Non-asymptotic convergence bounds for Sinkhorn iterates and their gradients: a coupling approach - Archive ouverte HAL
Communication Dans Un Congrès Année : 2023

Non-asymptotic convergence bounds for Sinkhorn iterates and their gradients: a coupling approach

Résumé

Computational optimal transport (OT) has recently emerged as a powerful framework with applications in various fields. In this paper we focus on a relaxation of the original OT problem, the entropic OT problem, which allows to implement efficient and practical algorithmic solutions, even in high dimensional settings. This formulation, also known as the Schrödinger Bridge problem, notably connects with Stochastic Optimal Control (SOC) and can be solved with the popular Sinkhorn algorithm. In the case of discrete-state spaces, this algorithm is known to have exponential convergence; however, achieving a similar rate of convergence in a more general setting is still an active area of research. In this work, we analyze the convergence of the Sinkhorn algorithm for probability measures defined on the d-dimensional torus , that admit densities with respect to the Haar measure. In particular, we prove pointwise exponential convergence of Sinkhorn iterates and their gradient. Our proof relies on the connection between these iterates and the evolution along the Hamilton-Jacobi-Bellman equations of value functions obtained from SOC-problems. Our approach is novel in that it is purely probabilistic and relies on coupling by reflection techniques for controlled diffusions on the torus.
Fichier principal
Vignette du fichier
2304.06549.pdf (358.6 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04402316 , version 1 (18-01-2024)

Identifiants

Citer

Giacomo Greco, Maxence Noble, Giovanni Conforti, Alain Durmus. Non-asymptotic convergence bounds for Sinkhorn iterates and their gradients: a coupling approach. 36th Annual Conference on Learning Theory, 2023, Bengalore, India. ⟨hal-04402316⟩
27 Consultations
12 Téléchargements

Altmetric

Partager

More