Improved convergence rates for the Difference-of-Convex algorithm

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Rotaru, Teodor, Patrinos, Panagiotis, Glineur, François
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909149505257472
author Rotaru, Teodor
Patrinos, Panagiotis
Glineur, François
author_facet Rotaru, Teodor
Patrinos, Panagiotis
Glineur, François
contents We consider a difference-of-convex formulation where one of the terms is allowed to be hypoconvex (or weakly convex). We first examine the precise behavior of a single iteration of the Difference-of-Convex algorithm (DCA), giving a tight characterization of the objective function decrease. This requires distinguishing between eight distinct parameter regimes. Our proofs are inspired by the performance estimation framework, but are much simplified compared to similar previous work. We then derive sublinear DCA convergence rates towards critical points, distinguishing between cases where at least one of the functions is smooth and where both functions are nonsmooth. We conjecture the tightness of these rates for four parameter regimes, based on strong numerical evidence obtained via performance estimation, as well as the leading constant in the asymptotic sublinear rate for two more regimes.
format Preprint
id arxiv_https___arxiv_org_abs_2403_16864
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Improved convergence rates for the Difference-of-Convex algorithm
Rotaru, Teodor
Patrinos, Panagiotis
Glineur, François
Optimization and Control
We consider a difference-of-convex formulation where one of the terms is allowed to be hypoconvex (or weakly convex). We first examine the precise behavior of a single iteration of the Difference-of-Convex algorithm (DCA), giving a tight characterization of the objective function decrease. This requires distinguishing between eight distinct parameter regimes. Our proofs are inspired by the performance estimation framework, but are much simplified compared to similar previous work. We then derive sublinear DCA convergence rates towards critical points, distinguishing between cases where at least one of the functions is smooth and where both functions are nonsmooth. We conjecture the tightness of these rates for four parameter regimes, based on strong numerical evidence obtained via performance estimation, as well as the leading constant in the asymptotic sublinear rate for two more regimes.
title Improved convergence rates for the Difference-of-Convex algorithm
topic Optimization and Control
url https://arxiv.org/abs/2403.16864