Smoothing alternating direction methods for fully nonsmooth constrained convex optimization

DOI10.1007/978-3-319-97478-1_4MaRDI QIDQ2415200zbMATH OpenOpenAlexFDO

Authors Volkan Cevher, Quoc Tran Dinh

Publication date 21 May 2019

Full work available at URL https://arxiv.org/abs/1507.03734, https://infoscience.epfl.ch/record/261196/files/Smoothing%20Alternating%20Direction%20Methods%20for%20Fully%20Nonsmooth%20Constrained%20Convex%20Optimization.pdf

zbMATH Keywords

gap reduction techniquehomotopy smoothing techniquenonsmooth constrained convex optimizationsmooth alternating direction method of multiplierssmooth alternating minimization

Mathematics Subject Classification ID

Special problems of linear programming (transportation, multi-index, data envelopment analysis, etc.) (90C08) Convex programming (90C25)

Abstract: We propose two new alternating direction methods to solve "fully" nonsmooth constrained convex problems. Our algorithms have the best known worst-case iteration-complexity guarantee under mild assumptions for both the objective residual and feasibility gap. Through theoretical analysis, we show how to update all the algorithmic parameters automatically with clear impact on the convergence performance. We also provide a representative numerical example showing the advantages of our methods over the classical alternating direction methods using a well-known feasibility problem.

Review text

Two alternating direction methods by means of smoothing techniques for solving nonsmooth constrained convex problems are proposed, that under mild hypotheses turn out to have the best known worst-case iteration-complexity guarantee for both the objective residual and feasibility gap in the literature. Special cases such as the one where one of the involved functions is strongly convex or composed with a linear operator are investigated as well. Computational results on a feasibility problem are presented, too, for illustrating the new methods, and the advantages of the latter are stressed. A long appendix containing the proofs of the technical results closes the paper. For the entire collection see [Zbl 1407.90006].

Recommendations

• Adaptive smoothing algorithms for nonsmooth composite convex minimization
• Proximal alternating penalty algorithms for nonsmooth constrained convex optimization
• An adaptive primal-dual framework for nonsmooth convex minimization
• A class of alternating linearization algorithms for nonsmooth convex optimization
• An approximate ADMM for solving linearly constrained nonsmooth optimization problems with two blocks of variables

Cited in

(4)

• Proximal alternating penalty algorithms for nonsmooth constrained convex optimization
• An adaptive primal-dual framework for nonsmooth convex minimization
• A smooth primal-dual optimization framework for nonsmooth composite convex minimization
• Adaptive smoothing algorithms for nonsmooth composite convex minimization

This page was built for publication: Smoothing alternating direction methods for fully nonsmooth constrained convex optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2415200)