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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mielke, Alexander, Zhu, Jia-Jie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917904311648256
author Mielke, Alexander
Zhu, Jia-Jie
author_facet Mielke, Alexander
Zhu, Jia-Jie
contents 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, $χ^2$) 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-)Łojasiewicz-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.
format Preprint
id arxiv_https___arxiv_org_abs_2501_17049
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Hellinger-Kantorovich Gradient Flows: Global Exponential Decay of Entropy Functionals
Mielke, Alexander
Zhu, Jia-Jie
Analysis of PDEs
Machine Learning
Optimization and Control
49Q22 (Primary) 35Q49 (Secondary)
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, $χ^2$) 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-)Łojasiewicz-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.
title Hellinger-Kantorovich Gradient Flows: Global Exponential Decay of Entropy Functionals
topic Analysis of PDEs
Machine Learning
Optimization and Control
49Q22 (Primary) 35Q49 (Secondary)
url https://arxiv.org/abs/2501.17049