Entropy-Transport distances between unbalanced metric measure spaces

Abstract

Inspired by the recent theory of Entropy-Transport problems and by the $\mathbf{D}$-distance of Sturm on normalised metric measure spaces, we define a new class of complete and separable distances between metric measure spaces of possibly different total mass. We provide several explicit examples of such distances, where a prominent role is played by a geodesic metric based on the Hellinger-Kantorovich distance. Moreover, we discuss some limiting cases of the theory, recovering the "pure transport" $\mathbf{D}$-distance and introducing a new class of "pure entropic" distances. We also study in detail the topology induced by such Entropy-Transport metrics, showing some compactness and stability results for metric measure spaces satisfying Ricci curvature lower bounds in a synthetic sense.

arXiv comment39 pages. Added a section with a comparison with the conic Gromov-Wasserstein distance. Some other minor improvements. Final version, to appear on "Probability Theory and Related Fields"

References in corpus (14)

• Neural Message Passing for Quantum Chemistry
• Metric measure spaces with Riemannian Ricci curvature bounded from below
• Learning with a Wasserstein Loss
• Riemannian Ricci curvature lower bounds in metric measure spaces with $\sigma$-finite measure
• Generalized Wasserstein distance and its application to transport equations with source
• Optimal Entropy-Transport problems and a new Hellinger-Kantorovich distance between positive measures
• Optimal transport in competition with reaction: the Hellinger-Kantorovich distance and geodesic curves
• Sinkhorn Divergences for Unbalanced Optimal Transport
• A new optimal transport distance on the space of finite Radon measures
• The Unbalanced Gromov Wasserstein Distance: Conic Formulation and Relaxation
• Global convergence of neuron birth-death dynamics
• Classification of atomic environments via the Gromov-Wasserstein distance
• Parallel Unbalanced Optimal Transport Regularization for Large Scale Imaging Problems
• Gromov-Wasserstein Distance based Object Matching: Asymptotic Inference

Cited by in corpus (2)

• Generalized Spectral Clustering via Gromov-Wasserstein Learning
• Regularity theory and geometry of unbalanced optimal transport