Heavy-ball dynamics with Hessian-driven damping for non-convex optimization under the Łojasiewicz condition

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Apidopoulos, Vassilis, Mavrogeorgou, Vasiliki, Tsironis, Theodoros G.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916793353764864
author Apidopoulos, Vassilis
Mavrogeorgou, Vasiliki
Tsironis, Theodoros G.
author_facet Apidopoulos, Vassilis
Mavrogeorgou, Vasiliki
Tsironis, Theodoros G.
contents In this paper, we examine the convergence properties of heavy-ball dynamics with Hessian-driven damping in smooth non-convex optimization problems satisfying a Łojasiewicz condition. In this general setting, we provide a series of tight, worst-case optimal convergence rate guarantees as a function of the dynamics' friction coefficients and the Łojasiewicz exponent of the problem's objective function. Importantly, the linear rates that we obtain improve on previous available rates and they suggest a different tuning of the dynamics' damping terms, even in the strongly convex regime. We complement our analysis with a range of stability estimates in the presence of perturbation errors and inexact gradient input, as well as an avoidance result showing that the dynamics under study avoid strict saddle points from almost every initial condition,
format Preprint
id arxiv_https___arxiv_org_abs_2506_11705
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Heavy-ball dynamics with Hessian-driven damping for non-convex optimization under the Łojasiewicz condition
Apidopoulos, Vassilis
Mavrogeorgou, Vasiliki
Tsironis, Theodoros G.
Optimization and Control
90C26, 34D05, 46N10, 65K05, 65B99
In this paper, we examine the convergence properties of heavy-ball dynamics with Hessian-driven damping in smooth non-convex optimization problems satisfying a Łojasiewicz condition. In this general setting, we provide a series of tight, worst-case optimal convergence rate guarantees as a function of the dynamics' friction coefficients and the Łojasiewicz exponent of the problem's objective function. Importantly, the linear rates that we obtain improve on previous available rates and they suggest a different tuning of the dynamics' damping terms, even in the strongly convex regime. We complement our analysis with a range of stability estimates in the presence of perturbation errors and inexact gradient input, as well as an avoidance result showing that the dynamics under study avoid strict saddle points from almost every initial condition,
title Heavy-ball dynamics with Hessian-driven damping for non-convex optimization under the Łojasiewicz condition
topic Optimization and Control
90C26, 34D05, 46N10, 65K05, 65B99
url https://arxiv.org/abs/2506.11705