Functional inequalities for perturbed measures with applications to log-concave measures and to some Bayesian problems

We study functional inequalities (Poincaré, Cheeger, log-Sobolev) for probability measures obtained as perturbations. Several explicit results for general measures as well as log-concave distributions are given. The initial goal of this work was to obtain explicit bounds on the constants in view of statistical applications. These results are then applied to the Langevin Monte-Carlo method used in statistics in order to compute Bayesian estimators.

[1]  Piotr Nayar,et al.  Concentration inequalities and geometry of convex bodies , 2013 .

[2]  E. Lieb,et al.  On extensions of the Brunn-Minkowski and Prékopa-Leindler theorems, including inequalities for log concave functions, and with an application to the diffusion equation , 1976 .

[3]  Yuansi Chen.  An Almost Constant Lower Bound of the Isoperimetric Coefficient in the KLS Conjecture , 2020, 2011.13661.

[4]  Patrick Cattiaux,et al.  On the Poincaré Constant of Log-Concave Measures , 2018, Lecture Notes in Mathematics.

[5]  E. Milman.  Isoperimetric Bounds on Convex Manifolds , 2010, 1004.0485.

[6]  Tyler Maunu,et al.  Exponential ergodicity of mirror-Langevin diffusions , 2020, NeurIPS.

[7]  R. Douc,et al.  Subgeometric rates of convergence of f-ergodic strong Markov processes , 2006, math/0605791.

[8]  M. Ledoux,et al.  Analysis and Geometry of Markov Diffusion Operators , 2013 .

[9]  P. Cattiaux,et al.  A note on Talagrand’s transportation inequality and logarithmic Sobolev inequality , 2008, 0810.5435.

[10]  S. Meyn,et al.  Stability of Markovian processes III: Foster–Lyapunov criteria for continuous-time processes , 1993, Advances in Applied Probability.

[11]  Arnak S. Dalalyan,et al.  Sparse Regression Learning by Aggregation and Langevin Monte-Carlo , 2009, COLT.

[12]  David Alonso-Gutiérrez,et al.  Approaching the Kannan-Lovász-Simonovits and Variance Conjectures , 2015 .

[13]  F. Malrieu,et al.  Functional inequalities for Gaussian convolutions of compactly supported measures: explicit bounds and dimension dependence , 2015, 1507.02389.

[14]  P. Cattiaux,et al.  Self-improvement of the Bakry-Emery criterion for Poincaré inequalities and Wasserstein contraction using variable curvature bounds , 2020, Journal de Mathématiques Pures et Appliquées.

[16]  Spectral gap of positive operators and applications , 2000 .

[17]  B. Klartag,et al.  Spectral gaps, symmetries and log-concave perturbations , 2019, 1907.01823.

[18]  P. Cattiaux,et al.  Semi Log-Concave Markov Diffusions , 2013, 1303.6884.

[19]  N. Berestycki,et al.  Concentration of measure , 2019, High-Dimensional Statistics.

[20]  Yee Whye Teh,et al.  Bayesian Learning via Stochastic Gradient Langevin Dynamics , 2011, ICML.

[21]  E. Milman.  Isoperimetric and Concentration Inequalities - Equivalence under Curvature Lower Bound , 2009, 0902.1560.

[22]  E. Milman.  On the role of convexity in isoperimetry, spectral gap and concentration , 2007, 0712.4092.

[23]  P. Cattiaux,et al.  Hitting times, functional inequalities, lyapunov conditions and uniform ergodicity , 2016, 1604.06336.

[24]  S. Bobkov.  Isoperimetric and Analytic Inequalities for Log-Concave Probability Measures , 1999 .

[25]  P. Cattiaux.  Hypercontractivity for perturbed diffusion semigroups , 2005, math/0510258.

[26]  M. Ledoux.  From Concentration to Isoperimetry: Semigroup Proofs , 2011 .

[27]  Miklós Simonovits,et al.  Isoperimetric problems for convex bodies and a localization lemma , 1995, Discret. Comput. Geom..

[28]  S. Meyn,et al.  Exponential and Uniform Ergodicity of Markov Processes , 1995 .

[29]  Michel Bonnefont,et al.  A NOTE ON SPECTRAL GAP AND WEIGHTED POINCAR ´ E INEQUALITIES FOR SOME ONE-DIMENSIONAL DIFFUSIONS , 2014, 1411.5982.

[30]  D. Bakry,et al.  Rate of convergence for ergodic continuous Markov processes : Lyapunov versus Poincaré , 2007, math/0703355.

[31]  P. Cattiaux,et al.  Entropic multipliers method for Langevin diffusion and weighted log Sobolev inequalities , 2017, Journal of Functional Analysis.

[32]  F. Barthe,et al.  Transference Principles for Log-Sobolev and Spectral-Gap with Applications to Conservative Spin Systems , 2012, 1202.5318.

[33]  Arnak S. Dalalyan,et al.  On sampling from a log-concave density using kinetic Langevin diffusions , 2018, Bernoulli.

[34]  Patrick Cattiaux,et al.  Poincaré inequalities and hitting times , 2010, 1012.5274.

[35]  J. Wellner,et al.  Log-Concavity and Strong Log-Concavity: a review. , 2014, Statistics surveys.

[36]  D. Bakry,et al.  A simple proof of the Poincaré inequality for a large class of probability measures , 2008 .

[37]  M. Ledoux,et al.  Perturbations of functional inequalities using growth conditions , 2007 .

[38]  Alain Durmus,et al.  High-dimensional Bayesian inference via the unadjusted Langevin algorithm , 2016, Bernoulli.

[39]  R. Latala.  On Some Problems Concerning Log-Concave Random Vectors , 2017 .

[40]  P. Cattiaux.  A Pathwise Approach of Some Classical Inequalities , 2004 .

[41]  B. Klartag,et al.  A Berry-Esseen type inequality for convex bodies with an unconditional basis , 2007, 0705.0832.

[42]  S. Aida,et al.  Logarithmic Sobolev Inequalities and Spectral Gaps: Perturbation Theory , 1994 .

[43]  David Zimmermann,et al.  Logarithmic Sobolev inequalities for mollified compactly supported measures , 2013, 1411.1688.

[44]  H. Weinberger,et al.  An optimal Poincaré inequality for convex domains , 1960 .

[45]  Santosh S. Vempala,et al.  Eldan's Stochastic Localization and the KLS Hyperplane Conjecture: An Improved Lower Bound for Expansion , 2016, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[46]  A. Dalalyan.  Theoretical guarantees for approximate sampling from smooth and log‐concave densities , 2014, 1412.7392.