On fine properties of mixtures with respect to concentration of measure and Sobolev type inequalities

DOI10.1214/08-AIHP309MaRDI QIDQ974769zbMATHFDO

Authors Djalil Chafaï, Florent Malrieu

Publication date 7 June 2010

Published in Annales de l'Institut Henri Poincaré. Probabilités et Statistiques (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/0805.0987

zbMATH Keywords

mixtures of distributionsconcentration of measureGaussian boundsfunctional inequalitiesmass transportationdeviation inequalitiestransportation of measureband isoperimetryfinite Gaussian mixturesGross logarithmic Sobolev inequalitiesMallows or Wasserstein distancetails probabilitiestransportation cost distancesPoincaré inequalities

Mathematics Subject Classification ID

Variational problems in a geometric measure-theoretic setting (49Q20) Inequalities; stochastic orderings (60E15) Statistical distribution theory (62E99) Sobolev spaces and other spaces of ``smooth functions, embedding theorems, trace theorems (46E35)

Abstract: Mixtures are convex combinations of laws. Despite this simple definition, a mixture can be far more subtle than its mixed components. For instance, mixing Gaussian laws may produce a potential with multiple deep wells. We study in the present work fine properties of mixtures with respect to concentration of measure and Sobolev type functional inequalities. We provide sharp Laplace bounds for Lipschitz functions in the case of generic mixtures, involving a transportation cost diameter of the mixed family. Additionally, our analysis of Sobolev type inequalities for two-component mixtures reveals natural relations with some kind of band isoperimetry and support constrained interpolation via mass transportation. We show that the Poincar'e constant of a two-component mixture may remain bounded as the mixture proportion goes to 0 or 1 while the logarithmic Sobolev constant may surprisingly blow up. This counter-intuitive result is not reducible to support disconnections, and appears as a reminiscence of the variance-entropy comparison on the two-point space. As far as mixtures are concerned, the logarithmic Sobolev inequality is less stable than the Poincar'e inequality and the sub-Gaussian concentration for Lipschitz functions. We illustrate our results on a gallery of concrete two-component mixtures. This work leads to many open questions.

Recommendations

• On boundary conditions for a certain class of problems in mixture theory
• Dimension-free log-Sobolev inequalities for mixture distributions
• On certain convex sets of measures and on phases of reacting mixtures
• scientific article; zbMATH DE number 4076186
• scientific article; zbMATH DE number 1405932
• On some properties of \(\alpha\)-mixtures
• Mixture estimate in fractional sense and its application to the well-posedness of the Boltzmann equation with very soft potential
• scientific article; zbMATH DE number 861303
• scientific article; zbMATH DE number 6840955
• Existence theorems in the theory of mixtures

Cites work

• scientific article; zbMATH DE number 1815382 (Why is no real title available?)
• scientific article; zbMATH DE number 3731128 (Why is no real title available?)
• scientific article; zbMATH DE number 49698 (Why is no real title available?)
• scientific article; zbMATH DE number 3636997 (Why is no real title available?)
• scientific article; zbMATH DE number 1069282 (Why is no real title available?)
• scientific article; zbMATH DE number 1790018 (Why is no real title available?)
• scientific article; zbMATH DE number 194758 (Why is no real title available?)
• scientific article; zbMATH DE number 1909499 (Why is no real title available?)
• scientific article; zbMATH DE number 1405932 (Why is no real title available?)
• scientific article; zbMATH DE number 3894218 (Why is no real title available?)
• A Topological Application of the Isoperimetric Inequality
• A new criterion for the logarithmic Sobolev inequality and two applications
• A simple proof of the blowing-up lemma (Corresp.)
• Bounding \(\bar d\)-distance by informational divergence: A method to prove measure concentration
• Concentration of normalized sums and a central limit theorem for noncorrelated random variables
• Elementary bounds on Poincaré and log-Sobolev constants for decomposable Markov chains
• Exponential integrability and transportation cost related to logarithmic Sobolev inequalities
• Finite mixture models
• Generalized symmetric polynomials and an approximate de Finetti representation
• Infinite dimensional convexity.
• Interpolated inequalities between exponential and Gaussian, Orlicz hypercontractivity and isoperimetry
• Lectures on Choquet's theorem
• Logarithmic Sobolev Inequalities
• Logarithmic Sobolev inequalities and stochastic Ising models
• Logarithmic Sobolev inequalities for finite Markov chains
• Markov chain decomposition for convergence rate analysis
• Measure concentration for compound Poisson distributions
• Monotonicity properties of optimal transportation and the FKG and related inequalities
• On logarithmic Sobolev inequalities. With a preface of Dominique Bakry and Michel Ledoux
• Optimal Transport
• Perturbations of functional inequalities using growth conditions
• Probability Inequalities for Sums of Bounded Random Variables
• Sobolev inequalities for probability measures on the real line
• Some inequalities satisfied by the quantities of information of Fisher and Shannon
• Spectral gaps for spin systems: Some non-convex phase examples
• Statistical analysis of finite mixture distributions
• Superadditivity of Fisher's information and logarithmic Sobolev inequalities
• The concentration of measure phenomenon
• Transportation cost for Gaussian and other product measures
• Transportation cost-information inequalities and applications to random dynamical systems and diffusions.
• Weighted Csiszár-Kullback-Pinsker inequalities and applications to transportation inequalities

Cited in

(13)

• Rapid convergence of the unadjusted Langevin algorithm: isoperimetry suffices
• Dimension-free log-Sobolev inequalities for mixture distributions
• Stein's density method for multivariate continuous distributions
• Poincaré and logarithmic Sobolev inequalities by decomposition of the energy landscape
• Variational methods for evolution. Abstracts from the workshop held December 5--8, 2023
• Bounds on the Poincaré constant for convolution measures
• Moment estimates derived from Poincaré and logarithmic Sobolev inequalities
• Convergence rate for geometric statistics of point processes having fast decay of dependence
• On some properties of \(\alpha\)-mixtures
• Some recent developments in functional inequalities
• Transportation inequalities: from Poisson to Gibbs measures
• On the convex Poincaré inequality and weak transportation inequalities
• On certain convex sets of measures and on phases of reacting mixtures

This page was built for publication: On fine properties of mixtures with respect to concentration of measure and Sobolev type inequalities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q974769)