[Optimalité du critère varentropique sur les expanseurs]
The cutoff phenomenon is an abrupt transition from out of equilibrium to equilibrium undergone by certain Markov processes in the limit where the size of the state space tends to infinity: instead of decaying gradually over time, their distance to equilibrium remains close to the maximal value for a while and suddenly drops to zero as the time parameter reaches a critical threshold. Despite the accumulation of many examples, this phenomenon is still far from being understood, and identifying the general conditions that trigger it has become one of the biggest challenges in the quantitative analysis of finite Markov chains. Very recently, the author proposed a general sufficient condition for the occurrence of a cutoff, based on a certain information-theoretical statistics called varentropy. In the present paper, we demonstrate the sharpness of this approach by showing that the cutoff phenomenon is actually equivalent to the varentropy criterion for all sparse, fast-mixing chains. Reversibility is not required.
Le cutoff désigne une transition abrupte dans la convergence à l’équilibre de certains processus de Markov, dans la limite où le nombre d’états tend vers l’infini : au lieu de décroître graduellement au cours du temps, la distance à l’équilibre reste longtemps proche de sa valeur initiale, puis chute brutalement à zéro lorsque le paramètre temporel atteint un seuil critique. Malgré l’accumulation de nombreux exemples, ce phénomène est toujours largement incompris, et l’identification des mécanismes généraux qui le sous-tendent constitue l’un des défis majeurs dans l’analyse quantitative des chaînes de Markov. Récemment, l’auteur a proposé une condition suffisante générale pour le cutoff, basée sur une quantité informationelle appelée varentropie. Dans cet article, nous démontrons l’optimalité de ce critère en montrant qu’il est en réalité équivalent au cutoff pour toutes les chaînes rapides et parcimonieuses. La réversibilité en temps n’est pas requise.
Accepté le :
Publié le :
DOI : 10.5802/ahl.199
Keywords: Cutoff phenomenon, varentropy, expanders
Salez, Justin  1
CC-BY 4.0
@article{AHL_2024__7__239_0,
author = {Salez, Justin},
title = {The varentropy criterion is sharp on expanders},
journal = {Annales Henri Lebesgue},
pages = {239--250},
year = {2024},
publisher = {\'ENS Rennes},
volume = {7},
doi = {10.5802/ahl.199},
mrnumber = {4765357},
zbl = {1544.60080},
language = {en},
url = {https://www.numdam.org/articles/10.5802/ahl.199/}
}
Salez, Justin. The varentropy criterion is sharp on expanders. Annales Henri Lebesgue, Tome 7 (2024), pp. 239-250. doi: 10.5802/ahl.199
[AD86] Shuffling cards and stopping times, Am. Math. Mon., Volume 93 (1986), pp. 333-348 | MR | DOI | Zbl
[Ald83] Random walks on finite groups and rapidly mixing Markov chains, Seminar on probability, XVII (Lecture Notes in Mathematics), Volume 986, Springer, 1983, pp. 243-297 | Numdam | MR | Zbl
[Ald89] Hitting times for random walks on vertex-transitive graphs, Math. Proc. Camb. Philos. Soc., Volume 106 (1989) no. 1, pp. 179-191 | DOI | MR | Zbl
[BCS18] Random walk on sparse random digraphs, Probab. Theory Relat. Fields, Volume 170 (2018) no. 3-4, pp. 933-960 | DOI | MR | Zbl
[BCS19] Cutoff at the “entropic time” for sparse Markov chains, Probab. Theory Relat. Fields, Volume 173 (2019) no. 1-2, pp. 261-292 | DOI | MR | Zbl
[BH20] A threshold for cutoff in two-community random graphs, Ann. Appl. Probab., Volume 30 (2020) no. 4, pp. 1824-1846 | DOI | MR | Zbl
[BHS17] Cutoff for nonbacktracking random walks on sparse random graphs, Ann. Probab., Volume 45 (2017) no. 3, pp. 1752-1770 | DOI | MR | Zbl
[BL22] Cutoff at the entropic time for random walks on covered expander graphs, J. Inst. Math. Jussieu, Volume 21 (2022) no. 5, pp. 1571-1616 | DOI | MR | Zbl
[BLPS18] Random walks on the random graph, Ann. Probab., Volume 46 (2018) no. 1, pp. 456-490 | DOI | MR | Zbl
[BT06] Modified logarithmic Sobolev inequalities in discrete settings, J. Theor. Probab., Volume 19 (2006) no. 2, pp. 289-336 | DOI | MR | Zbl
[Dia96] The cutoff phenomenon in finite Markov chains, Proc. Natl. Acad. Sci. USA, Volume 93 (1996) no. 4, pp. 1659-1664 | DOI | MR | Zbl
[DS81] Generating a random permutation with random transpositions, Z. Wahrscheinlichkeitstheor. Verw. Geb., Volume 57 (1981) no. 2, pp. 159-179 | DOI | MR | Zbl
[DSC96] Logarithmic Sobolev inequalities for finite Markov chains, Ann. Appl. Probab., Volume 6 (1996) no. 3, pp. 695-750 | DOI | MR | Zbl
[Her17] Cutoff for Ramanujan graphs via degree inflation, Electron. Commun. Probab., Volume 22 (2017), 45 | DOI | MR | Zbl
[HLW06] Expander graphs and their applications, Bull. Am. Math. Soc., Volume 43 (2006) no. 4, pp. 439-561 | DOI | MR | Zbl
[HSS22b] Universality of cutoff for graphs with an added random matching, Ann. Probab., Volume 50 (2022) no. 1, pp. 203-240 | DOI | MR | Zbl
[HŠS22a] Cutoff for random walk on random graphs with a community structure (2022) | arXiv | DOI
[KV13] Optimal lossless compression: Source varentropy and dispersion, 2013 IEEE International Symposium on Information Theory (IEEE International Symposium on Information Theory - Proceedings), IEEE (2013), pp. 1739-1743 | DOI
[LP16] Cutoff on all Ramanujan graphs, Geom. Funct. Anal., Volume 26 (2016) no. 4, pp. 1190-1216 | DOI | MR | Zbl
[LP17] Markov chains and mixing times, American Mathematical Society, 2017 (Second edition of [ MR2466937], with contributions by Elizabeth L. Wilmer, with a chapter on “Coupling from the past” by James G. Propp and David B. Wilson) | Zbl | DOI | MR
[LS10] Cutoff phenomena for random walks on random regular graphs, Duke Math. J., Volume 153 (2010) no. 3, pp. 475-510 | MR | DOI | Zbl
[LS11] Explicit expanders with cutoff phenomena, Electron. J. Probab., Volume 16 (2011), 15, pp. 419-435 | DOI | MR | Zbl
[MT06] Mathematical aspects of mixing times in Markov chains, Found. Trends Theor. Comput. Sci., Volume 1 (2006) no. 3 | MR | Zbl
[Oza20] An entropic proof of cutoff on Ramanujan graphs, Electron. Commun. Probab., Volume 25 (2020), 77 | MR | Zbl | DOI
[Per04] AIM Research Workshop on Sharp Thresholds for Mixing Times, 2004 http://www.aimath.org/wwn/mixingtimes.html
[Sal21] Cutoff for non-negatively curved Markov chains (2021) (to appear in J. Eur. Math. Soc.) | arXiv | DOI
[TT20] Sharp relations between volume growth, isoperimetry and resistance in vertex-transitive graphs (2020) | arXiv | DOI
[TT21] A finitary structure theorem for vertex-transitive graphs of polynomial growth, Combinatorica, Volume 41 (2021) no. 2, pp. 263-298 | MR | DOI | Zbl
Cité par Sources :





