Linesearch-free adaptive Bregman proximal gradient for convex minimization without relative smoothness

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ou, Hongjia, Latafat, Puya, Themelis, Andreas
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916877792444416
author Ou, Hongjia
Latafat, Puya
Themelis, Andreas
author_facet Ou, Hongjia
Latafat, Puya
Themelis, Andreas
contents This paper introduces adaptive Bregman proximal gradient algorithms for solving convex composite minimization problems without relying on global relative smoothness or strong convexity assumptions. Building upon recent advances in adaptive stepsize selections, the proposed methods generate stepsizes based on local curvature estimates, entirely eliminating the need for backtracking linesearch. A key innovation is a Bregman generalization of Young's inequality, which allows controlling a critical inner product in terms of the same Bregman distances used in the updates. Our theory applies to problems where the differentiable term is merely locally smooth relative to a distance-generating function, without requiring the existence of global moduli or symmetry coefficients. Numerical experiments demonstrate their competitive performance compared to existing approaches across various problem classes.
format Preprint
id arxiv_https___arxiv_org_abs_2508_01353
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Linesearch-free adaptive Bregman proximal gradient for convex minimization without relative smoothness
Ou, Hongjia
Latafat, Puya
Themelis, Andreas
Optimization and Control
65K05, 90C06, 90C25, 90C30, 49M29
This paper introduces adaptive Bregman proximal gradient algorithms for solving convex composite minimization problems without relying on global relative smoothness or strong convexity assumptions. Building upon recent advances in adaptive stepsize selections, the proposed methods generate stepsizes based on local curvature estimates, entirely eliminating the need for backtracking linesearch. A key innovation is a Bregman generalization of Young's inequality, which allows controlling a critical inner product in terms of the same Bregman distances used in the updates. Our theory applies to problems where the differentiable term is merely locally smooth relative to a distance-generating function, without requiring the existence of global moduli or symmetry coefficients. Numerical experiments demonstrate their competitive performance compared to existing approaches across various problem classes.
title Linesearch-free adaptive Bregman proximal gradient for convex minimization without relative smoothness
topic Optimization and Control
65K05, 90C06, 90C25, 90C30, 49M29
url https://arxiv.org/abs/2508.01353