Squared quadratic Wasserstein distance: optimal couplings and Lions differentiability

DOI10.1051/PS/2020013MaRDI QIDQ5140338zbMATH OpenOpenAlexFDO

Authors Aurélien Alfonsi, Benjamin Jourdain

Publication date 15 December 2020

Published in ESAIM: Probability and Statistics (Search for Journal in Brave)

Copyright license File:Creative commons license with attribution.png Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/1811.07787

zbMATH Keywords

convex ordercouplings of probability measuresdifferentiabilityoptimal transportWasserstein distance

Mathematics Subject Classification ID

Fréchet and Gateaux differentiability in optimization (49J50) Differentiability questions for infinite-dimensional manifolds (58B10) Inequalities; stochastic orderings (60E15) Martingales with discrete parameter (60G42) Special problems of linear programming (transportation, multi-index, data envelopment analysis, etc.) (90C08)

Abstract: In this paper, we remark that any optimal coupling for the quadratic Wasserstein distance $W_2^2(mu,u)$ between two probability measures $mu$ and $u$ with finite second order moments on $mathbb{R}^d$ is the composition of a martingale coupling with an optimal transport map $mathcalT$. We check the existence of an optimal coupling in which this map gives the unique optimal coupling between $mu$ and $$. Next, we give a direct proof that $sigmamapsto{W}_2^2(sigma,u)$ is differentiable at $mu$ in the Lions sense iff there is a unique optimal coupling between $mu$ and $u$ and this coupling is given by a map. It was known combining results by Ambrosio, Gigli and Savar'e and Ambrosio and Gangbo that, under the latter condition, geometric differentiability holds. Moreover, the two notions of differentiability are equivalent according to the recent paper of Gangbo and Tudorascu. Besides, we give a self-contained probabilistic proof that mere Fr'echet differentiability of a law invariant function $F$ on $L^2(Omega,mathbbP;mathbb{R}^d)$ is enough for the Fr'echet differential at $X$ to be a measurable function of $X$.

Recommendations

• Optimal Couplings of Kantorovich-Rubinstein-Wasserstein Lp-distance
• Non-monotone convergence in the quadratic Wasserstein distance
• Distances de Wasserstein et de Zolotarev
• Approximation de la L1-distance de Wasserstein
• The Wasserstein distance and approximation theorems
• Asymptotics of smoothed Wasserstein distances
• 𝐿₁-distortion of Wasserstein metrics: A tale of two dimensions
• Wasserstein distance and the rectifiability of doubling measures. II
• Conformal Wasserstein distance: II. Computational aspects and extensions

Cites work

• A new class of costs for optimal transport planning
• Characterization of a class of weak transport-entropy inequalities on the line
• Existence, duality, and cyclical monotonicity for weak transport costs
• Foundations of Modern Probability
• Hamiltonian ODEs in the Wasserstein space of probability measures
• scientific article; zbMATH DE number 699438 (Why is no real title available?)
• scientific article; zbMATH DE number 2152346 (Why is no real title available?)
• On a mixture of Brenier and Strassen theorems
• On differentiability in the Wasserstein space and well-posedness for Hamilton-Jacobi equations
• On the inverse implication of Brenier-McCann theorems and the structure of \((\mathcal P_{2}(M),W_{2})\)
• Optimal Transport
• Optimal transport for applied mathematicians. Calculus of variations, PDEs, and modeling
• Probabilistic theory of mean field games with applications I. Mean field FBSDEs, control, and games
• Sampling of probability measures in the convex order by Wasserstein projection
• Stochastic finance. An introduction in discrete time.
• The Existence of Probability Measures with Given Marginals

Cited in

(14)

• Solving mean field rough differential equations
• An introduction to mean field game theory
• Optimal Couplings of Kantorovich-Rubinstein-Wasserstein Lp-distance
• Monotone solutions of the master equation for mean field games with idiosyncratic noise
• Optimal vector quantization in terms of Wasserstein distance
• Dynamical Systems and Hamilton–Jacobi–Bellman Equations on the Wasserstein Space and their L² Representations
• A smooth variational principle on Wasserstein space
• An optimal transport-based characterization of convex order
• Deterministic optimal control on Riemannian manifolds under probability knowledge of the initial condition
• Propagation of chaos for mean field Schrödinger problems
• Reachability for multiagent control systems via Lyapunov functions
• Stochastic optimal transport and Hamilton-Jacobi-Bellman equations on the set of probability measures
• Wasserstein projections in the convex order: regularity and characterization in the quadratic Gaussian case
• Rearranged stochastic heat equation: ergodicity and related gradient flow on \(\mathcal{P}(\mathbb{R})\)

This page was built for publication: Squared quadratic Wasserstein distance: optimal couplings and Lions differentiability

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5140338)