A new envelope function for nonsmooth DC optimization

DOI10.1109/CDC42340.2020.9304514MaRDI QIDQ6504221arXivFDO

Authors Andreas Themelis, Ben Hermans, Panagiotis Patrinos

Published in arXiv

Full work available at URL https://doi.org/10.1109/CDC42340.2020.9304514

Mathematics Subject Classification ID

Large-scale problems in mathematical programming (90C06) Nonconvex programming, global optimization (90C26) Methods of quasi-Newton type (90C53)

Abstract: Difference-of-convex (DC) optimization problems are shown to be equivalent to the minimization of a Lipschitz-differentiable "envelope". A gradient method on this surrogate function yields a novel (sub)gradient-free proximal algorithm which is inherently parallelizable and can handle fully nonsmooth formulations. Newton-type methods such as L-BFGS are directly applicable with a classical linesearch. Our analysis reveals a deep kinship between the novel DC envelope and the forward-backward envelope, the former being a smooth and convexity-preserving nonlinear reparametrization of the latter.

This page was built for publication: A new envelope function for nonsmooth DC optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6504221)