Long-time behaviour of entropic interpolations

TL;DR

This paper studies the long-time behavior of entropic interpolations, showing their convergence to the heat equation and providing explicit rates under curvature conditions, thus extending understanding beyond short-time limits.

Contribution

It establishes the long-time convergence of entropic interpolations to the heat equation and provides explicit convergence rates assuming only the CD(0, n) curvature condition.

Findings

Entropic interpolations converge to the heat equation over long time.

Explicit convergence rates are derived under the CD(0, n) condition.

Asymptotic analysis of the entropic cost is performed.

Abstract

In this article we investigate entropic interpolations. These measure valued curves describe the optimal solutions of the Schr{\"o}dinger problem [Sch31], which is the problem of finding the most likely evolution of a system of independent Brownian particles conditionally to observations. It is well known that in the short time limit entropic interpolations converge to the McCann-geodesics of optimal transport. Here we focus on the long-time behaviour, proving in particular asymptotic results for the entropic cost and establishing the convergence of entropic interpolations towards the heat equation, which is the gradient flow of the entropy according to the Otto calculus interpretation. Explicit rates are also given assuming the Bakry-{\'E}mery curvature-dimension condition. In this respect, one of the main novelties of our work is that we are able to control the long time behavior of entropic interpolations assuming the CD(0, n) condition only.