Asymptotics of smoothed Wasserstein distances

DOI10.1007/S11118-020-09895-9zbMATH Open1485.58026arXiv2005.00738OpenAlexW3121620948MaRDI QIDQ2118845FDOQ2118845

Jonathan Niles-Weed, Hongbin Chen

Publication date: 23 March 2022

Published in: Potential Analysis (Search for Journal in Brave)

Abstract: We investigate contraction of the Wasserstein distances on $mathbb{R}^d$ under Gaussian smoothing. It is well known that the heat semigroup is exponentially contractive with respect to the Wasserstein distances on manifolds of positive curvature; however, on flat Euclidean space---where the heat semigroup corresponds to smoothing the measures by Gaussian convolution---the situation is more subtle. We prove precise asymptotics for the $2$-Wasserstein distance under the action of the Euclidean heat semigroup, and show that, in contrast to the positively curved case, the contraction rate is always polynomial, with exponent depending on the moment sequences of the measures. We establish similar results for the $p$-Wasserstein distances for $peq2$ as well as the $ch{i}^2$ divergence, relative entropy, and total variation distance. Together, these results establish the central role of moment matching arguments in the analysis of measures smoothed by Gaussian convolution.

Full work available at URL: https://arxiv.org/abs/2005.00738

Recommendations

• A class of Wasserstein metrics for probability distributions
• Entropy-regularized 2-Wasserstein distance between Gaussian measures
• Wasserstein geometry of Gaussian measures
• Limit laws of the empirical Wasserstein distance: Gaussian distributions
• A smoothed dual approach for variational Wasserstein problems

zbMATH Keywords

Wasserstein distanceheat semigroupEuclidean spacelong time asymptoticspolynomial rate

Mathematics Subject Classification ID

Diffusion processes (60J60) Diffusion processes and stochastic analysis on manifolds (58J65)

Cites Work

• The Malliavin Calculus and Related Topics
• Optimal Transport
• A class of Wasserstein metrics for probability distributions
• The concentration of measure phenomenon
• An introduction to the regularity theory for elliptic systems, harmonic maps and minimal graphs
• A convexity principle for interacting gases
• Ricci curvature for metric-measure spaces via optimal transport
• On the geometry of metric measure spaces. I
• On the geometry of metric measure spaces. II
• THE GEOMETRY OF DISSIPATIVE EVOLUTION EQUATIONS: THE POROUS MEDIUM EQUATION
• Title not available (Why is that?)
• On the Volume Elements on a Manifold
• Bounding \(\bar d\)-distance by informational divergence: A method to prove measure concentration
• On Divergences and Informations in Statistics and Information Theory
• Title not available (Why is that?)
• Generalization of an inequality by Talagrand and links with the logarithmic Sobolev inequality
• Analysis and Geometry of Markov Diffusion Operators
• Exponential convergence in \(L^p\)-Wasserstein distance for diffusion processes without uniformly dissipative drift
• Analysis for diffusion processes on Riemannian manifolds
• Transport inequalities, gradient estimates, entropy and Ricci curvature
• Reflection couplings and contraction rates for diffusions
• Reflection coupling and Wasserstein contractivity without convexity
• Transportation cost for Gaussian and other product measures
• A measure concentration inequality for contracting Markov chains
• Synthetic theory of Ricci curvature bounds
• Title not available (Why is that?)
• A survey on dynamical transport distances
• Dimensional contraction via Markov transportation distance
• Comparison between W₂ distance and Ḣ⁻¹ norm, and Localization of Wasserstein distance
• Optimal estimation of Gaussian mixtures via denoised method of moments
• A PDE approach to a 2-dimensional matching problem
• Dimensional improvements of the logarithmic Sobolev, Talagrand and Brascamp-Lieb inequalities
• Optimal rates of estimation for multi-reference alignment
• Convergence of Smoothed Empirical Measures With Applications to Entropy Estimation

Cited In (10)

• A smooth variational principle on Wasserstein space
• Squared quadratic Wasserstein distance: optimal couplings and Lions differentiability
• On smooth approximations in the Wasserstein space
• Berry-Esseen smoothing inequality for the Wasserstein metric on compact Lie groups
• Limit distribution theory for smooth \(p\)-Wasserstein distances
• Entropic regularization of Wasserstein distance between infinite-dimensional Gaussian measures and Gaussian processes
• Cutoff ergodicity bounds in Wasserstein distance for a viscous energy shell model with Lévy noise
• Optimal transport methods for combinatorial optimization over two random point sets
• Statistical inference with regularized optimal transport
• Optimal Couplings of Kantorovich-Rubinstein-Wasserstein Lp-distance

This page was built for publication: Asymptotics of smoothed Wasserstein distances

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2118845)