones dominated by distinct populations radio emitters region l- c - band emission overlap evidence for ongoing non - thermal activity energetic short - lived particles monitoring found flux density variability in overlapping region variation of spectral index explained by flux density variations sources appearance of new sources sne seen higher lower edges of overlapping region less energetic long - lived particles dominated by old population radio emitters probably clumps of snrs found two candidates in third l - band evn epoch agree with classification of iras23365 as composite system radio source lifetime l- c - bands limited by compton losses sne and snrs evidence likely providing mechanism of re - acceleration new electrons halt radio energy depletionfound equivalent magnetic field strength at galactic nuclear scales , 18 175@xmath30 g correspond to galaxy in advanced stage interaction @xcite magnetic field in nuclear circumnuclear regions stronger galactic scales lifetime of electrons synchrotron losses shorter @xmath31myr in innermost nuclear regions of iras23365 larger @xmath33myr in outer regions @xmath34kpc study of iras23365 shown high - resolution high - sensitivity observations needed understanding of nuclear circumnuclear starbursts resolution attained maximum baseline length 7,000 km not enough to resolve individual compact sources sne snrs agn within nuclear region iras23365 possible to infer activity sources by monitoring variations total flux density spectral index distribution @xcite sn activity in b1-nucleus of arp299 monitoring variations @xmath36 observations estimated lower limit for @xmath37 case iras23365 larger number of sne expected year several observations per year needed indirect study of sn population nuclear region between agn outbursts and sn explosions iras23365 good example of situation observing galaxies at higher redshiftsexpected square kilometer array ska ) maximum baseline length @xmath38 km detection sources faint 50njy ccsne exploding at @xmath39 angular resolution factor nuclear circumnuclear regions sn activity host galaxy resolved ska s high sensitivity detect sn activity flux density variations pixel by pixel nuclear zone composed non - thermal sources sne agn ) peak spectral index distribution varies central region steeper towards edges representing older radio emission sources ec1 el1 el2 marked with crosses section thank referee robert beswick for comments suggestions manuscript grateful editor suggestions improving presentation results acknowledge financial support spanish micinn grant aya2009 - 13036-c02 - 01 co - funded with feder funds support autonomic government of andalusia grants p08-tic-4075 tic-126 work benefited from research funding european community framework programme 7 advanced radio astronomy in europe grant agreement no . 227290 sixth framework programme radionet r113ct 2003 5058187 authors grateful to jive zsolt paragi bob campbell for assistanceeuropean vlbi network joint facility european chinese south african radio astronomy institutes funded national research councils article based on observations merlin national facility university of manchester jodrell bank observatory large array vla national radio astronomy observatory nrao facility national science foundation cooperative agreement universities | 0.505872 | 3,008 | 1,500 | , Saving $0.1 in GPT-4. |
55,544 | we have carried out a survey of the ngc 2068 region in the orion b molecular cloud using harp on the jcmt , in the and ( @xmath0 3 2 ) and ( @xmath0 4 3 ) lines .
we used to map the outflows in the region , and matched them with previously defined scuba cores .
we decomposed the and into gaussian clumps , finding 26 and 8 clumps respectively .
the average deconvolved radii of these clumps is @xmath1 au and @xmath2 au for and respectively .
we have also calculated virial and gas masses for these clumps , and hence determined how bound they are .
we find that the clumps are more bound than the clumps ( average gas mass to virial mass ratio of 4.9 compared to 1.4 ) .
we measure clump internal velocity dispersions of @xmath3 and @xmath4 for and respectively , although the values are heavily weighted by a majority of the clumps being protostellar , and hence having intrinsically greater linewidths .
we suggest that the starless clumps correspond to local turbulence minima , and we find that our clumps are consistent with formation by gravoturbulent fragmentation .
we also calculate inter - clump velocity dispersions of @xmath5 and @xmath6 for and respectively . the velocity dispersions ( both internal and external ) for our clumps match results from numerical simulations of decaying turbulence in a molecular cloud . however , there is still insufficient evidence to conclusively determine the type of turbulence and timescale of star formation , due to the small size of our sample .
[ firstpage ] ism : clouds ism : individual ( ngc 2068 ) stars : formation . | an understanding of the properties of molecular gas is important in the investigation of star formation in our galaxy .
a complete theory of star formation should be able to explain certain basic properties and characteristics of dense cores and protostars .
the kinematics of the dense cores and their surrounding natal environment , as well as the relation of the core mass function ( cmf ) to the stellar initial mass function ( imf ) , are all linked to the initial conditions of star formation .
most stars form in clusters , although @xcite find that the percentage of stars found in clusters appears to be dependent on the definitions used .
many studies of the less dense gas ( e.g. co low - j transitions ) in these clusters exist ( e.g. @xcite , @xcite , @xcite ) .
however , there are fewer surveys of the very dense gas tracers ( e.g. hcn or @xmath7 4 3 transitions ) , which reveal the density peaks in the gas most closely associated with the protostars . with the advent of receivers like harp on the jcmt @xcite ,
, large - scale dense gas surveys are now feasible . in recent years
, a general consensus has been reached that star formation is a dynamical process , occurring within a few freefall times ; the method by which molecular gas clouds collapse to form protostellar cores is also generally accepted to be a combination of gravity and turbulence .
this _ gravoturbulent fragmentation _ is essentially a two - phase process @xcite : the turbulence produces density enhancements in the molecular gas cloud through a network of interacting shocks ; these density enhancements decrease the jeans mass locally @xcite allowing the densest regions ( formed at the shock intersection points ) to collapse , accrete mass and form stars .
although turbulence plays such an important part in star formation , its origins , length- and timescales are still not understood .
it is unclear whether turbulence in star formation regions decays freely in a crossing time @xcite , or is driven @xcite and therefore persists for a longer timescale .
most competitive accretion simulations @xcite feature molecular clouds initially supported by turbulence that decays on a timescale comparable to the crossing time of the molecular cloud .
this allows competitive accretion to start on small scales and grow progressively larger , fed by loss of kinetic energy from the ambient cloud .
@xcite however have shown evidence for competitive accretion in an environment with driven turbulence ; as long as the turbulence is driven on large enough scales , clusters can form where gas and stars are virialised .
they have also produced representative core mass functions from their numerical simulations ; however they find that with these alone they are unable to distinguish between the driven and decaying modes of turbulence . indeed ,
the general properties of the cores produced in differing turbulent environments at low resolution are insufficently different @xcite to dismiss a particular environment .
we have therefore decided to use an alternative method examining the _ kinematics _ of the dense gas cores , rather than their spatial and structural properties . @xcite
have performed comparisons of the kinematic properties of protostellar and starless cores produced in simulations that utilise both driven and decaying turbulence modes . comparing the intercore velocity dispersions ( i.e. the dispersions of the core centroid velocities ) , they found that protostellar cores in decaying turbulence simulations have a higher intercore velocity dispersion than the ambient gas dispersion ; protostellar cores and starless cores in driven turbulence simulations have subvirial intercore velocity dispersions and intercore velocity dispersions close to that of the gas respectively .
they therefore suggest that comparing the starless and protostellar intercore velocity dispersions to the net gas dispersion could potentially distinguish between the two environments observationally .
we have chosen to observe the @xmath8 4 3 transition as it is optically thin , traces the densest regions ( n@xmath9 @xmath10 ) and will be most likely to trace the dense inner regions of prestellar cores .
we can calculate linewidths and virial ratios of cores to determine how well bound they are ; we can also potentially use the core - to - core velocity dispersions to distinguish between the two turbulence scenarios .
we have also observed the @xmath0 3 2 transition which is optically thin over most of the region and will also trace more of the less dense core envelopes for comparison with the . in addition , we observed the @xmath0 3 2 transition to investigate the large - scale kinematics , , as well as providing evidence for outflows which would pinpoint protostellar cores .
the orion b molecular cloud complex is the closest region of high - mass star formation , and lies at a distance of around 415 pc .
it includes 5 major star formation regions , one of which , ngc 2068 , is the subject of this paper .
ngc 2068 is a bright reflection nebula containing an infrared cluster ; submillimetre emission associated with the nebula has been observed to the south , as shown in figure [ ir_overlay ] .
the first unbiased molecular gas survey of ngc 2068 was carried out by @xcite , hereafter lbs90 , using cs @xmath0 2 1 emission .
they identified 2 clumps lbs 10 and lbs 17 at 1.7 resolution .
the region has since been mapped by many others , both in molecular line emission ( e.g. @xcite , in @xmath11 1 0 ) as well as in scuba 850 @xmath12 m dust emission ( e.g. @xcite , hereafter nwt ) , breaking the two lbs regions into smaller clumps and cores .
@xcite identified 18 smaller clumps with diameters of @xmath13 25 using both 850 @xmath12 m dust emission and co isotopologue mapping .
the clumps they identified are labelled oribsmm ( for orion b submmillimetre emission ) and then numbered by decreasing declination .
, hereafter m01 , identified 31 cores with fwhm diameters of @xmath1313 from 850 @xmath12 m dust emission ; their cores are labelled lbsx - mmy , where x is the corresponding lbs clump that the core belongs to ( either 10 or 17 ) and y begins at 1 and increases with ra . the ngc 2068 region has been extensively studied and well - documented at different densities and resolutions , and
is a fairly small , self - contained region with obvious dust and gas structure .
we have therefore chosen this region for investigations of the kinematics of pre- and protostellar cores and their surrounding gas , in order to examine the initial conditions for star formation and determine the possible mechanisms for star formation in this region .
we present an analysis of spectral line data from ngc 2068 , examining the gaseous structure of the region , as well as its kinematics .
we break the region down into gaussian clumps using two molecules that trace different densities , to examine the kinematics and mechanisms leading to the fragmentation and formation of prestellar cores .
section [ sec : obs ] gives an overview of the observations and data reduction procedure .
section [ sec : res ] describes the reduced data sets , highlighting the multiple - velocity - component structure of the gas , and several high - velocity outflows .
section [ sec : clfind ] presents the clumpfinding analysis of the region , using and , while section [ sec : clprop ] examines the characteristic properties of the clumps , including shape and size , masses and velocity dispersions .
we also present a comparison of observational data with numerical simulations of star formation .
we summarise the results and conclude in section [ sec : concl ] .
the observations were made over a total of 12 nights between 6 september and 2 october 2010 , using harp at the jcmt @xcite . in total ,
11.5 hours of data were taken , with an area of 400 @xmath14 600 ( centred on postion @xmath15 , + 00@xmath160000 ) being mapped .
all the standard telescope observing procedures were followed for each night of observations , with regular pointings and focusing of the jcmt s secondary mirror .
standard spectra were also taken towards various well - known calibration sources , to verify that the intensity in the tracking receptor matched recorded standards to within calibration tolerance before subsequent observations were allowed to continue .
the fully - sampled maps were taken in raster position - switched mode , an ` on - the - fly ' data collection method where the harp array continuously scans in a direction parallel to the sides of the map @xcite .
the telescope is pointed at an off - position after every row in the map to obtain background values that are subtracted from the raw data .
a low - resolution @xmath7 3 2 ` stare ' observation was first performed towards this position to verify that it was indeed emission - free .
data are presented in units of corrected antenna temperature , which is related to the main beam temperature ( @xmath17 ) using @xmath17 = /@xmath18 @xcite .
a value of @xmath18 = 0.61 was used , following @xcite .
emission from 3 molecules , and was observed and details of the transitions and frequencies are given in table [ mol_trans ] .
the frequencies and energies quoted are taken from lambda , the leiden atomic and molecular database .
the / data were observed simultaneously , utilising the acsis high - resolution dual sub - band mode .
each sub - band was divided into 4096 61 khz channels , giving a velocity resolution of 0.055 .
the data was observed in single sub - band mode , splitting the 250 mhz bandwidth into 8192 channels , giving a velocity resolution of 0.026 .
.molecules observed in this particular run for ngc 2068 , with their transitions ( column 2 ) , frequencies ( column 3 ) and energies of the upper level above ground ( column 4 ) .
the critical densities of each transition ( column 5 ) are taken from @xcite . [
cols="^,^,^,^,^",options="header " , ] | understanding properties molecular gas important in star formation complete theory star formation should explain basic properties characteristics of dense cores protostars kinematics dense cores surrounding natal environment relation core mass function to stellar initial mass function imf linked to initial conditions star formation most stars form in clusters percentage on definitions studies of less dense gas co low - j transitions ) in clusters exist fewer surveys of very dense gas tracers hcn or @xmath7 4 3 transitions ) reveal density peaks gas associated with protostars advent receivers like harp jcmt large - scale dense gas surveys feasible consensus star formation dynamical process within few freefall times molecular gas clouds collapse to form protostellar cores combination of gravity turbulence gravoturbulent fragmentation two - phase process turbulence produces density enhancements in molecular gas cloud density enhancements decrease jeans mass locally densest regions shock intersection to collapse accrete mass form stars turbulence important part in star formation origins length timescales not understoodunclear whether turbulence in star formation regions decays in crossing time or driven persists longer competitive accretion simulations feature molecular clouds supported by turbulence comparable to crossing time allows competitive accretion small scales grow larger fed by loss kinetic energy from ambient cloud @xcite shown evidence for competitive accretion in driven turbulence clusters can form where gas stars virialised produced representative core mass functions from simulations unable to distinguish between driven and decaying modes turbulence general properties of cores in turbulent environments low insufficently different environment decided alternative method examining kinematics dense gas cores @xcite performed comparisons of kinematic properties of protostellar starless cores in driven and decaying turbulence modes intercore velocity dispersions found protostellar cores in decaying turbulence have higher than ambient gas dispersion starless cores in driven turbulence simulations have subvirial close to gas suggest comparing starless protostellar intercore velocity dispersions to net gas dispersion could distinguish two environmentsobserve @xmath8 4 3 transition optically thin traces densest regions @xmath10 ) likely trace dense inner regions prestellar cores calculate linewidths virial ratios cores use core - to - core velocity dispersions distinguish turbulence scenarios observed @xmath0 3 2 transition optically thin less dense core envelopes observed @xmath0 3 2 transition large - scale kinematics evidence outflows protostellar cores orion b molecular cloud complex closest region high - mass star formation distance around 415 pc includes 5 major star formation regions ngc 2068 subject paper 2068 bright reflection nebula infrared cluster submillimetre emission observed south figure first unbiased molecular gas survey of ngc 2068 @xcite @xmath0 2 1 emission identified 2 clumps lbs 10 lbs 17 at 1.7 resolution region mapped by others molecular line emission scuba 850 @xmath12 m dust emission breaking two regions smaller clumps cores @xcite identified 18 smaller clumps diameters @xmath13 25 850 @xmath12 m dust emission isotopologue mappingclumps identified labelled oribsmm orion b submmillimetre emission numbered by decreasing declination m01 identified 31 cores diameters @xmath1313 from 850 @xmath12 m dust emission cores labelled lbsx - mmy x corresponding lbs clump 10 or 17 y begins at 1 increases with ra ngc 2068 region extensively studied at different densities resolutions small self - contained dust gas structure chosen for kinematics of pre- protostellar cores surrounding gas examine initial conditions for star formation mechanisms analysis of spectral line data from ngc 2068 gaseous structure kinematics region down into gaussian clumps different densities mechanisms fragmentation formation prestellar cores obs overview observations data reduction procedure res reduced data sets multiple - velocity - component structure gas high - velocity outflows clfind clumpfinding analysis clprop characteristic properties clumps shape size masses velocity dispersions comparison observational data with numerical simulations star formation summarise results conclude : concl observations made over 12 nights between 6 september 2 october 2010 using harp at jcmt @xcite .11.5 hours data taken area 400 @xmath14 600 postion @xmath15 + 00@xmath160000 mapped standard telescope observing procedures followed regular pointings focusing jcmt s secondary mirror spectra taken towards calibration sources intensity before observations fully - sampled maps taken raster position - switched mode harp array scans parallel sides map telescope pointed off - position after row background values subtracted data low - resolution @xmath7 3 2 stare observation position verify emission - free data presented units corrected antenna temperature related main beam temperature @xmath17 @xmath17 = /@xmath18 @xcite value @xmath18 = 0.61 used emission from 3 molecules observed transitions frequencies in table [ mol_trans ] frequencies energies from lambda leiden atomic molecular database data observed simultaneously acsis high - resolution dual sub - band mode sub - band divided into 4096 61 khz channels velocity resolution 0.055 single sub - band mode 250 mhz bandwidth 8192 channels velocity resolution 0.026 molecules observed ngc 2068 transitions column 2 frequencies column 3 energies upper level ground column 4 critical densities transition column 5 from @xcite cols="^,^^",options="header " | 0.353681 | 2,201 | 1,089 | , Saving $0.1 in GPT-4. |
55,554 | stars are usually faint point sources and investigating their surfaces and interiors observationally is very demanding . here
i give a review on the state - of - the - art observing techniques and recent results on studying interiors and surface features of active stars . | the only star we can study easily with any spatial resolution is the sun , and now with the stereo spacecrafts we can do it even in 3-d . for other stars
we are usually restricted to using indirect methods for getting information of their surface structures . here a short review on methods for studying stellar interiors and surfaces and the main recent results are presented .
kepler has not only revolutionised the exoplanet research , but also stellar astrophysics . the short , one minute ,
cadence data is uniquely suited for studying stellar oscillations and thus the interiors of stars ( see , e.g. , ( * ? ? ? * gilliland 2010 ) ) .
recent kepler results , based on a sample of 500 solar - like stars , show that even though distribution of observed radii is similar to the one predicted by models of synthetic stellar populations in the galaxy , the distribution of observed masses seems to deviate significantly from the predicted distribution ( ( * ? ? ?
* chaplin 2011a ) ) .
this result raises interesting questions on the star formation rate and initial mass function .
studying the same sample of solar - like stars further reveals that the number of stars showing oscillations decreases significantly with increasing levels of activity ( ( * ? ? ?
* chaplin 2011b ) ) .
this on the other hand implies that the magnetic fields can inhibit the amplitudes of solar - like oscillations .
the strength of the dynamo - created stellar magnetic fields and the frequency of the associated phenomena ( e.g. , starspots and flares ) is strongly linked to the rotation of the star , with rapid rotation enhancing the field generation ( e.g. , ( * ? ? ? * pallavicini 1981 ) ) .
many active stars , rapid rotation usually induced either by youth or binarity , show large starspots which are detectable even using photometric observations from small ground - based telescopes ( e.g. , ( * ? ? ? * kron 1947 ) ) . in some cases
the monitoring has become almost nightly after the implementation of automatic photometric telescopes in the late 1980 s ( e.g. , ( * ? ? ?
* rodon & cutispoto 1992 ) ; ( * ? ? ?
* strassmeier 1997 ) ) .
these observations give crucial information on the stellar activity cycles ( e.g. , ( * ? ? ?
* olh 2009 ) ) and long - term spot longitudes ( e.g. , ( * ? ? ?
* lehtinen 2012 ) ) , i.e. , 2-d information where one dimension is longitude and the other time . with observations using multiple filters
crude spot latitudes can be obtained ( e.g. , ( * ? ? ?
* roettenbacher 2011 ) ) , making the reconstruction 3-d ( longitude , latitude and time ) .
similarly , the continuous , high precision , space - based light - curves allow for recovering latitudinal and longitudinal information in addition to the temporal variations . for obtaining more detailed stellar surface structures
doppler imaging techniques have to be used ( e.g. , ( * ? ? ?
* vogt , penrod & hatzes 1987 ) ) , giving a 2-d snapshot of the surface temperature with a reasonable resolution ( e.g. , ( * ? ? ?
* korhonen 2010 ) ) . combining many observations from different epochs
adds the third dimension , namely time ( e.g. , ( * ? ? ?
* skelly 2010 ) ; ( * ? ? ?
* hackman 2012 ) ) .
doppler images can also be used for studying stellar surface rotation and its dependence on latitude ( for a recent review see ( * ? ? ?
* korhonen 2012 ) ) . if spectral lines formed at different heights in the stellar atmosphere are used in the reconstruction , also vertical structures in starspots can be mapped ( ( * ? ? ?
* berdyugina 2011 ) ) , coming closer to what is possible in the solar case .
obtaining multi - dimensional infromation on the active stars , which are just point sources , is challenging but not impossible .
observations give us detailed information , both temporal and spatial , on occurrence of starspots . in the future long baseline infrared / optical interferometry can offer a direct check on some of the results .
furthermore , while the interferometric imaging is restricted to the brightest and nearest stars , the method does not require large projected rotational velocities , like doppler imaging does .
this will open up new targets that have not been approachable earlier , significantly widening the parameter space for activity studies . | only star we can study with spatial resolution is sun now with stereo spacecrafts even in 3-d for other stars restricted to indirect methods for surface structures short review on methods for studying stellar interiors surfaces main recent results kepler revolutionised exoplanet research stellar astrophysics short one minute cadence data suited for studying stellar oscillations interiors of stars gilliland 2010 recent kepler results on sample 500 solar - like stars show distribution observed radii similar to stellar distribution observed masses deviate significantly from chaplin 2011a raises questions on star formation rate initial mass function same sample reveals number of stars showing oscillations decreases with increasing activity chaplin 2011b implies magnetic fields can inhibit amplitudes solar - like oscillations strength of dynamo - created stellar magnetic fields frequency of associated phenomena starspots flares linked to rotation star rapid rotation field generation pallavicini 1981 many active stars rapid rotation youth binarity show large starspots detectable even using photometric observations from small telescopes kron 1947monitoring almost nightly after automatic photometric telescopes late 1980 s * rodon & cutispoto 1992 ; * strassmeier 1997 observations give crucial information on stellar activity cycles olh 2009 long - term spot longitudes lehtinen 2012 2-d information one dimension longitude time with observations using multiple filters crude spot latitudes obtained * roettenbacher 2011 reconstruction 3-d ( longitude latitude time continuous high precision space - based light - curves allow latitudinal longitudinal information temporal variations for detailed stellar surface structures doppler imaging techniques used vogt penrod hatzes 1987 2-d snapshot of surface temperature with reasonable resolution * * korhonen 2010 ) combining many observations from different epochs adds third dimension time * skelly 2010 ; hackman 2012 doppler images used for stellar surface rotation dependence on latitude * * korhonen 2012 ) if spectral lines formed at different heights in stellar atmosphere used in reconstruction vertical structures in starspots can be mapped berdyugina 2011 closer to solar casemulti - dimensional on active stars point sources challenging not impossible observations give detailed information temporal spatial on starspots future long baseline infrared / optical interferometry direct check results interferometric imaging restricted to brightest nearest stars require large rotational velocities like doppler imaging new targets widening parameter space for activity studies | 0.183572 | 997 | 495 | , Saving $0.0 in GPT-4. |
55,557 | in this letter , we announce the discovery of a new dwarf satellite of the milky way , located in the constellation canes venatici .
it was found as a stellar overdensity in the north galactic cap using sloan digital sky survey data release 5 ( sdss dr5 ) .
the satellite s color - magnitude diagram shows a well - defined red giant branch , as well as a horizontal branch . as judged from the tip of the red giant branch
, it lies at a distance of @xmath0 kpc . based on the sdss data , we estimate an absolute magnitude of @xmath1 , a central surface brightness of @xmath2 mag arcsecond@xmath3 , and a half - light radius of @xmath4 ( @xmath5 pc at the measured distance ) .
the outer regions of canes venatici appear extended and distorted .
the discovery of such a faint galaxy in proximity to the milky way strongly suggests that more such objects remain to be found . | there are ten known dwarf spheroidal ( dsph ) companions of the milky way galaxy .
together with the two dwarf irregulars ( the large and small magellanic clouds ) , this makes up all the known satellite galaxies of the milky way .
the dsphs have such low surface brightness that they have often been found serendipitously .
for example , while sextans @xcite was found as part of an automated search , the intrinsically far more luminous sagittarius dsph was first identified kinematically from a radial velocity survey of stars in galactic center fields @xcite . as such galaxies are resolvable into individual stars because of their proximity , they are detectable as enhancements in the stellar number density in large photometric surveys .
for example , @xcite analyzed overdensities of m giants in the 2mass all sky survey and claimed the detection of a new disrupted dsph in canis major , although this remains controversial .
@xcite systematically surveyed @xmath6 square degrees of the sloan digital sky survey ( sdss ; * ? ? ?
* ) and identified a dsph companion to the milky way in the constellation of ursa major .
subsequent spectroscopic observations @xcite showed that ursa major was a dark matter dominated dsph .
meanwhile , @xcite announced the discovery of andromeda ix and x , new dsph satellites of m31 found in sdss data with similar methods .
very recently , @xcite mapped out the stars satisfying the color cut @xmath7 in almost all of sdss data release 5 ( dr5 ) .
they dubbed this plot of the high - latitude galactic northern hemisphere the `` field of streams '' , owing to its wealth of prominent stellar substructure
. visible by eye in the plot is a heretofore undetected enhancement of the stellar density , located in the constellation canes venatici . here
, we show that this corresponds to a new dsph companion the eleventh of the milky way galaxy .
-0.90 cm lc coordinates ( j2000 ) & 13:28:03.5 + 33:33:21.0@xmath8 + galactic coordinates ( _ l , b _ ) & 74.3 , 79.8 + position angle & @xmath9 + ellipticity & @xmath10 + @xmath11 ( plummer ) & @xmath12 + @xmath11 ( exponential ) & @xmath13 + a@xmath14 & @xmath15 + @xmath16 ( plummer ) & @xmath17 + @xmath16 ( exponential ) & @xmath18 + v@xmath19 & @xmath20 + ( m@xmath21m)@xmath22 & @xmath23 + m@xmath24 & @xmath25 [ tab : struct ]
the sdss is an imaging and spectroscopic survey .
sdss imaging data are produced in five photometric bands , namely @xmath26 , @xmath27 , @xmath28 , @xmath29 , and @xmath30 @xcite , and are automatically processed through pipelines to measure photometric and astrometric properties @xcite .
dr5 primarily covers @xmath31 square degrees around the north galactic pole ( ngp ) . in the process of analyzing this area of the ngp @xcite
, we visually identified a stellar overdensity in the constellation canes venatici .
the upper left panel of figure [ fig : cvn_disc ] shows a grayscale sdss image of the sky centered on the stellar overdensity ; note that no obvious object can be seen . however , a roughly elliptical overdensity of objects classified by the sdss pipeline as stars is readily visible in the photometric data ( upper middle and right panels ) .
plotting these stars in a color - magnitude diagram ( cmd ) reveals a clear red giant branch ( rgb ) and horizontal branch ( lower panels ) .
this cmd appears similar to that seen in sdss data for the sextans dsph ( see , e.g. , fig . 1 of * ? ? ?
* ) , although considerably more distant .
based on its apparent morphology and on the presence of a distinct stellar population , we conclude that this is most likely a hitherto unknown dsph galaxy . as is customary , we name it after its constellation , canes venatici ( cvn ) .
the distance of a distinct , metal - poor stellar population can be estimated from the @xmath32-band magnitude of the tip of the red giant branch ( trgb ; e.g. , * ? ? ?
* ; * ? ? ?
we applied the transforms from @xcite to convert dereddened sdss @xmath33 magnitudes to @xmath34 magnitudes .
the cumulative @xmath32-band luminosity function for the rgb shows a sharp rise starting at @xmath35 ( which corresponds to @xmath36 ) .
the horizontal branch fit in the lower right panel of figure [ fig : cvn_disc ] provides supporting evidence for the reliability of our procedure . assuming a metallicity of [ fe / h]@xmath37 ( see below ) and a trgb color of @xmath38 , from the calibration of @xcite we obtain the relation @xmath39 , yielding a distance modulus of @xmath40 ( @xmath41 kpc ) .
this is comparable to the most distant milky way dsph satellites previously known , leo i and ii ( 250 and 215 kpc , respectively : * ? ? ?
* ; * ? ? ?
. the age and metallicity of the stars in cvn can be estimated by comparison with the cmds of single epoch , single metallicity populations , e.g. , galactic globular clusters . in the lower right panel of figure [ fig : cvn_disc ] , we show a field - star - subtracted hess diagrams ( color - magnitude density plots ) in @xmath42 .
fiducial ridgelines of three globular clusters ngc 2419 ( [ fe / h ] @xmath43 ) , m 3 ( [ fe / h ] @xmath44 ) , and m 71 ( [ fe / h ] @xmath45 ) are overplotted from left to right .
the best match is to ngc 2419 , and in combination with the prominent horizontal branch , the similarity indicates that cvn is dominated by an old , metal - poor ( [ fe / h ] @xmath46 ) stellar population .
the horizontal branch extends very far to the blue , although it is predominantly red .
such red - dominated horizontal branches , hinting at a range of ages in the stellar population , are also seen in other metal - poor dsphs ( e.g. , * ? ? ?
* ) . -0.20 cm -0.6 cm -0.7 cm to investigate the spatial morphology , we select candidate members by applying a mask built as follows .
first , we find the class conditional probabilities @xmath47 and @xmath48 ( see e.g. , * ? ? ?
* ) of a star belonging to cvn and the background , respectively .
these are just the normalized hess diagrams .
the ratio of the class conditional probabilities @xmath49 is converted to a probability of membership of cvn via @xmath50 .
a cut of @xmath51 is used to select stars plotted in the top panel of fig .
[ fig : cvn_morph ] .
these objects are binned into @xmath52 bins , each @xmath53 , and smoothed with a gaussian with fwhm of @xmath54 to yield the plot in the lower panel .
the density contours , representing 2 , 3 , 5 , 7 , 10 , 15 , 20 , and 25@xmath55 above the background level , are markedly irregular , particularly in the outer parts somewhat reminiscent of the ursa minor dsph .
the black crosses are candidate blue horizontal branch stars ( bhbs ) selected with @xmath56 .
the distribution of bhbs is elongated in the same manner as the density contours and traces the same underlying morphology . to estimate the properties listed in table [ tab : struct ] , we first analyze the morphology shown in figure [ fig : cvn_morph ] to derive the centroid from the density - weighted first moment of the distribution , and the average ellipticity and position angle using the three density - weighted second moments ( e.g. , * ? ? ?
the radial profile shown in figure [ fig : cvn_prof ] is derived by computing the average density within elliptical annuli after first subtracting a constant asymptotic background level ( 0.25 arcminute@xmath3 ) reached at large radii .
we then fit the radial profile with standard plummer and exponential laws ( figure [ fig : cvn_prof ] , see also * ? ? ?
the best - fitting position angle , ellipticity and half - light radii are listed in table [ tab : struct ] . at the distance of cvn
, the approximately @xmath57 half - light radius corresponds to @xmath5 pc .
the overall extent is @xmath58 kpc along the major axis , making cvn one of the largest of the milky way dsphs .
although plummer and exponential laws provide reasonable fits to the data , there are some discrepant datapoints in figure [ fig : cvn_prof ] ( cf .
sextans profile in * ? ? ?
. the overall luminosity is computed by masking the stellar locus of cvn and computing the total flux within the mask and within the elliptical half - light radius .
a similar mask , but covering a larger area to minimize shot - noise , well outside the main body of cvn , is scaled by relative area and used to compute the foreground contamination within the half - light radius . after correcting for this contamination ,
the remaining flux is scaled to the total , assuming the fitted profiles are a fair representation of the overall flux distribution .
we also apply a correction of @xmath59 magnitudes for unresolved / faint stars , based on the stellar luminosity functions of other low metallicity , low surface brightness dsphs . the resulting luminosity and central surface brightness estimates ,
m@xmath60 and @xmath2 mag arcsec@xmath61 , are roughly consistent with the central surface brightness absolute magnitude relation noted for local group dsph satellites ( e.g. , * ? ? ?
however , cvn lies far from the central surface brightness
host galaxy distance correlation found for other local group dsph satellites ( e.g. , fig .
7 of * ? ? ?
* ) , suggesting that this correlation may be at least partially due to observational biases ( e.g. , * ? ? ?
* ) . a search of the sdss dr5 spectroscopic database revealed the spectrum of a ch carbon star , sdss j132755.56 + 333521.7 , in close proximity to the center of cvn ( lower panel of figure [ fig : cvn_morph ] ) .
as shown in figure [ fig : spec ] , the star s spectral features indicate that it is most likely a giant , rather than a dwarf .
there are no 2mass point sources within @xmath62 , and a comparison of the sdss coordinates with earlier - epoch apm data is consistent with zero proper motion , both of which would argue in favor of its being a distant object .
finally , the carbon star s dereddened @xmath28-band magnitude is @xmath63 ; at the calculated distance modulus for cvn , @xmath64 , the star would have an absolute magnitude @xmath65 , fairly typical of a ch carbon giant .
hence we conclude that sdss j132755.56 + 333521.7 is a carbon star in cvn .
we measure a heliocentric radial velocity of @xmath66 km s@xmath67 for the carbon star from cross - correlation with the spectrum of another ch carbon star of known velocity ( 1249 + 0146 , see * ? ? ?
the carbon star s radial velocity in the line of sight direction ( after correction to the galactic rest frame ) is @xmath68 km s@xmath67 .
given the distance of cvn , this is also approximately the radial velocity that would be seen from the galactic center .
we plan to obtain follow - up spectroscopy of rgb stars in cvn to supplement this first estimate of its systemic velocity with a more detailed analysis of the new dwarf s kinematics .
we have discovered a new dwarf spheroidal milky way satellite in the constellation canes venatici .
it has an absolute magnitude of @xmath1 and a surface brightness of @xmath69 mag arcsecond@xmath3 .
we find that cvn is dominated by an old , metal - poor ( [ fe / h ] @xmath70 ) stellar population . at a distance of @xmath0 kpc ,
cvn is one of the most remote of the known dsph companions to the milky way .
it is important to carry out a well - defined census of milky way dwarf galaxies to the faintest possible luminosities .
numerical simulations in cold dark matter cosmological models predict that the milky way s halo should contain about 500 satellites comparable to the dsphs @xcite .
this contrasts with the handful of known milky way dsphs .
the new dsph in canes venatici , taken with the recent discovery of ursa major @xcite , emphasizes how incomplete our current knowledge is .
further analysis of photometric data from large - area digital surveys such as sdss will thus likely lead to the discovery of more dsphs around the milky way in the near future .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . | ten known dwarf spheroidal dsph ) companions milky way galaxy two dwarf irregulars magellanic clouds known satellite galaxies dsphs low surface brightness often found serendipitously sextans @xcite found automated search luminous sagittarius dsph identified from radial velocity survey stars galactic center fields @xcite galaxies resolvable into stars detectable as enhancements stellar number density in large photometric surveys @xcite analyzed overdensities m giants 2mass sky survey claimed new disrupted dsph in canis major controversial @xcite surveyed @xmath6 square degrees survey identified dsph companion constellation ursa major spectroscopic observations showed ursa major dark matter dominated dsph @xcite announced discovery andromeda ix and x new dsph satellites of m31 sdss data @xcite mapped stars satisfying color cut @xmath7 in sdss data release 5 plot `` field of streams '' prominent stellar substructure visible undetected enhancement stellar density constellation canes venatici corresponds new dsph companion eleventh milky way galaxy -0.90 cm lc coordinates j2000 13:28:03.5 + 33:33:21.0@xmath8 + galactic coordinates 74.3 , 79.8 + position angle @xmath9 + ellipticity @xmath10 + @xmath11 plummer @xmath12 + @xmath11 exponential @xmath13 +@xmath14 @xmath15 + @xmath16 plummer @xmath17 + @xmath16 exponential @xmath18 +@xmath19 @xmath20 +@xmath21m@xmath22 @xmath23 + m@xmath24 @xmath25 sdss imaging spectroscopic survey data produced five photometric bands @xmath26 @xmath30 @xcite processed pipelines measure photometric astrometric properties dr5 covers @xmath31 square degrees north galactic pole identified stellar overdensity constellation canes venatici upper left panel figure grayscale sdss image sky stellar overdensity no obvious object elliptical overdensity objects stars visible photometric data stars color - magnitude diagram reveals clear red giant branch ) horizontal branchcmd similar to sdss data sextans dsph fig . 1 more distant morphology distinct stellar population likely unknown dsph galaxy name after constellation canes venatici cvn distance of distinct metal - poor stellar population estimated from @xmath32-band magnitude of tip red giant branch applied transforms from @xcite convert sdss @xmath33 magnitudes to @xmath34 magnitudes cumulative @xmath32-band luminosity function rgb shows rise starting at @xmath35 corresponds to @xmath36 ) horizontal branch lower right panel figure evidence for reliability procedure assuming metallicity of@xmath37 trgb color of @xmath38 calibration @xcite obtain relation @xmath39 distance modulus of @xmath40 ( @xmath41 kpc ) comparable to distant milky way dsph satellites leo i and ii 250 215 kpc age metallicity of stars in cvn estimated by comparison with cmds of single epoch single metallicity populations galactic globular clusters field - star - subtracted hess diagrams color - magnitude density plots in @xmath42 fiducial ridgelines of globular clusters ngc 2419 m 3 m 71 overplotted left to right best match ngc 2419 prominent horizontal branch similarity indicates cvn dominated by old metal - poor @xmath46 ) stellar population horizontal branch extends to blue predominantly red red - dominated horizontal branches range of ages seen in other metal - poor dsphs -0.20 -0.6 cm -0.7 cm investigate spatial morphology select candidate members mask find class conditional probabilities @xmath47 and @xmath48 of star belonging cvn background normalized hess diagrams ratio of class conditional probabilities @xmath49 converted to probability membership cvn via @xmath50 cut of @xmath51 select stars plotted top panel fig . objects binned into @xmath52 bins @xmath53 smoothed with gaussian fwhm @xmath54 plot lower panel density contours 2 3 5 7 10 15 20 25@xmath55 above background level irregular outer parts reminiscent of ursa minor dsph black crosses are candidate blue horizontal branch stars selected with @xmath56distribution of bhbs elongated same as density contours traces morphology estimate properties table analyze morphology figure derive centroid from density - weighted first moment average ellipticity position angle using three density - weighted second moments radial profile derived by average density within elliptical annuli after subtracting constant asymptotic background level ( 0.25 arcminute@xmath3 ) at large radii fit radial profile with standard plummer exponential laws best - fitting position angle ellipticity half - light radii listed in table distance of cvn @xmath57 half - light radius corresponds to @xmath5 pc overall extent is @xmath58 kpc along major axis cvn largest of milky way dsphs fits discrepant datapoints in figure overall luminosity computed by masking stellar locus of cvn computing total flux within mask elliptical half - light radius similar mask larger area minimize noise outside main body cvn scaled by relative area compute foreground contamination within half - light radius after correcting contamination remaining flux scaled to total profiles flux distributionapply correction @xmath59 magnitudes for unresolved faint stars based on stellar luminosity functions low surface brightness dsphs luminosity central surface brightness estimates@xmath60 @xmath2 arcsec@xmath61 consistent with central surface brightness absolute magnitude relation local group dsph satellites cvn far from central surface brightness host galaxy distance correlation dsph satellites due to observational biases search sdss dr5 spectroscopic database revealed spectrum ch carbon star sdss j132755.56 + 333521.7 center cvn star s spectral features indicate likely giant no 2mass point sources within @xmath62 comparison sdss coordinates with earlier - epoch apm data consistent with zero proper motion distant object carbon star s dereddened @xmath28-band magnitude is @xmath63 calculated distance modulus cvn @xmath64 star absolute magnitude @xmath65 typical of ch carbon giant sdss j132755.56 + 333521.7 is carbon star in cvnmeasure heliocentric radial velocity @xmath66 km s@xmath67 carbon star cross - correlation spectrum another ch carbon star carbon star s radial velocity line sight direction correction galactic rest frame @xmath68 km s@xmath67 distance cvn approximately radial velocity seen from galactic center plan follow - up spectroscopy rgb stars cvn supplement estimate velocity analysis new dwarf s kinematics discovered new dwarf spheroidal milky way satellite canes venatici absolute magnitude @xmath1 surface brightness @xmath69 mag arcsecond@xmath3 cvn dominated old metal - poor @xmath70 ) stellar population distance @xmath0 kpc cvn remote dsph companions milky way important census milky way dwarf galaxies milky way s halo 500 satellites comparable dsphs @xcite contrasts known milky way dsphs new dsph canes venatici discovery ursa major @xcite emphasizes incomplete knowledge further analysis photometric data surveys sdss likely discovery more dsphs milky way funding sdss sdss - ii provided alfred p. sloan foundation participating institutions national science foundation u.department energy national aeronautics space administration japanese monbukagakusho max planck society higher education funding council england sdss site.sdss.org managed astrophysical research consortium institutions american museum natural history astrophysical institute potsdam university basel cambridge western reserve chicago drexel university fermilab institute advanced study japan participation group johns hopkins university joint institute nuclear astrophysics kavli institute particle astrophysics korean scientist group chinese academy sciences los alamos national laboratory - institute new mexico state university ohio state university pittsburgh portsmouth princeton university united states naval observatory university washington | 0.505705 | 3,248 | 1,668 | , Saving $0.1 in GPT-4. |
55,558 | using the hipparcos data , absolute magnitudes and thus the evolutionary status for the group of stars were derived .
the origin of this small group of non - magnetic , chemically peculiar stars , still remains a matter of debate .
using new evolutionary tracks , an age determination could be provided to distinguish between the two competing theories - the diffusion / mass - loss and the accretion .
the results establish the members of this group as objects which are _ very close to the main sequence_. this is supported by pre - main sequence evolutionary tracks as well as by observational results .
this contradicts prior conclusions that most of these stars are in the middle of their main sequence lifetime .
the new results strongly support the predictions of the accretion theory . | using the hipparcos parallaxes , absolute magnitudes for members of the group were determined .
the origin of these non - magnetic , a to f - type metal - deficient dwarfs still remains controversial .
the two main competing theories involve diffusion , either in combination with mass - loss ( michaud & charland 1986 ) , or accretion of interstellar matter as in post - agb stars ( turcotte & charbonneau 1993 ) .
the latter model requires that stars are very close to the zero - age main sequence .
the recent discovery of stars in the young orion ob1 association and in ngc 2264 ( paunzen & gray 1997 ) seems to support the predictions of the accretion theory . but
an age determination of galactic field stars by iliev & barzova ( 1995 , hereafter ib95 ) , on the other hand , resulted in evolved members .
these stars were recalibrated using the accurate new absolute magnitudes as well as new stellar evolutionary tracks ( cesam ; morel 1997 ) . furthermore
, pre - main sequence models ( palla & stahler 1993 ) were used to confirm the results .
candidates as well as members of the group were taken from paunzen et al .
( 1997 ) and paunzen & gray ( 1997 ) .
the hipparcos data for the programme stars were extracted with the help of simbad .
the observed visual magnitudes were used to calculate the absolute magnitudes @xmath0(h ) .
possible correlations of the observed parallaxes with other astrophysical quantities ( e.g. apparent distance , effective temperature , metallicity , etc . ) were examined ( see also paunzen 1997 ) .
no systematic trend between the ( old ) photometrically calibrated and ( new ) absolute magnitudes has been detected .
although there are some individual differences , the overall validity of the `` standard '' calibration for the ( chemically peculiar ) stars is proven .
ib95 presented an age and mass determination for 20 well established stars ( and vega ) .
they concluded that most of the investigated stars are in the middle of their main sequence evolution , which is believed to be inconsistent with the much favoured accretion theory .
only one star of their sample ( hd 290799 , a member of the young orion ob1 association ) seems to fulfill the predictions of the accretion theory .
the hipparcos data ( available for 18 stars from ib95 ) were used to test their conclusions .
after the calibration of the programme stars in a @xmath1 versus @xmath2 diagram , the new cesam models ( morel 1997 ) were used to determine the ages and masses .
the initial parameters of the evolutionary tracks were x=0.7 and z=0.02 ( solar abundance ) ; these values were found to be valid for the study of ( chemically peculiar ) stars by ib95 .
this seems to be appropriate because the main contribution to the overall metallicity is due to c , n and o ( solar abundant in stars ) .
furthermore , there are strong indications that the phenomenon is restricted to the stellar surface ( holweger & rentzsch - holm 1995 ) . due to the individual corrections to the absolute magnitude ,
all programme stars ( except hd 193256 and hd 193281 , a distant close binary system ) are , within the errors , significantly _ younger _ and _ less massive_. in order to test a possible pre - main sequence hypothesis and thus the consistency with the accretion theory , the evolutionary tracks from palla & stahler ( 1993 ) were applied .
it turned out that six stars ( hd 30422 , hd 31295 , hd 107233 , hd 110411 , hd 125162 ( itself ) and hd 183324 ) are indeed _ very close to the main sequence_. this is proven by the individual results derived from the pre- and main sequence tracks . for three additional stars ( hd 38545 , hd 111786 and hd 221756 ) , both models are very close , resulting in the same conclusion with a high confidence .
these findings contradict the results from ib95 and strongly support the accretion theory .
the remaining nine programme stars are on the main sequence ( using the appropriate models ) but still significantly less evolved than reported by ib95 ( see fig . 2 therein ) .
these results are consistent with values for other low - mass `` dusty '' pre - main sequence objects ( see also gerbaldi & faraggiana 1993 ) such as herbig ae / be stars or @xmath3 pictoris .
the explanation for stars as true pre - main sequence object could also lead to a solution for the apparent small number of members .
a star with 2m@xmath4 needs only a few 10@xmath5 years to reach the main sequence .
the probability to find such objects is therefore very small compared to the lifetime on the main sequence ( there is also only a similarly small number of herbig ae stars known ) .
this conclusion is further strengthened by the lack of stars in open clusters older than 10@xmath6 years ( paunzen & gray 1997 ) .
with the hipparcos data , absolute magnitudes and evolutionary status ( mass and age ) for the group of stars were estimated .
no systematic influence of the apparent distance , effective temperature , metallicity and rotational velocity was found on the difference between the photometrically calibrated and the `` new '' absolute magnitudes , thus proving the validity of the `` standard '' photometric calibration ( e.g. in the strmgren system ) for these ( chemically peculiar ) stars .
it turned out that six stars ( e.g. itself ) are definitely _ very close to the main sequence _ , and this is also true , with a high probability , of three additional programme stars .
these results contradict the conclusions of ib95 and support the accretion theory .
the small number of stars ( statistical effect due to the short `` lifetime '' on the pre - main sequence ) and the lack of them in open clusters older than 10@xmath6 years further strengthen the accretion hypothesis .
this research was carried outwithinthe working group _
_ seismology - ams _ with funding from the fonds zur frderung der wissenschaftlichen forschung ( project _ s7303-ast _ ) .
use was made of the simbad database , operated at cds , strasbourg , france . | hipparcos parallaxes absolute magnitudes for members group determined origin of non - magnetic metal - deficient dwarfs controversial theories involve diffusion with mass - loss or accretion of interstellar matter post - agb stars latter model requires stars close to zero - age main sequence discovery of stars in young orion ob1 association ngc 2264 accretion theory age determination of galactic field stars by iliev & barzova ib95 resulted in evolved members stars recalibrated using new absolute magnitudes new stellar evolutionary tracks 1997 pre - main sequence models ( palla & stahler 1993 ) confirm results candidates members group taken from paunzen et al . 1997 & gray 1997 ) hipparcos data for programme stars extracted with simbad observed visual magnitudes used to calculate absolute magnitudes correlations of parallaxes with astrophysical quantities examined no systematic trend between old calibrated and new absolute magnitudes detected validity of standard calibration for chemically peculiar stars proven ib95 age and mass determination for 20 established starsconcluded investigated stars main sequence evolution inconsistent with accretion theory only one star ( hd 290799 member young orion ob1 association predictions accretion theory hipparcos data for 18 stars from ib95 used conclusions after calibration stars new cesam models morel 1997 used ages masses initial parameters evolutionary x=0.7 z=0.02 ( solar abundance valid for study chemically peculiar stars by ib95 main contribution to metallicity due to c n o solar abundant indications phenomenon restricted to stellar surface ( holweger holm 1995 ) due corrections magnitude all programme stars except hd 193256 hd 193281 significantly younger less massive test pre - main sequence hypothesis consistency with accretion theory evolutionary tracks from palla & stahler 1993 ) applied six stars ( hd 30422 hd 31295 hd 107233 hd 110411 hd 125162 hd 183324 ) close to main sequence proven by results from pre- main sequence tracks for three additional stars ( hd 38545 hd 111786 hd 221756 ) both models close same conclusion high confidencefindings contradict results ib95 support accretion theory remaining nine programme stars on main sequence less evolved than ib95 fig . 2 results consistent with low - mass pre - main sequence objects herbig ae / be stars @xmath3 pictoris explanation for stars as pre - main sequence object could lead to solution for small number members star with 2m@xmath4 needs 10@xmath5 years to reach main sequence probability to find objects small compared to lifetime on main sequence small number of herbig ae stars conclusion strengthened by lack of stars in open clusters older than 10@xmath6 years hipparcos data absolute magnitudes evolutionary status for stars estimated no systematic influence of distance temperature metallicity rotational velocity found on absolute magnitudes proving validity of standard photometric calibration for stars six stars close to main sequence high probability of three additional programme stars results contradict conclusions ib95 support accretion theory small number of stars short lifetime on pre - main sequence lack in open clusters older than 10@xmath6 years strengthen accretion hypothesisresearch working group seismology - ams funding fonds frderung wissenschaftlichen forschung project s7303-ast use simbad database operated cds strasbourg france | 0.263041 | 1,393 | 691 | , Saving $0.0 in GPT-4. |
55,559 | a partial wave analysis of the centrally produced @xmath0 and @xmath1 systems shows that the @xmath2 has j = 0 .
in addition , a study of central meson production as a function of the difference in transverse momentum ( @xmath3 ) of the exchanged particles shows that undisputed @xmath4 mesons are suppressed at small @xmath3 whereas the glueball candidates are enhanced and that the production cross section for different resonances depends strongly on the azimuthal angle between the two outgoing protons .
= 14.5pt | there is considerable current interest in trying to isolate the lightest glueball .
several experiments have been performed using glue - rich production mechanisms .
one such mechanism is double pomeron exchange ( dpe ) where the pomeron is thought to be a multi - gluonic object .
consequently it has been anticipated that production of glueballs may be especially favoured in this process@xcite . the wa102 experiment at the cern omega spectrometer studies centrally produced exclusive final states formed in the reaction @xmath5 where the subscripts @xmath6 and @xmath7 refer to the fastest and slowest particles in the laboratory frame respectively and @xmath8 represents the central system
the isolation of the reaction @xmath9 has been described in detail in a previous publication@xcite .
a partial wave analysis ( pwa ) of the centrally produced system has been performed , using the reflectivity basis@xcite , in 40 mev intervals of the mass spectrum using an event - by - event maximum likelihood method@xcite .
the @xmath10 and @xmath11-waves from the physical solution are shown in fig .
[ fi:1 ] .
the @xmath10-wave shows a threshold enhancement ; the peaks at 1.5 gev and 1.7 gev are interpreted as being due to the @xmath12 and @xmath2 with j = 0 .
a fit has been performed to the @xmath10 wave using three interfering breit - wigners to describe the @xmath13 , @xmath12 and @xmath2 and a background of the form @xmath14 , where @xmath15 is the mass , @xmath16 is the threshold mass and a , b , c , d are fit parameters .
the resulting fit is shown in fig .
[ fi:1 ] and gives for the @xmath13 m = 985 @xmath17 10 mev , @xmath18 = 65 @xmath17 20 mev , for the @xmath12 m = 1497 @xmath17 10 mev , @xmath18 = 104 @xmath17 25 mev and for the @xmath19 m = 1730 @xmath17 15 mev , @xmath18 = 100 @xmath17 25 mev parameters which are consistent with the pdg@xcite values for these resonances .
the @xmath11-wave shows peaks in the 1.3 and 1.5 gev regions , presumably due to the @xmath20 and @xmath21 and a wide structure above 2 gev .
there is no evidence for any significant structure in the d - wave in the region of the @xmath2 .
in addition , there are no statistically significant structures in any of the other waves .
a fit has been performed to the @xmath11 wave above 1.2 gev using three incoherent relativistic spin 2 breit - wigners to describe the @xmath20 , @xmath21 and the peak at 2.2 gev and a background of the form described above .
the resulting fit is shown in fig .
[ fi:1 ] and gives for the @xmath20 m = 1305 @xmath17 20 mev , @xmath18 = 132 @xmath17 25 mev , for the @xmath21 m = 1515 @xmath17 15 mev , @xmath18 = 70 @xmath17 25 mev and for the @xmath22 m = 2130 @xmath17 35 mev , @xmath18 = 270 @xmath17 50 mev .
a study has also been made of the centrally produced channel@xcite .
this channel has lower statistics than the channel but has the advantage that only even spins can contribute , which also means that there are only two ambiguous solutions to the pwa .
the physical solution resulting from the pwa is the same as for the final state ; namely that the @xmath10-wave shows a threshold enhancement and peaks at 1.5 gev and 1.7 gev interpreted as being due to the @xmath12 and @xmath2 with j = 0 .
the isolation of the reaction @xmath23 has been described in detail in a previous publication@xcite .
the resulting centrally produced system consists of 2.87 million events .
a pwa of the centrally produced system has been performed , using the reflectivity basis@xcite , in 20 mev intervals of the mass spectrum using an event - by - event maximum likelihood method@xcite .
the @xmath10 , @xmath24 and @xmath11-waves from the physical solution are shown in fig .
the @xmath10-wave spectrum shows a clear threshold enhancement followed by a sharp drop at 1 gev .
there is clear evidence for the @xmath25 in the @xmath24 wave and for the @xmath26 in the @xmath11 wave .
an interesting feature of the @xmath11 wave is the presence of a structure below 1 gev . in order to see
if this effect is due to acceptance problems or problems due to non - central events , we have reanalysed the data using a series of different cuts but after acceptance correction no cut has been found that can remove the low mass structure . in order to investigate any systematic effects we have also analysed the central @xmath27 data and a similar structure is also found@xcite .
this structure does indeed seem to be a real effect which is present in other centrally produced @xmath1 systems@xcite . in order to obtain a satisfactory fit to the @xmath10 wave from threshold to 2 gev
it has been found to be necessary to use three interfering breit - wigners to describe the @xmath13 , @xmath28 and @xmath12 and a background of the form @xmath14 , where @xmath15 is the mass , @xmath16 is the threshold mass and a , b , c , d are fit parameters .
the fit is shown in fig .
[ fi:2]a ) for the entire mass range and in fig .
[ fi:2]b ) for masses above 1 gev .
the resulting parameters are for the @xmath13 m = 982 @xmath17 3 mev , @xmath18 = 80 @xmath17 10 mev , for the @xmath28 m = 1308 @xmath17 10 mev , @xmath18 = 222 @xmath17 20 mev and for the @xmath12 m = 1502 @xmath17 10 mev , @xmath18 = 131 @xmath17 15 mev which are consistent with the pdg @xcite values for these resonances .
as can be seen , the fit describes the data well for masses below 1 gev .
it was not possible to describe the data above 1 gev without the addition of both the @xmath28 and @xmath12 resonances .
however , even with this fit using three breit - wigners it can be seen that the fit does not describe well the 1.7 gev region .
this could be due to a decay mode of the @xmath2 with j = 0 . including a fourth breit - wigner in this mass region decreases the @xmath29 from 256 to 203 and yields for the @xmath2 m = 1750 @xmath17 20 mev and @xmath18 = 160 @xmath17 30 mev parameters which are consistent with the pdg @xcite values for the @xmath2 .
the fit is shown in fig .
[ fi:2]c ) for masses above 1 gev .
the wa102 experiment studies mesons produced in double exchange processes .
however , even in the case of pure dpe the exchanged particles still have to couple to a final state meson .
the coupling of the two exchanged particles can either be by gluon exchange or quark exchange .
assuming the pomeron is a colour singlet gluonic system if a gluon is exchanged then a gluonic state is produced , whereas if a quark is exchanged then a @xmath30 state is produced@xcite . in order to describe
the data in terms of a physical model , close and kirk@xcite , have proposed that the data be analysed in terms of the difference in transverse momentum ( @xmath3 ) between the particles exchanged from the fast and slow vertices .
the idea being that for small differences in transverse momentum between the two exchanged particles an enhancement in the production of glueballs relative to @xmath4 states may occur . the contribution of each resonance as a function of @xmath3 has been calculated .
figure [ fracratio ] shows the ratio of the number of events for @xmath3 @xmath31 0.2 gev to the number of events for @xmath3 @xmath32 0.5 gev for each resonance considered .
it can be observed that all the undisputed @xmath4 states which can be produced in dpe , namely those with positive g parity and @xmath33 , have a very small value for this ratio ( @xmath34 ) .
some of the states with @xmath35 or g parity negative , which can not be produced by dpe , have a slightly higher value ( @xmath36 ) .
however , all of these states are suppressed relative to the the glueball candidates the @xmath12 , @xmath2 , and @xmath37 , together with the enigmatic @xmath13 , which have a large value for this ratio .
the azimuthal angle ( @xmath38 ) is defined as the angle between the @xmath39 vectors of the two protons . naively it may be expected that this angle would be flat irrespective of the resonances produced . fig .
[ fi : phidep ] shows the @xmath38 dependence for two @xmath40 = @xmath41 final states ( the @xmath42 and @xmath43 ) , two @xmath40 = @xmath44 final states ( the @xmath45 and @xmath46 ) and two @xmath40 = @xmath47 final states ( the @xmath48 and @xmath49 systems ) . the @xmath38 dependence is clearly not flat and considerable variation is observed between final states with different @xmath40s .
in conclusion , a partial wave analysis of the centrally produced @xmath0 system has been performed .
the striking feature is the observation of peaks in the @xmath10-wave corresponding to the @xmath12 and @xmath2 with j = 0 .
in addition , a partial wave analysis of a high statistics sample of centrally produced events shows that the @xmath10-wave is composed of a broad enhancement at threshold , a sharp drop at 1 gev due to the interference between the @xmath13 and the s - wave background , the @xmath28 , the @xmath12 and the @xmath2 with j = 0 .
a study of centrally produced pp interactions show that there is the possibility of a glueball-@xmath4 filter mechanism ( @xmath3 ) .
all the undisputed @xmath30 states are observed to be suppressed at small @xmath3 , but the glueball candidates @xmath12 , @xmath2 , and @xmath37 , together with the enigmatic @xmath13 , survive .
in addition , the production cross section for different resonances depends strongly on the azimuthal angle between the two outgoing protons .
99 d.robson , nucl . phys . * b130 * , 328 ( 1977 ) ; + f.e .
close , rep .
. phys . * 51 * , 833 ( 1988 ) .
d. barberis _ et al .
, _ hep - ex/9903042 .
chung , phys .
rev . * d56 * , 7299 ( 1997 ) .
particle data group , european physical journal * c3 * , 1 ( 1998 ) .
d. barberis _ et al .
, _ hep - ex/9903043 .
d. barberis _ et al .
, _ hep - ex/9903044 .
d. alde _ et al . , _
( na12/2 collaboration ) in preparation ; + m.c .
et al . , _ ( e690 collaboration ) proceedings of hadron 97 , aip conf .
( 1997 ) 36 ; + t. @xmath50kesson _ et al . , _
( afs collaboration ) nucl .
* b264 * , 154 ( 1986 ) .
f.e . close and a. kirk , phys .
b397 * , 333 ( 1997 ) . | interest lightest glueball experiments performed using glue - rich production mechanisms mechanism double pomeron exchange pomeron multi - gluonic object anticipated production glueballs favoured wa102 experiment at cern omega spectrometer studies final states reaction @xmath5 subscripts @xmath6 @xmath7 fastest slowest particles @xmath8 central system of reaction @xmath9 described previous publication partial wave analysis of system performed reflectivity basis in 40 mev intervals mass event - by - event maximum likelihood method @xmath10 @xmath11-waves solution shown in fig . @xmath10-wave shows threshold enhancement peaks at 1.5 gev 1.7 gev due to @xmath12 @xmath2 with j = 0 fit performed to @xmath10 wave using three interfering breit - wigners @xmath13 form @xmath14 @xmath15 mass @xmath16 threshold mass d fit parameters resulting fit shown in fig .fi:1 ] @xmath13 m = 985 10 = 65 20 mev @xmath12 m = 1497 10 = 104 25 mev @xmath19 m = 1730 15 = 100 25 consistent with pdg@xcite values @xmath11-wave shows peaks 1.3 1.5 gev due to @xmath20 @xmath21 wide structure above 2 gev no evidence significant structure d - wave @xmath2 no statistically significant structures other waves fit performed @xmath11 wave above 1.2 gev using three incoherent relativistic spin 2 breit - wigners @xmath20 @xmath21 peak 2.2 gev form resulting fit fig . fi:1 @xmath20 m = 1305 20 mev = 132 25 mev @xmath21 m = 1515 15 mev = 70 25 mev @xmath22 m = 2130 35 mev = 270 50 mev study centrally produced channel@xcite lower statistics even spins contribute two ambiguous solutions pwaphysical solution from pwa same as final state @xmath10-wave shows threshold enhancement peaks at 1.5 1.7 gev due to @xmath12 @xmath2 with j = 0 isolation of reaction @xmath23 described in previous publication centrally produced system 2.87 million events pwa performed reflectivity basis in 20 mev intervals event - by - event maximum likelihood method @xmath10-waves from solution shown in fig . @xmath10-wave shows threshold enhancement sharp drop at 1 gev clear evidence for @xmath25 @xmath26 in wave @xmath11 wave structure below 1 gev acceptance problems or non - central events reanalysed data cuts no cut found remove low mass structure analysed central @xmath27 data similar structure found structure real effect present in other centrally produced @xmath1 systems satisfactory fit to @xmath10 wave from threshold to 2 gev to use three interfering breit - wigners @xmath13 @xmath28 @xmath12 background of form @xmath14 @xmath15 is mass @xmath16 threshold mass a b c d are fit parametersfit shown in fig .:2]a entire mass range:2]b masses above 1 gev parameters @xmath13 m = 982 3 mev = 80 10 mev @xmath28 m = 1308 10 = 222 20 mev @xmath12 m = 1502 10 = 131 15 mev consistent with pdg @xcite values resonances fit describes data masses below 1 gev not possible describe data above 1 gev without addition @xmath28 @xmath12 resonances three breit - wigners fit describe 1.7 gev region decay mode @xmath2 with j = 0 fourth breit - wigner decreases @xmath29 from 256 to 203 yields @xmath2 m = 1750 @xmath17 20 mev @xmath18 = 160 30 mev parameters consistent with pdg @xcite values @xmath2 fit shown fig .:2]c masses above 1 gev wa102 experiment studies mesons double exchange processes exchanged particles couple to final state meson coupling gluon or quark exchange gluon gluonic state produced quark @xmath30 state produceddata physical model close kirk@xcite proposed data difference transverse momentum @xmath3 ) between particles exchanged from fast and slow vertices small differences momentum enhancement production glueballs @xmath4 states may occur contribution of each resonance @xmath3 calculated figure shows ratio events for @xmath3 @xmath31 0.2 gev to @xmath3 @xmath32 0.5 gev for each resonance undisputed @xmath4 states produced in dpe positive g parity and @xmath33 have small value for ratio ( @xmath34 ) some states with @xmath35 or g parity negative have higher value ( @xmath36 ) states suppressed relative to glueball candidates @xmath12 @xmath2 @xmath37 @xmath13 large value for ratio azimuthal angle ( @xmath38 ) defined as angle between @xmath39 vectors of two protons angle flat irrespective of resonances produced fig . shows @xmath38 dependence for two @xmath40 = @xmath41 final states = @xmath44 = @xmath47@xmath38 dependence not flat variation between final states @xmath40s partial wave analysis of centrally @xmath0 system performed peaks in @xmath10-wave @xmath12 @xmath2 with j = 0 partial wave analysis high statistics sample events shows @xmath10-wave broad enhancement at threshold sharp drop at 1 gev interference between @xmath13 s - wave background @xmath28 @xmath12 @xmath2 with j = 0 study interactions possibility of glueball-@xmath4 filter mechanism @xmath30 states suppressed at small @xmath3 glueball candidates @xmath12 @xmath2 @xmath37 @xmath13 survive production cross section for resonances depends on azimuthal angle between two outgoing protons.robson nucl . . b130, 328 ( 1977 ) phys 51 833 ( 1988 ) barberis et al hep - ex/9903042 chung phys . rev d56 7299 ( 1997 ) particle data group european physical journal c3 1 ( 1998 ) barberis - ex/9903043na12/2 collaboration preparation m.c. et al e690 collaboration proceedings hadron 97 aip conf 1997 ) 36 @xmath50kesson et al afs collaboration nucl b264 154 1986 ) close kirk b397 333 1997 ) | 0.428199 | 2,897 | 1,454 | , Saving $0.1 in GPT-4. |
55,571 | we present the analysis of cp - violating effects in non - diagonal chargino pair production in @xmath0 collisions .
these effects appear only at the one - loop level .
we show that cp - odd asymmetries in chargino production are sensitive to the phases of @xmath1 and @xmath2 parameters and can be of the order of a few % . | supersymmetry ( susy ) is one of the most promising extensions of the standard model ( sm ) @xcite since , among other things , it solves the hierarchy problem , provides a natural candidate for dark matter _ etc_.
it also introduces many new sources of cp violation that may be needed to explain baryon asymmetry of the universe .
these phases , if large @xmath3 , can cause problems with satisfying experimental bounds on lepton , neutron and mercury edms @xcite .
this can be overcome by pushing sfermion spectra above a tev scale or arranging internal cancelations @xcite .
most unambiguous way to detect the presence of cp - violating phases would be to study cp - odd observables measurable at future accelerators the lhc and the ilc .
such observables in the chargino sector are , for example , triple products of momenta of initial electrons , charginos and their decay products @xcite .
however they require polarized initial electron / positron beams or measurement of chargino polarization . in this talk we present another possibility of detecting cp - violating phases in the chargino sector .
as it was recently pointed out @xcite , in non - diagonal chargino pair production @xmath4 a cp - odd observable can be constructed beyond tree - level from production cross section without polarized @xmath0 beams or measurement of chargino polarization .
we show here the results of the full one - loop calculation of this effect . in the reaction ( [ produkcja ] )
the cp violation can be induced by the complex higgsino mass parameter @xmath1 or complex trilinear coupling in top squark sector @xmath2 . since these asymmetries
can reach a few percent , they can be detected in simple event - counting experiments at future colliders .
in @xmath0 collisions charginos are produced at tree - level via the @xmath5-channel @xmath6 exchange and @xmath7-channel @xmath8 exchange . as it was shown in @xcite no cp violation effects
can be observed at the tree - level for the production processes of diagonal @xmath9 and non - diagonal @xmath10 chargino pairs without the measurement of polarization of final chargino .
however the situation is different for non - diagonal production if we go beyond tree - level approximation .
radiative corrections to the chargino pair production include the following generic one - loop feynman diagrams : the virtual vertex corrections , the self - energy corrections to the @xmath11 , @xmath12 and @xmath13 propagators , and the box diagrams contributions .
we also have to include corrections on external chargino legs .
one - loop corrected matrix element squared is given by @xmath14 accordingly , the one - loop cp asymmetry for the non - diagonal chargino pair is defined as @xmath15 where @xmath16 , @xmath17 denote cross sections for production of @xmath18 and @xmath19 , respectively .
since the asymmetry vanishes at tree - level it has to be finite at one loop , hence no renormalization is needed .
vs. center of mass energy .
the selectron mass is 403 gev.[fig : thrs_scan],title="fig : " ] vs. center of mass energy .
the selectron mass is 403 gev.[fig : thrs_scan],title="fig : " ] the cp asymmetry eq .
( [ cpasym ] ) arises due to the interference between complex couplings , which in our case are due to complex mixing matrices of charginos or stops , and non - trivial imaginary part from feynman diagrams the absorptive part .
such contributions appear when some of the intermediate state particles in loop diagrams go on - shell .
this is illustrated in fig .
[ fig : thrs_scan ] where the contribution to @xmath20 from double selectron exchange appears at the threshold for selectron pair production at @xmath21 gev .
for the numerical results in this section we use two parameter sets ( a ) and ( b ) with gaugino / higgsino mass parameters defined as follows at the low scale : @xmath22 and with @xmath23 .
this gives the following chargino masses : @xmath24 for the sfermion mass parameters in scenario ( a ) we assume @xmath25 and for the sfermion trilinear coupling : @xmath26 .
scenario ( b ) is for comparison with ref .
@xcite for which we take @xmath27 in our numerical analysis we consider the dependence of the asymmetry ( [ cpasym ] ) on the phase of the higgsino mass parameter @xmath28 and soft trilinear top squark coupling @xmath29 . in fig .
[ fig2 ] we show the cp asymmetry in scenario ( a ) as a function of the phase of @xmath1 and @xmath2 , left and middle panel , respectively .
contributions due to box corrections , vertex corrections and self energy corrections have been plotted in addition to the full result . in this scenario
the asymmetry can reach @xmath30 for the @xmath1 parameter and @xmath31 for @xmath2 , respectively .
we note that for the asymmetry due to the non - zero phase of the higgsino mass parameter there are significant cancelations among various contributions .
in addition , we also show in the right panel of fig .
[ fig2 ] the dependence of the asymmetry due to @xmath2 as a function of @xmath32 . for the asymmetry generated by the @xmath1 parameter
all possible one - loop diagrams containing absorptive part contribute .
the situation is different for the phase of the trilinear coupling @xmath2 when chargino mixing matrices remain real . in this case
only vertex and self - energy diagrams containing stop lines contribute to the asymmetry @xcite . in scenario ( a ) as a function of the phase of @xmath1 parameter ( left ) , the phase of @xmath2 ( middle ) , and as a function of @xmath32 with @xmath33 ( right ) .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
[ fig2],title="fig : " ] in scenario ( a ) as a function of the phase of @xmath1 parameter ( left ) , the phase of @xmath2 ( middle ) , and as a function of @xmath32 with @xmath33 ( right ) .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
[ fig2],title="fig : " ] in scenario ( a ) as a function of the phase of @xmath1 parameter ( left ) , the phase of @xmath2 ( middle ) , and as a function of @xmath32 with @xmath33 ( right ) .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
[ fig2],title="fig : " ] we present also the results for the heavy sfermion scenario ( b ) .
this is to compare with @xcite where only box diagrams with @xmath13 , @xmath34 , @xmath12 exchanges have been calculated neglecting all sfermion contributions . as can be seen in the left panel of fig .
[ fig3 ] these gauge - box diagrams constitute the main part of the asymmetry @xmath20 , however this is due to partial cancelation of vertex and self - energy contributions . for lower values of the universal scalar mass @xmath35
the discrepancy between full and approximate result of @xcite increases significantly .
this is illustrated in the middle and right panel of fig .
[ fig3 ] where we show two paths of approaching of the full result to the gauge - box approximation as the function of @xmath35 .
as can be seen these paths depend strongly on the center of mass energy . in scenario ( b ) as a function of the phase @xmath36 .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
middle and right : asymmetry @xmath20 as a function of the universal scalar mass @xmath35 with @xmath37 at different cms .
the full lines denote full result and dashed lines show only the box contributions after neglecting diagrams with slepton exchange.[fig3],title="fig : " ] in scenario ( b ) as a function of the phase @xmath36 .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
middle and right : asymmetry @xmath20 as a function of the universal scalar mass @xmath35 with @xmath37 at different cms .
the full lines denote full result and dashed lines show only the box contributions after neglecting diagrams with slepton exchange.[fig3],title="fig : " ] in scenario ( b ) as a function of the phase @xmath36 .
different lines denote full asymmetry ( full line ) and contributions from box ( dashed ) , vertex ( dotted ) and self energy ( dash - dotted ) diagrams .
middle and right : asymmetry @xmath20 as a function of the universal scalar mass @xmath35 with @xmath37 at different cms .
the full lines denote full result and dashed lines show only the box contributions after neglecting diagrams with slepton exchange.[fig3],title="fig : " ]
it has been shown that cp - odd asymmetry can be generated in non - diagonal chargino pair production with unpolarized electron / positron beams .
the asymmetry is pure one - loop effect and is generated by interference between complex couplings and absorptive parts of one loop integrals .
the effect is significant for the phases of the higgsino mass parameter @xmath1 and the trilinear coupling in stop sector @xmath2 . at future linear collider
it may give information about cp violation in chargino and stop sectors .
h. p. nilles , phys .
* 110 * ( 1984 ) 1 ; h. e. haber and g. l. kane , phys .
rept . *
117 * ( 1985 ) 75 .
v. barger , t. falk , t. han , j. jiang , t. li and t. plehn , phys .
d * 64 * ( 2001 ) 056007 [ arxiv : hep - ph/0101106 ] .
t. ibrahim and p. nath , arxiv:0705.2008 [ hep - ph ] and references therein .
o. kittel , a. bartl , h. fraas and w. majerotto , phys .
d * 70 * ( 2004 ) 115005 [ arxiv : hep - ph/0410054 ] .
p. osland and a. vereshagin , phys
. rev .
d * 76 * ( 2007 ) 036001 [ arxiv:0704.2165 [ hep - ph ] ] .
k. rolbiecki and j. kalinowski , arxiv:0709.2994 [ hep - ph ] . s. y. choi , a. djouadi , h. s. song and p. m. zerwas , eur .
j. c * 8 * ( 1999 ) 669 [ arxiv : hep - ph/9812236 ] . | supersymmetry ( ) promising standard model solves hierarchy problem candidate for dark matter introduces new sources of cp violation explain baryon asymmetry phases large cause problems experimental bounds on lepton neutron mercury edms overcome by pushing sfermion spectra above tev scale or arranging internal cancelations detect cp - violating phases study cp - odd observables at future accelerators ilc observables in chargino sector products of momenta initial electrons charginos decay products require polarized initial electron / positron beams or measurement chargino polarization another possibility detecting cp - violating phases chargino non - diagonal chargino pair production cp - odd observable constructed beyond tree - level without polarized beams or measurement polarization results of full one - loop calculation cp violation induced by complex higgsino mass parameter or complex trilinear coupling in top squark sector asymmetries detected in event - counting experiments at future colliders collisions charginos produced at tree - level via exchangeshown in @xcite no cp violation effects at tree - level diagonal @xmath9 non - @xmath10 chargino pairs without measurement polarization different for non - diagonal production beyond tree radiative corrections chargino production include - loop feynman diagrams virtual vertex corrections self - energy corrections @xmath11 propagators box diagrams contributions corrections on external chargino legs one - loop corrected matrix element by @xmath14 one - loop cp asymmetry non - diagonal chargino pair defined as @xmath15 @xmath16 cross sections for production @xmath18 @xmath19 asymmetry vanishes at tree - level finite at one loop no renormalization needed selectron mass 403 cp asymmetry arises interference between complex couplings mixing matrices non - trivial imaginary part feynman diagrams absorptive part contributions appear when intermediate state particles loop diagrams on - shell illustrated in fig contribution to @xmath20 from double selectron exchange threshold for selectron pair production at @xmath21 gevnumerical results use two parameter sets a b ) gaugino higgsino mass parameters low scale @xmath22 @xmath23 chargino masses : @xmath24 sfermion mass parameters scenario a assume @xmath25 sfermion trilinear coupling @xmath26 scenario b ) comparison @xcite take @xmath27 numerical analysis dependence asymmetry cpasym phase higgsino mass parameter @xmath28 soft trilinear top squark coupling @xmath29 fig . fig2 ] cp asymmetry scenario a ) function phase @xmath1 @xmath2 contributions box corrections vertex corrections self energy corrections plotted result asymmetry reach @xmath30 @xmath1 @xmath31 @xmath2 asymmetry non - zero phase higgsino mass parameter significant cancelations among contributions dependence asymmetry @xmath2 function @xmath32 asymmetry generated @xmath1 parameter one - loop diagrams absorptive part contribute different phase trilinear coupling @xmath2 chargino mixing matrices real vertex self - energy diagrams stop lines contribute asymmetry @xcitescenario ( a ) function of phase @xmath1 @xmath2 @xmath32 @xmath33 right ) lines denote full asymmetry contributions from box vertex self energy - diagrams fig2] scenario ( a ) function phase @xmath1 @xmath2 @xmath32 @xmath33 lines denote full asymmetry contributions from box vertex self energy dash - dotted diagrams fig2] scenario a function phase @xmath1 @xmath2 @xmath32 @xmath33 different lines denote full asymmetry full line ) contributions from box vertex dotted self energy dash - dotted ) diagrams results for heavy sfermion scenario ( b ) compare with @xcite box diagrams with @xmath13 @xmath34 @xmath12 exchanges calculated neglecting sfermion contributions left panel fig . fig3 gauge - box diagrams main part of asymmetry @xmath20 due to partial cancelation of vertex self - energy contributions lower values of universal scalar mass @xmath35 discrepancy between full approximate result of @xcite increases illustrated in middle right panel of fig .fig3 two paths full result gauge - box approximation @xmath35 paths depend on center mass energy scenario ( b ) phase @xmath36 lines denote full asymmetry contributions from box vertex self energy diagrams asymmetry @xmath20 universal scalar mass @xmath35 with @xmath37 different cms full lines denote result dashed lines show box contributions after neglecting diagrams exchange scenario b ) @xmath36 lines denote full asymmetry contributions from box vertex self energy diagrams asymmetry @xmath20 scalar mass @xmath35 with @xmath37 different cms full lines denote result dashed lines show box contributions after neglecting diagrams exchange scenario ( b ) phase @xmath36 lines denote full asymmetry contributions from box vertex self energy diagrams asymmetry @xmath20 universal scalar mass @xmath35 with @xmath37 different cms full lines denote result dashed lines show box contributions after neglecting diagrams slepton exchange cp - odd asymmetry generated in non - diagonal chargino pair production with unpolarized electron / positron beamsasymmetry one - loop effect generated interference complex couplings absorptive integrals significant phases higgsino mass parameter @xmath1 trilinear coupling stop sector @xmath2 future linear collider information cp violation chargino stop sectors nilles. 110 * 1984 ) 1. haber kane 117 * 1985 ) 75 barger falk han jiang li plehn phys 64 2001 ) 056007 hep - ph/0101106 t. ibrahim p. nath arxiv:0705.2008 hep - ph. kittel bartl fraas majerotto phys 70 2004 ) 115005 hep - ph/0410054 ] p. osland vereshagin. 76 * 2007 ) 036001:0704.2165 hep - ph k. rolbiecki j. kalinowski arxiv:0709.2994 hep - ph choi djouadi song zerwas 8 1999 ) 669 hep - ph/9812236 ] | 0.357866 | 2,571 | 1,276 | , Saving $0.1 in GPT-4. |
55,577 | fermilab s hadron physics research continues in all its accelerator - based programs .
these efforts will be identified , and the optimization of the fermilab schedules for physics will be described .
in addition to the immediate plans , the fermilab long range plan will be cited , and the status and potential role of a new proton source , the proton driver , will be described . . | hadron physics is found in three areas of accelerator physics at fermilab : the tevatron collider experiments , the neutrino program , and fixed target physics with beams from the main injector . at the collider , cdf and dzero
are in the midst of a multi - year run now , and btev is scheduled to take over as soon as these programs are ended . in the neutrino program , we have miniboone using beam from the booster accelerator now , and will be turning on numi , the neutrino program using beam from the main injector , in a few weeks .
the minos experiment will operate with the neutrino beams as soon as they are available , and the minerva experiment has been approved to join as soon as it can be built . finally , there is the mipp experiment , e907 - main injector particle production , finishing its commissioning run now , and about to get its physics data . also approved for the 120 gev main injector program is a drell - yan experiment ( e906 ) , which is starting to build its detector this year .
the possibility of returning to run for kaon physics using beam from the main injector is retained .
the fermilab accelerator experiment schedules through 2009 are available on the web at ` http://www.fnal.gov/directorate/program_planning/schedule/index.html ` optimization of running conditions is a challenge with such a diverse program , especially since each component of the fermilab program has its own preference for how we should operate the accelerator complex , particularly the proton source . the tevatron collider benefits from as rapid a cycling of the upstream accelerators as is consistent with antiproton beam - cooling times in the accumulator and recycler rings , with as much beam as acceptable for losses .
the neutrino program can use even more rapid cycle times , with as much beam as acceptable for losses .
for neutrinos , there is no lower limit on how fast the proton source accelerators should cycle . at the opposite extreme
is the so - called sy120/fixed target program , which depends on experiment and test - beam user status . during set up , multiple , somewhat short duration spills per minute
is optimal . during data taking ,
long cycles with long spills are best for maximizing integrated spill seconds by minimizing end effects .
the office of program planning is charged with organizing all this by coordinating between experiments and the accelerator , under given guidelines from director . at the moment , for example , there is the following guideline on the use of beam time for sy120 : .... the integrated effect of running sy120 beam will not reduce the antiproton stacking rate by more than 5% globally , with the details of scheduling to be worked out between the experimenters and the office of program planning . .... note that this guideline is flexible , leaving the details of spill length and rate open to maximize the physics of the sy120 program .
we also have the flexibility of mixing the various spill configurations in the accelerator time line sequence , adding an element of flexibility - and complexity .
the fermilab director established the fermilab long range planning committee ( lrpc ) in the spring of 2003 . an excerpt from the charge to the lrpc states :
.... i would like the long - range planning committee to develop in detail a few realistically achievable options for the fermilab program in the next decade under each possible outcome for the linear collider .
.... it was clear from the start that a new intense proton source to serve long baseline neutrino experiments and to provide other new physics options at fermilab was one such option .
a lrpc working group , with bob kephart as chairman , was charged to explore this option .
the full lrpc received recommendations from this working group .
the recommendations were subsequently adopted in the final flrpc report and by the director .
the fermilab linear accelerator and booster are old machines . at the same time
, there is a desire for significantly more intense proton sources for long baseline neutrino physics . in order to meet the new demands
, a new proton source will need to be constructed .
the high level parameters desired for such a new facility at fermilab are : * @xmath0 @xmath1 beam power at 8 gev * @xmath2 @xmath1 beam power at 120 gev , 6 times the power of the current main injector there are two possible implementations being considered at fermilab ; an 8 gev synchrotron and an 8 gev superconducting rf linear accelerator . a web page has been established to track the progress on such a new proton source , the proton driver : http://www-bd.fnal.gov/pdriver/ currently , most of the accelerator effort is focused on bringing design for the linear accelerator to the same level as that for the synchrotron so that a final choice and proposal can be made .
however , it is safe to say that there is particular appreciation for some linear accelerator advantages : better performance , flexibility , and the connection and possible synergies with an international linear collider ( ilc ) , now that `` cold '' technology has been chosen for the latter . for either choice of 8 gev injector ( synchrotron or linear accelerator ) the beam in the main injector will increase by a factor of 5 from its design value of @xmath3 @xmath4 protons per pulse to @xmath5 @xmath6 protons per cycle .
the main injector beam power could also be increased by shortening the main injector ramp time .
this would require additional magnet power supplies , and could be done prior to a proton driver as a first step .
both more protons per cycle and faster ramp times imply more rf power , meaning money is required .
the recent international technology recommendation panel opted for a `` cold '' technology for the international linear collider .
this recommendation , and its wide acceptance , has provided a huge boost for a super - conducting rf linear - accelerator - based proton driver at fermilab .
fermilab has been working on both warm and cold linear - collider - related technology . in fy 2005
, all this effort will be dedicated to cold technology , also advancing a linear - accelerator proton driver .
so , fy 2005 will see as much as a factor of two increase in super - conducting r&d spending at fermilab relative to fy 2004 .
plans are forming for a scrf module test facility ( smtf ) to be built in the meson east area at fermilab .
long lead - time items like modulators are already being ordered .
it seems likely that a new , more intense proton source will be proposed for construction at fermilab in the relatively near future .
such a project would be similar in scope to the main injector project ( i.e. , for cost and schedule ) .
an 8 gev synchrotron and a superconducting rf linear accelerator both appear to be technically possible .
the linear accelerator is strongly preferred by many , if it can be made affordable .
the fermilab management has requested that an 8 gev linear accelerator design be developed , including cost and schedule information .
fermilab s bill foster is leading a team to develop the technical design .
it is always hard to guess about time scales . a technically limited schedule ( when was the last time this happened ? )
could have the following doe construction project critical decision ( cd ) schedule : * cd-0 ( statement of need ) submitted in 2005 * cd-1 ( preliminary acquisition strategy , project execution plan , conceptual design report , project scope , baseline cost / schedule range , project management plan , hazard analysis , etc . )
submitted in 2006 * cd-2/3a ( project baseline approved , approval to start construction ) in 2007 - 8 .
as usual , the availability of funding from doe may push this schedule later . once funding is approved , typical projects of this scale ( the fermilab main injector , slac b - factory , spallation neutron source ) have construction times of 4 - 5 years
. the time scale will also depend on how the international linear collider plays out over the next few years ( e.g. , could the proton driver be seen as an ilc engineering test facility ? ) .
it is up to users to make the physics case that a proton driver is required , and that it should be built as soon as possible .
making the physics case is crucial in all of this ! a workshop to think about the broadest physics potential of a proton driver was just held .
the program , with links to talks can be found at the above proton driver web page .
steve geer has been asked to organize the documentation of the physics case .
a strong physics case , coupled to a machine technical design , will make it possible to submit a proton driver project to the doe for approval and funding .
the recent proton driver workshop was a step in this process .
information on the workshop and talks can be found linked to the web page : http://www-td.fnal.gov/projects/pd/physicsincludes/workshop/index.html three of the working groups at the workshop may be most interesting for hadron physics : * wg2 on neutrino interactions jorge morfin ( fermilab ) , rex tayloe ( indiana ) , ron ransome ( rutgers ) , conveners * wg4 on kaons / pions hogan nguyen ( fermilab ) and taku yamanaka ( osaka ) , conveners * wg5 on antiprotons dave christian ( fermilab ) and mark mandelkern ( uc - irvine ) , conveners while not represented at the workshop , there is also a working group on neutrons . the workshop , of course , had additional working groups less connected to hadron physics . in each case , the working groups identified physics measurements of interest to the particle and nuclear physics communities in the era of a proton driver .
however , in principle , some of the measurements discussed at the workshop could be made sooner .
organizing to make these earlier measurements may focus a part of the research community in anticipation of a proton driver .
it would be easier then to do the experiments which depend on the flux available from a proton driver .
the neutrino working group included the following physics topics in its discussion of goals : * strange quark contribution and axial form factor through parity violation * duality * coherent pion production * exclusive resonant final states * parton distribution functions , particularly at high @xmath7 * * generalized parton distribution functions including quark flavor dependence through the weak analog of deeply - virtual compton scattering @xmath8 * * detailed study of nuclear effects with neutrinos that are expected to be significantly different than those measured with charged leptons * spin - dependent distribution measurements using polarized targets * a - dependence and more detailed nuclear effects in the case of the kaon and pion working group , the following were among the topics discussed : * production dynamics for exotic / rare quark and gluon states * rare pion decays , e.g. , improving @xmath9 from pion beta decay in flight * ckm parameters via @xmath10 * pushing further in @xmath11 in the search for new physics many of these measurements can be expected to help define the nature of the high mass objects which are anticipated to be found at the tevatron or the lhc .
the proton driver working group draft reports will go to a proton driver science advisory committee ( pdsac)@xcite , whose chair is peter meyers ( princeton ) .
the pdsac is composed of particle and nuclear physicists , and is international in composition .
the plan is to have the pdsac review conveners presentations and the reports in order to advise the laboratory on the strongest physics motivations for a proton driver .
a draft statement of the physics case ( a statement of need ) could be ready for reading over the winter holidays .
then , a final draft statement of the physics case could be prepared for submission to doe .
i have focused on a only part of the hadron physics in fermilab s future .
this part is dominantly proposal driven .
i have not said much about the collider or mipp , which are covered in separate talks at this first meeting of the aps topical group on hadron physics . in its vision of the future at fermilab , the long range plan says .... fermilab will remain the primary site for accelerator - based particle physics in the u.s . in the next decade and beyond .
.... defining the future of much of hadron physics at fermilab , e.g. , use of neutrinos , hadron beams , and even tevatron beams will depend on the users .
what will be the total available beam ?
will we build a proton driver ?
what fraction of the available protons will go to hadron physics ?
will a stretcher ring be proposed and built ?
a robust program depends on robust interest , and eventually , compelling physics - driven experiment proposals .
we may hope this meeting of the topical group on hadron physics , and related gatherings in the future , will contribute directly to that robust program .
much of the material on the proton driver is taken from the presentation by bob kephart on october 6 , 2004 , to the proton driver workshop at fermilab .
thanks also go to david christian , hugh montgomery , hogan nguyen , and jorge morfin for comments on a draft of this paper .
1 the membership of the proton driver science advisory committee is : * peter meyers , princeton university , usa , chair * ed blucher , university of chicago , usa * gerhard buchalla , ludwig - maximilians university , munich , germany * john dainton , university of liverpool , uk * yves declais , lapp - in2p3-cnrs annecy , france * lance dixon , slac , usa * umberto dosselli , infn and university of padua , italy * don geesaman , argonne national laboratory , usa * geoff greene , oak ridge national laboratory , usa * taka kondo , kek , japan * marvin marshak , university of minnesota , usa * bill molzon , university of california - irvine , usa * hitoshi murayama , university of california - berkeley , usa * jim siegrist , university of california - berkeley , usa * tony thomas , thomas jefferson national laboratory , usa * taku yamanaka , osaka university , japan | hadron physics in three accelerator at fermilab tevatron collider experiments neutrino program fixed target physics with beams main injector collider cdf dzero multi - year run btev take over neutrino program miniboone using beam from booster accelerator turning on numi beam main injector in weeks minos experiment with neutrino beams available minerva experiment approved to join mipp experiment e907 - main injector particle production finishing commissioning run about to physics data approved for 120 gev drell - yan experiment ( e906 ) starting build detector possibility returning for kaon physics using beam main injector retained fermilab accelerator experiment schedules through 2009 available on web http://www.fnal.gov/directorate/program_planning/schedule/index optimization of running conditions challenge with diverse program each fermilab preference accelerator complex particularly proton source tevatron collider benefits rapid cycling upstream accelerators with antiproton beam - cooling times accumulator recycler beam acceptable for losses neutrino program rapid cycle times beam for losses no lower limit proton source acceleratorsopposite extreme sy120/fixed target program depends on experiment test - beam user status during set up multiple short duration spills per minute optimal data taking long cycles with long spills best for maximizing integrated spill seconds minimizing end effects office of program planning experiments accelerator under guidelines from director guideline on use beam time for sy120 integrated effect running sy120 beam reduce antiproton stacking rate by more than 5% globally details scheduling worked between experimenters office program planning guideline flexible details spill length rate open to maximize physics sy120 program flexibility of mixing spill configurations in accelerator time line sequence adding flexibility complexity fermilab director established long range planning committee spring 2003 develop realistically achievable options for fermilab program next decade linear collider new intense proton source long baseline neutrino experiments new physics options fermilab one option lrpc working group bob kephart chairman option received recommendations recommendations adopted in final flrpc report by director fermilab linear accelerator booster old machines desire for more intense proton sources for long baseline neutrino physicsmeet new demands new proton source parameters for new facility fermilab are @xmath0 @xmath1 beam power 8 gev @xmath2 @xmath1 power 120 gev 6 times power current main injector two implementations 8 gev synchrotron and 8 gev superconducting rf linear accelerator web page established progress new proton source http://www-bd.fnal.gov/pdriver/ accelerator effort design linear accelerator same level as synchrotron final choice proposal appreciation for linear accelerator advantages better performance flexibility connection synergies with international linear collider cold '' technology chosen 8 gev injector beam main injector increase by factor 5 from @xmath3 to @xmath6 protons per cycle main injector beam power could increased shortening ramp time additional magnet power supplies proton driver more protons per cycle faster ramp times imply more rf power money required international technology recommendation panel opted for cold '' technology for international linear collider boost for super - conducting rf linear - accelerator - based proton driver at fermilab fermilab working on warm and cold linear - collider technologyfy 2005 effort dedicated to cold technology advancing linear - accelerator proton driver fy 2005 factor two increase in super - conducting r&d spending at fermilab relative fy 2004 plans forming for scrf module test facility meson east area at fermilab long lead - time items modulators ordered likely new intense proton source proposed for construction fermilab future project similar to main injector project cost schedule 8 gev synchrotron superconducting rf linear accelerator technically possible linear accelerator preferred if affordable fermilab management requested 8 gev linear accelerator design including cost schedule information fermilab s bill foster leading team develop technical design hard guess time scales technically limited schedule doe construction project critical decision cd ) schedule cd-0 statement of need submitted 2005 cd-1 preliminary acquisition strategy execution plan design report cost schedule management plan hazard analysis 2006 cd-2/3a project baseline approved approval start construction 2007 - 8 availability funding doe may push schedule later once funding approved typical projects main injector construction times 4 - 5 years time scale on international linear collider next proton driver as ilc engineering test facilityup to users make physics case proton driver required should be built soon possible physics case crucial workshop physics potential of proton driver held program links to talks at proton driver web page steve geer asked to organize documentation physics case strong physics case machine technical design submit proton driver project to doe for approval funding recent proton driver workshop step in process information on workshop talks web page : http://www-td.fnal.gov/projects/pd/physicsincludes/workshop/index.html three working groups interesting for hadron physics : wg2 on neutrino interactions jorge morfin rex tayloe ron ransome wg4 on kaons / pions wg5 on antiprotons dave christian mark mandelkern working group on neutrons workshop had additional working groups less connected to hadron physics working groups identified physics measurements of interest to particle nuclear physics communities in era proton driver measurements discussed could be made sooner organizing earlier measurements may focus research community in anticipation proton driver easier to do experiments on flux from proton driverneutrino working group included physics topics strange quark contribution axial form factor parity violation duality coherent pion production exclusive resonant final states parton distribution functions at high @xmath7 generalized parton distribution functions quark flavor dependence compton scattering detailed study nuclear effects with neutrinos different than charged leptons spin - dependent distribution measurements polarized targets a - dependence detailed nuclear effects kaon and pion working group discussed production dynamics for exotic rare quark gluon states rare pion decays @xmath9 pion beta decay flight ckm parameters via @xmath10 @xmath11 search new physics measurements define high mass objects at tevatron lhc proton driver working group draft reports to proton driver science advisory committee@xcite chair peter meyers princeton international presentations reports advise laboratory on physics motivations for proton driver draft statement physics case ready for reading over winter holidays final draft statement for submission to doe focused on part hadron physics in fermilab s future proposal drivennot said about collider mipp covered separate talks first meeting aps topical group hadron physics vision future fermilab long range plan primary site for accelerator - particle physics next decade defining future hadron physics use neutrinos hadron beams tevatron beams on users total available beam ? build proton driver ? protons to hadron physics ? stretcher ring proposed built ? robust program depends on interest compelling physics - driven experiment proposals hope meeting group hadron related gatherings contribute to robust program material proton driver from presentation bob kephart october 6 2004 proton driver workshop fermilab thanks to david christian hugh montgomery hogan nguyen jorge morfin for comments draft papermembership proton driver science committee peter meyers princeton university ed blucher university chicago gerhard buchalla ludwig - maximilians university munich john dainton university liverpool yves declais lance dixon slac umberto dosselli university padua italy don geesaman argonne national laboratory geoff greene oak ridge laboratory taka kondo japan marvin marshak university minnesota bill molzon university california - irvine hitoshi murayama berkeley jim siegrist tony thomas thomas jefferson laboratory taku yamanaka osaka university japan | 0.507568 | 3,069 | 1,508 | , Saving $0.1 in GPT-4. |
55,584 | we measure a value for the cosmic expansion of @xmath0(stat ) @xmath1 44(syst ) km s@xmath2 mpc@xmath2 at a redshift of @xmath3 based on the differential age technique .
this technique , also known as cosmic chronometers , uses the age difference between two redshifts for a passively evolving population of galaxies to calculate the expansion rate of the universe .
our measurement is based on analysis of high quality spectra of luminous red galaxies ( lrgs ) obtained with the southern african large telescope ( salt ) in two narrow redshift ranges of @xmath4 and @xmath5 as part of an initial pilot study .
ages were estimated by fitting single stellar population models to the observed spectra .
this measurement presents one of the best estimates of @xmath6 via this method at @xmath7 to date .
[ firstpage ] galaxies : evolution cosmology : cosmological parameters cosmology : observations | in recent times , numerous observations and methods have been used and proposed for constraining cosmological parameters . these include observations of the cosmic microwave background ( cmb ) , from the _ wmap _ team @xcite and _
planck _ team @xcite , baryonic acoustic oscillations ( bao ) peaks @xcite , type ia supernovae ( sne , * ? ? ?
* ; * ? ? ?
* ) , gravitational lensing @xcite and the abundance of clusters of galaxies @xcite .
many of these methods use quantities which require the hubble parameter to be integrated along the line of sight ( e.g. the luminosity distance in supernovae observations ) and thus probe average expansion over a long period . a complementary technique known as _ cosmic chronometers _ ( cc ) , originally proposed by @xcite , uses passively evolving early type galaxies to measure the hubble parameter at a specific redshift , rather than a quantity which gives an integrated measurement of @xmath6 out to redshift @xmath8 .
this technique is independent of the cosmological model and allows a determination of the expansion rate at a given redshift without relying on the nature of the metric between the chronometers and observers , and therefore can provide tighter constraints on cosmological parameters of the model .
the expansion rate @xmath6 at redshifts @xmath9 0 can be obtained by @xmath10 in the cc method , @xmath11 is approximated by determining the time interval @xmath12 corresponding to a given @xmath13 , where @xmath13 is centered at redshift @xmath8 . if one assumes that most stars in luminous red galaxies ( lrgs ) formed near the beginning of the universe at a similar time ( as supported by observations cited in @xcite ) , then measuring the age difference between ensembles of lrgs at two different redshifts provides the differential quantity @xmath14 required to estimate @xmath6 .
the biggest advantage of this method is the differential approach where the relative ages @xmath12 are used instead of the absolute ages .
the relatives ages diminish the systematic effects underlying the absolute age estimates .
furthermore , the use of the small redshift slices @xmath13 , which corresponds to a short time evolution of these massive and passive galaxies , controls the effect of some observation biases such as the progenitor bias .
more discussions about avoiding this effect can be found in @xcite .
the cc method has been used to measure the hubble parameter up to redshift @xmath15 and these results have been used to constrain the nature of dark energy and to recover the local hubble constant @xcite . in each measurement
, the authors assumed that lrgs are massive , passively - evolving elliptical galaxies that are homogeneous populations forming their stars at high redshift , and they fit single
burst equivalent ages to the galaxies .
these galaxies represent the ideal population to trace the differential age evolution of the universe . in many of the papers ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the authors have attempted to improve this method by pointing out the need for better fitting of stellar population models , better selection of targets , larger samples , and better quality data to precisely determine @xmath6 .
the age dating of galaxies is an important topic in the study of the galaxy evolution . during the last few decades
, it has been widely exploited by the creation of synthetic stellar population tools .
there are number of stellar population synthesis codes available ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) which can be used to generate synthetic spectra to determine different parameters such as age , metallicity , chemical abundance and star formation history of a galaxy .
typically two approaches are used for fitting models to data : ( i ) full spectrum fitting ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and ( ii ) techniques which fit features in the spectra such as lick index fitting ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we explored one of these techniques which is the full spectrum fitting , as further outlined in section [ section : analysis ] .
@xcite found that a measurement of @xmath6 within a 3% accuracy would be viable from a large redshift program targeting lrgs , and explored optimal observational set
ups for the experiment using the southern african large telescope ( salt ) .
the potential to obtain wide optical wavelength coverage and high resolution spectra from a 10meter class telescope like salt provides an excellent opportunity to explore the potential of the cc method . in this paper
, we describe the results of several observations which use the robert stobie spectrograph ( rss ) on salt to obtain 16 spectra of massive lrgs at @xmath4 and @xmath5 .
we age date the lrgs and hence determine @xmath16 ) .
the paper is organized as follows : we first describe the lrg sample selection from the 2df sdss lrgs catalogue ( section [ section : selectionandreduction ] ) .
we then describe the data reduction and some details about the method adopted to estimate the ages of these galaxies ( section [ section : selectionandreduction ] and [ section : analysis ] ) . in section [ section : hz ] we discuss the derived ages and estimate @xmath17 .
our targets are lrgs from 2df sdss lrg and qso ( 2slaq ) catalogue . ]
@xcite where their spectroscopic redshifts are already available , and from megaz
lrg photometric redshift catalogue @xcite
. those galaxies must be visible with salt , massive , without emission lines , the brightest and the reddest in the catalogue .
detailed description on each criterion is described below .
many galaxies meet the selection criteria , but for a pilot survey , only few galaxies including those 16 galaxies listed in table [ tab : tabchar ] were aligned for the observations .
we estimated the stellar masses of lrgs using the method outlined below , requiring potential targets to have stellar masses larger than @xmath18 m@xmath19 .
we also required targets to have redshifts within 0.01 of @xmath20 or @xmath21 , and to have magnitude @xmath22 .
we derived the stellar masses of lrgs in 2slaq by first cross matching the catalogue with the _ _ wise _ _ data and detected 13518 sources in at least two of the four mid infrared bands centered at 3.4 @xmath23 m ( w1 ) and 4.6 @xmath23 m ( w2 ) .
the details on cross matching the catalogue are provided in appendix [ appendix : sedfitting ] .
the use of the _ wise _ data helped us to achieve our goal of selecting most massive lrgs which have typically formed at high redshifts .
moreover , the combination of sdss and _ wise _ photometry has been investigated to produce a new technique to target lrgs at high redshift ( @xmath9 0.6 ) .
we then combined the optical and mid infrared flux densities to build the spectral energy distribution ( sed ) of these detected sources .
next , we performed sed fitting with cigale @xcite .
details of the mechanisms of cigale can be found in @xcite and @xcite .
we adopted basic input parameters as suggested by the code developer ( giovannoli , private communication ) to extract the derived stellar mass of each galaxy given in table [ tab : tabchar ] .
appendix [ appendix : sedfitting ] provides the details on how we processed the sed fitting by using our combined @xmath24 photometry plus _ wise _ 3.4 @xmath23 m and 4.6 @xmath23 m fluxes .
the sed fitting was used to obtain only stellar masses of our galaxies , whereas the stellar population fits to the salt spectra using full spectrum fitting over our entire wavelength range were used to derive the galaxy ages ( see section [ section : analysis ] ) . and those derived masses were then used to select our sample .
we further constrained the selection criteria by requiring that targets should show no optical emission lines by visual inspection .
we also included sdss j013403.82 + 004358.8 in our sample .
this target was selected from the megaz
lrg photometric redshift catalogue @xcite , and it has been spectroscopically confirmed as a central cluster galaxy of a very rich cluster @xcite .
the mass of this object was derived by fitting only its model @xmath25 fluxes .
the spectroscopic observations of this sample were carried out with the robert stobie spectrograph ( rss , * ? ? ?
* ; * ? ? ?
* ) at salt .
long slit optical spectra of the sample were obtained between december 2011 and september 2012 under proposal codes 2011 - 3-rsa_oth-026 and 2012 - 1-rsa_oth-013 ( pi : a. ratsimbazafy ) .
the observation was made using the pg0900 grating at two different grating angles to cover the @xmath26 4000 6000 range ( rest frame wavelength range ) at both redshifts with a spectral resolution of @xmath26 46 .
the log of observations is given in table [ tab : obslog ] .
a slit width of 1 `` , yielding a resolution of @xmath27 , was used during the 2011 observations ; this was increased to a slit width of 1.5 '' that yielded a resolution of @xmath28 during the second semester of 2012 observations . for wavelength calibration , spectra of a neon arc lamp
were taken after each observation , as well as flat
field images in order to perform a standard reduction of two dimensional long slit spectra .
the spectrophotometric standard stars were observed for flux calibration .
for all observations , fidelity checking and basic calibrations such as the overscan , gain , cross talk corrections and mosaicking were already performed by the automated salt pyraf pipeline called pysalt @xcite .
further reduction was performed by following the standard long slit data reduction technique with iraf .
cosmic rays were removed from the two dimensional spectrum using the lacosmic ( laplacian cosmic ray identification ) software @xcite .
the two dimensional science , flat field and arc images were trimmed to exclude nonuseful areas before performing the rest of the reduction .
the two ccd gaps were filled with interpolated pixel values .
all the spectra were flat fielded by using calibrating flat
fields obtained for the science frames .
this process helps to correct for the differences in sensitivity across the field and between detector pixels .
the flat fields were also used to remove fringing at the far red portion of the ccd chips . the long package from iraf
was then used for the wavelength calibration , extracting the spectrum , and subtracting the night sky spectrum .
spectrophotometric standard stars were reduced according to the same procedure , and a relative flux calibration was applied to the science spectrum to recover the spectral shape . due to the nature of the salt telescope , the unfilled entrance pupil of the telescope moves during the observation and an absolute flux calibration
is not possible .
we combined the individual frames to create a master science image as well as an error image .
this procedure was done before extracting the one dimensional spectra .
all spectra were corrected for foreground galactic extinction using the reddening maps of @xcite and the extinction curve of the @xcite with @xmath29 . the galactic @xmath30 parameter for each galaxy
is given in table [ tab : tabchar ] . | recent observations methods used for constraining cosmological parameters include observations cosmic microwave background wmap planck baryonic acoustic oscillations peaks type ia supernovae gravitational lensing abundance of clusters of galaxies methods use quantities hubble parameter integrated along line sight probe average expansion over long period complementary technique cosmic chronometers _ proposed by @xcite uses passively evolving early type galaxies to measure hubble parameter at specific redshift technique independent of cosmological model allows determination expansion rate at redshift without relying on nature metric provide tighter constraints on cosmological parameters model expansion rate @xmath6 at redshifts @xmath9 obtained by @xmath10 cc method @xmath11 approximated by determining time interval @xmath12 corresponding to @xmath13 centered at redshift @xmath8 most stars in luminous red galaxies formed near beginning universe similar time supported observations measuring age difference between ensembles lrgs at two redshifts provides differential quantity @xmath14 required to estimate @xmath6advantage method is differential approach relative ages @xmath12 used instead of absolute ages relatives ages diminish systematic effects absolute age estimates use small redshift slices @xmath13 short time evolution massive passive galaxies controls observation biases progenitor bias discussions about avoiding effect in @xcite cc method used hubble parameter up to redshift @xmath15 results constrain dark energy recover local hubble constant @xcite authors assumed lrgs are massive passively - evolving elliptical galaxies homogeneous populations forming at high redshift fit single burst equivalent ages to galaxies galaxies represent ideal population trace differential age evolution universe authors attempted to improve method for better fitting stellar population models selection of targets larger samples better quality data determine @xmath6 age dating of galaxies important topic in galaxy evolution exploited by synthetic stellar population tools stellar population synthesis codes available* ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? ; * ? ? ? ; * ? ? ? ; * ? ? ? ; * ? ? ? ; * ? ? ? * ; * ? ? ? * ; * ? ? ? ; * ? ? ? ; * ? ? ? ; * ? ? ? ; * ? ? ? * ? ? ? ? ? ? generate synthetic spectra age metallicity chemical abundance star formation history galaxy two approaches fitting models full spectrum fitting ? ? techniques lick index fitting * ? ? ? ? ? ? explored full spectrum fitting section analysis @xcite found measurement of @xmath6 3% accuracy viable from large redshift program targeting lrgs explored observational set ups southern african large telescope ( salt ) potential wide optical wavelength coverage high resolution spectra from 10meter class telescope salt cc method results observations robert stobie spectrograph on salt 16 spectra of massive lrgs at @xmath4 and @xmath5 age date lrgs determine @xmath16paper organized describe lrg sample selection from 2df sdss lrgs catalogue selectionandreduction data reduction method estimate ages analysis hz discuss derived ages estimate @xmath17 targets are lrgs from 2df sdss lrg qso 2slaq catalogue @xcite megaz lrg photometric redshift catalogue galaxies visible with salt massive without emission lines brightest reddest many galaxies meet selection criteria pilot survey few galaxies aligned for observations estimated stellar masses of lrgs using method targets stellar masses larger than @xmath18 m@xmath19 required targets redshifts within 0.01 of @xmath20 or @xmath21 magnitude @xmath22 derived stellar masses lrgs in 2slaq by cross matching catalogue with _ _ wise _ data detected 13518 sources in two four mid infrared bands at 3.4 @xmath23 m and 4.6 @xmath23 m w2 ) details cross matching catalogue in appendix sedfitting _ wise _ data goal selecting massive lrgs formed at high redshiftscombination sdss wise photometry investigated new technique target lrgs at high redshift ( @xmath9 0.6 ) combined optical mid infrared flux densities spectral energy distribution sed ) of sources performed sed fitting with cigale @xcite mechanisms in @xcite adopted basic input parameters code developer extract derived stellar mass each galaxy table appendix sedfitting sed fitting @xmath24 photometry plus _ wise _ 3.4 @xmath23 m 4.6 @xmath23 m fluxes sed fitting stellar masses stellar population salt spectra full spectrum fitting derive galaxy ages derived masses select sample selection criteria targets show no optical emission lines included sdss j013403.82 + 004358.8 in sample target selected from megaz lrg photometric redshift catalogue @xcite confirmed as central cluster galaxy of rich cluster @xcite mass derived by fitting only model @xmath25 fluxes spectroscopic observations with robert stobie spectrograph at salt long optical spectra obtained between december 2011 september 2012 under proposal codes 2011 - 3-rsa_oth-026 and 2012 - 1-rsa_oth-013ratsimbazafy observation using pg0900 grating two angles @xmath26 4000 6000 range redshifts spectral resolution @xmath26 46 log observations in table obslog slit width 1 `` resolution @xmath27 used 2011 observations increased to 1.5 '' resolution @xmath28 second semester 2012 wavelength calibration spectra neon arc lamp taken flat field images standard reduction two dimensional long slit spectra spectrophotometric standard stars observed for flux calibration fidelity checking basic calibrations overscan gain cross talk corrections mosaicking performed by automated pyraf pipeline pysalt @xcite reduction standard long slit data reduction technique with iraf cosmic rays removed from spectrum using lacosmic software @xcite science arc images trimmed nonuseful before reduction two ccd gaps filled with interpolated pixel values spectra flat fielded calibrating flat fields science frames differences detector pixels flat fields fringing at far red portion ccd chips long package from iraf used for wavelength calibration extracting spectrum subtracting night sky spectrumspectrophotometric standard stars reduced relative flux calibration applied science spectrum recover spectral shape salt telescope unfilled entrance pupil moves absolute flux calibration possible combined frames master science image error image procedure before extracting one dimensional spectra spectra corrected for galactic extinction using reddening maps @xcite extinction curve @xmath29 galactic @xmath30 parameter each galaxy given in table | 0.435423 | 2,714 | 1,339 | , Saving $0.1 in GPT-4. |
55,588 | we demonstrate the efficiency of high quality optical spectroscopic observations of two compact radio sources , pks 1549 - 79 and pks 1345 + 12 , as a probe of the kinematics and physical conditions in the circumnuclear gas in the early stages of radio source evolution .
we outline a schematic model for pks 1345 + 12 based on the model for pks 1549 - 79 proposed by tadhunter et al .
( 2001 ) in which the young radio source is expanding through a dense and dusty enshrouding cocoon , sweeping material out of the circumnuclear regions .
pks 1345 + 12 ; pks 1549 - 79 ; ism kinematics & physical conditions . | gigahertz - peaked spectrum radio sources ( gps : d @xmath0 1 kpc ) and the larger compact steep spectrum radio sources ( css : d @xmath0 15 kpc ) account for a significant fraction of the radio source population ( @xmath1 40% ) though their nature is not fully understood ( see odea 1998 and references therein ) .
currently , we believe they are _ young _ radio sources ( fanti et al . 1995 ) supported by estimates of dynamical ages : t@xmath2 @xmath1 10@xmath3 - 10@xmath4 years ( owsianik et al . 1998 ) ; and radio spectral ages : t@xmath5 @xmath0 10@xmath6 years ( murgia et al .
this is in preference to the _ frustration scenario _ where the ism is so dense , the radio jets can not escape and the radio source remains confined and frustrated for its entire lifetime ( van breugel 1984 ) .
if compact radio sources are young radio sources , we will observe them relatively recently after the event(s ) which triggered the activity ( e.g. a merger ; heckman et al .
indeed , many compact radio sources exhibit features attributed to mergers such as double nuclei , tidal tails , arcs of emission and distorted isophotes ( e.g. stanghellini et al . 1993 ) .
hence , the host galaxy will still retain the dense ism injected by the merger .
the young radio jets will also be relatively small and so will readily interact with the dense ism . in these proceedings
we demonstrate the efficiency of high quality optical spectroscopic observations as a probe of the kinematics and physical conditions of the ism in the host galaxies of radio sources , particularly in the early stages of radio source evolution .
we discuss observations of two compact radio sources , the flat spectrum radio source pks 1549 - 79 and the gps source pks 1345 + 12 , and outline a schematic model for pks 1345 + 12 .
for detailed discussions see tadhunter et al .
( 2001 ) and holt et al .
( 2002 , 2003 ) .
from low resolution spectra of the compact flat - spectrum radio source pks 1549 - 79 , tadhunter et al .
( 2001 ) report : * highly broadened forbidden emission lines ( fwhm @xmath1 1350 km s@xmath7 ) .
* 2 distinct redshift systems : the high ionisation forbidden lines ( e.g. [ o iii ] ) are blueshifted by @xmath1 600 km s@xmath7 with respect to the low ionisation lines ( e.g. [ o ii ] ) and hi 21 cm absorption .
tadhunter et al .
( 2001 ) interpret the blueshifted emission lines as material in outflow . due to the small scale radio source ( @xmath1 300 pc ) , the radio morphology ( core - jet ) and the flat radio spectrum , they proposed a schematic model in which the young radio source is enshrouded by a cocoon of dense gas and dust completely obscuring the central qso nucleus .
the young radio jets are expanding and sweeping material out of the nuclear regions with the direction of propagation close to the observer s line - of - sight ( see fig 1 ) .
we present here a summary of results from intermediate resolution ( @xmath1 4 ) spectra over a wide spectral range ( @xmath1 4500 ) from isis on the 4.2 m wht on la palma . for a detailed discussion see holt et al .
( 2003 ) . from modelling of
the highly extended [ o ii]@xmath83726,3729 emission line ( @xmath1 20 kpc ) using between 1 and 3 gaussian components , we believe that the narrowest of these represents the galaxy rest frame : * it is the only component observed across the entire spatial range ; * it is consistent with the deep hi absorption ( mirabel 1989 ; morganti et al .
2002 ) ( see fig 2 ) and the stellar absorption lines ( grandi 1977 ) ; * it is the least kinematically disturbed component ( fwhm @xmath1 340 km s@xmath7 ) ; * it has a small velocity amplitude ( @xmath9 @xmath0 250 km s@xmath7 ; see fig 2 ) consistent with gravitational motions ( tadhunter et al .
1989 ) ; * it is the only component consistently observed in all emission lines , irrespective of the model required to reproduce the emission line profile . in the nuclear aperture ,
the emission lines are broad with strong blue asymmetries .
it is essential to model all emission lines with 3 gaussian components ( see fig 2 , also holt et al .
the narrowest of these is consistent with the narrow component of [ o ii ] and the rest frame of the galaxy .
the intermediate and broad components are blueshifted by up to @xmath1 2000 km s@xmath7 with respect to this narrow component . due to large reddening in the nucleus ( see below ) , we interpret this as material in outflow .
however , one model does not reproduce all emission lines different velocity widths and shifts for the broad and intermediate components are required for [ o i]@xmath86300,6363 and [ s ii]@xmath86716,6731 ( see holt et al .
2003 ) .
we estimated the density in the nucleus using the density diagnostic , the [ s ii]@xmath86716,6731 doublet .
the intermediate and broad components have densities consistent with the high density limit . by varying the fit parameters
, we obtained lower limits on on the density of n@xmath10 @xmath11 5300 @xmath12 and n@xmath10 @xmath11 4200 @xmath12 for the intermediate and broad components respectively .
the narrow component is consistent with the low density limit ( n@xmath10 @xmath0 150 @xmath12 ) .
we have investigated reddening in the nucleus of pks 1345 + 12 using three independent techniques , each assuming a simple foreground screen for interstellar extinction and the seaton ( 1979 ) extinction law . *
_ balmer line ratios_. the h@xmath13/[n ii ] blend is highly complex .
we measure h@xmath13/h@xmath14 ratios of 3.32 @xmath15 0.33 , 5.25 @xmath15 0.28 and 18.81 @xmath15 4.74 corresponding to e(b - v ) values of 0.06 @xmath15 0.05 , 0.42 @xmath15 0.10 and 1.44 @xmath15 0.50 for the narrow , intermediate and broad components respectively . by varying the model fitting parameters
, we estimate a lower limit on the reddening in the broadest component of e(b - v ) @xmath11 0.92 . * _ comparison with the infra - red_. veilleux et al .
( 1997 ) present measurements of pa@xmath13 .
the broadest component is consistent in velocity width and shift with the broadest component in [ o iii ] . by varying the quoted flux of pa@xmath13 by a factor of 2 , to account for possible slit loss differences between the observations
, pa@xmath13/h@xmath14 gives a range of e(b - v ) values : 1.60 @xmath0 e(b - v ) @xmath0 2.00 .
this is consistent , within the errors , with the measured value for h@xmath13/h@xmath14 .
* _ the nebular continuum_. before modelling the faint lines in the nucleus , we subtracted both the nebular ( see dickson et al .
1995 ) and stellar continuum .
if zero reddening is assumed for all components , a large discontinuity , corresponding to the balmer edge at @xmath1 3645 , remains when subtracting the nebular continuum from the spectrum ( see fig 7 in holt et al .
2003 ) . however , if the measured e(b - v ) values are assumed , the discontinuity disappears .
the strength of this discontinuity is particularly sensitive to the amount of flux in the intermediate component and we estimate a lower limit of e(b - v ) @xmath11 0.3 for the intermediate component .
the nucleus of pks 1345 + 12 is highly reddened , with reddening increasing with broadness of component .
we estimate an upper limit on the mass of line emitting gas of @xmath0 10@xmath16 m@xmath17 ( from the density and reddening - corrected h@xmath14 luminosity ) which is consistent with the radio source being young .
combining these results , we have developed the model from tadhunter et al .
fig 3 shows the schematic model which retains the small scale radio jets expanding out through an enshrouding cocoon giving rise to emission line outflows .
new features include stratification of the ism - all emission lines in the nucleus require 3 gaussian components though the velocity width and shift of the intermediate and broad components varies between emission lines .
however , gradients ( e.g. density , velocity , ionisation potential ) must exist across the regions emitting the intermediate and broad components . the narrow component remains consistent in all emission lines and represents the relatively quiescent halo .
` strata ' positioning is determined by the reddening - the reddening increases with broadness of component and so the broadest component must originate from the region closest to the nucleus .
jh acknowledges a pparc phd studentship and a marie curie fellowship .
the william herschel telescope is operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofisica de canarias .
999 r. d. dickson , c. n. tadhunter , m. shaw , n. clark , r. morganti , 1995 , mnras , 273 , 29 . c. fanti , r. fanti , r. t. schilizzi , r. e. spencer , c. stanghellini , 1995 , a&a , 302 , 317 . s. a. grandi , 1977 , aj , 215 , 446 . t. m. heckman , e. p. smith , s. a. baum , w. j. m. van breugel , g. k. miley , g. d. illingworth , g. d. bothun , b. balick , 1986 , apj , 311 , 526 .
j. holt , c. n. tadhunter , r. morganti , 2002 , pasa , proc .
3rd gps & css workshop , kerastari , greece , astro - ph/0210092 .
j. holt , c. n. tadhunter , r. morganti , 2003 , mnras , submitted .
i. f. mirabel , 1989 , apj , 340 , l13 .
r. morganti , c. n. tadhunter , t. a. oosterloo , j. holt , a. tzioumis , k. wills , 2002 , pasa , proc .
3rd gps & css workshop , kerastari , greece , astro - ph/0212321 .
m. murgia , c. fanti , r. fanti , l. gregorini , u. klein , k .- h .
mack , m. vigotti , 1999 , a&a , 345 , 769 . c. p. odea , 1998 , pasp , 110 , 493 .
i. owsianik , j. e. conway , a. g. polatidis , 1998 , a&a , 336 , l37 .
c. stanghellini , c. p. odea , s. a. baum , e. laurikainen , 1993 , apjs , 388,1 . c. n. tadhuhnter , r. a. e. fosbury , p. j. quinn , 1989 , mnras , 24 , 225 . c. n. tadhunter , k. wills , r. morganti , t. oosterloo , r. dickson , 2001 , mnras , 327 , 227 .
w. j. m van breugel , 1984 , ius symp .
, 110 , 59 . m. villar - martn , c. n. tadhunter , r. morganti , r. a. e. fosbury , 1999 , a&a , 336 , 7 . | gigahertz - peaked spectrum gps @xmath0 1 kpc larger compact steep spectrum css : d @xmath0 15 kpc fraction radio source population @xmath1 40% ) nature not understood odea 1998 believe young radio sources fanti et 1995 supported dynamical ages t@xmath2 @xmath1 10@xmath3 - 10@xmath4 years owsianik 1998 radio spectral ages t@xmath5 @xmath0 10@xmath6 years preference frustration scenario ism dense radio jets escape remains confined frustrated lifetime van breugel 1984 compact sources young recently after merger exhibit features mergers double nuclei tidal tails arcs of emission distorted isophotes 1993 host galaxy dense ism merger young radio jets small interact with dense ism efficiency of high quality optical spectroscopic observations kinematics physical conditions ism in host galaxies radio early stages radio source evolution discuss observations two compact radio sources flat spectrum pks 1549 - 79 gps source pks 1345 + 12 outline schematic model for pks 1345 + 12 detailed see tadhunter et al .2001 holt et al 2002 2003 low resolution spectra compact flat - spectrum radio source pks 1549 - 79 tadhunter report broadened forbidden emission lines @xmath1 1350 km s@xmath7 2 redshift systems high ionisation forbidden lines blueshifted @xmath1 600 km s@xmath7 low ionisation 21 cm absorption interpret blueshifted emission lines material outflow small scale radio source @xmath1 300 pc flat radio spectrum schematic model young radio source enshrouded dense gas dust obscuring central qso nucleus radio jets expanding sweeping material nuclear regions close observer s line sight fig 1 summary results intermediate resolution @xmath1 4 spectra wide spectral range @xmath1 4500 ) isis 4.2 m wht la palma discussion holt et al . 2003 modelling extended ii]@xmath83726,3729 emission line @xmath1 20 kpc 1 3 gaussian components narrowest represents galaxy rest frame only component observed spatial range consistent deep hi absorption mirabel 19892002 fig 2 stellar absorption lines grandi 1977 least kinematically disturbed component @xmath1 340 km s@xmath7 small velocity amplitude @xmath9 @xmath0 250 km s@xmath7 consistent with gravitational motions tadhunter 1989 only component observed in emission lines nuclear aperture emission lines broad strong blue asymmetries essential model lines 3 gaussian components fig 2 narrowest consistent with narrow component [ o ii ] rest frame galaxy intermediate broad components blueshifted by @xmath1 2000 km s@xmath7 large reddening nucleus material in outflow model reproduce different velocity widths shifts broad intermediate components required o i]@xmath86300,6363]@xmath86716,6731 holt 2003 estimated density nucleus density diagnostic@xmath86716,6731 doublet intermediate broad components consistent high density limit varying fit parameters obtained lower limits density n@xmath10 @xmath11 5300 @xmath12 4200 @xmath12 intermediate broad components narrow component consistent with low density limit n@xmath10 @xmath0 150 @xmath12investigated reddening nucleus pks 1345 + 12 three techniques assuming foreground screen interstellar extinction seaton 1979 extinction law balmer line ratios h@xmath13/[n ii ] blend complex h@xmath13/h@xmath14 ratios 3.32 0.33 5.25 0.28 18.81 4.74 e(b - v ) values 0.06 0.05 0.42 0.10 1.44 0.50 narrow intermediate broad components varying model fitting parameters estimate lower limit reddening broadest component e(b - v ) @xmath11 0.92 comparison infra - red veilleux et al 1997 measurements pa@xmath13 broadest component consistent velocity width shift [ o iii ] varying flux@xmath13 factor 2@xmath13/h@xmath14 range e(b - v ) values 1.60 2.00 consistent measured value h@xmath13/h@xmath14 nebular continuum modelling subtracted nebular stellar continuum zero reddening assumed all components large discontinuity balmer edge @xmath1 3645 remains subtracting nebular continuum fig 7 2003measured e(b - v ) values assumed discontinuity disappears strength sensitive to flux intermediate component estimate lower limit e(b - v ) @xmath11 0.3 nucleus pks 1345 + 12 highly reddened reddening increasing with broadness upper limit mass line emitting gas @xmath0 10@xmath16 m@xmath17 density luminosity consistent with radio source young developed model from tadhunter et al 3 shows model small scale radio jets expanding emission line outflows new features include stratification ism emission lines require 3 gaussian components velocity width shift varies gradients density velocity ionisation potential must exist across regions narrow component consistent quiescent halo strata positioning determined by reddening increases with broadness broadest component region closest to nucleus acknowledges pparc phd studentship marie curie fellowship william herschel telescope operated island of la palma isaac newton group spanish observatorio del roque de muchachos instituto de astrofisica canarias dickson tadhunter shaw clark morganti 1995 mnras , 273 , 29fanti schilizzi spencer stanghellini 1995 a&a 302 , 317 grandi 1977 215 , 446 heckman smith s. baum van breugel miley illingworth bothun balick 1986 apj 311 , 526 j. tadhunter. morganti 2002 pasa 3rd gps css workshop kerastari greece astro - ph/0210092 j. holt tadhunter. morganti 2003 mnras submitted f. mirabel 1989 apj 340 , l13 morganti tadhunter. oosterloo j. holt tzioumis k. wills 2002 pasa 3rd gps css workshop kerastari greece astro - ph/0212321. murgia fanti fanti gregorini klein mack vigotti 1999 a&a 345 , 769 p. odea 1998 110 , 493 owsianik j e. conway. polatidis 1998 a&a 336 , l37. stanghellini. odea s. baum laurikainen 1993 apjs, 388,1 tadhuhnter fosbury quinn 1989 mnras 24 , 225 tadhunter wills. morganti oosterloo dickson 2001 mnras 327 , 227 van breugel 1984 symp 110 , 59 villar - martn tadhunter. morganti fosbury 1999 a&a 336 , 7 | 0.433673 | 2,850 | 1,535 | , Saving $0.1 in GPT-4. |
55,596 | we propose the first continuous model with long range screening ( shadowing ) that described columnar growth in one space dimension , as observed in plasma sputter deposition .
it is based on a new continuous partial derivative equation with non - linear diffusion and where the shadowing effects apply on all the different processes .
plasma sputtering is a common process for film growth which often exhibits wide columns more or less close packed separated by thin deep grooves @xcite .
this columnar growth mainly results from a shadowing instability@xcite , where the elevated parts of the surface are more exposed to the sputtering while they shadow the incoming particles to the lower parts .
the modelization of this shadowing instability has been well described by probabilistic monte - carlo methods ( mc)@xcite and also with continuous models based on partial derivative equations ( pde)@xcite including the seminal work of bales and zangwill @xcite .
however , both approach fail to describe at long times the strongly nonlinear columnar microstructures observed recently ( see @xcite for instance ) . in fact , although the continuous models gives tall and well separated columns at early time , only few sharp peaks remain later on@xcite .
columnar structure using pde has already been obtained by gillet et al.@xcite but in that case no shadowing effect was taken into account ! on the other hand , discrete approaches using mc methods including shadowing have been developped and showed a fair description of the columnar structure , particularly through the formation of sharp column sides . however
, these models can not avoid the coarsening of the columnar structures showing larger and larger plateau as time increases , in contrast with experimental observations .
the goal of this paper is to present a new continuous non - local model which includes both non - linear shadowing and diffusion effects to simulate columnar - like growth .
we consider a two dimensional model where the one dimensional ( 1d ) surface described by @xmath0 is subjected to receive particles from all directions not shadowed by the surface itself .
our starting point is deduced from the models developped initially by bales and zangwill@xcite and by karunasiri et al @xcite : @xmath1 where the deterministic deposition term @xmath2 is multiplied by the solid angle @xmath3 which modelizes the shadowing effect as a long range screening ( see figure [ mc2 - 1 ] ) .
@xmath4 is the diffusion / relaxation coefficient while @xmath5 is the usual noise with zero mean @xmath6 and its correlation given by @xmath7 . + for small surface angles , we retrieve a kpz - like equation @xcite with shadowing effects ( defining @xmath8 ) : @xmath9 _ shadowing process : interface @xmath0 and solid angle @xmath3 .
_ ] a complete study of these equations has shown that it is unable to reproduce columnar shapes corresponding to experiments and mc simulations @xcite . indeed , in the most favourable situation ,
only broad peaks emerge instead of flat columns .
experiments suggest thus that the diffusion should be enhanced in the region more exposed to the flux .
moreover , we will assume that the flux also increases ( greater than for normal shadowing ) on the top of the columns compared to the grooves .
although we have no strong argument for it , we expect some point effect near the sharp edge to be responsible of this process .
we then propose the following stochastic differential equation where the main ingredients are non - linear shadowing effects and diffusion : @xmath10 in this equation @xmath11 is a given function of the solid angle @xmath12 .
therefore , in order to increase the shadowing effect and the diffusion from top to edges and bottom , @xmath11 has to be stronger than linear , and we will take later on for the numerics @xmath13 . the fact that this function , which modelizes the shadowing , is in factor to the right hand term will obviously increase the deposit rate for surfaces which are not shadowed ( mainly for large value of @xmath14 ) and make it smaller for shadowed one ( for small value of @xmath14 ) .
the diffusion is also affected by the shadowing in the same way . + a plane - wave analysis performed on equation ( [ eq2.0 ] ) shows that the solutions are unstable for large enough wavelengths @xmath15 , i.e. @xmath16 , with @xmath17 and the growth rate @xmath18 .
then , starting from a flat substrate , the noise will trigger the instability and will drive the system into a strong non - linear regime .
figure [ shadow1 ] shows the evolution of the interface profile for different times for @xmath19 , @xmath20 and @xmath21 .
it exhibits the desired columnar shape .
this shape is characterised by flat column tops and vertical sides as compared to previous monte - carlo simulations and columnar growth experiments .
the shadowed deposition favors the columnar growth and the anisotropic diffusion smoothes the top of the column very efficiently and leads to vertical sides .
the competition between these two effects leads to a columnar regime as expected .
moreover , most of the columns formed at the beginning of the simulation are still present at the end .
this is also the case for `` poisson / wedding cakes '' morphologies for which initial columns always remains @xcite and for step meander process @xcite .
+ for numerical simulation , equation ( [ eq2.0 ] ) is integrated using the following explicit scheme : @xmath22 \label{eq3.0}\end{aligned}\ ] ] with the notation @xmath23 .
@xmath24 is a random number picked with the uniform distribution between @xmath25 . to obtain a discrete form of the gradient term , we follow the scheme proposed by lim _ et al . _
although their study strictly applied for the kpz equation @xcite .
@xmath3 is evaluated following reference @xcite .
the time evolution of the roughness @xmath26 of the interface is given figure [ shadow3 ] .
it shows the existence of different regimes . the first one , for @xmath27 is driven by the fluctuations and @xmath26 scales as @xmath28 .
for the second one ( @xmath29 ) , diffusion induced a relative reduction of the roughness which scales as @xmath30 .
then , because of the shadowing instability described above , sharp canyons appear and the roughness quickly increases . finally , after @xmath31 , the columnar regime appears which leads to @xmath32 as in the discrete model @xcite .
even if @xmath33 shows the same scaling as obtained in previous studies on continuous columnar growth model @xcite , the column shapes are rather different and are now in closer agreement with the mc models and more important with the experiments @xcite .
indeed , experiments display relative constant column width , while mc simulations lead to a column width which is increasing with time . in that respect ,
our model ( [ eq2.0 ] ) exhibits a better qualitative agreement with experiement than mc simulations .
we have presented the first continuous model , to our knowledge , that exhibits a columnar growth without the coarsening dynamics of the structures , in good agreement with experimental observations on sputtering deposition . for reproducing these wide flat columns with sharp edges
, we have introduced an increase of the relaxation and of the flux on the top of the column compared to the grooves . by sake of simplicity ,
we have considered a 1d surface and we have taken the same non local and nonlinear multiplicative factor @xmath34 for all the terms of the dynamics .
further on , we have considered a simple power law behaviour for this function @xmath35 .
we argue that @xmath36 is needed to enhance the shadowing effects on the protuberances .
we have tested numerically @xmath37 and @xmath38 with no loss of properties of the results . however
, a better choice of the shadowing function @xmath39 should be obtained through further experimental comparisons .
similarly , different shadowing functions should be considered in the future for the diffusion term and the deposition term .
finally , this new continuous model , with @xmath37 , considered as a minimal model , already correctly reproduces the formation of flat wide columns , with sharp edges and thin separating grooves , as usually encountered in sputtering deposition .
further works should perform such approaches to two dimensional surfaces in particular . | french ministry of research , conseil rgional du centre , communaut dagglomration du drouais , conseil gnral de leure et loir and crt plasma - laser are acknowledged for a research grant .
c.j . wants to thank the financial suport of the dga .
99 a. g. dirks , r. a. wolters , a. e. m. de veirman , _ columnar microstructures in magnetron - sputtered refractory metal thin films of tungsten , molybdenum and w
ti(n ) _ thin solid films * 208 * 2 , pp 181188 ( 1992 ) .
j. krug , _ origin of scale invariance in growth processes _ , adv .
phys . * 46 * , 1997 , pp 139282 ; j. krug and m. schimschak , _ metastability of step flow growth in 1 + 1 dimensions _ , j. phys .
i france * 5 * , pp 10651086 ( 1995 ) . | french ministry research conseil rgional centre communaut drouais conseil gnral leure loir crt plasma - laser acknowledged research grant financial suport dga dirks wolters de veirman columnar microstructures magnetron - sputtered refractory metal films tungsten molybdenum 181188 1992 j. krug origin scale invariance growth processes adv . . 46 1997 139282 krug schimschak metastability step flow growth 1 + 1 dimensions j phys . france 5 10651086 1995 | 0.070346 | 232 | 121 | , Saving $0.0 in GPT-4. |
55,610 | the axial strange form factor @xmath0 of the nucleon is assumed to be dominated at low momentum transfer by the isoscalar axial vector mesons @xmath1 and @xmath2 .
the importance of the @xmath3-triangular vertex correction is demonstrated .
= -1.3 truecm 15 truecm | the emc measurements of the longitudinally polarized deep inelastic structure function @xmath4 of the proton@xcite raised the question of the spin distribution among quarks and gluons . in the framework of the `` naive '' parton model the proton spin is expressed in terms of the spin polarization distribution of quarks and antiquarks @xmath5 as @xmath6 in fact , the emc has measured the integral @xmath7 note that @xmath8 const . in the perturbative regime of qcd because of chiral symmetry arguments .
in combining ( 2 ) with the experimental value of the isovector axial coupling constant @xmath9 known from neutron @xmath10 decay , @xmath11 on the one side , and with the @xmath12 ratio extracted from hyperon @xmath10 decay @xmath13 on the other side , the result is obtained , that the fraction of the proton spin carried by the quarks is surprisingly small ( see @xcite for details ) . the integral in ( 2 ) can alternatively be expressed as a combination of the axial isovector ( @xmath14 ) , axial hypercharge ( @xmath15 ) , and axial flavor singlet ( @xmath16 ) form factors as follows : @xmath17 where ( 5 ) is based on the ( presumably rather strong ) assumption of su(3 ) symmetry leading to the relations @xmath18 the nucleon matrix elements of the axial hypercharge ( @xmath19 ) and flavor singlet ( @xmath20 ) quark currents are normalized according to : @xmath21 where @xmath22 denote the corresponding gell - mann matrices , @xmath23 the nucleon dirac spinor@xcite , and @xmath24 stands for the flavor quark su(3 ) spinor .
the strange axial nucleon form factor @xmath25 is introduced as @xmath26 in the case of a vanishing strange axial form factor , eq .
( 5 ) reduces to the so called ellis - jaffe sum rule@xcite .
now the surprise is the significant deviation of the measured value for @xmath27 from the ellis - jaffe sum rule , attributed to a small flavor singlet axial form factor or , alternatively , to a significant nonvanishing value of @xmath28 .
different scenarios have been discussed in the literature in this respect . here , we would like to point out the one presented in@xcite where due to a non - trival structure of the hadronic vacuum the flavor singlet axial charge is `` eaten '' by instanton effects .
the axial form factors introduced above enter the neutral axial vector current according to ( for details see , e.g. , ref .
@xcite ) @xmath29 here , the notation @xmath30 has been used and the @xmath31 content of the nucleon has been neglected in the last line of ( 9 ) . thus in the light of the proton spin problem , the theoretical investigation of the strange axial nucleon form factor becomes important .
in this study we present a @xmath32-vector meson dominance ( vmd ) model for this form factor and demonstrate the importance of a lowest 1-loop vertex correction .
in the spirit of the vmd ansatz@xcite and in analogy to the @xmath34 dominance of the isotriplet axial current exploited in@xcite we suggest the following representation for the strange axial form factor of the nucleon at low energies : @xmath35 where @xmath36 . here , `` @xmath37 '' and `` @xmath38 '' label respectively the constants related to the @xmath1 and @xmath2 mesons . in ( 10 ) the deviation of the mixing angle between the isoscalar of the octet and the flavor singlet state in the axial vector meson nonet from the ideal mixing angle of @xmath39 has been made explicit according to : @xmath40 the angle @xmath41 has been determined experimentally in@xcite to be @xmath42 .
the latter equation shows that a significant violation of the ozi rule is observed in the axial vector nonet@xcite . by chiral symmetry arguments it is possible to parametrize the coupling constants of the @xmath32 mesons as @xmath43 and @xmath44 .
the values for @xmath45 and @xmath46 are taken from@xcite .
dispersion relations with one subtraction have now been assumed for the @xmath47 and @xmath48 strange form factors , respectively , @xmath49 now , the imaginary part of @xmath50 can be calculated from the feynman graphs of fig . 1 by means of the cutkosky rules@xcite .
the most important @xmath33 vertex correction is due to the ( @xmath51 ) decay channel with a @xmath52 branching ratio .
it is especially important for the modification of the one - body isoscalar axial vector current of in - medium nucleons , a fact demonstrated in@xcite .
there it was shown for the case of the fermi gas model that the @xmath53 exchange current reduces the one - body matrix element of the isoscalar axial vector current by about 10% .
the effect is determined by the ratio @xmath54 and acquires importance mainly because of the above mentioned significant violation of the ozi rule in the axial vector meson nonet .
we use the value @xmath55 in accordance with@xcite .
to describe the @xmath56 vertex we use the following effective lagrangian density @xmath57 the value of @xmath58 extracted from the corresponding partial width is up to a sign @xmath59 .
the coupling constant of the @xmath60 meson to the nucleon is identified to that of the scalar - isotriplet channel of the bonn potential@xcite as @xmath61 , and a pseudoscalar form of the @xmath62 vertex is assumed . from the graphs in fig . 1
we obtain for the absorptive part of the @xmath33 form factor the following expressions for its low frequency part ( @xmath63 ) : @xmath64 where @xmath65 the high frequency part ( @xmath66 ) is simply obtained from ( [ i m ] ) by replacing @xmath67 by @xmath68 .
it is small for unitarity arguments . in fig . 2 the ratio of @xmath69 including the vertex contribution to the one ( @xmath70 ) without
is presented . at higher momentum transfers ,
the vertex contribution is on the level of several percent and thus non negligible .
we expect that the @xmath3 triangular vertex correction to the strange axial form factor will increase in case the @xmath71 interaction is included .
the relevant t - matrix amplitude can be evaluated within the framework presented in@xcite .
the main conclusion of our study is that the strong violation of the ozi rule in the axial vector meson nonet implies j. ashman et al .
* 206b * ( 1988 ) 364 ; nucl .
* b328 * ( 1990 ) 1 j. ellis and m. karliner , phys .
* b313 * ( 1993 ) 131 e. reya , lecture notes in physics * 426 * ( springer , berlin - heidelberg 1994 ) j. ellis and r. l. jaffe , phys .
* d9 * ( 1974 ) 1444 | emc measurements longitudinally polarized deep inelastic structure function proton@xcite raised question spin distribution among quarks gluons naive parton model proton spin expressed spin polarization distribution of quarks antiquarks emc measured integral @xmath7 @xmath8 in perturbative regime qcd chiral symmetry arguments combining ( 2 ) with experimental value isovector axial coupling constant @xmath9 from neutron decay @xmath12 ratio from hyperon result fraction proton spin carried by quarks small see details integral in ( 2 ) expressed combination axial isovector hypercharge flavor singlet ) form factors ( 5 ) based su(3 ) symmetry relations nucleon matrix elements axial hypercharge flavor singlet ) quark currents normalized according @xmath21 @xmath22 gell - mann matrices nucleon dirac spinor flavor quark su(3 ) spinor strange axial nucleon form factor @xmath25 introduced as @xmath26 vanishing strange axial form factor ( 5 ) reduces to ellis - jaffe sum rule@xcitesurprise significant deviation @xmath27 from ellis - jaffe sum rule attributed to small flavor singlet axial form factor or significant nonvanishing value @xmath28 scenarios discussed point out one in@xcite non - trival structure hadronic vacuum flavor singlet axial charge eaten by instanton effects axial form factors enter neutral axial vector current @xmath29 notation @xmath30 used @xmath31 content nucleon neglected in last line of 9 ) proton theoretical investigation of strange axial nucleon form factor important present @xmath32-vector meson dominance model importance of lowest 1-loop vertex correction @xmath34 dominance representation for strange axial form factor nucleon at low energies : @xmath35 @xmath36 @xmath37 @xmath38 label constants related to @xmath1 and @xmath2 mesons deviation of mixing angle between isoscalar octet flavor singlet state axial vector meson from ideal mixing angle of @xmath39 explicit according angle @xmath41 determined in@xcite to be @xmath42 equation shows violation ozi rule in axial vector@xcite chiral symmetry coupling constants @xmath32 mesons @xmath43 values @xmath45 @xmath46 from@xcite dispersion relations subtraction assumed for @xmath47 @xmath48 form factors imaginary part @xmath50 calculated from feynman graphs cutkosky rules@xcite important @xmath33 vertex correction due to @xmath51 decay channel @xmath52 branching ratio important for modification one - body isoscalar axial vector current in - medium nucleons demonstrated in@xcite @xmath53 exchange current reduces one - body matrix element axial vector current 10% effect determined by ratio @xmath54 violation ozi rule use value @xmath55 with@xcite @xmath56 vertex effective lagrangian density @xmath57 value @xmath58 extracted partial width up to @xmath59 coupling constant @xmath60 meson to nucleon identified scalar - isotriplet channel bonn potential@xcite @xmath61 pseudoscalar form @xmath62 vertex assumed graphs figobtain absorptive @xmath33 form factor expressions low frequency @xmath63 @xmath64 @xmath65 high frequency @xmath66 obtained replacing @xmath67 @xmath68 small unitarity arguments fig . 2 ratio @xmath69 vertex contribution @xmath70 ) presented higher momentum transfers vertex contribution several percent non negligible @xmath3 triangular vertex correction axial form factor increase @xmath71 interaction included t - matrix amplitude evaluated framework@xcite conclusion strong violation ozi rule axial vector meson implies ashman et al . 206b 1988 ) 364 b328 1990 ) j ellis m karliner phys b313 1993 ) 131 reya lecture notes physics 426 springer 1994 j ellis l jaffe phys 1974 ) 1444 | 0.263468 | 1,608 | 828 | , Saving $0.0 in GPT-4. |
55,615 | we have developed mggpod , a user - friendly suite of monte carlo codes built around the widely used geant ( version 3.21 ) package .
the mggpod monte carlo suite and documentation are publicly available for download .
mggpod is an ideal tool for supporting the various stages of gamma - ray astronomy missions , ranging from the design , development , and performance prediction through calibration and response generation to data reduction . in particular
, mggpod is capable of simulating _ ab initio _ the physical processes relevant for the production of instrumental backgrounds .
these include the build - up and delayed decay of radioactive isotopes as well as the prompt de - excitation of excited nuclei , both of which give rise to a plethora of instrumental gamma - ray background lines in addition to continuum backgrounds .
[ 2001/04/25 1.1 ( pwd ) ] | intense and complex instrumental backgrounds , against which the much smaller signals from celestial sources have to be discerned , are a notorious problem for low and intermediate energy gamma - ray astronomy ( @xmath050 kev 10 mev ) . therefore a detailed qualitative and quantitative understanding of instrumental line and continuum backgrounds is crucial for most stages of gamma - ray astronomy missions , ranging from the design and development of new instrumentation through performance prediction to data reduction .
a promising approach for obtaining quantitative estimates of instrumental backgrounds is _ ab initio _
monte carlo simulation ( see e.g. * ? ? ?
we have developed a suite of monte carlo packages , named mggpod @xcite , that supports this type of simulation .
the mggpod monte carlo suite ( version 1.0 ) and documentation are publicly available for download from http://sigma-2.cesr.fr / spi / mggpod/. in this paper we provide an overview of the capabilities , functioning , and structure of the mggpod package , and give examples of applications to past and present gamma - ray missions .
the mggpod monte carlo suite allows _ ab initio _ simulations of instrumental backgrounds including the many gamma - ray lines arising from interactions of the various radiation fields within the instrument and spacecraft materials .
it is possible to simulate both prompt instrumental backgrounds , such as energy losses of cosmic - ray particles and their secondaries , as well as delayed instrumental backgrounds , which are due to the decay of radioactive isotopes produced in nuclear interactions .
of course , mggpod can also be used to study the response of gamma - ray instruments to astrophysical and calibration sources .
the mggpod suite is therefore an ideal monte carlo tool for gamma - ray astronomy .
a detailed description of the physics of the mggpod monte carlo suite can be found in @xcite .
the documentation available from the mggpod web site provides comprehensive practical advice for users .
first applications of mggpod have been presented in @xcite .
mggpod is a suite of five closely integrated monte carlo packages , namely * mg*eant , * g*calor , * p*rompt , * o*rihet , and * d*ecay .
the mggpod suite resulted from a combination of the nasa / gsfc mgeant @xcite and the university of southampton ggod @xcite monte carlo codes , which we supplemented with the newly developed prompt package .
all these packages are based on the widely used geant detector description and simulation tool ( version 3.21 ) created and supported at cern , which is designed to simulate the passage of elementary particles through an experimental setup . in a nutshell ,
the capabilities and functions of the five packages that constitute the mggpod suite are as follows : * mgeant is a multi - purpose simulation package that was created to increase the versatility of the geant simulation tool . a modular , `` object oriented '' approach was pursued , allowing for rapid prototyping of detector systems and easy generation of most of the radiation fields relevant to gamma - ray astronomy .
within the mggpod suite , mgeant ( i.e. geant ) stores and transports all particles , and treats electromagnetic interactions from about 10 kev to a few tev .
mgeant provides the option of using the glecs package to take into account the energy of bound electrons in compton scatterings .
the mgeant simulation package and a user manual are available at a nasa / gsfc web site .
* gcalor @xcite simulates hadronic interactions down to 1 mev for nucleons and charged pions and down to thermal energies ( @xmath1 ev ) for neutrons .
equally important , this package provides access to the energy deposits from all interactions as well as to isotope production anywhere in the simulated setup .
* prompt simulates prompt photon emission associated with the de - excitation of excited nuclei produced by neutron capture , inelastic neutron scattering , and spallation . * orihet , originally developed for the ggod suite and improved for mggpod , calculates the build - up and decay of activity in any system for which the nuclide production rates are known .
hence orihet can be used to convert nuclide production rates , determined from simulations of cosmic - ray irradiation , to decay rates .
these are required input for simulating the radioactive decays giving rise to delayed background .
* decay , again originally developed for ggod and improved for our purposes , enables mggpod to simulate radioactive decays .
the overall structure of the mggpod package is illustrated in fig .
[ mggpod_flow_chart ] . depending on the simulated radiation field or gamma - ray source distribution one or three steps , requiring two or three input files ,
are needed to obtain the resulting energy deposits in the detector system under study .
in general , it is advisable to simulate each component of the radiation environment separately .
mggpod distinguishes two classes of radiation fields .
class i comprises radiation fields for which only prompt energy deposits are of interest , such as celestial or laboratory gamma - ray sources or cosmic - ray electrons .
class ii comprises radiation fields for which in addition delayed energy deposits resulting from the activation of radioactive isotopes need to be considered
. examples for class ii fields are cosmic - ray protons , or geomagnetically trapped protons . for both of these classes , the simulation of the prompt energy deposits requires two input files : a mass model , and a model of the simulated radiation field .
the mass model is a detailed computer description of the experimental setup under study .
it specifies the geometrical structure of instrument and spacecraft , the atomic and/or isotopic composition of materials , and sets parameters that influence the transport of particles in different materials .
each component of the radiation environment ( and analogously for gamma - ray sources ) to which the instrument is exposed is characterized by three quantities : the type of the incident particles , and their spectral and angular distributions .
the prompt energy deposits are written to an output event file ; in case of a class ii radiation field there is an additionial output file in which all the nuclei produced in hadronic interactions are recorded . to simulate delayed energy deposits ( class ii radiation field ) two additonal steps need to be taken .
these require as input the time history of the radiation field which is responsible for the activation , and the previously calculated isotope production rates . based on this information first the activity of each isotope produced in each structural element of the mass model
is determined .
then these activities are used to simulate the delayed energy deposits due to radioactive decays in the instrument . combining prompt and delayed energy deposits from each component of the radiation environment and gamma - ray sources , it is possible to obtain the total energy deposited in the system as a function of position and time .
this section addresses some of the practical considerations a potential user of mggpod might have .
it is a synopsis of the documentation available from the mggpod web site .
installation of the complete mggpod package ( software and data files ) requires little more than 100 mb of disk space .
the documentation provides detailed installation instructions , including the installation of the required cernlib and other libraries which are necessary to build , but are not included in , the mggpod package .
the mggpod software is written in the fortran 77 and c programming languages ; in addition there are a few c shell scripts .
the source code is completely open and transparent , allowing the user to adapt , change , and improve the code .
in fact , a few customizations and adjustments of the code are inevitable when simulating different instruments .
these unavoidable changes are described in the documentation .
the format of both the event and activation output files is fits .
mggpod inherited the interactive display capabilities of ( m)geant , which are based on cern s paw++ package .
this is a very convenient feature when creating a mass model , or when specifying the parameters of a radiation field ( `` beam '' ) . to facilitate the use of mggpod by novice users
the release contains examples for all simulation steps outlined in sec .
[ structure ] .
the examples include all input and output files , and instructions on how to build and run the necessary executables . as described in sec .
[ cap_func ] , different modules of mggpod simulate different physical processes which can be treated over different energy ranges .
the mggpod input file allows the user to define low - energy cutoff values for the tracking of five different particle types : photons , electrons , neutral and charged hadrons , and muons .
we recommend to use as default low - energy cutoff values 10 kev for photons , electrons , and muons , @xmath2 ev for neutral hadrons , and 1 mev for charged hadrons .
experienced users can lower these thresholds , in particular for photons . for many applications of mggpod the computation time needed to complete
a simulation is an important consideration .
the speed of mggpod monte carlo simulations varies greatly with the type and energy of the incident particle ( which affect e.g. the number of secondary particles that are produced and tracked ) , with the physical volume and mass described in the mass model ( but less with geometrical complexity ) , and of course with the speed of the computer employed . as an example , for the tgrs and spi mass models , using a 1.6 ghz cpu , typically about @xmath3 cpu seconds and @xmath4 cpu seconds are needed to simulate the interactions of a cosmic - ray proton , respectively .
the difference in processing time is mainly due to the fact that integral / spi is a much heavier and complex instrument than wind / tgrs . for both models
the time needed to simulate an incident gamma - ray photon , or a radioactive decay , is about ten times shorter than the respective proton processing time .
the computation time for a simulation also depends on the desired statistics in the result ; better statistics require the simulation of more particles and consequently more processing time . for spi , about @xmath5 photons , requiring a computation time of about 8 cpu hours , in a homogeneous beam covering the 19 ge detectors are sufficient for simulating the instrument response at a given energy .
when simulating the activation of integral and spi by cosmic - ray protons , @xmath5 protons , requiring a computation time of about 3 cpu days ( and representing about 21 s of actual in - orbit irradiation ) , provide acceptable statistics for calculating nuclide production rates , and enough statistics in the detector count rate to obtain a crude spectrum .
detailed studies of prompt background line production by cosmic - ray protons require multiple simulation runs .
the mggpod web site also provides off - the - shelf model spectra for two common radiation fields : cosmic - ray protons ( at solar minumum and maximum , based on * ? ? ? * ) , and the diffuse cosmic x - ray and gamma - ray emission ( based on * ? ? ?
* ) . for missions in low - earth orbit albedo radiations (
most importantly neutrons and gamma rays ) and geomagnetically trapped paricles need to be considered in addition .
the web site provides links to the esa space information system ( spenvis ) and the nasa space ionizing radiation environments and shielding tools ( sirest ) , which can be used to obtain radiation field models for missions in a low - earth environment .
users should be aware that the space environment is not stable .
detailed simulations require that input particle spectra reflect conditions in a given orbit and at a given phase of the solar cycle ( including the polarity of the heliosphere for cosmic - ray protons , * ? ? ?
currently , gcalor does not treat hadronic interactions for alpha particles or heavier cosmic rays .
if diffuse cosmic x - rays are an important background component , the uncertainty in the simulation result is usually dominated by uncertainties in the photoelectric absorption in materials surrounding the detectors as defined in the mass model .
the mggpod monte carlo suite has been , and is being , applied to model the instrumental backgrounds of several gamma - ray missions . using the mggpod codes , very good agreement between the monte carlo results and the actual data of the transient gamma - ray spectrometer ( tgrs ) on board _ wind _
@xcite has been obtained , as discussed in @xcite . as an example , in fig .
[ comp_tgrs ] we show a comparison of mggpod simulations with tgrs data . first modelling results for the spi spectrometer on board the esa integral observatory @xcite again yielded good agreement , but also indicate remaining deficiencies with respect to the production and thermalization of secondary neutrons in a massive spacecraft and/or instrument @xcite .
both of these instruments operate in highly elliptical orbits above the earth s radiation belts .
recently , the mggpod suite has been applied to modelling the instrumental background of the reuven ramaty high - energy solar spectroscopic imager ( rhessi , described in * ? ? ?
this is the first time mggpod is used for modelling the instrumental background of an instrument in low - earth orbit .
this case is particularly difficult because activation during passages through the south - atlantic anomaly ( saa ) gives rise to a strong and time - variable background component . unlike all other components of the radiation environment
, the time - variability of the saa - induced backgrounds needs to be accounted for in the simulations .
preliminary modelling results are presented by @xcite at this workshop .
the mggpod monte carlo suite is an ideal tool for supporting the various stages of gamma - ray astronomy missions , ranging from the design , development , and performance prediction through calibration and response generation to data reduction .
the mggpod software and documentation are publicly available for download at cesr .
the package has been , and is being , successfully applied to several past and present gamma - ray missions .
dean , a. j. , et al .
2003 , space sci .
105 , 285 gruber , d.e .
, et al . 1999 , apj 520 , 124 moskalenko , i. , et al . , 2002 , apj 565 , 280 owens , a. , et al . 1995 , space sci .
71 , 273 smith , d.m . , et al . , solar physics 210 , 33 , 2002 sturner , s. , et al .
2000 , _ the fifth compton symposium _ ( aip 510 ) , 814 vedrenne , g. , et al .
2003 , a&a 411 , l63 , 2003 weidenspointner , g. , et al .
2003 , new astr . rev . 48 , 227 weidenspointner , g. , et al .
2004 , in preparation wunderer , c. , et al . , 2004 ,
these proceedings zeitnitz , c. , & gabriel , t. a. 1994 , nim a 349 , 106 | complex instrumental backgrounds smaller signals from celestial sources problem for low intermediate energy gamma - ray astronomy detailed understanding of instrumental line continuum backgrounds crucial for gamma - ray astronomy missions design new instrumentation performance prediction data reduction approach for quantitative estimates backgrounds is monte carlo simulation mggpod @xcite supports this simulation mggpod monte carlo suite version 1.0 documentation available for download from http://sigma-2.cesr.fr / spi / mggpod/ overview of capabilities functioning structure mggpod package examples of applications to past present gamma - ray missions mggpod suite allows simulations of instrumental backgrounds gamma - ray lines from interactions radiation fields instrument spacecraft materials simulate prompt instrumental backgrounds delayed instrumental backgrounds due to decay radioactive isotopes mggpod response of gamma - ray instruments to astrophysical calibration sources mggpod suite ideal monte carlo tool for gamma - ray astronomy detailed description of physics mggpod suite in @xcite documentation mggpod web site provides advice first applications of mggpod presented in @xcite mggpod is five monte carlo packages mg*eant g*calor p*rompt o*rihet d*ecay from nasa / gsfc mgeant @xcite university of southampton ggod @xcite carlo codes supplemented with prompt package packages based on geant detector simulation tool version 3.21 ) cern passage elementary particles through experimental setup capabilities functions of five packages mgeant multi - purpose simulation package versatility geant simulation tool modular object oriented approach rapid prototyping detector easy generation of radiation fields gamma - ray astronomy mgeant stores transports particles treats electromagnetic interactions from 10 kev to tev glecs package energy of bound electrons in compton scatterings simulation package user manual available at nasa / gsfc web site gcalor @xcite simulates hadronic interactions to 1 mev for nucleons charged pions thermal energies for neutrons provides access to energy deposits from interactions isotope production simulated setup prompt simulates photon emission de - excitation of excited nuclei neutron capture inelastic neutron scattering spallationorihet developed for ggod suite improved for mggpod calculates build - up decay activity in system nuclide production rates known nuclide production rates cosmic - ray irradiation to decay rates required input for simulating radioactive decays delayed background decay developed for ggod improved enables mggpod to simulate radioactive decays structure mggpod package illustrated in fig . mggpod_flow_chart ] depending simulated radiation field one or three steps input files needed to obtain energy deposits in detector system simulate each component radiation environment separately mggpod distinguishes two classes of radiation fields class i prompt energy deposits class ii delayed energy deposits from activation radioactive isotopes examples cosmic - ray protons geomagnetically trapped protons simulation prompt energy deposits requires two input files mass model model of simulated radiation field mass model description of experimental setup specifies geometrical structure of instrument spacecraft atomic isotopic composition of materials sets parameters transport of particlescomponent radiation environment gamma - ray sources instrument exposed characterized by three quantities type incident particles spectral angular distributions prompt energy deposits written to output event file class ii radiation field additionial output file nuclei hadronic interactions recorded simulate delayed energy deposits class ii radiation field two additonal steps require time history radiation field calculated isotope production rates activity of each isotope in structural element mass model determined activities used to simulate delayed energy deposits due to radioactive decays combining prompt delayed energy deposits radiation environment gamma - ray sources obtain total energy deposited system function of position and time section addresses practical considerations user mggpod synopsis of documentation from mggpod web site installation complete mggpod package software data requires than 100 mb disk space documentation provides detailed installation instructions required cernlib other libraries mggpod software written in fortran 77 c programming languages few c shell scripts source code open transparent user adapt change improve code customizations adjustments inevitable when simulating different instruments changes described in documentation format of event activation output files fits .mggpod inherited interactive display capabilities of m)geant based on cern s paw++ package convenient feature creating mass model specifying parameters radiation field use novice users release contains examples for simulation steps in structure examples include input output files instructions to build run executables different modules mggpod simulate different physical processes over different energy ranges input file allows define low - energy cutoff values for five particle types photons electrons neutral charged hadrons muons recommend default low - energy cutoff values 10 kev for photons electrons muons @xmath2 ev for neutral hadrons 1 mev for charged hadrons experienced users can lower thresholds for photons computation time simulation important speed of simulations varies with type energy of incident particle physical volume mass model geometrical complexity speed computer for tgrs spi mass models 1.6 ghz cpu about @xmath3 cpu seconds and @xmath4 cpu seconds needed to simulate interactions of cosmic - ray proton difference processing time due to integral / spi heavier complex instrument than wind / tgrsmodels time incident gamma - ray photon radioactive decay ten times shorter than proton processing time computation time depends on desired statistics better statistics require more particles more processing time spi @xmath5 photons computation time 8 cpu hours 19 ge detectors sufficient instrument response energy activation of integral spi by cosmic - ray protons @xmath5 protons computation time 3 cpu days 21 s in - orbit irradiation provide acceptable statistics nuclide production rates enough detector count rate crude spectrum studies background line production by cosmic - ray protons require multiple simulation runs mggpod web site provides model spectra for radiation fields cosmic - ray protons diffuse cosmic x - ray gamma - ray emission missions low - earth orbit albedo radiations neutrons gamma rays geomagnetically trapped paricles web site links to esa space information system nasa space ionizing radiation environments shielding tools radiation field models for low - earth environment space environment not stablesimulations require input particle spectra reflect conditions orbit phase solar cycle including polarity heliosphere for cosmic - ray protons gcalor treat hadronic interactions for alpha particles heavier cosmic rays if diffuse cosmic x - rays important uncertainty simulation result dominated by photoelectric absorption materials surrounding detectors mggpod monte carlo suite applied to instrumental backgrounds of gamma - ray missions good agreement between monte carlo results data transient gamma - ray spectrometer @xcite obtained fig comparison mggpod simulations with tgrs data modelling results for spi spectrometer esa integral observatory @xcite yielded agreement indicate deficiencies production thermalization of secondary neutrons in spacecraft instrument @xcite instruments operate in elliptical orbits above earth s radiation belts mggpod suite applied to instrumental background of reuven ramaty high - energy solar spectroscopic imager first time mggpod used for modelling instrumental background instrument in low - earth orbit case difficult activation during south - atlantic anomaly time - variable background component - variability saa - induced backgrounds needs formodelling results presented by @xcite workshop mggpod monte carlo suite gamma - ray astronomy missions design development performance prediction calibration response generation data reduction mggpod software documentation available at cesr package applied to past gamma - ray missions dean a. j. . 2003 space sci . 105 , 285 gruber 1999 apj 520 , 124 moskalenko 2002 apj 565 , 280 owens 1995 space sci . 71 , 273 smith solar physics 210 , 33 , 2002 sturner. 2000 fifth compton symposium aip 510 ) , 814 vedrenne 2003 a&a 411 , l63 weidenspointner new astr . rev . 48 , 227 2004 wunderer .. 2004 proceedings zeitnitz gabriel 1994 nim a 349 , 106 | 0.511063 | 3,219 | 1,580 | , Saving $0.1 in GPT-4. |
55,620 | recent studies have shown the possibility of water transport across carbon nanotubes , even in the case of nanotubes with small diameter ( 0.822 nm ) . in this case
, water shows subcontinuum transport following an ordered 1d structure stabilized by hydrogen bonds . in this work ,
we report md simulations describing the effect of a perpendicular electric field in this single - file water transport in carbon nanotubes .
we show that water permeation is substantially reduced for field intensities of 2 - 3 v / nm and it is no longer possible under perpendicular fields of 4 v / nm . | despite its strong hydrophobic character , single walled carbon nanotubes ( swcnts ) are spontaneously filled by water @xcite .
transport of water in these systems has novel and promising properties which have been extensively studied by molecular dynamics ( md ) simulations ( see for example @xcite for a review ) .
quite interestingly , the water transport trough a carbon nanotube depends strongly on its radius and md simulations predict a transition between continuum to subcontinuum transport as the radius decreases @xcite , as recently confirmed experimentally @xcite . in the case of the nanotubes with small radius ,
the nanotubes are filled by a one - dimensional ordered chain of water molecules which maintain 2 hydrogen bonds per molecule inside the cnt @xcite .
this effective 1d water system has extremely interesting properties and it has been studied experimentally @xcite , by md simulations @xcite and analytical treatments based on the 1d ising model @xcite .
interestingly , this system has been analyzed as a model for water transport in biological pores @xcite .
a recent md simulation study has shown the strong response of this single - file water system inside cnts to electric fields parallel to the axis of the nanotube @xcite , due to the peculiarities of the 1d hydrogen bonding network .
motivated by this previous study , in this work we consider a related but different question , namely the effect of applying an electric field perpendicular to the single file water transport across carbon nanotubes . in this case , a competition arises between the hydrogen bonding network ( which tends to maintain the alignment of the chain of water molecules in the nanotube axis direction ) and the perpendicular electric field ( which tends to align the water dipole in the perpendicular direction ) .
a simple numerical calculation proposed in @xcite suggests that this effect will appear for field intensities between 2 - 3 v / nm .
noting that a single water - water hydrogen bond has a free energy of about 5@xmath0 and the dipole of the water molecule in liquid phase is about 2.5 d we can say that h - bonding will compete with external fields with intensities up to @xmath1 v / nm . albeit high , these field intensities can be studied in simulations and experiments . in this work ,
we report a preliminary set of md simulations in order to study the effect on water transport of this competition between hydrogen bonding and electric field .
we have performed a total of 33 molecular dynamics ( md ) simulations of water transport across carbon nanotubes in both equilibrium and nonequilibrium conditions in presence and in absence of an electric field perpendicular to the axis of the nanotubes .
all simulations described here were performed using the 2.7 version of the namd program @xcite .
the preparation of the system and analysis of the results were made using the visual molecular dynamics ( vmd ) software @xcite .
the simulated system is shown in figure [ fig : system ] .
it consisted of a periodic simulation box of dimensions @xmath2 @xmath3 containing 248 water molecules and a porous membrane made of four carbon nanotubes with their axis aligned in the @xmath4 direction .
each carbon nanotubes has 144 carbon atoms and has a radius of 0.411 nm and length 1.34 nm .
water molecules were modelled using the tip3p model as implemented in the charmm force field and carbon atoms from nanotubes were modelled as lennard - jones spheres with parameters @xmath5 kcal / mol and @xmath6 .
carbon atoms were maintained fixed during all the simulation as in previous works@xcite .
lennard - jones interactions were computed using a smooth ( 10 - 12 ) cutoff , as it is customary done in namd2 simulations .
the electrostatic interactions were calculated using the particle - mesh ewald ( pme ) method .
we employ a multiple time step .
the equations of motion are solved with a 1 fs time step , nonbonded interactions are updated each 2 time steps and electrostatic interactions are updated each 4 time steps .
the temperature was maintained at 300 k in all simulations employing a langevin thermostat with a 5 ps@xmath7 dumping constant .
the langevin thermostat is applied to the water oxygen atoms only .
some of our simulations were performed in presence of an external pressure difference .
this external pressure was applied in namd2 as a tcl force ( as implemented in @xcite ) which acts on oxygen atoms in a slab of 5.4 thickness ( all oxygen atoms with coordinates @xmath8 or @xmath9 ) .
this methodology is also described in detail in ref @xcite .
the applied pressure @xmath10 is computed from the force @xmath11 by @xmath12 where @xmath13 is the number of water molecules in the slab and @xmath14 its transversal area . in most of our simulation runs
, we applied an uniform electric field @xmath15 , perpendicular to the axis of the nanotubes , a feature which is implemented in namd2 .
the field is acting in all the simulation box . in all simulations , we compute the number of permeation events , defined as the number of water molecules observed to completely cross the nanotube ( entering from one end and leaving the tube from the other end ) .
obviously this quantity depends on the simulation time , so it was normalized to the total production time ( in ns ) .
the length of the production runs was decided by monitoring this quantity , i.e. simulations were performed until this quantity reached an stable average value .
depending on the particular simulation runs , simulation times are between 5 and 40 ns .
first , we conducted equilibrium simulations ( no applied external fields or pressure gradients ) . in this case , the average flux of water is negligible , as expected in equilibrium . in equilibrium
we obtain an average rate of permeation events of water molecules of 9 molecules / ns in any of the directions . and
@xmath16 direction .
filled triangles correspond to simulations with a water flow induced under an external pressure of @xmath17 atm . in this case ,
permeation events occur only in the direction of decreasing pressure.,width=340 ] now , we consider the effect of applying an external electric field @xmath15 perpendicular to the axis of the nanotubes .
the results are shown in figure [ fig : results ] .
fields up to 1 v / nm have no significant impact on water transport .
however , for larger fields , the number of permeation events decreases substantially as the field strength increases . for fields about 2 v / nm ,
the number of permeation events has been reduced to less than its equilibrium value ( see fig . [
fig : results ] ) .
we also observe that chains inside nanotubes start to break and we even observe crossing of isolated ( not hydrogen bonded ) water molecules across the nanotube , as shown in the snapshot in figure [ fig : dehydration ] .
larger fields reduce the permeation events even more and it can be say that water permeation is not possible for fields larger than 4 - 5 v / nm . direction .
atoms from water molecules are shown with van der waals radius and carbon atoms from the four swcnt are not shown for simplicity .
note that one of the tubes is empty and in another nanotube there is only a single , isolated water molecule inside one of the nanotubes oriented in the field direction . in another of the tubes the hydrogen - bonded water chain is still intact and water molecules fill the tube.,width=340 ] we have also performed simulations under a large external pressure @xmath18 atm ( see fig.[fig : results ] ) . in absence of external electrical field , we observe about 133 permeation events per ns . in all cases permeation
occurs in the direction of decreasing pressure , as should be expected in this case . under the application of external , perpendicular electric field
we obtain results strongly resembling to those obtained in the absence of a pressure gradient .
again , the effect of fields smaller than 1 v / nm is negligible , and substantial reduction of the flow of water requires fields of the order of 2 - 3 v / nm .
it can be said that the presence of the external pressure has no effect on the competition between hydrogen bonding and the external electric field .
in this work , we have performed md simulations of single - file transport of water in carbon nanotubes of small radius .
simple arguments suggest that the application of an electric field perpendicular to the axis of the carbon nanotube will disrupt the hydrogen bond structure of this 1d water system and affect the transport properties of water inside the nanotube .
our simulations show that water transport is not affected under perpendicular fields up to 1 v / nm .
larger fields substantially reduce the flow of water , and in fact transport is not possible for perpendicular fields above @xmath19 v / nm .
our work shows the possibility of using perpendicular electric fields as nanofluidic `` switches '' to regulate water transport in carbon nanotubes .
it is also interesting to note that a previous work @xcite has shown the advantages of pumping water across nanotubes using parallel electric fields ( between 0.01 v / nm - 1 v / nm ) instead of using a pressure gradient .
it could be interesting to analyze also this parallel case with higher fields ( 2 - 4 v / nm ) which could also somewhat influence the hydrogen bonding network .
our own ongoing work indicates that under parallel fields @xmath20 4 v / nm there are significant structural changes in the single file of water molecules .
overall , all these results show the interesting possibilities arising from using electric fields to control water flow in carbon nanotubes .
this work is supported by the spanish government ( grants fis2009 - 13370-c02 - 02 and consolider - nanoselect - csd2007 - 00041 ) and generalitat de catalunya ( grant 2009sgr164 ) .
11 k. wu , b. zhou , p. xiu , w. qi , r. wan and h. fang , _ kinetics of water filling the hydrophobic channels of narrow carbon nanotubes studied by molecular dynamics simulations _ , j. chem .
( 2010 ) , art .
204702 .
j. faraudo _ the missing link between the hydration force and interfacial water : evidence from computer simulations _ ,
current opinion colloid and interf .
doi:10.1016/j.cocis.2011.04.009 ( accepted for publication ) | strong hydrophobic character single walled carbon nanotubes filled by water transport water has novel properties studied by molecular dynamics md simulations water transport trough carbon nanotube depends on radius simulations predict transition between continuum to subcontinuum transport as radius decreases confirmed experimentally nanotubes with small radius filled by one - chain of water molecules maintain 2 hydrogen bonds per molecule effective 1d water system has interesting properties studied experimentally md simulations analytical treatments model system analyzed as model for water transport in biological pores recent md simulation study strong response of single - file water system to electric fields parallel to axis nanotube due to 1d hydrogen bonding network related question effect of applying electric field perpendicular to single file water transport across carbon nanotubes competition arises between hydrogen bonding network alignment water molecules nanotube axis and perpendicular electric field water dipole in numerical calculation suggests effect for field intensities between 2 - 3 v / nm single water - water hydrogen bond free energy 5@xmath0 dipole water molecule liquid phase 2.5 d h - bonding with fields intensities up to @xmath1 v / nm field intensities studied in simulations experiments preliminary md simulations effect water transport competition between hydrogen bonding electric field performed 33 molecular dynamics simulations of water transport across carbon nanotubes in equilibrium nonequilibrium conditions electric field perpendicular 2.7 version namd program @xcite preparation analysis visual molecular dynamics software simulated system in figure periodic simulation box dimensions @xmath2 @xmath3 248 water molecules porous membrane four carbon nanotubes aligned @xmath4 direction each carbon 144 carbon atoms radius 0.411 nm length 1.34 nm water molecules modelled using tip3p model carbon atoms as lennard - jones spheres parameters @xmath5 kcal / mol @xmath6 carbon atoms maintained fixed lennard - jones interactions computed smooth ( 10 - 12 ) cutoff electrostatic interactions calculated particle - mesh ewald method multiple time step equations of motion solved 1 fs time step nonbonded interactions updated 2 electrostatic interactions 4 temperature maintained at 300 k simulations langevin thermostat 5 ps@xmath7 dumping constant applied to water oxygen atoms simulations performed external pressure difference external pressure applied in namd2 as tcl force @xcite acts on oxygen atoms in slab 5.4 thickness oxygen atoms coordinates @xmath8 or @xmath9 methodology described in ref @xcite applied pressure @xmath10 computed from force @xmath11 by @xmath12 @xmath13 number water molecules slab @xmath14 transversal area applied uniform electric field @xmath15 perpendicular to axis nanotubes implemented in namd2 field acting in simulation box compute number permeation events water molecules cross nanotube depends on simulation time normalized to total production time in ns length production runs decided monitoring quantity simulations performed until stable average value simulation times between 5 and 40 ns conducted equilibrium simulations no external fields pressure gradients average flux of water negligible equilibrium average rate of permeation events of water molecules 9 molecules / ns@xmath16 direction filled triangles correspond to simulations water flow under external pressure @xmath17 atm permeation events occur decreasing pressure consider effect applying external electric field @xmath15 perpendicular to axis nanotubes results shown in figure fields up to 1 v / nm no significant impact on water transport larger fields permeation events decreases as field strength increases fields 2 v / nm permeation events reduced to less than equilibrium value chains inside nanotubes break crossing of isolated water molecules across nanotube snapshot larger fields reduce permeation events water permeation not possible for fields larger than 4 - 5 v / nm atoms from water molecules shown with van der waals radius carbon atoms from four swcnt not shown for simplicity one empty another single isolated water molecule field direction hydrogen - bonded water chain intact water molecules fill tube performed simulations under large external pressure @xmath18 atm absence of external electrical field 133 permeation events per ns permeation occurs decreasing pressureapplication external perpendicular electric field results resembling pressure gradient effect fields smaller than 1 v / nm negligible substantial reduction flow water requires fields 2 - 3 v / nm presence external pressure no effect competition hydrogen bonding external electric field performed simulations single - file transport water in carbon nanotubes small radius application electric field perpendicular hydrogen bond structure transport properties water simulations show water transport not affected under perpendicular fields up to 1 v / nm larger fields reduce flow water transport not possible fields above @xmath19 v / nm work shows possibility perpendicular electric fields as nanofluidic switches regulate water transport previous work shown advantages pumping water across nanotubes using parallel electric fields 0.01 v - 1 v / nm instead pressure gradient analyze parallel case with higher fields ( 2 - 4 v / nm ) influence hydrogen bonding network parallel fields @xmath20 4 v / nm significant structural changes in single file water molecules results show possibilities electric fields control water flow in carbon nanotubes work supported by spanish government grants - generalitat de catalunya grant 2009sgr164xiu qi wan fang kinetics water filling hydrophobic channels carbon nanotubes molecular dynamics simulations j chem. 2010 art 204702 faraudo missing link hydration force interfacial water computer simulations current opinion colloid interf.1016.2011.04.009 accepted publication | 0.354094 | 2,255 | 1,092 | , Saving $0.1 in GPT-4. |
55,621 | superfluidity or superconductivity with mismatched fermi momenta appears in many systems such as charge neutral dense quark matter , asymmetric nuclear matter , and in imbalanced cold atomic gases .
the mismatch plays the role of breaking the cooper pairing , and the pair - breaking state can not be properly described in the framework of standard bcs theory .
i give a brief review on recent theoretical development in understanding unconventional color superconductivity , including gapless color superconductor , the chromomagnetic instabilities and the higgs instability in the gapless phase .
i also introduce a possible new framework for describing unconventional color superconductor . | the research of charge neutral cold dense quark matter has driven the color superconductivity theory into a new territory beyond standard bcs theory .
the requirement of charge neutrality condition induces a substantial mismatch between the fermi surfaces of the pairing quarks , which reduces the available phase space for cooper pairing , and eventually breaks the cooper - pairing state .
superfluidity or superconductivity with mismatched fermi momenta also appears in other systems , e.g. , electronic superconductor in a strong external magnetic field , asymmetric nuclear matter , and in imbalanced cold atomic systems .
however , it has been an unsolved old problem how a bcs superconductivity is destroyed as the mismatch is increased .
the results based on conventional bcs framework brings us puzzles .
for example , in the standard bcs framework , it was found that at moderate mismatch , homogeneous gapless superconducting phases @xcite can be formed . however , gapless superconducting phases exhibit anti - meissner screening effect , i.e. , chromomagnetic instability @xcite , which is in contradict with the meissner effect in standard bcs superconductivity .
i give a brief report on recent theoretical development of understanding the puzzles in g2sc phase .
firstly , i explain that the gapless phase is the result under the conventional bcs framework in mean - field approximation , then i introduce a possible proper new framework .
in the mean - field approximation , the order parameter of 2sc phase takes a constant and has the form of @xmath0 , and the free energy for @xmath1 quarks in @xmath2-equilibrium has the form @xcite : @xmath3 .
\label{pot - m}\end{aligned}\ ] ] the sum over @xmath4 runs over all ( 6 quark and 6 antiquark ) quasi - particles . with the increase of mismatch
, the ground state will be in the gapless 2sc phase when @xmath5 .
the gapless phase is in principle a metastable sarma state @xcite , i.e. , the free energy is a local maximum with respect to the gap parameter @xmath6 , and one has @xmath7 in the gapless phase . as we already knew that g2sc phase exhibits anti - meissner screening effect or chromomagnetic instability @xcite , which is in contradict with the meissner effect in standard bcs superconductivity .
in recent years , there have been lots of effort trying to understand the puzzle of anti - meissner effect , here i give a brief report from my own point of view . in the 2sc phase , the color symmetry @xmath8 breaks to subgroup @xmath9 .
the full order parameter of the 2sc phase is characterized by @xmath10 ( 0 , \ , 0 , \ , \delta + h(x ) ) , \label{ii}\ ] ] where @xmath11 are five nambu - goldstone diquarks describing the phase fluctuation of the order parameter , and @xmath12 is the higgs field describing the spatial fluctuation of the order parameter .
expanding around the ground state : @xmath13 , the free energy of the system takes the expression as @xmath14 , where @xmath15 is the contribution from the mean - field approximation and was given in eq .
( [ pot - m ] ) , @xmath16 and @xmath17 are contributions from the fluctuation of nambu - goldstone currents and higgs field , respectively .
the quadratic action of the goldstone modes in the long wavelength limit can be written down with the aid of the meissner masses @xmath18 evaluated in ref.@xcite , it takes the form of @xmath19 \left [ { \vec { \bf a}}^a- \frac{1}{g } { \vec \triangledown } \varphi^a \right ] + higher \ , orders \ , .\ ] ] it was found that at zero temperature , with the increase of mismatch , for five gluons with @xmath20 corresponding to broken generator of @xmath21 , their meissner screening mass squares become negative @xcite .
this indicates the spontaneous generation of nambu - goldstone currents @xmath22 @xcite or gluon condensate state @xmath23 @xcite .
the ng current state can also be interpreted as a colored larkin - ovchinnikov - fulde - ferrel ( loff ) state with the single plane - wave order parameter @xmath24 .
the loff state @xcite was proposed in 1960s to describe a possible ground state of the pair - breaking state of an electronic superconductor when applying a strong external magnetic field .
however , it has still not yet been confirmed experimentally in electronic superconduting systems , and it still remains being persued after more than 40 years .
imbalanced cold atom systems offer another intriguing experimental possibility to understand how cooper pairing is destroyed . due to the absence of both the orbital effects and impurities
, it seems very promising to search for the loff state in imbalanced cold atom systems .
however , recent experiments @xcite in imbalanced cold atom systems did not show evidence of the formation of the loff state , rather indicated a non - uniform state of phase separation state .
obviously , there is something missing in our understanding of the pairing breaking state if we failed to observe the ( lo)ff state in the mismatch regime where the magnetic or superfluid density instability develops .
the free energy from the higgs field can be evaluated and takes the form of @xmath25 it was found in ref .
@xcite that the self - energy of the higgs field , @xmath26 , becomes negative in the gapless phase when @xmath27 , the same type of instability was also discussed in refs .
@xcite .
negative @xmath26 indicates the higgs mode is unstable and will decay .
the numerical results in ref .
@xcite shows that @xmath26 reaches its minimum at a momentum , i.e. , @xmath28 , which indicates that a stable state may develop around this minimum , we characterize this momentum as @xmath29 .
the inverse @xmath30 is the typical wavelength for the unstable mode .
if mixed phase can be formed , the typical size @xmath31 of the 2sc bubbles should be as great as @xmath30 , i.e. , @xmath32 , which turns out to be comparable to the coherence length of 2sc , @xmath33 .
considering that the coherence length @xmath34 of a superconductor is proportianl to the inverse of the gap magnitude , i.e. , @xmath35 , therefore , a rather large ratio of @xmath36 means a rather small ratio of @xmath37 .
when @xmath38 , a phase separation state is more favorable . in the system of imbalanced neutral atoms ,
the higgs instability persists and induces spatial non - uniform phase separation state .
this explains why imbalanced cold atom experiments did not observe loff state rather showed strong evidence of phase separation .
for the 2-flavor quark matter being considered , it was found in ref .
@xcite that , the electric coulomb interaction is not strong enough to compete with the higgs instability .
i reviewed the recent theoretical development for describing unconventional color superconductor .
this review is based on my own understanding , and i apologize if i missed some important work in this field . from my own point of view , i think with the increase of mismatch , it is very essential to consider the contribution from the phase fluctuation and the amplitude fluctuation of the order parameter .
the instability from the phase part of the order parameter induces the nambu - goldstone currents generation and forms the single plane - wave state , and the instability from the amplitude part indues the spatial inhomogeneity and forms the mixed state .
the true ground state of a mismatched pairing state should be determined by the competition of the instabilities .
i thank w. q. chao , i. giannakis , d.f .
hou , h.c .
ren , i. shovkovy , and p.f .
zhuang for collaboration .
the work is supported by the institute of high energy physics , chinese academy of sciences(cas ) , and cas key project under grant no .
kjcx3-syw - n2 .
10 huang m , zhuang p and chao w , 2003 _ phys .
_ d * 67 * 065015 ; shovkovy i and huang m 2003 _ phys .
_ b * 564 * 205 ; huang m and shovkovy i 2003 _ nucl . phys . _ a * 729 * 835 .
alford m , kouvaris c and rajagopal k 2004 phys .
lett . * 92 * 22200 ; 2005 _ phys .
_ d * 71 * 054009 .
huang m and shovkovy i 2004 _ phys .
_ d(r ) * 70 * 051501 ; 2004 _ phys .
_ d * 70 * 094030 .
casalbuoni r , gatto r , mannarelli m , nardulli g and ruggieri m 2005 _ phys .
_ b * 605 * 362 ; alford m and wang q 2005 _ j. phys . _ g * 31 * 719 .
sarma g 1963 _ j. phys .
solids _ * 24 * 1029 .
hong d _ preprint _ hep - ph/0506097 .
kryjevski a _ preprint _
hep - ph/0508180 ; schafer t 2006 _ phys .
lett . _ * 96 * , 012305 ; gerhold a and schafer t 2006 _ phys .
_ d * 73 * 125022 .
gorbar e , hashimoto m and miransky v 2006 _ phys .
_ b * 632 * 305 .
fulde p and ferrell r 1964 _ phys .
* 135 * a550 ; larkin a and ovchinnikov yu 1965 _ sov .
. jetp _ * 20 * 762 . zwierlein m , schirotzek a , schunck c and ketterle w 2006 _ science _ * 311 * 492 ; partridge g , li w , kamar r , liao y , and hulet r _ preprint _ cond - mat/0511752 ; zwierlein m , schunck c , schirotzek a , and ketterle w 2006 _ nature _ * 442 * 54 . | research charge neutral cold dense quark matter driven color superconductivity theory beyond bcs theory charge neutrality induces mismatch between fermi surfaces pairing quarks reduces phase space cooper pairing breaks cooper - pairing state superfluidity superconductivity with mismatched fermi momenta appears in systems electronic superconductor magnetic field asymmetric nuclear matter imbalanced cold atomic systems unsolved problem bcs superconductivity mismatch conventional bcs framework puzzles moderate mismatch homogeneous gapless superconducting phases formed exhibit anti - meissner screening effect chromomagnetic instability meissner effect standard bcs superconductivity report on theoretical development understanding puzzles g2sc phase gapless phase result under conventional bcs framework mean - field approximation framework order parameter of 2sc phase constant form @xmath0 free energy for @xmath1 quarks @xmath2-equilibrium form @xcite : @xmath3 sum over @xmath4 runs over 6 quark antiquark quasi - particles increase mismatch ground state in gapless 2sc phase when @xmath5 gapless phase is metastable sarma state @xcite free energy local maximum gap parameter @xmath6 @xmath7 in gapless phase g2sc phase exhibits anti - meissner screening effect chromomagnetic instability with meissner effect in standard bcs superconductivity anti - meissner effect report in 2sc phase color symmetry @xmath8 breaks to subgroup @xmath9 full order parameter characterized by @xmath10 ( 0\delta + h(x ) ) @xmath11 five nambu - goldstone diquarks phase fluctuation @xmath12 higgs field spatial fluctuation expanding around ground state : @xmath13 free energy expression as @xmath14 @xmath15 contribution from mean - field approximation @xmath16 @xmath17 contributions from fluctuation of nambu - goldstone currents and higgs field quadratic action of goldstone modes in long wavelength limit written down with meissner masses @xmath18 evaluated in ref.@xcite form of @xmath19 \left\vec\triangledown^a\right + higher ordersfound zero temperature increase mismatch five gluons @xmath20 broken generator @xmath21 meissner screening mass squares negative indicates generation nambu - goldstone currents @xmath22 gluon condensate state @xmath23 ng current state interpreted colored larkin - ovchinnikov - fulde - ferrel loff ) state single plane - wave order parameter @xmath24 loff state proposed 1960s describe possible ground state pair - breaking state electronic superconductor strong external magnetic field not confirmed in electronic superconduting systems after 40 years imbalanced cold atom systems experimental possibility understand cooper pairing destroyed absence orbital effects impurities promising search for loff state recent experiments show evidence formation loff state non - uniform state phase separation state missing understanding pairing breaking state observe lo)ff state in mismatch regime magnetic superfluid density instability develops free energy higgs field evaluated form @xmath25 self - energy higgs field @xmath26 becomes negative in gapless phase @xmath27 negative @xmath26 indicates higgs mode unstable decay numerical results ref@xcite shows @xmath26 reaches minimum @xmath28 stable state around minimum momentum as @xmath29 inverse @xmath30 typical wavelength unstable mode if mixed phase formed typical size @xmath31 of 2sc bubbles as great as @xmath30 @xmath32 comparable to coherence length 2sc @xmath33 coherence length @xmath34 superconductor proportianl to inverse gap magnitude @xmath35 large ratio of @xmath36 means small ratio @xmath37 @xmath38 phase separation state favorable imbalanced neutral atoms higgs instability persists induces non - uniform phase separation explains imbalanced cold atom experiments loff state showed strong evidence phase separation 2-flavor quark matter electric coulomb interaction not strong enough compete with higgs instability reviewed theoretical development unconventional color superconductor based on understanding increase of mismatch essential to consider contribution phase amplitude fluctuation of order parameter instability from phase induces nambu - goldstone currents forms single plane - wave state instability amplitude indues spatial inhomogeneity forms mixed statetrue ground state mismatched pairing determined by competition instabilities thank chao giannakis hou ren shovkovy zhuang collaboration work supported by institute high energy physics chinese academy sciences cas key project grant kjcx3-syw - n2 huang m zhuang p chao w 2003 phys . 67 * 065015 shovkovy huang m 564 * 205 729 * 835 alford m kouvaris c rajagopal k 2004 phys 92 * 22200 2005 71 * 054009 huang m shovkovy i 2004 70 * 051501 70 * 094030 casalbuoni r gatto r mannarelli m nardulli g ruggieri m 2005 605 * 362 alford m wang q 2005 31 * 719 sarma g 1963 24 * 1029 hong d hep - ph/0506097 kryjevski a ph/0508180 schafer t 2006 phys 96 * 012305 gerhold schafer t 73 * 125022 gorbar e hashimoto m miransky v 2006 632 * 305 ferrell 1964 phys 135 a550 larkin ovchinnikov 1965 sov 20 762 zwierlein schirotzek schunck c ketterle 2006 science 311 * 492 partridge li kamar r liao hulet r preprint cond mat/0511752 zwierlein schunck schirotzek ketterle 2006 nature 442 * 54 | 0.354535 | 2,334 | 1,222 | , Saving $0.1 in GPT-4. |
55,633 | one of the greatest challenges in cosmology today is to determine the nature of dark energy , the source of the observed present acceleration of the universe .
high precision experiments are being developed to reduce the uncertainties in the observations .
recently , we showed that the agreement to an accuracy of @xmath0 of measurements of the present density fluctuations @xmath1 , derived from galaxy distribution ( gd ) data and cosmic microwave background ( cmb ) anisotropies in the @xmath2cdm model , puts very strong limits on the possible decay of the vacuum energy into cold dark matter . using this agreement , combined with the evidence that the matter density @xmath3 and that the universe is approximately flat , we show that the vacuum metamorphosis model ( vmm ) and the popular brane - world model ( bwm ) , both used to explain dark energy , can be discarded .
when we relax the @xmath4 requirement , we find that an agreement within @xmath0 can be obtained only with @xmath5 for the vmm and @xmath6 for the bwm , both of which are not consistent with observations .
the agreement of the cmb and gd data and previous constraints from snia data exclude , or put strong limits on , other dark energy models , which have been suggested , that can be described by the parametrized equation of state ( eos ) @xmath7 , where @xmath8 and @xmath9 are constants , @xmath10 is the cosmological scale factor and @xmath11 is the pressure ( energy density ) of the dark energy .
we find that the supergravity ( sugra ) model with @xmath12 and @xmath13 can be discarded .
in general , we find best values @xmath14 with @xmath15 . for redshifts @xmath16 , where the supernova data is sensitive , @xmath17 for this parametrized eos . | many dark energy models have been suggested to explain the recent acceleration of the universe , first indicated by snia observations @xcite .
the nature of dark energy is one of the major problems in cosmology .
theories in which gravity is modified as well as parametrizations of the dark energy equation of state ( eos ) @xmath18 , where @xmath11 is the pressure ( energy density ) of the dark energy , have been suggested @xcite . based on observations ,
various constraints have been put on the eos for a variety of models ( e.g. , @xcite ) .
we begin by analyzing two interesting models that have been suggested for modifying gravity , which allow for a description of dark energy in terms of an effective eos .
the first is the five - dimensional brane - world model ( bwm ) of defayet et al .
@xcite , where gravity is modified by adding a five - dimensional einstein - hilbert action that dominates at distances which are greater than the crossover distance @xmath19 , where @xmath20 is the planck mass and @xmath21 is a five - dimensional planck mass .
the second model is the vacuum metamorphosis model ( vmm ) of parker and raval @xcite , which assumes the existence of a quantized non - interacting scalar field coupled to the ricci scalar curvature . in the vmm ,
the quantum vacuum undergoes a phase transition at a redshift @xmath22 : from a zero value at @xmath23 to a non - zero value for @xmath24 .
linder @xcite analyzed the linear growth of a density perturbation @xmath25 and the gravitational potential for both of these models , comparing them with the simplest linear parametrization of the dark energy eos as a function of @xmath10 : @xmath26 , where @xmath8 and @xmath9 are constants and @xmath10 is the cosmological scale factor .
measurements of the present density fluctuations @xmath1 , derived from the cosmic microwave background ( cmb ) anisotropies in the @xmath2cdm model , have been compared with those derived from the 2df galaxy redshift survey ( 2dfgrs ) @xcite .
it was found that their difference @xmath27 is no more than @xmath0 @xcite .
we recently showed that an agreement within @xmath0 ( i.e. , @xmath28 ) of the two sets of @xmath1 puts strong limits on a possible decay of the vacuum energy into cdm @xcite .
here we make a similar analysis to show that the bwm and the vmm as dark energy candidates can be discarded , in the face of the present evidence that @xmath29 , the matter density @xmath3 and that the universe is flat , indicated by recent cmb data .
we use the agreement to within @xmath0 of the present observed @xmath1 between the cmb and galaxy distribution ( gd ) data and the constraints from the gold snia data @xcite to restrict the parameters of the eos , @xmath26 .
in particular , we analyze a model suggested by supergravity ( sugra ) @xcite , studied by linder @xcite and recently by solevi et al .
@xcite ) . in @xmath30
ii , we discuss the effect of dark energy on the linear growth of @xmath25 in the models : 1 ) the vacuum energy decay into cdm ; 2 ) the bwm and vmm ; 3 ) dark energy models parametrized by @xmath26 ; and 4 ) the sugra model .
our conclusions are presented in section iii .
the nature of dark energy is still unknown and there are many alternative models to explain it .
one possibility is that instead of a constant vacuum energy , described by a cosmological constant , we have a vacuum energy which is decaying .
other possibilities are models that modify gravity and phenomenological models that parametrize the dark energy eos in the form @xmath31 , setting values for @xmath8 and @xmath9 . in this section , all of the above models will be analyzed .
for these models , the friedmann equation can be written in a general form in terms of an effective eos @xcite .
modeling the dark energy as an ideal fluid in a flat universe , we can write the friedmann equation as @xmath32}\ , , \label{friedeq}\ ] ] or @xmath33 where @xmath34 is the present value for the hubble parameter , @xmath4 is the present normalized matter density and @xmath35 depends on the phenomenological model @xcite .
the eos @xmath36 for the dark energy can be written as @xmath37 the linear growth of a density fluctuation , @xmath38 , depends on the eos .
we define the growth factor @xmath39 , where @xmath40 , the cosmological scale factor , and @xmath41 is normalized to unity at @xmath42 , the recombination epoch . in terms of @xmath41
, we have @xmath43\frac{g^{\prime}(a)}{a}+ \frac{3}{2}\frac{1-w(a)}{1+x(a)}\frac{g(a)}{a^2}=0\ , , \label{ggrowth}\ ] ] where @xmath44 is defined as @xmath45 the linear mass power spectrum is proportional to @xmath46 and we define the deviation from the standard @xmath2cdm model by @xmath47 where @xmath48 is the density fluctuation in the standard @xmath2cdm model . since @xmath46 derived from gd data differs from @xmath48 derived from the cmb anisotropies by no more than 10 per cent , the maximum value of @xmath49 is @xmath29 .
we apply the above description to the following models used to explain the recent acceleration of the universe , suggested in the literature : a phenomenological vacuum energy decay model ; two models in which gravity is modified , bwm and vmm ; and the sugra model . in a previous paper ,
we put limits on the rate of a possible decay of the vacuum energy ( i.e. , the decay of the cosmological constant ) into cdm from the observed agreement , to within @xmath0 , of the @xmath1 derived from the cmb and the galaxy survey data @xcite .
let us consider the vacuum energy decay model described , in a flat universe , by a power law dependence @xmath50 where @xmath51 . from conservation of energy and eq.([mainl ] ) , we have @xmath52\ , , \,\label{rhopeeb}\ ] ] where @xmath53 .
eq.([rhopeeb ] ) modifies the friedmann equation [ eq.([igno ] ) ] by a factor @xmath54 \label{delta1}\,.\ ] ] comparing the vacuum energy decay model with the @xmath2cdm model , the deviation @xmath55 is shown in table [ tabvdec ] , assuming @xmath56 , the observed value .
it is to be noted that @xmath57 occurs only for @xmath58 .
.the deviation @xmath55 of the vacuum energy decay model from the @xmath2cdm model as a function of the power @xmath59 , assuming @xmath56 . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] the sugra model @xcite is an attractive model to possibly explain the acceleration of the universe .
this model can be described by the eos of @xmath30 ii - c with @xmath12 and @xmath13 @xcite .
this equation of state is in agreement with observations for the low redshift snia dataset @xcite and gd data @xcite .
$ ] , for the @xmath2cdm model ( top curve ) and the sugra model ( bottom curve ) with the matter density @xmath60 ( dashed lines ) . ]
[ sugrafig ] shows that the growth of @xmath61 is smaller for the sugra model than for the @xmath2cdm model .
the @xmath49 for the sugra model is @xmath62 for @xmath63 , which is appreciably greater than the maximum allowed value @xmath29 .
we calculate the value of @xmath49 numerically for well - known dark energy models from the growth equation for @xmath61 . from observations ,
the maximum value of @xmath49 is @xmath28 . a @xmath2-decay into cdm model , described by a power law dependence ( studied in our previous paper @xcite ) ,
was first considered .
it was found that the factor @xmath49 increases as the exponent @xmath59 increases .
the maximum possible value of @xmath59 was found to be ( @xmath64 ) .
we combined the constraints from the gold snia data @xcite and the condition that @xmath28 to restrict the values of the parameters of the linear eos for dark energy , @xmath26 .
it was found that the best fit values of @xmath8 and @xmath9 are @xmath14 with @xmath66 . for @xmath67 , where the supernova data is sensitive , @xmath17 for this parametrized eos .
finally , we also analyzed the sugra model for the above parametrized eos with @xmath12 and @xmath13 .
@xmath49 was found to be very large : @xmath62 for @xmath63 , which is appreciably greater than the maximum value @xmath28 .
* acknowledgments . *
thanks the brazilian agencies fapesp ( grant 00/06770 - 2 ) and cnpq ( grant 300414/82 - 0 ) for partial support
. a.m.p thanks fapesp for financial support ( grants 03/04516 - 0 and 00/06770 - 2 ) . | dark energy models suggested explain recent acceleration universe indicated by snia observations @xcite nature dark energy major in cosmology theories gravity modified parametrizations dark energy equation state eos pressure density energy suggested @xcite observations constraints on eos for models @xcite analyzing two models for modifying gravity description dark energy effective eos first five - dimensional brane - world model of defayet et al gravity modified adding five - dimensional einstein - hilbert action at distances greater than crossover distance @xmath19 planck mass five - dimensional planck mass second model vacuum metamorphosis model of parker and raval assumes quantized non - interacting scalar field ricci scalar curvature quantum vacuum phase transition at redshift from zero to non - zero @xmath24 analyzed linear growth of density perturbation @xmath25 gravitational potential for models with simplest linear parametrization of dark energy eos as function of @xmath10 : @xmath26 @xmath8 constants cosmological scale factorpresent density fluctuations @xmath1 from cosmic microwave background cmb anisotropies in @xmath2cdm model compared with 2df galaxy redshift survey 2dfgrs ) @xcite difference @xmath27 no more than @xmath0 @xcite agreement within @xmath0 @xmath28 ) of two sets @xmath1 decay vacuum energy into cdm similar analysis bwm vmm as dark energy candidates be discarded evidence @xmath29 matter density @xmath3 universe flat cmb data agreement within @xmath0 @xmath1 between cmb galaxy distribution data constraints from gold snia data to restrict parameters eos @xmath26 analyze model by supergravity ( sugra ) @xcite studied by linder solevi et al . @xmath30 ii effect of dark energy on linear growth @xmath25 models vacuum energy decay into cdm bwm vmm dark energy models parametrized by @xmath26 sugra model conclusions presented in section iii nature of dark energy unknown alternative models possibility constant vacuum energy decayingpossibilities models gravity parametrize dark energy eos @xmath31 setting values for @xmath8 @xmath9 models analyzed friedmann equation written effective eos @xcite modeling dark energy ideal fluid flat universe equation as @xmath32 @xmath33 @xmath34 present value hubble parameter @xmath4 present normalized matter density @xmath35 depends on model @xcite eos @xmath36 dark energy written as @xmath37 linear growth density fluctuation @xmath38 depends on eos define growth factor @xmath39 @xmath40 cosmological scale factor @xmath41 normalized to unity at @xmath42 recombination epoch @xmath41 @xmath43 @xmath44 defined as @xmath45 linear mass power spectrum proportional to @xmath46 deviation from standard @xmath2cdm model by @xmath47 @xmath48 density fluctuation standard @xmath2cdm model @xmath46 differs from @xmath48 10 per cent maximum value of @xmath49 is @xmath29apply description to models explain recent acceleration universe phenomenological vacuum energy decay model two models gravity modified bwm vmm sugra model previous paper limits on rate possible decay vacuum energy cosmological into cdm from observed agreement @xmath0 @xmath1 from cmb galaxy survey data @xcite consider vacuum energy decay model described flat universe by power law @xmath50 @xmath51 conservation energy @xmath52 @xmath53 modifies friedmann equation by factor @xmath54 comparing vacuum energy decay model with @xmath2cdm model deviation @xmath55 shown in table assuming @xmath56 observed value @xmath57 occurs only for @xmath58 deviation @xmath55 vacuum energy decay model from @xmath2cdm model function of power @xmath59 assuming @xmath56 sugra model @xcite attractive model explain acceleration universe model described by eos @xmath30 ii - c with @xmath12 and @xmath13 @xciteequation of state with low redshift snia dataset @xcite and gd data @xmath2cdm model sugra model matter density @xmath60 growth of @xmath61 smaller for sugra model than @xmath2cdm model @xmath49 for sugra is @xmath62 for @xmath63 greater than maximum value @xmath29 calculate value @xmath49 for dark energy models from growth equation @xmath61 maximum value @xmath49 is @xmath28 @xmath2-decay into cdm model power law dependence factor @xmath49 increases as exponent @xmath59 increases maximum value of @xmath59 @xmath64 ) combined constraints from gold snia data condition @xmath28 restrict values parameters linear eos for dark energy best fit values of @xmath8 and @xmath9 are @xmath14 with @xmath66 @xmath67 @xmath17 for parametrized eos analyzed sugra model for parametrized eos with @xmath12 and @xmath13 @xmath49 large : @xmath62 for @xmath63 greater than maximum value @xmath28 thanks brazilian agencies fapesp cnpq for supportthanks financial support grants 03/04516 00/06770 2 | 0.347016 | 2,171 | 1,104 | , Saving $0.1 in GPT-4. |
55,650 | the results from a large - field far - infrared ( fir ) and sub - millimeter ( sub - mm ) survey of our neighbor galaxy m31 are presented .
we have obtained _ herschel _ images of a @xmath0 degree area centered on andromeda . using 21 cm atomic hydrogen maps , we are able to disentangle genuine emission from m31 from that for foreground galactic cirrus , allowing us to recognize dusty structures out to @xmath1 kpc from the center .
we first characterize the fir and sub - mm morphology and then , by de - projecting _ herschel _ maps and running an _ ad hoc _ source extraction algorithm , we reconstruct the intrinsic morphology and the spatial distribution of the molecular complexes .
finally , we study the spatially resolved properties of the dust ( temperature , emissivity , mass , etc . ) , by means of a pixel by pixel sed fitting approach . | we have observed the local galaxy andromeda ( m31 ) , the nearest spiral galaxy in the local group , with _
herschel_. simultaneous observations were conducted in five bands from 100 to 500 @xmath2 m , of a @xmath3 degree area @xcite .
we study , for the first time at these wavebands , the characteristics of the extended dust emission , focusing on larger scales than ever done before for m31 .
furthermore , thanks to the spatial resolution that the instruments onboard _ herschel _ can reach , we are able to study the geometrical distribution and the properties of dust on a sub - kpc scale . here , we summarize the results from a series of papers whose goal is the study of the morphology of the interstellar medium in andromeda , with a particular focus on the physical properties of dust as a function of position and galactocentric distance .
in the left hand panel of fig . [ fig : fig1 ] , we show a view of m31 at 250 @xmath2 m .
this wavelength gives the best compromise between sensitivity and spatial resolution .
we have tried to identify structures belonging to m31 out to the largest distances , but this task turned out to be quite challenging due to the presence of a substantial amount of galactic dust sitting on the line of sight between us and andromeda .
we have used the 21 cm hydrogen maps from @xcite and @xcite to disentangle the dust emission coming from m31 from that of the galactic cirrus .
we exploit the physical link between gas and dust , and the fact that hi observations carry also the information on the gas velocity : this is used to tell which gas ( and hence dust ) component belongs to m31 and which one to the milky way . in this way
, we manage to remove the galactic cirrus emission from our data even though the north - east portion of our maps still remains quite confused , retaining some contamination .
this only marginally affects the outskirts of m31 , at _ herschel _ wavebands .
[ cols= " < , < " , ] we first looked for the presence of dust at large distances from m31 at the high - velocity clouds location @xcite .
a quick check of the most massive one , davis cloud , was enough to make us realize that our _ herschel _ data are not sensitive enough for the detection of a possible dust component here .
hence , we concentrated our search in the disk outskirts , specifically in the south
west region : the fact that there is no cirrus contamination makes the dust detection here quite clear , out to galactocentric distances of @xmath1 kpc .
three ring - like structures are revealed , at distances of 21 ( e ) , 26 ( f ) and 31 ( g ) kpc , respectively , from the nucleus .
even though quite faint , their detection is confirmed by the perfect coincidence with similar features observed in the 0-moment 21 cm map from @xcite .
we have tried to check whether these features have a counterpart , at similar distances , on the opposite side of the galaxy . due to the high degree of confusion of the north
east region , we are not able to tell whether this is the case or not , with the only exception of structure `` e '' , the closest one , even though impossible to follow along all the disk of the galaxy .
these newly discovered structures add to the already well known morphological features of m31 , such as the 10 kpc star forming ring and the 15 kpc ring ( already known for a while ; see e.g. * ? ? ?
* ) and the more recently discovered 1.5 kpc , inner ring @xcite .
overall , we find evidence of a recurrent pattern in the morphological features of the dust in m31 : rings , or more in general , arc - like structures are found at distances of 5 kpc each .
if we also take into account a structure defined by @xcite as an arm - crossing region , located at @xmath4 kpc from the center , we note that m31 has structures at radii of 5 , 10 , 15 , 21 , 26 and 31 kpc .
while the presence of these structures was never investigated before , the appearance of multiple rings in a galaxy was argued to be compatible with dissipative cloud collisions and weak bar perturbation , which would give rise to resonances , due to gravitational torques @xcite .
we adopted a different , more systematic approach to the study of the ism morphology in @xcite , where we de - projected our _ herschel _ images and applied csar @xcite , a source extraction algorithm suited for taking into account the hierarchical tree which builds up the arms and disks structure of the ism in m31 .
this yields a catalog of point sources which correspond to giant molecular clouds ( gmcs ) . following a similar approach as in @xcite
, we fit 2 rings to the positions of the molecular clouds , ending up with a best fit radius of @xmath5 and @xmath6 kpc , respectively , reproducing the 10 and 15 kpc rings ( see sect . 2 ) .
we then fit 2 logarithmic spiral arms to the remaining points , letting the pitch angle as a free parameters .
such spiral arms are indeed suited to reproduce the gmcs dislocation , and they turn to have an angle of @xmath7 . consistently with the findings by @xcite , the best fit models for these two rings are displaced with respect to the optical centre of m31 .
it was argued that this displacement might be the result of a head on collision with andromeda s compact satellite m32 , as dynamical simulations from the latter work tend indeed to confirm .
in @xcite we took advantage of _ herschel _ high resolution data , and used pacs and spire images to build the fir spatially - resolved sed of andromeda .
we fit a modified black
body function to the sed of each of the pixels , limiting ourselves to those having a flux above @xmath8 in all bands . despite this quite severe requirement ,
we are left with @xmath9 pixels .
our fitting procedure considers 3 free parameters : the dust emissivity index @xmath10 , the dust mass and its temperature ( t@xmath11 ) . like this
, we assume that dust emits at one single temperature , while it is more likely that a range of temperatures , in the same position , would be a more realistic description . in spite of this simplification ,
we find a @xmath12 value less than 2.73 for more than 90% of the pixels , showing that our model is an adequate description of the dust sed at a pixel scale .
the result of the fits are then be used to trace the properties of dust on a physical scale of about 140 pc , as a function of the location .
interestingly , we find the highest values of t@xmath11 ( @xmath13 k ) inside the bulge , which is dominated by old stars , while in the 10 kpc ring , where star formation is currently occurring , the temperature is constant , about 17 k only . furthermore ,
if the radial profile is considered , we notice that the temperature monotonically decreases from the center out to @xmath14 kpc , and then it assumes a slightly positive slope .
this trend is mirrored by that of the emissivity index @xmath10 which increases out to the same distance , and then it decrease again towards the outskirts of the galaxy . as there is a
well known degeneracy between dust temperature and the @xmath10 parameter , we used montecarlo simulations to check whether these trends are spurious effect caused by this , and found that this is not the case .
the fact that the dust temperature is much higher in the bulge region , which is dominated by old stellar stars and where star formation activity is minimal if not absent at all , while it is significantly lower in the star forming ring , rises a question about the dust heating sources and mechanism .
it had already been argued that dust in some nearby galaxies and in m31 itself , might be significantly heated by a old stellar population .
we investigated this , and we found a linear dependency of t@xmath11 on the 3.6 @xmath2 m emission in the inner 3.1 kpc region ( fig.[fig : fig1 ] , right panel ) . as this band traces very well the presence of old stars , this correlation suggests that indeed the old bulge population is the one which is responsible for the dust heating in the central region .
we argue that different values of @xmath10 could maybe be due to differences in the composition of the dust , or substantial changes in the radiation field leading to changes in the grain size ( from sputtering ) or mantle loss .
we have used _ herschel _ pacs and spire observations to probe the dust properties and morphology in the nearby galaxy m31 .
we performed analysis both on a global and on a spatially resolved scale .
we can summarize our results as follows : 1 .
m31 s ir morphology is dominated by 2 rings and 2 spiral arms ; 2 .
there is a recurrent radial pattern in the presence of structures : arc - like features are detected at intervals of @xmath4 kpc , out to a galactocentric distance of @xmath1 kpc ; 3 .
a variable emissivity index @xmath10 is required to fit the spatially resolved ir sed ; 4 .
there is a transition region at a distance of @xmath15 kpc , marking different dust properties , or witnessing a change in the radiation field ; 5 .
t@xmath11 is the highest in the bulge ( 30 k ) , only reaching 17 k in the starforming ring ; 6 .
the strong correlation between the 3.6 @xmath2 m flux and t@xmath11 in the inner regions strongly suggest that old stars in the bulge are the responsible for these `` high '' temperatures .
baade , w. , & gaposchkin , c. h. p. 1963
, evolution of stars and galaxies .
block , d. l. , bournaud , f. , combes , f. , et al .
2006 , , 443 , 832 braun , r. , thilker , d. a. , walterbos , r. a. m. , & corbelli , e. 2009 , , 695 , 937 combes , f. 1988 , nato asic proc . 232 : galactic and extragalactic star formation , 475 fritz , j. , gentile , g. , smith , m. w. l. , et al . 2012 , , 546 , a34 gordon , k. d. , bailin , j. , engelbracht , c. w. , et al .
2006 , , 638 , l87 haas , m. , lemke , d. , stickel , m. , et al .
1998 , , 338 , l33 jungwiert , b. , & palous , j. 1996 , , 311 , 397 kirk , j. m. , ward - thompson , d. , palmeirim , p. , et al .
2013 , , 432 , 1424 kirk , j. m. , gear , w. k. , fritz , j. , et al . 2013 ,
arxiv:1306.2913 smith , m. w. l. , eales , s. a. , gomez , h. l. , et al .
2012 , , 756 , 40 thilker , d. a. , braun , r. , walterbos , r. a. m. , et al .
2004 , , 601 , l39 | observed local galaxy andromeda ( m31 ) nearest spiral galaxy with _ herschel_ observations conducted in five bands from 100 to 500 @xmath2 m @xmath3 degree area @xcite study characteristics extended dust emission larger scales for m31 thanks to spatial resolution instruments herschel study geometrical distribution properties of dust on sub - kpc scale summarize results from papers study morphology of interstellar medium in andromeda focus on physical properties of dust position galactocentric distance fig view of m31 at 250 @xmath2 m wavelength compromise between sensitivity spatial resolution tried to identify structures to m31 largest distances challenging substantial galactic dust on line of sight between us andromeda used 21 cm hydrogen maps from @xcite to disentangle dust emission from m31 from galactic cirrus exploit physical link between gas dust observations carry information on gas velocity gas dust component to m31 milky way remove galactic cirrus emission from data north - east portion maps confused retaining contaminationmarginally affects outskirts m31 at _ herschel _ wavebands looked for dust at large distances from m31 at high - velocity clouds location @xcite quick check davis cloud _ herschel data not sensitive for detection possible dust component concentrated search in disk outskirts south west region no cirrus contamination makes dust detection clear to galactocentric distances of @xmath1 kpc three ring - like structures revealed at distances 21 26 f 31 ( g ) kpc from nucleus detection confirmed by coincidence with similar features in 0-moment 21 cm map from @xcite tried to check features counterpart at similar distances opposite side galaxy high confusion north east region not tell exception of structure `` e '' closest impossible follow newly discovered structures add to morphological features m31 10 kpc star forming ring 15 kpc ring 1.5 kpc inner ring @xcite evidence recurrent pattern in morphological features dust in m31 rings arc - like structures found at distances 5 kpc each structure @xcite arm - crossing region at @xmath4 kpc from center m31 has structures at radii 5 10 15 21 26 31 kpc presence structures multiple rings compatible with dissipative cloud collisions weak bar perturbation resonances due to gravitational torques adopted systematic approach ism morphology de - projected herschel images applied csar @xcite source extraction algorithm for hierarchical tree arms disks structure ism in m31 yields catalog of point sources to giant molecular clouds fit 2 rings to positions molecular clouds best fit radius of @xmath5 and @xmath6 kpc reproducing 10 and 15 kpc rings fit 2 logarithmic spiral arms to remaining points pitch angle free parameters spiral arms reproduce gmcs dislocation angle of @xmath7 best fit models for two rings displaced optical centre of m31 displacement might result of collision with andromeda s compact satellite m32 dynamical simulations confirm @xcite _ herschel high resolution data used pacs spire images to build fir spatially - resolved sed of andromedafit modified black body function to sed pixels limiting to flux above @xmath8 in all bands left with @xmath9 pixels fitting procedure considers 3 parameters dust emissivity index @xmath10 dust mass temperature ( t@xmath11 ) assume dust emits at one single temperature likely range of temperatures same position realistic description @xmath12 value less than 2.73 for 90% pixels model adequate description dust sed at pixel scale result fits used to trace properties dust on physical scale about 140 pc function location highest values of t@xmath11 ( @xmath13 k ) inside bulge dominated by old stars in 10 kpc ring star formation occurring temperature constant about 17 k radial profile temperature decreases from center to @xmath14 kpc assumes slightly positive slope trend mirrored by emissivity index @xmath10 increases same distance towards outskirts galaxy degeneracy between dust temperature @xmath10 parameter used montecarlo simulations to check trends not casedust temperature higher in bulge region dominated by old stellar stars star formation activity minimal lower in star forming ring question about dust heating sources mechanism argued dust in galaxies m31 might be heated by old stellar population found linear dependency of t@xmath11 on 3.6 @xmath2 m emission in inner 3.1 kpc region fig old stars correlation suggests old bulge population responsible for dust heating central region different values of @xmath10 due to differences in composition dust changes in radiation field grain size mantle loss used herschel pacs spire observations dust properties morphology in galaxy m31 analysis global spatially resolved scale results m31 s ir morphology dominated by 2 rings 2 spiral arms recurrent radial pattern in presence structures arc - like features at intervals of @xmath4 kpc distance @xmath1 kpc variable emissivity index @xmath10 required to fit spatially resolved ir sed transition region at @xmath15 kpc different dust properties change in radiation field t@xmath11 highest in bulge ( 30 k ) 17 k in starforming ringcorrelation 3.6 @xmath2 m flux t@xmath11 inner regions suggest old stars bulge responsible high temperatures baade gaposchkin 1963 evolution stars galaxies block d. bournaud combes 2006 443 , 832 braun thilker walterbos corbelli 2009 , 937 combes 1988 nato asic proc . 232 galactic extragalactic star formation 475 fritz j. gentile smith w l.. 2012 546 , a34 gordon bailin engelbracht 2006 638 , l87 haas lemke stickel 1998 338 , l33 jungwiert palous j 1996 311 , 397 kirk ward - thompson palmeirim 2013 432 , 1424 kirk gear fritz j. 2013 arxiv:1306.2913 smith w. l. eales gomez h. 2012 756 , 40 thilker braun walterbos r a. m. 2004 601 l39 | 0.442533 | 2,587 | 1,318 | , 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.