Riemannian conditional gradient methods for composite optimization problems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Kangming, Fukuda, Ellen H.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914572857769984
author Chen, Kangming
Fukuda, Ellen H.
author_facet Chen, Kangming
Fukuda, Ellen H.
contents In this paper, we propose Riemannian conditional gradient methods for minimizing composite functions, i.e., those that can be expressed as the sum of a smooth function and a retraction-based convex function. We analyze the convergence of the proposed algorithms, utilizing three types of step-size strategies: adaptive, diminishing, and those based on the Armijo condition. We establish the convergence rate of \(\mathcal{O}(1/k)\) for the adaptive and diminishing step sizes, where \(k\) denotes the number of iterations. Additionally, we derive an iteration complexity of \(\mathcal{O}(1/ε^2)\) for the Armijo step-size strategy to achieve \(ε\)-optimality, where \(ε\) is the optimality tolerance. Finally, the effectiveness of our algorithms is validated through some numerical experiments performed on the sphere and Stiefel manifolds.
format Preprint
id arxiv_https___arxiv_org_abs_2412_19427
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Riemannian conditional gradient methods for composite optimization problems
Chen, Kangming
Fukuda, Ellen H.
Optimization and Control
In this paper, we propose Riemannian conditional gradient methods for minimizing composite functions, i.e., those that can be expressed as the sum of a smooth function and a retraction-based convex function. We analyze the convergence of the proposed algorithms, utilizing three types of step-size strategies: adaptive, diminishing, and those based on the Armijo condition. We establish the convergence rate of \(\mathcal{O}(1/k)\) for the adaptive and diminishing step sizes, where \(k\) denotes the number of iterations. Additionally, we derive an iteration complexity of \(\mathcal{O}(1/ε^2)\) for the Armijo step-size strategy to achieve \(ε\)-optimality, where \(ε\) is the optimality tolerance. Finally, the effectiveness of our algorithms is validated through some numerical experiments performed on the sphere and Stiefel manifolds.
title Riemannian conditional gradient methods for composite optimization problems
topic Optimization and Control
url https://arxiv.org/abs/2412.19427