Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient
Mathematical Programming · 2025-09 · 213 (1-2) · 433-471
DOI: 10.1007/s10107-024-02143-7 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematical Programming, online erschienen am 28.10.2024. Verfasst von Puya Latafat und 3 weiteren Autor:innen. Die Arbeit wurde bisher 24-mal zitiert (Crossref) und nennt 71 Literaturangaben. Gefördert von: Fonds Wetenschappelijk Onderzoek, Japan Society for the Promotion of Science.
Abstract (Originalsprache)
Backtracking linesearch is the de facto approach for minimizing continuously differentiable functions with locally Lipschitz gradient. In recent years, it has been shown that in the convex setting it is possible to avoid linesearch altogether, and to allow the stepsize to adapt based on a local smoothness estimate without any backtracks or evaluations of the function value. In this work we propose an adaptive proximal gradient method, adaPGM, that uses novel estimates of the local smoothness modulus which leads to less conservative stepsize updates and that can additionally cope with nonsmooth terms. This idea is extended to the primal-dual setting where an adaptive three-term primal-dual algorithm, adaPDM, is proposed which can be viewed as an extension of the PDHG method. Moreover, in this setting the “essentially” fully adaptive variant adaPDM+\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\textsf {adaPDM}^{\textsf {+}}$$\end{document} is proposed that avoids evaluating the linear operator norm by invoking a backtracking procedure, that, remarkably, does not require extra gradient evaluations. Numerical simulations demonstrate the effectiveness of the proposed algorithms compared to the state of the art.
Förderung
- Fonds Wetenschappelijk Onderzoek
- Japan Society for the Promotion of Science
Zitieren
APA
Latafat, P., Themelis, A., Stella, L., & Patrinos, P. (2025). Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient. Mathematical Programming, 213(1-2), 433-471. https://doi.org/10.1007/s10107-024-02143-7
Vancouver
Latafat P, Themelis A, Stella L, Patrinos P. Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient. Mathematical Programming. 2025;213(1-2):433-471. doi:10.1007/s10107-024-02143-7
BibTeX
@article{Latafat2025,
author = {Latafat, Puya and Themelis, Andreas and Stella, Lorenzo and Patrinos, Panagiotis},
title = {Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient},
journal = {Mathematical Programming},
year = {2025},
volume = {213},
number = {1-2},
pages = {433--471},
doi = {10.1007/s10107-024-02143-7}
}
EUVIMED