chunk_uid
stringlengths
40
40
chunk_type
stringclasses
2 values
chunk_index
int64
0
6.71k
total_chunks
int64
1
6.71k
section_title
stringlengths
1
157
embed_text
stringlengths
1
83.3k
spans
dict
paper_doi
stringlengths
0
63
paper_id_arxiv
stringlengths
9
16
title
stringlengths
7
245
authors
listlengths
1
768
categories
listlengths
1
7
year
int64
2k
2.02k
language
stringclasses
2 values
discipline
stringclasses
8 values
dense_vector
listlengths
1.02k
1.02k
a2bdc816e98acacfa9c628b80ac2c6ac9fe3f2e7
subsection
46
92
Rates and Domains of Convergence
Lastly,the final row shows the large q\gg 1 leading order approximationsof \alpha _{c} and Q^{*}.]In order to asses the applicability of SCE for anharmonicities with q\ne 4, we list the values of the critical \alpha _{c} and the optimal \alpha ^{*} for the first few integer perturbation powers q>2 in REF . \alpha _{c} ...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ 0.0067984312772750854, 0.041202612221241, -0.040195439010858536, -0.006325364112854004, -0.0109186926856637, -0.003099733730778098, 0.03451863303780556, 0.026125509291887283, 0.03549528867006302, 0.039951276034116745, -0.007149416487663984, 0.016893072053790092, -0.03158866986632347, 0.028...
43e01ad36375afcdffab3c80ba16a6b0f74a7310
subsection
47
92
Rates and Domains of Convergence
However, it should be noted that these bounds are not particularly tight as compared with the numerically fitted values (as demonstrated in for q=4; additional numerical results are listed in the table). Furthermore, they also do not reflect the additional convergent factor of \left(1-1/G\right)^{N}, giving a stretched...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.05525742843747139, 0.05791198089718819, -0.02119065821170807, -0.02120591513812542, -0.00689192209392786, -0.007685237098485231, 0.042747464030981064, 0.03307512402534485, -0.01687319576740265, 0.011129290796816349, -0.05010088160634041, 0.009153632447123528, -0.03392946347594261, 0.004...
f4db80e2a2cf2d3cbbaab5a03b6a0abb84349b03
subsection
48
92
SCE in the Complex Plane: Oscillatory Integrals and Stokes Phenomenon
Following the success of the SCE in treating the anharmonic oscillator, we wish to elucidate other properties of this technique. We would like to explore how the SCE carries over to complex functions, and in particular, oscillatory integrals. Using the results of the previous section for q=3, we will treat the Airy fun...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1142/p345", "end": 463, "openalex_id": "https://openalex.org/W1552566358", "raw": "O. Vallée and M. Soares, Airy Functions and Applications to Physics (Imperial College Press, London, 2010).", "source_ref_id": "1f105dd802fca08d5714...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.013920068740844727, 0.026573289185762405, -0.047804445028305054, -0.0005456606158986688, 0.03293805569410324, -0.037059132009744644, 0.002991593675687909, -0.01604165881872177, 0.002054812852293253, 0.019903255626559258, -0.005857265554368496, 0.026756448671221733, -0.018651671707630157, ...
850912496192738e005bb0e22f4680aec90c8698
subsection
49
92
The SCE of
The Airy function can be put into the integral representation \mathrm {Ai}\left(z\right) & =\frac{e^{-\frac{2}{3}z^{3/2}}}{\pi }\int _{0}^{\infty }e^{-z^{1/2}t^{2}}\cos \left(\frac{t^{3}}{3}\right)dt\sim \frac{e^{-\frac{2}{3}z^{3/2}}}{\pi z^{\frac{1}{4}}}\int _{-\infty e^{\frac{1}{4}\arg z}}^{\infty e^{\frac{1}{4}\arg ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 399, "openalex_id": "https://openalex.org/W3089116201", "raw": "DLMF, “NIST Digital Library of Mathematical Functions,” http://dlmf.nist.gov/.", "source_ref_id": "1bad14f3312569fc1768be8ccfc8645c155416c3", "start": 0 ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.023677121847867966, 0.05302942916750908, -0.03450879454612732, 0.008848409168422222, 0.03707178309559822, -0.0019279614789411426, 0.02387544885277748, -0.01978687383234501, 0.003297176444903016, 0.04177059233188629, -0.026041783392429352, -0.0253705233335495, -0.013211590237915516, -0.0...
90ce36059837790fcd38000905fcf307b0772b77
subsection
50
92
The SCE of
As we saw in the case of the anharmonic oscillator, it was this independence that allowed us to demonstrate uniform and exponential convergence. A lesson that is learned from this is that one must take care not to introduce artificial symmetries into the problem when applying the SCE: Originally, the relation between t...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.052843790501356125, 0.04289744049310684, -0.02012152411043644, 0.025338780134916306, 0.014995798468589783, -0.005533800460398197, 0.032523948699235916, 0.005358366295695305, 0.008344558998942375, 0.06050186604261398, -0.023645460605621338, -0.005747372284531593, -0.035391915589571, -0.0...
7dd791ce7e199145d7d0ab3f6f69aaade337dd34
subsection
51
92
The SCE of
The integrand e^{-\left(t^{\prime }\right)^{2}} is entire and decays to zero at infinity for \left|\arg t^{\prime }\right|<\frac{\pi }{4}, and since we assumed \left|\arg G_{\Delta }\right|<\pi , we may deform the integration path to again run over the positive real axis. This leads to\tilde{\mathrm {Ai}}_{\Delta }^{\l...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.016629301011562347, 0.015263867564499378, -0.029047880321741104, -0.013074597343802452, 0.008352790959179401, 0.016202125698328018, 0.02462357096374035, 0.009046949446201324, 0.013318696990609169, 0.03392987698316574, -0.036157287657260895, -0.03170246630907059, 0.007086523342877626, -0...
e8a42cced32d07a52c9e3e50b278066d5d316e80
subsection
52
92
The SCE of
In other words, we treat the SCE for the Airy function as the sum of two separate SCEs, whose combined numeric value gives \mathrm {Ai}\left(z\right). The condition for each G_{\Delta } is now0 & =M\left[1-\left(3i\Delta G_{\Delta }^{\frac{3}{4}}\left(1-\frac{\sqrt{z}}{\sqrt{G_{\Delta }}}\right)\right)^{-1}\frac{C_{3}\...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.03965122252702713, 0.021367095410823822, -0.044992994517087936, -0.044779326766729355, 0.028250351548194885, 0.02527422085404396, 0.007028247695416212, 0.021229734644293785, 0.010660653933882713, 0.018452012911438942, -0.03727031871676445, 0.001926854019984603, -0.005421900190412998, 0....
ca089f19f3fe9bdb91ea8632ad28dceb1170a365
subsection
53
92
The SCE of
This finally yields\tilde{\mathrm {Ai}}_{\Delta }^{\left(N\right)}\left(z\right) & =\frac{1}{2}\left(\frac{z}{G_{\Delta }}\right)^{\frac{1}{4}}\sum _{n=0}^{N}\frac{1}{n!}\left(1-\frac{\sqrt{z}}{\sqrt{G_{\Delta }}}\right)^{n}\sum _{l=0}^{n}\binom{n}{l}\left(-\frac{C_{3}\left(M\right)}{M}\right)^{-l}\Gamma \left(\frac{2n...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04731318727135658, 0.022389821708202362, -0.020497294142842293, -0.05433385446667671, -0.0013640697579830885, 0.018391093239188194, 0.020482031628489494, 0.021184097975492477, 0.019520506262779236, 0.015323367901146412, -0.05512749403715134, 0.011408583261072636, -0.01990206353366375, 0...
701f58ccb9c8f13bfd05a1ae2c1c99f825bb5682
subsection
54
92
The SCE of
For general z, one finds that G_{-}^{\frac{1}{4}}\left(z\right)=\left(G_{+}^{\frac{1}{4}}\left(z^{*}\right)\right)^{*}, which quickly leads to the conclusion that the expansion satisfies \tilde{\mathrm {Ai}}^{\left(N\right)}\left(z^{*}\right)=\left[\tilde{\mathrm {Ai}}^{\left(N\right)}\left(z\right)\right]^{*} for all ...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.03960714116692543, 0.03856966644525528, -0.01048155128955841, 0.004878422245383263, 0.0033489090856164694, -0.022092118859291077, 0.041956719011068344, 0.010008584707975388, 0.006694003473967314, 0.014425482600927353, -0.011519026011228561, -0.011618196964263916, -0.014394968748092651, ...
6715ec1987fa8031bef61c1dc4775a014602f6b9
subsection
55
92
Analytic Properties and Solutions for
Recall that each of the Airy SCEs \tilde{\mathrm {Ai}}_{\pm }\left(z\right) is in essence a complex extension of the partition function of an anharmonic oscillator with an x^{3} perturbation, as analyzed in . In particular, this implies that the coefficients of the SCE expansion, when viewed as power series in \left(1-...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1450, "openalex_id": "https://openalex.org/W1965926499", "raw": "G. G. Stokes, Trans. Camb. Phil. Soc. 10, 105 (1864).", "source_ref_id": "971a90dc504e06ef2838fe1f5e99aa972f0967a0", "start": 1077 }, { "ar...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.030288323760032654, 0.01279433723539114, -0.048308733850717545, -0.02329988405108452, 0.026870397850871086, -0.005950854625552893, -0.0020484672859311104, -0.01071916799992323, 0.01959204487502575, 0.05493096634745598, -0.030654530972242355, 0.024886779487133026, -0.02206393890082836, 0...
dd78ae7507f021135691ed53924b91720057a398
subsection
56
92
Analytic Properties and Solutions for
Thus, the lines of phase \arg z=0,\pm \frac{\pi }{3},\pm \frac{2\pi }{3} and \pi , which are called Stokes linesWe will not draw the distinction between Stokes and anti-Stokes lines. define six different wedges in the complex plane with three distinct behaviors of \mathrm {Ai}\left(z\right) — exponential decay, growth,...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ 0.0003614506567828357, -0.02052352949976921, -0.03936855494976044, -0.018982358276844025, 0.021164413541555405, -0.040436696261167526, 0.023499060422182083, -0.019073912873864174, 0.003788079135119915, 0.009086811915040016, -0.03805626928806305, 0.0035095999483019114, -0.015488017350435257, ...
6969d36bcf8b7b59cf067d0e0ee4bed3d9a5e207
subsection
57
92
Analytic Properties and Solutions for
Note that as M\rightarrow \infty this root tends to the principal root of \@root 3 \of {C_{3}\left(M\right)/3iM}. However, this property is suddenly violated once we cross the Stokes line at \arg z=\frac{2\pi }{3}. To see this, we substitute y=rz^{\frac{1}{4}} into () to obtain3\Delta ir(r^{2}-1)z^{\frac{3}{4}}=C_{3}\...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.015265177004039288, 0.007747020106762648, -0.02862316183745861, -0.017134226858615875, -0.003398618893697858, -0.03124745935201645, 0.018095452338457108, -0.0029714074917137623, 0.024198472499847412, -0.009520710445940495, -0.061243798583745956, 0.005839826539158821, -0.013167264871299267...
32cc20c10d95a4632a66ee16cd8c23400c8cd5e8
subsection
58
92
Analytic Properties and Solutions for
If instead \delta \varphi >0, then our privileged root scatters from r=1 up to r\propto e^{+\frac{i\pi }{3}} and G_{+}^{\frac{1}{4}} tends to the line e^{+\frac{i\pi }{3}+\frac{1}{4}\frac{2\pi i}{3}}=e^{+\frac{i\pi }{2}}. The same logic applies for the root G_{-}, with all the signs of the arguments above negated.We no...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04503119736909866, 0.019571484997868538, -0.009816251695156097, 0.00690646143630147, 0.02105116844177246, -0.04603799059987068, 0.004618290811777115, 0.019266396760940552, -0.00022702942078467458, 0.04124808683991432, -0.03398695960640907, 0.004900498315691948, -0.04439050704240799, -0....
e0ea8b58571a76762810e284e6df78d238dada49
subsection
59
92
Analytic Properties and Solutions for
REF and REF , respectively.Recalling that C_{3}\left(M\right)/3M\sim \sqrt{M}, it asymptotically behaves as\left(1-\frac{\sqrt{z}}{\sqrt{G_{\Delta }}}\right)^{N}\sim \left(1-\frac{\sqrt{z}}{M^{\frac{1}{3}}3^{-\frac{2}{3}}e^{-\frac{\Delta }{3}\pi i}}\right)^{N}\sim e^{-\left(\frac{9}{\alpha }\right)^{\frac{1}{3}}e^{\fra...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.02730037458240986, 0.005344861187040806, -0.04193484038114548, -0.04315565153956413, -0.003196997568011284, 0.010056426748633385, -0.00412786565721035, 0.0031321418937295675, 0.028978990390896797, 0.01680140569806099, -0.046940166503190994, 0.007576655130833387, -0.005184629932045937, 0...
0fb3dfcaabaa3e6947a40635cd6e9b171e2d9e9f
subsection
60
92
Analytic Properties and Solutions for
However, the actual exponential convergence rate is roughly 10^{-0.5N}, so the scaling of the cusp is closer to 10^{-0.20N}, or 10^{-0.27\left|z\right|^{\frac{3}{2}}} if we substitute N=M/\alpha , with M given approximately by (), and \alpha =\alpha ^{*}\approx 1. This implies that the exponential convergence is always...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.047300051897764206, 0.011886045336723328, -0.03875552862882614, -0.009635478258132935, 0.010360237210988998, -0.02093408815562725, -0.019164150580763817, -0.007457387167960405, 0.011283351108431816, 0.033750876784324646, -0.013747530989348888, 0.02558780275285244, 0.002395518822595477, ...
d720745cbd0b72a0b86e08cad9eb6f0dad66b3e8
subsection
61
92
Analytic Properties and Solutions for
Note theeye formed for N<30; for N>30, the error transitions smoothlyacross both Stokes lines, showing that the SCE smooths out the Stokeslines at large order. Lastly, the lower and upper dashed lines representthe uniform exponential convergence component in the \left|\arg z\right|<\frac{\pi }{3}and \frac{\pi }{3}<\lef...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ 0.006201537791639566, 0.0038864496164023876, -0.0659666657447815, -0.016232315450906754, 0.00004326488124206662, -0.023372093215584755, -0.012250516563653946, 0.001138474210165441, 0.0018278517527505755, 0.03585144877433777, -0.02678942307829857, 0.05409754812717438, -0.009153561666607857, ...
e2a5c12580b454d60e97bcf40f2dbcd16f2e9b50
subsection
62
92
Analytic Properties and Solutions for
It is only once N is reduced, that the roots split discontinuously, with \sqrt{G}\rightarrow \sqrt{z} or \sqrt{G}\rightarrow 0, depending on whether \arg z is larger or smaller than \frac{2\pi }{3}. This occurs abruptly, at the value of N at which the “collision event” depicted in REF takes place, and predicted by () ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 552, "openalex_id": "", "raw": "M. V. Berry, Proc. R. Soc. A 422, 7 (1989a).", "source_ref_id": "95044939ba5e54e77451504c22fbca1c647fc101", "start": 396 }, { "arxiv_id": "", "doi": "", "end": ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.038793738931417465, 0.024326208978891373, -0.04028932750225067, -0.010766712948679924, 0.019015341997146606, -0.03314713016152382, 0.008729354478418827, -0.023669980466365814, 0.007149829994887114, 0.04511183872818947, -0.010186790488660336, 0.03152945265173912, -0.02803465910255909, 0....
482f2a9f6d9d3efbf147c630d97d38c91530fd84
subsection
63
92
Analytic Properties and Solutions for
However for any z, it may be bounded by \left|1-\sqrt{z}/\sqrt{G}\right|^{N}<2^{N}\approx 10^{0.30N} [see REF ], which is weaker than the exponentially convergent component (both its bound and its rate in practice), and convergence is formally still uniform.(iii) For \frac{2\pi }{3}<\left|\arg z\right|\le \pi , SCE pro...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.026354489848017693, 0.038455888628959656, -0.04318657144904137, -0.016328491270542145, -0.016053806990385056, -0.024569038301706314, -0.006722149904817343, 0.00107203412335366, -0.00773314293473959, 0.022600464522838593, -0.008515232242643833, 0.04410218819975853, 0.011391793377697468, ...
88365840aaa46ee8b953ddfe48df452cda173702
subsection
64
92
Further Numerical Results
Recalling that z=0 corresponds to an infinite anharmonicity [cf. the rhs of ()], and so represents an extreme test case, we first examine the convergence of the SCE to \mathrm {Ai}\left(0\right). We take z=0 as having phase zero, that is, lying on the positive real line, and so it is contained in the first Stokes wedge...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.03130427002906799, 0.02680389955639839, -0.056414809077978134, 0.014393556863069534, 0.008741396479308605, -0.021342433989048004, 0.024393532425165176, -0.009694864973425865, 0.008054899983108044, 0.05208224803209305, -0.04155595973134041, 0.00023610256903339177, 0.002059491351246834, 0...
3e2efa5f8ad8c6cc25c3d5e75ba843b4d764bd84
subsection
65
92
Further Numerical Results
Similarly to , the SCE exhibits performance much better then superasymptotics, and comparable with hyperasymptotics, though in principle the hyperasymptotic expansion is of order 2N_{0}. [Figure: Comparison of the Airy SCE versus the Padé and Chebyshev \tau approximations (a) The relative accuracy of the three methods ...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.01658601686358452, 0.02813672088086605, -0.08605808019638062, 0.018859537318348885, 0.028869129717350006, -0.02233848161995411, 0.03967216610908508, 0.007812364958226681, -0.01551029086112976, 0.06567269563674927, -0.013122331351041794, -0.0002720308839343488, -0.014945725910365582, 0.0...
f5846ddd3f97fd4bbf6bd90a40e3339b57a763a3
subsection
66
92
Extension to Multiple Degrees of Freedom
In the case of an oscillator in more than one spatial dimension, or alternatively that of several coupled oscillators, we may examine a potential of the generic formV\left(\left\lbrace x_{i}\right\rbrace \right)=\frac{1}{2}\sum \gamma _{ij}x_{i}x_{j}+\sum g_{ijkl}x_{i}x_{j}x_{k}x_{l}\,,for which we assume that \gamma _...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.0855071023106575, 0.016173720359802246, -0.019911985844373703, -0.0038927551358938217, 0.00958216655999422, -0.026045793667435646, -0.008987096138298512, -0.029478894546628, 0.02036973275244236, 0.02760213240981102, -0.029860349372029305, 0.01769954338669777, -0.0024375016801059246, -0....
66f1d6b45cabe72d982369ea996bf177b07e99b7
subsection
67
92
Extension to Multiple Degrees of Freedom
\gamma \left(\Omega \right) and g\left(\Omega \right) are the quadratic and quartic coefficients of the potential along the ray defined by \Omega , which by assumption are positive and non-negative, respectively. By satisfying first-order self-consistency for \left\langle r^{2M}\right\rangle , we find that\mathcal {Z}_...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1694, "openalex_id": "", "raw": "D. S. Rosa, R. L. S. Farias, and R. O. Ramos, Physica A 464, 11 (2016).", "source_ref_id": "c7b1c92c9421e058a1091dc526d52637b1618d81", "start": 1551 } ] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04048910737037659, 0.03252553939819336, -0.03667513653635979, -0.027140213176608086, -0.0017630077200010419, -0.025080667808651924, 0.011617353186011314, -0.005015370901674032, 0.011525818146765232, 0.016598397865891457, -0.039848361164331436, 0.009039109572768211, -0.016552630811929703, ...
6ff06f56fa5e20fea1afb36847bb710dd5ff7732
subsection
68
92
Conclusions and Outlook
In this paper we have investigated the analytical properties of the SCE by applying it to the toy model of the classical anharmonic oscillator in thermal equilibrium. We utilized the benefit of an explicit closed-form expansion to show that for this model the SCE is exponentially and uniformly convergent for any positi...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 2894, "openalex_id": "", "raw": "I. R. C. Buckley, A. Duncan, and H. F. Jones, Phys. Rev. D 47, 2554 (1993).", "source_ref_id": "939c4a588d4b82d0d306dfa9bae7c7c6c042f3ca", "start": 2507 } ] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04700712859630585, 0.011011572554707527, -0.04886909946799278, 0.023015178740024567, 0.028707925230264664, -0.04322214052081108, 0.02704436145722866, -0.028463732451200485, 0.016040420159697533, 0.05182993784546852, 0.001891540945507586, 0.012041761539876461, -0.025624990463256836, 0.01...
0d4b90a3b249c3962bcc5988f14850c21b795004
subsection
69
92
Conclusions and Outlook
Its strength is exemplified by the remarkable result that in the SCE, optimal convergence is achieved repeatedly in the linear scaling M\left(N\right)\sim N, for any possible anharmonicity.These results provide fertile grounds for additional inquiry: We have seen that for a given system, the SCE is not unique — neither...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 2261, "openalex_id": "", "raw": "C. M. Bender, A. Duncan, and H. F. Jones, Phys. Rev. D 49, 4219 (1994).", "source_ref_id": "b203fbf35013eb199eb167eef3477df300270d64", "start": 2066 }, { "arxiv_id": "", ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04192996397614479, 0.015456714667379856, -0.056730568408966064, 0.04455440118908882, 0.012939086183905602, -0.044798534363508224, 0.028838293626904488, 0.0058477651327848434, 0.03829847276210785, 0.04699573665857315, -0.01358756609261036, 0.004920819774270058, -0.025206804275512695, 0.0...
5eaf9be27e500216d26602d8661fa2f013cac479
subsection
70
92
Conclusions and Outlook
Most recently, Serone, Spada, and Villadoro introduced Exact Perturbation Theory (EPT) , an approach in which the problem is framed as a particular realization in the parameter space of a more general model with a Borel-resummable PT, and the coupling-dependent interpolation within this space provides the non-perturbat...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 339, "openalex_id": "", "raw": "M. Serone, G. Spada, and G. Villadoro, J. High Energy Phys. 05, 56 (2017a).", "source_ref_id": "dba326469212552e32f2becc73c8fbb74e313ef3", "start": 0 }, { "arxiv_id": "", ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.01974194310605526, -0.010633797384798527, -0.0008882157853804529, 0.041192617267370224, 0.009115773253142834, -0.07262105494737625, 0.03002484142780304, 0.003394576720893383, 0.027629565447568893, 0.02413582243025303, 0.007571050431579351, 0.000021514084437512793, 0.016766920685768127, ...
d3e08287b2e8d0d053f5aa1b21d8768c3aeb3152
subsection
71
92
Direct Estimation of the Quartic SCE Coefficients
The specific case of the quartic anharmonicity admits a direct error estimation by inspection of the SCE series coefficients. Given by the sum over l in (), these coefficients may be expressed asS_{n,K} & =\sum _{l=0}^{n}\binom{n}{l}\frac{\Gamma \left(n+l+\frac{1}{2}\right)}{n!\left(-K\right)^{l}}=\frac{\left(-1\right)...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 500, "openalex_id": "https://openalex.org/W3089116201", "raw": "DLMF, “NIST Digital Library of Mathematical Functions,” http://dlmf.nist.gov/.", "source_ref_id": "1bad14f3312569fc1768be8ccfc8645c155416c3", "start": 126...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.039836201816797256, 0.03956146910786629, -0.03800465166568756, 0.014003721997141838, -0.0013107025297358632, -0.026771146804094315, 0.03974462300539017, 0.00632266141474247, -0.0219785925000906, 0.02921321429312229, -0.007288040593266487, 0.023626986891031265, -0.04279720410704613, 0.00...
8d393af7fd5759ed1d91640b97261bb2cbb658e8
subsection
72
92
Direct Estimation of the Quartic SCE Coefficients
This reproduces the bounding form we used to prove case 3 of Proposition REF at the end of .Writing down the sum represented by _{1}F_{1} explicitly, we haveS_{n,K} & =\frac{\left(-1\right)^{n}\pi e^{-K}}{K^{n}n!\Gamma \left(\frac{1}{2}-n\right)}\sum _{s=0}^{\infty }\frac{K^{s}\Gamma \left(-n+s+\frac{1}{2}\right)}{s!\...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.05520028620958328, 0.03536602109670639, -0.04867023229598999, -0.016782838851213455, -0.005942651070654392, -0.015852155163884163, 0.02625751495361328, 0.047663263976573944, -0.025159001350402832, 0.026165971532464027, -0.028790198266506195, 0.06517844647169113, -0.016538726165890694, 0...
b664f28de71a78ddf5d5949147648132fbec7153
subsection
73
92
Direct Estimation of the Quartic SCE Coefficients
They occur for integer values of s which are closest to the two solutions of\frac{K\left(n-s-\frac{1}{2}\right)}{\left(s+1\right)\left(2n-s-\frac{1}{2}\right)}=1\,,which are given bys_{max}^{\pm } & =\frac{1}{4}\left[\left(4n+1\right)+\left(2K-1\right)-3\pm \sqrt{\left(4n+1\right)^{2}+\left(2K-1\right)^{2}-1}\right]\ap...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.021472588181495667, 0.029438965022563934, -0.07655046135187149, -0.00734447967261076, -0.023624425753951073, -0.028996387496590614, 0.01184273511171341, 0.013269662857055664, -0.006863750517368317, 0.03284222632646561, 0.0005560820573009551, 0.0404423326253891, -0.023242894560098648, 0....
c76d058fd88a6fbdd4c73cab181933746777f5c0
subsection
74
92
Direct Estimation of the Quartic SCE Coefficients
Recalling that K=M+2 and substituting M=\alpha N into the approximate roots in (), we find the following limits:\ln Q_{-} & \equiv \lim _{N\rightarrow \infty }\frac{1}{N}\ln P_{+}=-\frac{1}{2}\left[\alpha +\sqrt{\alpha ^{2}+4}+4\mathrm {tanh}^{-1}\left(\frac{\alpha }{2}-\frac{1}{2}\sqrt{4+\alpha ^{2}}\right)\right]\,,\...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1539, "openalex_id": "", "raw": "I. R. C. Buckley, A. Duncan, and H. F. Jones, Phys. Rev. D 47, 2554 (1993).", "source_ref_id": "939c4a588d4b82d0d306dfa9bae7c7c6c042f3ca", "start": 1166 } ] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.023645924404263496, 0.02324928343296051, -0.06315750628709793, -0.003077784087508917, -0.02080841362476349, -0.020915202796459198, -0.005816134624183178, -0.005324147175997496, 0.015095253475010395, 0.02942773513495922, -0.017894625663757324, 0.011487343348562717, -0.044057697057724, 0....
a7e277f60e30943d9968a84e4f3ede6bd7fa1996
subsection
75
92
Direct Estimation of the Quartic SCE Coefficients
Thus, the expansion must diverge for this scaling of M with N.
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.05054677650332451, 0.023106226697564125, -0.028692010790109634, -0.022068429738283157, -0.015215923078358173, 0.016742093488574028, -0.04294644668698311, 0.009424104355275631, 0.017764627933502197, -0.004612851422280073, -0.005429352633655071, -0.034521982073783875, -0.02873779647052288, ...
d67e74a1fc6cddf8c6645f96614dd98f21bffdd8
subsection
76
92
Summary of Competing Asymptotic and Numerical Methods
In we compared the SCE with other asymptotic methods for the case of g\ll 1, while in the strongly coupled regime g\gg 1 we compared it against numerical approximation schemes. In this Appendix, we will briefly describe each.
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.04218010604381561, 0.017061274498701096, -0.0191214457154274, -0.031146744266152382, 0.005890564993023872, -0.07788974791765213, 0.0339394211769104, 0.02078484371304512, 0.014703521504998207, 0.03339004144072533, -0.006615440361201763, 0.005467085167765617, -0.018526284024119377, 0.0102...
ec0fd710147409deb8e748bbc10e0fb6dc9bd0e7
subsection
77
92
Superasymptotics
The superasymptotic expansion of \mathcal {Z} is defined by terminating its usual asymptotic series at its least term . This truncation usually depends on the value of g. We can find the general superasymptotic form of \mathcal {Z} by standard PT,\mathcal {Z}_{SA}\left(g\right) & =\sum _{n=0}^{N_{0}}\int _{-\infty }^{\...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 119, "openalex_id": "", "raw": "J. P. Boyd, Acta Applicandae Mathematica 56, 1 (1999).", "source_ref_id": "a1f0dc9b3e8b3359250931a8cd2526252333c171", "start": 0 }, { "arxiv_id": "", "doi": "10.1201/...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.08253248780965805, 0.022449202835559845, -0.004376144614070654, -0.014803349040448666, 0.010667568072676659, -0.056374818086624146, 0.055276013910770416, -0.00040561368223279715, -0.00306559051387012, 0.030049273744225502, -0.04331124201416969, 0.026813901960849762, -0.020556816831231117,...
797275bd1f38f8d7098d368e29371f8f88513eee
subsection
78
92
Hyperasymptotics of
Past the optimal truncation of superasymptotics, one may find an asymptotic expansion for the remainder. Truncation of this series at its least term will yield a new, smaller remainder. This process can be iterated systematically to improve upon superasymptotics. This technique it called hyperasymptotics, and was devel...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 347, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 430, 653 (1990).", "source_ref_id": "dfee5969e4d320f8d915ee156a5c53b119ff41a5", "start": 264 }, { "arxiv_id": "", "doi": ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.05461098253726959, 0.007558587472885847, -0.03520730510354042, 0.016032442450523376, 0.04094298183917999, -0.031302161514759064, 0.035847991704940796, -0.010815415531396866, 0.0007117161294445395, 0.012157808057963848, -0.05708220601081848, 0.02921229973435402, 0.001803840510547161, 0.0...
b7437d758f82344a704e249665294653d1aedc47
subsection
79
92
Hyperasymptotics of
Taking the limit of u\rightarrow +\infty with k=1, we see that all paths end at complex infinity,z_{1}\left(u\right)\rightarrow \left\lbrace -\infty ,-i\infty \right\rbrace ,\ z_{2}\left(u\right)\rightarrow \left\lbrace -\infty ,\infty \right\rbrace ,\ z_{3}\left(u\right)\rightarrow \left\lbrace \infty ,i\infty \right\...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 607, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 434, 657 (1991).", "source_ref_id": "e8b49f0b0149d162f576740c838d235a0040a939", "start": 417 }, { "arxiv_id": "", "doi": ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.06120406091213226, -0.003321579657495022, -0.014072355814278126, 0.017063874751329422, 0.004803984425961971, -0.013347369618713856, 0.01843753270804882, -0.007524589076638222, 0.03495194390416145, 0.03644770383834839, -0.006776709109544754, 0.034371957182884216, 0.017765967175364494, 0....
df8aa6ad9aae844c02135950a6a2f43720a3fc27
subsection
80
92
Hyperasymptotics of
We thus define\forall z\in C_{2}:\qquad \left[k\left(f\left(z\right)-f_{2}\right)\right]^{\frac{1}{2}}\equiv k^{\frac{1}{2}}z\sqrt{\frac{1}{2}+gz^{2}}\,,where k^{\frac{1}{2}} is taken on the smooth manifold of the square root (i.e., it is continuous in k and experiences no branch cuts, and thus is multi-valued), while ...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.06720577925443649, -0.00032785514486022294, -0.03656335920095444, 0.02087590843439102, 0.0107278972864151, -0.005371578969061375, 0.022600308060646057, -0.028383910655975342, -0.012162352912127972, 0.011200962588191032, -0.029024837538599968, 0.007473666686564684, -0.01061344612389803, ...
90f545211874d254c40ced2d6ef4085ee04dcaa7
subsection
81
92
Hyperasymptotics of
However, 2 is adjacent to both. It is easily verified that for k=-1 (or in general, k with argument \pm \pi ), the path z_{2}\left(u\right) in () coincides partially with z_{1,3}\left(u^{\prime }\right) for some real u^{\prime }<u (i.e. z_{2}\left(u\right) “arrives late” at the contours C_{1,3}\left(\theta _{k}=\pm \pi...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1371, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 434, 657 (1991).", "source_ref_id": "e8b49f0b0149d162f576740c838d235a0040a939", "start": 1143 }, { "arxiv_id": "", "doi"...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.034787263721227646, -0.005504169035702944, -0.029660720378160477, 0.02151317708194256, -0.019956905394792557, -0.014792216010391712, 0.0008835087646730244, 0.003322336357086897, -0.003175482153892517, 0.04821392893791199, -0.013892019167542458, -0.009009595960378647, 0.01730208657681942, ...
0d64835944b755ab62569c2e1be4438bf330f6d7
subsection
82
92
Hyperasymptotics of
Using our definition (REF ), one observes that the following holds:\forall z\in C_{1}\left(+\pi \right):\qquad \arg \left\lbrace z\sqrt{\frac{1}{2}+gz^{2}}\right\rbrace =-\frac{\pi }{2}\ ,\\ \forall z\in C_{3}\left(-\pi \right):\qquad \arg \left\lbrace z\sqrt{\frac{1}{2}+gz^{2}}\right\rbrace =+\frac{\pi }{2}\ .Thus, si...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.03052830696105957, -0.003125680610537529, -0.023067865520715714, 0.004287083633244038, 0.036005400121212006, -0.014425043947994709, 0.008444486185908318, -0.0026145868469029665, 0.006281874142587185, 0.014379275031387806, -0.03823285177350044, 0.006285688374191523, 0.012533235363662243, ...
a0efd98881eced2b96c88047c8ba51f2b0d19d1a
subsection
83
92
Hyperasymptotics of
The coefficients at z_{1} are thenT_{r}^{\left(1\right)} & =\frac{\left(r-\frac{1}{2}\right)!}{2\pi i}\oint _{z_{1}}\frac{\left[\left(f\left(z\right)-f_{1}\right)\right]^{\frac{1}{2}}dz}{\left[f\left(z\right)-f\left(z_{1}\right)\right]^{r+1}}\\ & =\frac{\left(r-\frac{1}{2}\right)!}{2\pi i}\oint _{z_{1}}\frac{\sqrt{g}\l...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.03751101717352867, 0.03088783659040928, -0.024295175448060036, 0.026492727920413017, 0.03662589192390442, -0.0011636351700872183, 0.04141778126358986, -0.00661555165424943, -0.009827948175370693, 0.013200582005083561, -0.04959756135940552, 0.010858050547540188, -0.04269968718290329, 0.0...
6a861cffa1eade59172ba4c8178ebba74ff580f1
subsection
84
92
Hyperasymptotics of
The saddles z_{1} and z_{3} are not adjacent to each other: The contour C_{3}\left(\theta _{k}\right) never leaves the top half of the complex plane, except for \arg \,k which is a multiple of 2\pi , when it coincides with half of the real line. The same applies to C_{1} in the lower half. Thus, they could only meet on...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 2167, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 434, 657 (1991).", "source_ref_id": "e8b49f0b0149d162f576740c838d235a0040a939", "start": 1923 } ] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.07428447157144547, -0.005192130338400602, -0.05408055707812309, -0.03491431102156639, -0.0046694837510585785, -0.03998054936528206, 0.022080861032009125, 0.010895462706685066, -0.004211691208183765, 0.006714290473610163, -0.042055875062942505, 0.032838985323905945, 0.011215916834771633, ...
e686f187ee73b535cb56a09e8ee5814726c87753
subsection
85
92
Hyperasymptotics of
We haveN\left(2\right)=\frac{1}{16g}=10,\quad N\left(21\right)=N\left(23\right)=5,\quad N\left(212\right)=N\left(232\right)=2,\\ N\left(2121\right)=N\left(2123\right)=N\left(2321\right)=N\left(2323\right)=1\,.where N\left(nm\ldots \right) is the truncation of the scattering path that originates at z_{n}, scatters to z_...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1820, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 430, 653 (1990).", "source_ref_id": "dfee5969e4d320f8d915ee156a5c53b119ff41a5", "start": 1184 }, { "arxiv_id": "", "doi"...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.012919257394969463, 0.013346592895686626, -0.0360182486474514, -0.032721664756536484, -0.016086116433143616, -0.02150411531329155, 0.01581140048801899, 0.016101378947496414, -0.051768604665994644, 0.021809356287121773, -0.011362536810338497, 0.026311635971069336, -0.03540777042508125, -...
ff22deb2dc8473ef24e42c3912b0337e227a1b73
subsection
86
92
Hyperasymptotics of
We thus obtain\mathcal {Z}_{HA} & =\sum _{r=0}^{N\left(2\right)}T_{r}^{\left(2\right)}+2\sum _{r=0}^{N\left(23\right)}K^{\left(23\right)}T_{r}^{\left(3\right)}+2\sum _{r=0}^{N\left(232\right)}K^{\left(232\right)}T_{r}^{\left(2\right)}+4\sum _{r=0}^{N\left(2323\right)}K^{\left(2323\right)}T_{r}^{\left(3\right)}\,.The re...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1067, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 434, 657 (1991).", "source_ref_id": "e8b49f0b0149d162f576740c838d235a0040a939", "start": 974 }, { "arxiv_id": "", "doi":...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.02196202240884304, -0.014254710637032986, -0.03668985515832901, -0.018818048760294914, 0.0006691434537060559, 0.0019001193577423692, 0.010553674772381783, 0.0008751804707571864, -0.0224198829382658, 0.017291849479079247, -0.046274393796920776, 0.05240971967577934, 0.0028005775529891253, ...
9e2454ac66c0f0de7979e7aa8f5160970d2ba3e8
subsection
87
92
Hyperasymptotics of
Setting g=\frac{1}{160} and S=3 (and dividing by the exact \mathcal {Z}\left(g\right)), we obtain an expected relative error of 4.0\cdot 10^{-12}, in close agreement with our numerical result.It is worth noting that for g much smaller than \frac{1}{160}, evaluation of the complete hyperseries becomes impractical: The a...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1727, "openalex_id": "", "raw": "M. V. Berry and C. J. Howls, Proc. R. Soc. A 430, 653 (1990).", "source_ref_id": "dfee5969e4d320f8d915ee156a5c53b119ff41a5", "start": 1344 }, { "arxiv_id": "", "doi"...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.05879266932606697, 0.014057125896215439, -0.03953089565038681, -0.013965548947453499, 0.016316033899784088, -0.03907300904393196, 0.03583727777004242, -0.008440380915999413, -0.0333341620862484, 0.011630325578153133, -0.05238225311040878, 0.0256721880286932, 0.009127210825681686, 0.0253...
d59f599cd686521113c5b25cf76a9df4f22b08ed
subsection
88
92
Chebyshev Polynomial Approximation by Lanczos's
With the definition (), we have\frac{d\mathcal {Z}}{dg}=-\int _{-\infty }^{\infty }x^{4}e^{-\left[\frac{1}{2}x^{2}+gx^{4}\right]}dx\,.However, performing the rescaling y=g^{\frac{1}{4}}x, we have\mathcal {Z}\left(g\right)g^{\frac{1}{4}}=\int _{-\infty }^{\infty }e^{-\left[\frac{1}{2}g^{-\frac{1}{2}}y^{2}+y^{4}\right]}d...
{ "cite_spans": [] }
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.036186929792165756, 0.029703183099627495, -0.00023360553313978016, -0.025889214128255844, 0.020046215504407883, 0.008192403241991997, 0.04738473892211914, 0.0031865702476352453, 0.0037052698899060488, 0.023661857470870018, -0.038963496685028076, -0.011045251041650772, 0.000746107485610991...
6627eea341b45e0f9c664b6aef153f9b37ccaaf0
subsection
89
92
Chebyshev Polynomial Approximation by Lanczos's
Note that g=0 is a singular point of (), so we do not necessarily require a second boundary condition.Next, we outline the procedure of the \tau method, due to Lanczos , : To obtain the N-th order approximation of \mathcal {Z}, instead of solving () approximately, we will obtain an exact solution to an approximate equa...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 783, "openalex_id": "", "raw": "J. P. Boyd, Acta Applicandae Mathematica 56, 1 (1999).", "source_ref_id": "a1f0dc9b3e8b3359250931a8cd2526252333c171", "start": 102 }, { "arxiv_id": "", "doi": "", ...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.008435499854385853, 0.029135270044207573, -0.013667644932866096, -0.0524129793047905, 0.012149865739047527, -0.015635419636964798, 0.03175897151231766, 0.014987122267484665, 0.036426715552806854, 0.03413860499858856, -0.006807127967476845, -0.004057582002133131, -0.008862613700330257, 0...
b6319681719df14d73d407fc7e4b6938d64913c5
subsection
90
92
Chebyshev Polynomial Approximation by Lanczos's
This \tau approximation can be computed systematically with ease by any computer algebra software.In the case of the Airy function, we note that \mathrm {Ai}\left(z\right) satisfies the second-order differential equation \frac{d^{2}\mathrm {Ai}\left(z\right)}{dz^{2}}=z\cdot \mathrm {Ai}\left(z\right)\,.Defining \tilde{...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1119/1.1972842", "end": 305, "openalex_id": "https://openalex.org/W2120062331", "raw": "M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions, Vol. 55 (Dover, New York, 1964).", "source_ref_id": "d88a70f14cea1e8d188fa5...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.026715926826000214, 0.035458486527204514, -0.003322326112538576, -0.017622441053390503, -0.0027940343134105206, 0.012656116858124733, 0.0555068776011467, 0.02663963846862316, 0.023771224543452263, 0.04314827919006348, -0.031094836071133614, -0.0219098050147295, -0.015585562214255333, 0....
64eb25c6a6097aadc669456d55df2c41f03f060a
subsection
91
92
Padé Approximants
With the usual perturbative expansion of \mathcal {Z} in Subsection REF , we define the Padé approximation of \mathcal {Z} of order N to be \mathcal {Z}_{\text{Padé}}^{\text{$\left(N\right)$}}\left(g\right)=\frac{P\left(g\right)}{Q\left(g\right)}\,,with P\left(x\right) and Q\left(x\right) polynomials of order \frac{N}{...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 449, "openalex_id": "https://openalex.org/W1657347807", "raw": "C. M. Bender and S. A. Orszag, Advanced Mathematical Methods for Scientists and Engineers (McGraw-Hill, New York, 1978).", "source_ref_id": "6cbeb287ae43209a9df...
10.1103/PhysRevD.98.056017
1803.06631
A Deal with the Devil: From Divergent Perturbation Theory to an Exponentially-Convergent Self-Consistent Expansion
[ "Benjamin Remez", "Moshe Goldstein" ]
[ "math-ph", "cond-mat.stat-mech", "hep-th", "math.MP" ]
2,018
en
Physics
[ -0.02306288480758667, 0.01031037699431181, -0.021399183198809624, -0.024970801547169685, 0.012195397168397903, -0.03391510993242264, 0.09255681186914444, -0.00543755991384387, -0.0022608800791203976, 0.05739010497927666, -0.05107108876109123, -0.0026882532984018326, -0.05012476444244385, 0...
5d2c42caad1c82d981f70c4565bae818a707b3e7
abstract
0
47
Abstract
We consider the problem of link prediction, based on partial observation of a large network, and on side information associated to its vertices. The generative model is formulated as a matrix logistic regression. The performance of the model is analysed in a high-dimensional regime under a structural assumption. The mi...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.03488929569721222, -0.0006290871533565223, -0.010210384614765644, 0.014102235436439514, -0.035469260066747665, -0.0073220208287239075, 0.03797225281596184, 0.039437420666217804, -0.007318205665796995, 0.03244734928011894, -0.0526544488966465, 0.031867388635873795, -0.0329052172601223, 0...
e9c417385da5f74fce12b1d19cf82c55940e51a0
subsection
1
47
Introduction
In the field of network analysis, the task of link prediction consists in predicting the presence or absence of edges in a large graph, based on the observations of some of its edges, and on side information. Network analysis has become a growing inspiration for statistical problems. Indeed, one of the main characteris...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1016/0166-218x(94)00103-k", "end": 1387, "openalex_id": "https://openalex.org/W2062307640", "raw": "Kučera, L. (1995). Expected complexity of graph partitioning problems. Discrete Appl. Math. 57 193–212. Combinatorial optimization 1992 (...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.07721707969903946, -0.010270177386701107, -0.04297298565506935, 0.031024174764752388, 0.005238095298409462, 0.004192765336483717, 0.024050220847129822, 0.0056424932554364204, 0.017015226185321808, 0.03268754854798317, -0.0491076223552227, -0.008301598951220512, -0.017595117911696434, 0....
b17b86e6b16c19d724a7234f0b2259bbde6ddd82
subsection
2
47
Introduction
The key assumption in this model is that the network is a consequence of the information, but not necessarily based on similarity: it is possible to model more complex interactions, e.g. where opposites attract.The focus on a high-dimensional setting is another aspect of this work that is also motivated by modern appli...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.09100852906703949, 0.009621206670999527, -0.02977156825363636, 0.031312793493270874, 0.005523991771042347, 0.016907688230276108, 0.015259645879268646, 0.023576153442263603, 0.030092021450400352, 0.030946562066674232, -0.03973611816763878, 0.03326602652668953, -0.03933936730027199, 0.055...
63ec92c490b4b54bd72be0f0f5b8896dd8bd1e49
subsection
3
47
Introduction
Furthermore, we show in Section  that the minimax rate cannot be attained by a (randomised) polynomial-time algorithm, and we identify a corresponding computational lower bound. The proof of this bound is based on a reduction scheme from the so-called dense subgraph detection problem. Technical proofs are deferred to t...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.056501176208257675, 0.007272391114383936, -0.02936413697898388, -0.018960842862725258, -0.017618482932448387, -0.0018991351826116443, 0.012805814854800701, 0.028174318373203278, 0.015238843858242035, 0.048080913722515106, -0.06791124492883682, 0.02544383518397808, 0.011410064995288849, ...
60e8e35c902b48c0cb10f8d801c2ee969b7c6813
subsection
4
47
Generative model
For a set of vertices V = [n] and explanatory variables X_i \in \mathbf {R}^d associated to each i \in V, a random graph G=(V,E) is generated by the following model. For all i,j \in V, variables X_i,X_j \in \mathbf {R}^d and an unknown matrix \Theta _\star \in \mathbf {S}_d, an edge connects the two vertices i and j in...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.07350249588489532, -0.01849772036075592, -0.025808287784457207, 0.013217017985880375, 0.028128134086728096, -0.0003476907731965184, -0.045542243868112564, 0.010988744907081127, 0.05317331477999687, 0.03437035158276558, -0.014834805391728878, 0.00323366723023355, -0.02003919705748558, 0....
60924cc3df18cd20fb67ca24553f962d12d3bc2e
subsection
5
47
Generative model
Indeed, writing \textbf {vec}(A) \in \mathbf {R}^{d^2} for the vectorized form of a matrix A\in \mathbf {S}^d, we have thatX_i^\top \Theta _\star X_j = \mathbf {Tr}(X_j X_i^\top \Theta _\star ) = \langle \textbf {vec}(X_j X_i^\top ), \textbf {vec}(\Theta _\star ) \rangle \, .The vector of observation Y \in \mathbf {R}^...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06999862194061279, -0.026028258726000786, -0.022259805351495743, 0.0253111831843853, -0.000736145069822669, 0.006640562321990728, -0.06536052376031876, 0.019620362669229507, 0.03777606412768364, 0.0048631345853209496, -0.020581547170877457, 0.009436387568712234, -0.00015399925177916884, ...
b69782d0e3e4114808550f3805c76c40bd97895b
subsection
6
47
Comparaison with other models
This model can be compared to other settings in the statistical and learning literature.Generalised linear model. As discussed above in the remark to (REF ), this is an example of a logistic regression model. We focus in this work on the case where the matrix \Theta _\star is block-sparse. The problem of sparse general...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/009053607000000929", "end": 426, "openalex_id": "https://openalex.org/W3102942031", "raw": "van de Geer, S. (2008). High-dimensional generalized linear models and the lasso. The Annals of Statistics 36 614–645.", "source_ref_i...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06323394924402237, -0.0032025156542658806, -0.036591459065675735, 0.0006466067861765623, 0.01901290938258171, 0.00605407077819109, 0.01112392544746399, 0.012222585268318653, 0.02211051993072033, 0.014992122538387775, -0.04510606825351715, 0.006446993909776211, -0.007057360373437405, 0.0...
468dbd39e3aba01f7b5f79f6d56f9a3b711431c9
subsection
7
47
Parameter space
The unknown predictor matrix \Theta _\star describes the relationship between the observed features X_i and the probabilities of connection \pi _{ij}(\Theta _\star ) = \sigma (X_i^\top \Theta _\star X_j) following Definition REF . We focus on the high-dimensional setting where d^2 \gg N: the number of features for each...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.039906054735183716, 0.008130705915391445, -0.0601031556725502, 0.04225526377558708, -0.0006001734873279929, 0.006612872704863548, -0.00038613227661699057, -0.0047899470664560795, 0.03453643247485161, 0.031149910762906075, -0.02491077594459057, 0.018488585948944092, -0.0038594158831983805,...
2671350bdbdff0231ead65efed0a6ccd42794b86
subsection
8
47
Parameter space
The effect of these projections on the affinity is weighted by the \lambda _\ell . By allowing for negative eigenvalues, we allow our model to go beyond a geometric description, where close or similar Xs are more likely to be connected. This can be used to model interactions where opposites attract.The assumption of si...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/12-aos1039", "end": 621, "openalex_id": "https://openalex.org/W3103211861", "raw": "Bunea, F., She, Y. and Wegkamp, M. (2012). Joint variable and rank selection for parsimonious estimation of high-dimensional matrices. The Annals of...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06090056151151657, 0.01449286937713623, 0.005247944500297308, 0.050008028745651245, -0.021861350163817406, 0.009755989536643028, 0.03276913985610008, 0.0147293321788311, 0.02141893468797207, 0.009305947460234165, -0.05098439007997513, 0.011449366807937622, 0.00017639347061049193, 0.0133...
9983fb612906361b91e6a6115caa5f82130a0e3d
subsection
9
47
Explanatory variables
As mentioned above, this problem is different from tasks such as metric learning, where the objective is to estimate the X_i with no side information. Here they are seen as covariates, allowing us to infer from the observation on the graph the predictor variable \Theta _\star . For this task to be even possible in a hi...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1109/tit.2005.858979", "end": 1821, "openalex_id": "https://openalex.org/W2129131372", "raw": "Candes, E. J. and Tao, T. (2005). Decoding by Linear Programming. IEEE Trans. Information Theory 51 4203–4215.", "source_ref_id": "b1d5e...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.04332619532942772, -0.0002972478396259248, -0.04247187823057175, 0.05278473347425461, -0.014401383697986603, -0.008306730538606644, 0.012318979948759079, 0.013974224217236042, 0.05818524956703186, 0.019725624471902847, -0.03664419800043106, -0.011449404992163181, 0.04225829616189003, 0....
1d4a8263ada8634f349db107f6fdbac7922a9168
subsection
10
47
Explanatory variables
These different measures of restricted isometry are related, as shown in the following lemmaLemma 5 For a matrix \mathbb {X}\in \mathbf {R}^{d \times n}, let \mathbb {D}_\Omega \in \mathbf {R}^{N\times d^2} be defined row-wise by \mathbb {D}_{\Omega \,(i,j)} = \textbf {vec}(X_j X_i^\top ) for all (i,j) \in \Omega . It...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/009053607000000929", "end": 2259, "openalex_id": "https://openalex.org/W3102942031", "raw": "van de Geer, S. (2008). High-dimensional generalized linear models and the lasso. The Annals of Statistics 36 614–645.", "source_ref_...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.021132564172148705, 0.00401671277359128, -0.03823697194457054, 0.017653701826930046, -0.011130833998322487, -0.018691256642341614, -0.008262298069894314, 0.03213370218873024, 0.04052569717168808, 0.01660088822245598, -0.02448936179280281, -0.00535561703145504, 0.010627314448356628, 0.04...
d9cc1a241d8bee126fc89be8a97daf2eb0d6f7a1
subsection
11
47
Explanatory variables
Similarly to , Assumption REF can be shown to be redundant for minimax optimal prediction, because the log-likelihood function in the matrix logistic regression model satisfies the so-called self-concordant property. Our analysis to follow can be combined with an analysis similar to to get rid of the assumption for min...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/09-ejs521", "end": 217, "openalex_id": "https://openalex.org/W2000483484", "raw": "Bach, F. (2010). Self-concordant analysis for logistic regression. Electronic Journal of Statistics 4 384–414.", "source_ref_id": "65f4b9afd339...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.057695429772138596, -0.007753490936011076, -0.03298189491033554, 0.034171804785728455, -0.023355819284915924, -0.0044507249258458614, 0.006430096458643675, 0.02655942551791668, 0.04643704369664192, 0.012905958108603954, -0.04567427933216095, -0.019816596060991287, -0.023782966658473015, ...
efe49b8c864389cb63526e0cc3bd6407392a883c
subsection
12
47
Random designs
For random designs, we require the block isometry property to hold with high probability. Then the results in this article carry over directly and thus we do not discuss it in full detail. It is well known that for sparse linear models with the dimension of a target vector \bar{p} and the sparsity \bar{k}, the classica...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/s00365-007-9005-8", "end": 640, "openalex_id": "https://openalex.org/W1984305442", "raw": "Mendelson, S., Pajor, A. and Tomczak-Jaegermann, N. (2008). Uniform uncertainty principle for bernoulli and subgaussian ensembles. Constructi...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.05570410192012787, 0.009568896144628525, -0.012376993894577026, 0.029271360486745834, -0.02012978307902813, -0.011102667078375816, 0.008241155184805393, -0.018374722450971603, 0.008332722820341587, 0.04148047789931297, -0.048256538808345795, 0.03201841190457344, 0.0240977481007576, 0.02...
a01a1ec321cc7115e0e60ab9102fb15b10d5d6a8
subsection
13
47
Matrix Logistic Regression
The log-likelihood for this problem is\ell _Y(\Theta ) = -\sum _{(i,j) \in \Omega } \xi (s_{(i,j)} X_i^\top \Theta X_j)\, ,where s_{(i,j)} = 2 Y_{(i,j)} -1 is a sign variable that depends on the observations Y and \xi : x \mapsto \log (1+e^x) is a softmax function, convex on \mathbf {R}. As a consequence, the negative ...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06371373683214188, 0.009978157468140125, -0.04143834114074707, 0.009848471730947495, 0.007258575409650803, 0.003614030545577407, -0.016477691009640694, 0.06011305749416351, 0.006545304786413908, -0.01687437668442726, -0.0002534112427383661, 0.03521343693137169, -0.02882070280611515, 0.0...
27a4b364531f4ac0e0f9cf996b61f865cf963ef8
subsection
14
47
Penalized logistic loss
In a classical setting where d is fixed and N grows, the maximiser of \ell _Y - the maximum likelihood estimator - is an accurate estimator of \Theta _\star , provided that it is possible to identify \Theta from \mathbf {P}_{\Theta } (i.e. if the X_i are well conditioned). We are here in a high-dimensional setting wher...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/s00440-006-0011-8", "end": 1192, "openalex_id": "https://openalex.org/W2167121755", "raw": "Birgé, L. and Massart, P. (2007). Minimal penalties for gaussian model selection. Probability theory and related fields 138 33–73.", "...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.07788308709859848, 0.0002451013133395463, -0.029007788747549057, 0.008552796207368374, -0.016357889398932457, -0.014275006018579006, -0.010841679759323597, -0.00490584084764123, 0.028977269306778908, 0.021683359518647194, -0.03062526509165764, 0.0296486746519804, -0.007477020379155874, ...
b0ab7b3ebbd90a670bf6956a1800a1010dfa10a7
subsection
15
47
Penalized logistic loss
It can formally be shown using standard conditioning arguments .Corollary 12 Assume the design matrix \mathbb {X} satisfies the block isometry property from Definition REF and \max _{(i,j) \in \Omega } |X_i^\top \Theta _\star X_j| < M for some M > 0 and all \Theta _\star in a given class, and the penalty term p(\Theta ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/13-aos1170", "end": 64, "openalex_id": "https://openalex.org/W2073772490", "raw": "Nickl, R. and van de Geer, S. (2013). Confidence sets in sparse regression. Ann. Statist. 41 2852–2876.", "source_ref_id": "a1a05630ba7556631d0...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.04696490243077278, -0.012877965345978737, -0.019118590280413628, 0.003940443508327007, -0.00478727649897337, -0.009063402190804482, -0.007621496915817261, 0.01713501662015915, 0.024535270407795906, 0.007896145805716515, -0.01863032579421997, 0.016951918601989746, -0.015510013327002525, ...
f8e9554a20a413f4c561dbe5e0535370387edb34
subsection
16
47
Penalized logistic loss
In particular, a naive MLE approach with p(\Theta ) = 0 in (REF ) yields a suboptimal estimator as it follows from Theorem REF .
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.051743071526288986, 0.012249317951500416, 0.003035634057596326, -0.012378980405628681, -0.008832322433590889, -0.014629011042416096, 0.017893461510539055, 0.0357259064912796, 0.004362770821899176, 0.022668102756142616, -0.041278522461652756, -0.017542609944939613, -0.03670218959450722, ...
a22a8348c27dc2264dc203bd22a7bbaeb21182cd
subsection
17
47
Convex relaxation
In practice, computation of the estimator (REF ) is often infeasible. In essence, in order to compute it, we need to compare the likelihood functions over all possible subspaces \mathcal {P}_{k,r}(M). Sophisticated step-wise model selection procedures allow to reduce the number of analysed models. However, they are not...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/009053607000000929", "end": 804, "openalex_id": "https://openalex.org/W3102942031", "raw": "van de Geer, S. (2008). High-dimensional generalized linear models and the lasso. The Annals of Statistics 36 614–645.", "source_ref_i...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.030753158032894135, -0.00830609817057848, -0.012348552234470844, -0.008199316449463367, 0.014735888689756393, -0.014140961691737175, -0.016047779470682144, 0.025032704696059227, 0.043566975742578506, 0.01984615996479988, -0.026085268706083298, 0.0204410869628191, -0.028037849813699722, ...
4613cf77e961917fbd02ef3a0cc4131bc1658030
subsection
18
47
Prediction
In applications, as new users join the network, we are interested in predicting the probabilities of the links between them and the existing users. It is natural to measure the prediction error of an estimator \hat{\Theta } by {\rm I}\hspace{-1.79993pt}{\rm E}\big [ \sum _{(i,j)\in \Omega } (\pi _{ij}(\hat{\Theta }) - ...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.05148906260728836, -0.005982131697237492, -0.0316198393702507, 0.011933742091059685, 0.014901917427778244, -0.015146085992455482, 0.008011783473193645, 0.018846767023205757, 0.03329849988222122, 0.034122567623853683, -0.04239378124475479, 0.028659293428063393, -0.02261612005531788, 0.01...
e92c5048bcee1566ef7e9b6f23f08273d19a94f4
subsection
19
47
Information-theoretic lower bounds
The following result demonstrates that the minimax lower bound on the rate of estimation matches the upper bound in Theorem REF implying that the rate of estimation is minimax optimal.Theorem 17 Let the design matrix \mathbb {X} satisfy the block isometry property. Then for estimating \Theta _\star \in \mathcal {P}_{...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06427591294050217, 0.007626278791576624, -0.031832750886678696, 0.011330689303576946, -0.012955899350345135, 0.024736763909459114, 0.010674501769244671, 0.011605372652411461, 0.04437662661075592, 0.028139786794781685, -0.03534259647130966, 0.010392188094556332, -0.01606897823512554, 0.0...
8742f1d089a45ded28e9261ec16d4763b0886e93
subsection
20
47
Computational lower bounds
In this section, we investigate whether the lower bound in Theorem REF can be achieved with an estimator computable in polynomial time. The fastest rate of estimation attained by a (randomised) polynomial-time algorithm in the worst-case scenario is usually referred to as a computational lower bound. Recently, the gap ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 948, "openalex_id": "https://openalex.org/W58673994", "raw": "Berthet, Q. and Rigollet, P. (2013a). Complexity theoretic lower bounds for sparse principal component detection. In Conference on Learning Theory.", "source_ref_...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.03900209814310074, 0.0010538272326812148, -0.00640116585418582, 0.027420489117503166, -0.00017524049326311797, 0.027603596448898315, 0.0150530394166708, 0.010528734885156155, 0.02716108411550522, 0.06677354127168655, -0.040924735367298126, 0.01670864410698414, -0.0299229696393013, 0.049...
71c8e2af440f8e34d77666ef269c29426be228ee
subsection
21
47
The dense subgraph detection problem
Although our work is related to the study of graphs, we recall for absolute clarity the following notions from graph theory. A graph G = (V,E ) is a non-empty set V of vertices, together with a set E of distinct unordered pairs \lbrace i,j \rbrace with i,j \in V, i \ne j. Each element \lbrace i,j \rbrace of E is an edg...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-1-4684-2001-2_9", "end": 870, "openalex_id": "https://openalex.org/W2401610261", "raw": "Karp, R. (1972). Reducibility among combinatorial problems. Springer.", "source_ref_id": "f7ac7dcf0077ac5ad31e27a7f242d716bb72464b", ...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.04270379617810249, 0.007470874581485987, -0.0074823214672505856, 0.020939815789461136, 0.014216791838407516, -0.04316166415810585, 0.008424765430390835, -0.004208567552268505, -0.009569435380399227, 0.030555037781596184, -0.037148334085941315, -0.0007902987417764962, -0.014430464245378971...
6c3d626562324690bcb86ffd0dcb9605efd456e2
subsection
22
47
The dense subgraph detection problem
Its variations have been used extensively as computational hardness assumptions in statistical problems, see , , , .The Planted Clique problem can be reduced to the so-called dense subgraph detection problem of testing the null hypothesis in (REF ) against the alternative H_1: A \sim G(n,1/2,k,q), where q \in (1/2,1]. ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/13-aos1127", "end": 116, "openalex_id": "https://openalex.org/W2006452405", "raw": "Berthet, Q. and Rigollet, P. (2013b). Optimal detection of sparse principal components in high dimension. The Annals of Statistics 41 1780–1815.", ...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06920530647039413, -0.020901953801512718, -0.005332286935299635, 0.019849227741360664, -0.007327126804739237, -0.0332905575633049, 0.021817369386553764, 0.010512004606425762, 0.012159749865531921, 0.05050034448504448, -0.054100971668958664, 0.012678484432399273, -0.03197846561670303, 0....
cb614d8be56f1ca19c85376a944dae8dafbbc8bf
subsection
23
47
Reduction to the dense subgraph detection problem and a computational lower bound
Consider the vectors of explanatory variables X_i = N^{1/4} e_i, i =1,..., n and assume without loss of generality that the observed set of edges \Omega in the matrix logistic regression model consists of the interactions of the n nodes X_i, i.e. it holds N = |\Omega | ={n 2} . It follows from the matrix logistic regre...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06816864758729935, -0.010023903101682663, -0.02059706300497055, 0.010725729167461395, 0.015402025543153286, -0.03448100760579109, -0.02558612823486328, -0.006118090357631445, 0.04180440679192543, 0.039759960025548935, -0.07085389643907547, -0.0019319280982017517, -0.027218636125326157, ...
aef8cbab88b1a343068858fbf097d893480b34f1
subsection
24
47
Reduction to the dense subgraph detection problem and a computational lower bound
Let c > 0 be a positive constant and f(k,d,N) be a real-valued function satisfying f(k,d,N) \le ck^2/N for k = k_{n} < n^{\beta }, 0 < \beta < 1/2 and a sequence d = d_{n}, for all n> m_0 \in {\rm I}\hspace{-1.79993pt}{\rm N}. If Conjecture REF holds, for some the design \mathbb {X} that fulfils the block isometry prop...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.036192119121551514, -0.002878051484003663, -0.054013531655073166, 0.0017451405292376876, 0.016005704179406166, -0.015265688300132751, -0.0025213942863047123, 0.013541528023779392, 0.03597850352525711, 0.0353681817650795, -0.031172215938568115, 0.0032308942172676325, 0.0017308361129835248,...
eaea17fbb88cb9a2762cd0f3e6887d223e786c7c
subsection
25
47
Reduction to the dense subgraph detection problem and a computational lower bound
Assume that there exists a hypothetical estimator \hat{\Theta } computable in polynomial time that attains the rate f(k,d,N) for the prediction error, i.e. such that it holds that\limsup _{n \rightarrow \infty } \frac{1}{f(k,d,N)} \sup _{\Theta _\star \in \mathcal {F}_k} \frac{1}{N}{\rm I}\hspace{-1.79993pt}{\rm E}\big...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.021075231954455376, -0.01918288692831993, -0.034764211624860764, -0.0032448379788547754, 0.03351282328367233, -0.021990882232785225, -0.011582983657717705, 0.022738663479685783, 0.02801891788840294, 0.029545001685619354, -0.06119600310921669, 0.0020373233128339052, -0.01869453862309456, ...
eeb57c4df6ffe7f1fa977843c47c3c7ba1c93df2
subsection
26
47
Reduction to the dense subgraph detection problem and a computational lower bound
For the type II error, we obtain\sup _{\Theta \in \mathcal {G}_k^{\alpha _N}} \mathbf {P}_{\Theta } (\psi = 0) & = \sup _{\Theta \in \mathcal {G}_k^{\alpha _N}} \mathbf {P}_{\Theta } \big (\Vert \hat{\Theta }\Vert _{F,\Omega } < \tau _{d,k}(u)\big ) \\ & \le \sup _{\Theta \in \mathcal {G}_k^{\alpha _N}} \mathbf {P}_{\T...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.039039742201566696, -0.008081135340034962, -0.014330851845443249, -0.008821333758533001, 0.007943779230117798, -0.007527894340455532, 0.008844226598739624, 0.052592240273952484, 0.01753583364188671, 0.021961763501167297, -0.02435786835849285, -0.0011446366552263498, -0.02901272475719452, ...
ca032493822e7fa1962a58a3357476e707ded86f
subsection
27
47
Concluding remarks
Our results shed further light on the emerging topic of statistical and computational trade-offs in high-dimensional estimation. The matrix logistic regression model is very natural to study the connection between statistical accuracy and computational efficiency as the model is based on the study of a generative model...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1214/15-aoas842", "end": 1320, "openalex_id": "https://openalex.org/W1916786071", "raw": "Bogdan, M., van den Berg, E., Sabatti, C., Su, W. and Candès, E. (2015). Slope—adaptive variable selection via convex optimization. The Annals of A...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.04076933488249779, -0.03037864901125431, -0.04357680678367615, 0.01563943549990654, -0.005672155879437923, 0.0043218242935836315, 0.009993980638682842, -0.0206745695322752, 0.033811695873737335, 0.040647272020578384, -0.04864245653152466, 0.011374827474355698, -0.018950417637825012, 0.0...
8d453cd4d8f4a2f36c4f1adde42d69e8953775f5
subsection
28
47
Some geometric properties of the likelihood
Let us recall the stochastic component of the likelihood function\zeta (\Theta ) = \ell _Y(\Theta ) - \ell (\Theta ) = \sum _{(i,j) \in \Omega } \big (Y_{(i,j)} - \pi _{ij}(\Theta _\star ) \big )X_i^\top \Theta X_j \,,which is a linear function in \Theta . The deviation of the gradient \nabla \zeta of the stochastic co...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.030063312500715256, 0.00907240528613329, -0.029834404587745667, -0.01747334562242031, 0.021761564537882805, -0.010659504681825638, -0.00015153264394029975, 0.04480501636862755, 0.0006237755878828466, -0.015901507809758186, -0.053808752447366714, 0.047155145555734634, -0.029056115075945854...
80c97962427f7379b6e2c334c394e5f774978978
subsection
29
47
Some geometric properties of the likelihood
In this notation, we have \zeta (\Theta ) = \savebox { }{\m@th {\langle }}\mathopen {\copy \hspace{1.111pt}\hspace{0.0pt}\usebox { }}\zeta , \Theta \savebox { }{\m@th {\rangle }}\mathclose {\copy \hspace{1.111pt}\hspace{0.0pt}\usebox { }}_F, with\nabla \zeta = \sum _{(i,j) \in \Omega } \varepsilon _{i,j} X_j X_i^\top...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.022798502817749977, 0.015519462525844574, -0.04877109453082085, 0.012551384046673775, 0.003626167308539152, 0.0001790670503396541, 0.019380252808332443, 0.05875115841627121, -0.0017634885152801871, 0.012017282657325268, -0.02275272272527218, 0.002977615687996149, -0.007721581030637026, ...
67ed0f90664cf77368678f3db6c7dcbdfe4c5f3e
subsection
30
47
Some geometric properties of the likelihood
Furthermore, using that \sup _{t\in \mathbf {R}}\sigma ^\prime (t) \le 1/4 , we obtain for all \Theta \in \mathcal {P}(M)\ell (\Theta _\star ) - \ell (\Theta )\le \frac{1}{8} \Vert \mathbb {X}^\top ( \Theta _\star - \Theta ) \mathbb {X}\Vert _{F,\Omega }^2 \,.We shall also be using the bounds\max _{(i,j) \in \Omega } \...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.008315708488225937, 0.03037904016673565, -0.05660785362124443, -0.0016393008409067988, -0.018569206818938255, -0.009231199510395527, -0.0025652816984802485, 0.05157265439629555, 0.0031546291429549456, -0.022765206173062325, 0.0010661653941497207, 0.04131915420293808, -0.015929540619254112...
b7df72d49cbb5ad156ef871f17a5337661cb601f
subsection
31
47
Entropy bounds for some classes of matrices
Recall that an \varepsilon -net of a bounded subset \mathbf {K} of some metric space with a metric \rho is a collection \lbrace K_1,...,K_{N_\varepsilon } \rbrace \in \mathbf {K} such that for each K \in \mathbf {K}, there exists i \in \lbrace 1,...,N_\varepsilon \rbrace such that \rho (K,K_i) \le \varepsilon . The \va...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1109/tit.2011.2111771", "end": 640, "openalex_id": "https://openalex.org/W2162451874", "raw": "Candes, E. and Plan, Y. (2011). Tight oracle inequalities for low-rank matrix recovery from a minimal number of noisy random measurements. IEE...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.018039749935269356, 0.012636981904506683, -0.003914717119187117, -0.003531258087605238, -0.03415647894144058, -0.07270843535661697, 0.010095849633216858, -0.02103111334145069, -0.026647549122571945, -0.004574801307171583, 0.0008055503712967038, -0.008714633993804455, -0.013171154074370861...
784fa8634095de313e816232a33b2a28e5da71fd
subsection
32
47
Proof of Theorem
It suffices to show the following uniform deviation inequality\sup _{\Theta _\star \in \mathcal {P}_{k,r}(M)} \mathbf {P}_{\Theta _\star }\big ( \ell (\Theta _\star ) -\ell (\hat{\Theta }) + p(\hat{\Theta }) > 2 p(\Theta _\star ) + R_t^2\big ) \le e^{- c R_t}\,,for any R_t > 0 and some numeric constant c >0 . Indeed, t...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.03918997943401337, -0.022891344502568245, -0.028003742918372154, -0.006451543886214495, -0.0030311953742057085, 0.04190642014145851, -0.033970754593610764, 0.03934258967638016, -0.0026649339124560356, 0.019747599959373474, -0.02875152789056301, 0.025760391727089882, -0.011056519113481045,...
a4620aa059109f512b6e51ec19be6074bbcc8c08
subsection
33
47
Proof of Theorem
In view of \ell _Y(\hat{\Theta }) - p(\hat{\Theta }) \ge \ell _Y(\Theta _\star ) - p(\Theta _\star ), we have on the complement:\savebox { }{\m@th {\langle }}\mathopen {\copy \hspace{1.111pt}\hspace{0.0pt}\usebox { }}\mathcal {E}_\Omega ,\mathbb {X}^\top (\hat{\Theta }- \Theta _\star )\mathbb {X}\savebox { }{\m@th {\r...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.032095711678266525, 0.023842964321374893, -0.027321016415953636, 0.007585357408970594, -0.005457338877022266, -0.027321016415953636, 0.014400357380509377, 0.07310011982917786, -0.004057727754116058, 0.007165855728089809, -0.00800867285579443, 0.02843460440635681, -0.022637849673628807, ...
04046d2262ef555b3dac0add8a44138e635232d8
subsection
34
47
Proof of Theorem
It follows,&\mathbf {P}_{\Theta _\star }\Big (\sup _{\tau (\Theta ; \Theta _\star ) \ge R_t} \frac{\savebox { }{\m@th {\langle }}\mathopen {\copy \hspace{1.111pt}\hspace{0.0pt}\usebox { }}\mathcal {E}_\Omega ,\mathbb {X}^\top (\Theta - \Theta _\star )\mathbb {X}\savebox { }{\m@th {\rangle }}\mathclose {\copy \hspace{...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.01443931832909584, 0.046712297946214676, -0.029015935957431793, -0.014866471290588379, -0.009778766892850399, -0.0184591393917799, 0.02552242949604988, 0.025537684559822083, -0.01465289480984211, 0.004443160258233547, -0.00974062830209732, -0.01465289480984211, -0.01801672950387001, 0.0...
1cc4dbb5f8213e267891390f4e2035f3e73aa78d
subsection
35
47
Proof of Theorem
Then following the lines of the proof of Lemma REF and using the singular value decomposition, we deriveH(\varepsilon , \lbrace \mathbb {X}^\top \Theta ^\prime \mathbb {X}:\Theta ^\prime \in \mathcal {G}_1, \Vert \mathbb {X}^\top \Theta ^\prime \mathbb {X}\Vert _{F,\Omega } \le B \rbrace , \Vert \cdot \Vert _{F,\Omega ...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.0034296719823032618, 0.01884602941572666, -0.018495049327611923, -0.018174590542912483, -0.008019099943339825, -0.026399699971079826, 0.0006847899057902396, -0.01893758773803711, 0.0005426817224361002, -0.01664859801530838, -0.023927589878439903, -0.005463059525936842, -0.0370816588401794...
cabe88d61e9b2e6a5da708d78f9a6d9e71e1bd0d
subsection
36
47
Proof of Theorem
By Dudley's entropy integral bound, see and for a more recent reference, we then have{\rm I}\hspace{-1.79993pt}{\rm E}\big [&\sup _{{\Theta \in G_{2^s R_t}(\Theta _\star ) \\k(\Theta )=K, \, \operatorname{\mathbf {rank}}(\Theta ) = R}} \savebox { }{\m@th {\langle }}\mathopen {\copy \hspace{1.111pt}\hspace{0.0pt}\usebo...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-1-4419-5821-1_11", "end": 868, "openalex_id": "https://openalex.org/W2092074494", "raw": "Dudley, R. (1967). The sizes of compact subsets of hilbert space and continuity of gaussian processes. Journal of Functional Analysis 1 29...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.02402760088443756, 0.05278444290161133, 0.006182339508086443, 0.004046552814543247, -0.03624735027551651, -0.03040444850921631, -0.013623268343508244, -0.05839851126074791, 0.01935938000679016, -0.010709444992244244, -0.018581343814730644, -0.0026068040169775486, 0.02787201665341854, -0...
8e488bd3666e57f6757e2cef15f048c5158583f4
subsection
37
47
Proof of Theorem
Furthermore, by Bousquet's version of Talagrand's inequality, see Theorem REF , in view of the bounds (REF ) and (), we have for all u > 0\mathbf {P}_{\Theta _\star }\Big (&\sup _{{\Theta \in G_{2^s R_t}(\Theta _\star ) \\k(\Theta )=K, \, \operatorname{\mathbf {rank}}(\Theta ) = R}} \savebox { }{\m@th {\langle }}\matho...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.02481619268655777, 0.03281353786587715, -0.04651891440153122, -0.0017436961643397808, 0.0030104515608400106, 0.016498342156410217, 0.022252157330513, -0.01758195273578167, 0.02103118970990181, 0.003384373150765896, -0.0344923697412014, 0.014643996022641659, -0.0024820007383823395, 0.003...
f6e1d8bb34e0ba20b33a0a09443e6734073561e2
subsection
38
47
Proof of Theorem
Plugging this back into (REF ) and using (REF ), we obtain\mathbf {P}_{\Theta _\star }\Big (\sup _{{\Theta \in G_{2^s R_t}(\Theta _\star ) \\k(\Theta )=K, \, \operatorname{\mathbf {rank}}(\Theta ) = R}} \savebox { }{\m@th {\langle }}\mathopen {\copy \hspace{1.111pt}\hspace{0.0pt}\usebox { }}\mathcal {E}_\Omega ,\mathb...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1016/s1631-073x(02)02292-6", "end": 1405, "openalex_id": "https://openalex.org/W2167784882", "raw": "Bousquet, O. (2002). A bennett concentration inequality and its application to suprema of empirical processes. Comptes Rendus Mathematiq...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.018613167107105255, 0.028163857758045197, -0.02538713812828064, -0.01392936147749424, -0.006007323041558266, 0.00015948012878652662, 0.020672820508480072, 0.006453581620007753, 0.004954611416906118, 0.0025936379097402096, -0.03408345580101013, 0.0050537800416350365, -0.007189717143774033,...
4c5c55f14dd142048e8db69735c8507a8106239d
subsection
39
47
Proof of Theorem
Let \mathcal {F} be a countable set of measurable real-valued functions on B such that f(\varepsilon _i)\le b < \infty a.s. and {\rm I}\hspace{-1.79993pt}{\rm E}f(\varepsilon _i) = 0 for all i = 1,...,n, f \in \mathcal {F} .
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.022754574194550514, -0.006943884305655956, -0.014940797351300716, 0.020221201702952385, 0.026600418612360954, -0.025715263560414314, -0.008035066537559032, -0.009141509421169758, -0.007748917210847139, -0.010301366448402405, -0.025715263560414314, -0.004341835156083107, -0.001058751600794...
1fc690d6db4bb794b78896c16a7866fe452f807e
subsection
40
47
Proof of Theorem
LetS := \sup _{f \in \mathcal {F}}\sum _{i=1}^{n}f(\varepsilon _i)\,, \quad \quad v:= \sup _{f \in \mathcal {F}}\sum _{i=1}^{n} {\rm I}\hspace{-1.79993pt}{\rm E}[f^2(\varepsilon _i)] .Then for all u > 0, it holds that\mathbf {P}\Big (S - {\rm I}\hspace{-1.79993pt}{\rm E}[S] \ge \sqrt{2(v + 2b{\rm I}\hspace{-1.79993pt}{...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.028513051569461823, 0.024271944537758827, -0.03661387041211128, -0.02122078835964203, 0.010190860368311405, -0.02428719960153103, -0.00481701223179698, -0.03365425020456314, 0.024455014616250992, -0.020808883011341095, -0.044974036514759064, 0.01080109179019928, -0.00030773767502978444, ...
5e47898508ea742a63ebbd2894682fbe42a17972
subsection
41
47
Proof of Theorem
Plugging this bound back into (REF ) and using the block isometry property yields the desired assertion.The proof is split into two parts. First, we show a lower bound of the order kr and then a lower bound of the order k \log (de/k). A simple inequality (a + b)/2 \le \max \lbrace a,b\rbrace for all a, b > 0 then comp...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/b13794", "end": 471, "openalex_id": "https://openalex.org/W1511694993", "raw": "Tsybakov, A. (2008). Introduction to Nonparametric Estimation. 1st ed. Springer.", "source_ref_id": "593662d8e29446ba66e0ad36346e2c2ebfc8eb41", ...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.04579683020710945, 0.009633810259401798, -0.05040397122502327, 0.00888629350811243, -0.001495033036917448, 0.0242561474442482, 0.01919134333729744, 0.01306628342717886, 0.0070251296274363995, 0.030861753970384598, -0.032585617154836655, 0.01720813475549221, -0.02061009779572487, 0.07792...
8451a9bc4f194b99211d2e3ab675d4d36fa9c78d
subsection
42
47
Proof of Theorem
Thus \mathcal {B}^\circ is a 2\delta -separated set in the Frobenius metric with \delta ^2 = \frac{kr}{64} \big ( \frac{\alpha _N}{2}\big )^2\lfloor \frac{k}{r} \rfloor . The Kullback-Leibler divergence between the measures \mathbf {P}_{\Theta _{u}} and \mathbf {P}_{\Theta _{v}}, \Theta _{u}, \Theta _{v} \in \mathcal ...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.09905534982681274, 0.008559773676097393, -0.05807219073176384, -0.021254481747746468, 0.00587053969502449, 0.007083555683493614, -0.01686015911400318, 0.029142901301383972, 0.001640241825953126, 0.03814515843987465, -0.05941489711403847, 0.029783738777041435, -0.018050288781523705, 0.01...
91a6c851015db5ed5ca3c67dac40d5f1060a0953
subsection
43
47
Proof of Theorem
Using simple calculations, we then have (2k -1)\alpha _N^2 \le \Vert \Theta _u - \Theta _v \Vert _F^2 \le 2 k^2\alpha _N^2 for all \Theta _u, \Theta _v \in \mathcal {G}_k^{\alpha _N}, u\ne v. Furthermore, according to Lemma REF to follow, there exists a subset \mathcal {G}^{\alpha _N, 0}_{k} \subset \mathcal {G}_k^{\al...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.06701536476612091, 0.0025424486957490444, -0.059508178383111954, -0.029204782098531723, 0.014877044595777988, 0.010795393958687782, 0.007827614434063435, 0.01954614743590355, -0.008033604361116886, 0.01256538089364767, -0.023192930966615677, 0.025954721495509148, -0.023864304646849632, ...
159171ec3053c1c65c58ddb17199d87d98843384
subsection
44
47
Proof of Theorem
As in the first part of the proof, taking \gamma > 0 small enough, we obtain\frac{1+ \Delta _{\Omega ,2k}(\mathbb {X})}{4} k^2\alpha _N^2N +\log 2 &= k \gamma \log (2) \log \big (\frac{de}{k}\big ) +\log 2 = \log (2^{k \gamma \log (de/k) + 1})\\ &< \log (2^{\rho k \log (de/k)} + 1) \,.The desired lower bound then follo...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.0355011448264122, 0.011347850784659386, -0.04334619641304016, 0.0026919664815068245, -0.014186721295118332, -0.001299241092056036, 0.03241806849837303, 0.01811687834560871, -0.03577587381005287, 0.04157571494579315, -0.030235497280955315, 0.034158021211624146, -0.023123057559132576, 0.0...
6502149a6ae6b12f5692afc9bef483bef274a173
subsection
45
47
Proof of Theorem
Then there exists a subset \mathcal {G}^0 = \lbrace E^{(0)},..., E^{(J)}\rbrace \subset \mathcal {G} of cardinality\log J := \log (\operatorname{\mathsf {card}}(\mathcal {G}^0 ))\ge \rho k \log (\frac{de}{k})\,,where \rho = \frac{\alpha }{- \log (\alpha \beta )}(- \log \beta +\beta -1 ) such that\rho _H(E^{(k)}, E^{(l)...
{ "cite_spans": [] }
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.016538800671696663, 0.009901920333504677, -0.016294684261083603, 0.019300352782011032, 0.001684012939222157, -0.029400616884231567, -0.00940606091171503, 0.014685050584375858, -0.014433307573199272, 0.0023705868516117334, -0.009993462823331356, 0.023129908367991447, -0.01901046559214592, ...
430cf471ee160cfd1fffd373649928a2c2d87834
subsection
46
47
Proof of Theorem
For instance, in order to get the distance between matrices 2k^2, i.e. c = 2 we need to shift all the k columns (and consequently rows) and so the number of distinct columns of a matrix is m = k, and in order to get the minimal possible distance 4k - 2, i.e. c = (4k - 2)/k^2 we need to shift only one column and a corre...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-3-540-48503-2", "end": 1191, "openalex_id": "https://openalex.org/W156419107", "raw": "Massart, P. (2007). Concentration inequalities and model selection, vol. 6. Springer.", "source_ref_id": "a37e3e039449b03ed507a8ba3c7c5...
1803.07054
Optimal link prediction with matrix logistic regression
[ "Nicolai Baldin", "Quentin Berthet" ]
[ "math.ST", "stat.ML", "stat.TH" ]
2,018
en
Mathematics
[ -0.016498509794473648, 0.01625431329011917, -0.04316169023513794, -0.0033996698912233114, -0.023213908076286316, 0.02561008371412754, -0.006604745984077454, -0.009142098017036915, -0.04825928807258606, 0.023671776056289673, -0.023244433104991913, 0.008325566537678242, -0.005036548245698214, ...
87727d5f3716d801c3a685abe211f6d9ca73ac94
abstract
0
41
Abstract
We study the question of whether coherent neutrino scattering can occur on macroscopic scales, leading to a significant increase of the detection cross section. We concentrate on radiative neutrino scattering on atomic electrons (or on free electrons in a conductor). Such processes can be coherent provided that the net...
{ "cite_spans": [] }
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ -0.022234583273530006, -0.010125357657670975, -0.029681717976927757, 0.009682802483439445, 0.00014199440192896873, -0.03921954706311226, -0.008194901049137115, 0.031619805842638016, 0.01642032340168953, 0.03369523584842682, 0.012109225615859032, -0.005757032427936792, -0.03171136975288391, ...
942a4b289a566d0c4a7113fd93bab68b86aa48dc
subsection
1
41
Introduction
Recently, the COHERENT collaboration has reported the first observation of coherent elastic neutrino–nucleus scattering , , a process predicted over forty years ago , . This observation completed the standard-model picture of neutrino interactions with nucleons and nuclei and opened up a new window to probe physics bey...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1126/science.aao0990", "end": 168, "openalex_id": "https://openalex.org/W3103990895", "raw": "D. Akimov et al. [COHERENT Collaboration], “Observation of Coherent Elastic Neutrino-Nucleus Scattering,” Science 357 (2017) no.6356, 1123 [arX...
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ 0.024674775078892708, 0.000513580976985395, -0.040346384048461914, 0.0182657428085804, 0.005695646163076162, -0.04672490060329437, -0.007568762172013521, 0.019471250474452972, 0.024262765422463417, 0.0365009680390358, 0.020585201680660248, -0.003971311263740063, -0.022904662415385246, 0.00...
612a56c86645a7afadc0196f355b93cb68557a70
subsection
2
41
Macroscopic coherence?
How about scattering with coherence on macroscopic scales? Clearly, this would require measuring even much smaller recoil energies and so does not look practical. It is interesting, however, to inquire what could be the increase of the detection cross sections if such measurements were possible, leaving for the moment ...
{ "cite_spans": [] }
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ 0.011177144013345242, 0.0006313369376584888, -0.025406066328287125, -0.02996847778558731, 0.024460015818476677, -0.03744533658027649, -0.018753187730908394, -0.01180275809019804, 0.013359165750443935, 0.04748569428920746, 0.018325937911868095, 0.015464894473552704, -0.02372758835554123, 0....
f9b7d5f68b4a469b9c8b02cbb90d36a34cf67403
subsection
3
41
Macroscopic coherence?
For q_0\sim (1~{\rm cm})^{-1}\simeq 2\times 10^{-5} eV and the total mass of particles in the coherent volume m_t\sim 1\,g, one finds E_{rec}\sim \vec{q_0}^2/2m_t\sim 10^{-43}~{\rm eV}, the quantity which is not going to be ever measured. To give just one reason for that, in order to measure recoil energy of this magni...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1689, "openalex_id": "", "raw": "J. Weber, “Gravitons, neutrinos, and anti-neutrinos,” Found. Phys. 14 (1984) 1185.", "source_ref_id": "c7af87d093b27031c136a3221ce2bbadd7a633c9", "start": 1620 }, { "arxiv...
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ -0.032690294086933136, 0.01412458810955286, -0.034704823046922684, -0.015444714576005936, -0.002701298799365759, -0.01287313923239708, -0.007203463464975357, 0.012392398901283741, 0.010072640143334866, 0.02852388471364975, 0.025578400120139122, 0.028081297874450684, -0.008981437422335148, ...
8030f965cc8a1e9d8fa07a936997220207d5fc00
subsection
4
41
Weber's approach and structure factors
Weber suggested to detect neutrinos through their coherent scattering on crystals in torsion balance experiments. This approach combines two interesting ideas. First, as the force coincides with momentum transfer per unit time, the force neutrinos impinge on a crystal is directly related to the momentum transfer to the...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-1-4684-8911-8_1", "end": 924, "openalex_id": "https://openalex.org/W95097360", "raw": "P. G. Debrunner and H. Frauenfelder, Introduction to the Mössbauer effect, in “An introduction to Mössbauer spectroscopy”, Ed. L. May, Plenum...
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ 0.023272115737199783, 0.009530122391879559, -0.061713073402643204, 0.02566799893975258, 0.007332623936235905, -0.027865497395396233, 0.009316476061940193, 0.014222696423530579, 0.043431106954813004, 0.027865497395396233, 0.016603318974375725, 0.015092539601027966, -0.04169142246246338, -0....
43203d8df8f1780a1170400efb6101f5057eb4ef
subsection
5
41
Weber's approach and structure factors
For elastic neutrino scattering the structure factor is given byF(\vec{k}-\vec{k}^{\prime }) =\sum _{i=1}^N e^{i(\vec{k}-\vec{k}^{\prime })\vec{r}_i}\,,where \vec{k} and \vec{k}^{\prime } are the momenta of the incident and scattered neutrinos, \vec{r}_i is the coordinate of the ith scatterer and N is the total number ...
{ "cite_spans": [] }
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ -0.0016220157267525792, -0.0062782783061265945, -0.030346283689141273, -0.024457059800624847, -0.0018203575164079666, -0.02837812341749668, 0.004805972333997488, 0.017743956297636032, 0.0017383508384227753, 0.028256068006157875, -0.0027519913855940104, 0.020185085013508797, -0.02940034680068...
3996ee64dad96ec382293c575815928a7f0f6b65
subsection
6
41
Weber's approach and structure factors
This leads to the well known Bragg condition for diffraction on crystals,2d\sin \vartheta = n\lambda \,,where d is the interplanar distance in the crystal, \vartheta is the angle between the neutrino momentum and the atomic plane (the scattering angle being \theta =2\vartheta ), \lambda =2\pi /|\vec{k}| and n is an int...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 763, "openalex_id": "https://openalex.org/W2010872619", "raw": "L. D. Landau and E. M. Lifshitz, Electrodynamics of Continuous Media. Second Edition revised and enlarged by E. M. Lifshitz and L. P. Pitaevsky, Pergamon Press, 1984,...
10.1007/JHEP10(2018)045
1806.10962
Coherent scattering and macroscopic coherence: Implications for neutrino, dark matter and axion detection
[ "Evgeny Akhmedov", "Giorgio Arcadi", "Manfred Lindner", "Stefan Vogl" ]
[ "hep-ph", "astro-ph.CO", "hep-ex", "nucl-ex" ]
2,018
en
Physics
[ 0.02658705972135067, -0.022206759080290794, -0.04102526977658272, -0.02020738646388054, 0.005265519954264164, -0.015216589905321598, 0.008974277414381504, 0.02399245649576187, 0.008615612052381039, 0.04108631983399391, 0.014224535785615444, 0.02112312987446785, -0.009332943707704544, -0.00...