Coupling Matrix Manifolds and Their Applications in Optimal Transport

TL;DR

This paper introduces the coupling matrix manifold (CMM) with a Riemannian geometric framework for optimal transport, enabling the development of efficient optimization algorithms that outperform existing methods in certain cases.

Contribution

The paper develops a novel Riemannian geometric approach to optimal transport by defining the coupling matrix manifold and applying it to create new optimization algorithms.

Findings

The proposed algorithms are comparable to Sinkhorn in standard cases.

They outperform other state-of-the-art algorithms without geometry considerations.

Effective in non-entropy optimal transport scenarios.

Abstract

Optimal transport (OT) is a powerful tool for measuring the distance between two defined probability distributions. In this paper, we develop a new manifold named the coupling matrix manifold (CMM), where each point on CMM can be regarded as the transportation plan of the OT problem. We firstly explore the Riemannian geometry of CMM with the metric expressed by the Fisher information. These geometrical features of CMM have paved the way for developing numerical Riemannian optimization algorithms such as Riemannian gradient descent and Riemannian trust-region algorithms, forming a uniform optimization method for all types of OT problems. The proposed method is then applied to solve several OT problems studied by previous literature. The results of the numerical experiments illustrate that the optimization algorithms that are based on the method proposed in this paper are comparable to the classic ones, for example, the Sinkhorn algorithm, while outperforming other state-of-the-art algorithms without considering the geometry information, especially in the case of non-entropy optimal transport.