English

Hellinger-Kantorovich Gradient Flows: Global Exponential Decay of Entropy Functionals

Analysis of PDEs 2025-01-29 v1 Machine Learning Optimization and Control Machine Learning

Abstract

We investigate a family of gradient flows of positive and probability measures, focusing on the Hellinger-Kantorovich (HK) geometry, which unifies transport mechanism of Otto-Wasserstein, and the birth-death mechanism of Hellinger (or Fisher-Rao). A central contribution is a complete characterization of global exponential decay behaviors of entropy functionals (e.g. KL, χ2\chi^2) under Otto-Wasserstein and Hellinger-type gradient flows. In particular, for the more challenging analysis of HK gradient flows on positive measures -- where the typical log-Sobolev arguments fail -- we develop a specialized shape-mass decomposition that enables new analysis results. Our approach also leverages the (Polyak-)\L{}ojasiewicz-type functional inequalities and a careful extension of classical dissipation estimates. These findings provide a unified and complete theoretical framework for gradient flows and underpin applications in computational algorithms for statistical inference, optimization, and machine learning.

Keywords

Cite

@article{arxiv.2501.17049,
  title  = {Hellinger-Kantorovich Gradient Flows: Global Exponential Decay of Entropy Functionals},
  author = {Alexander Mielke and Jia-Jie Zhu},
  journal= {arXiv preprint arXiv:2501.17049},
  year   = {2025}
}
R2 v1 2026-06-28T21:22:17.393Z