2021arXiv (Cornell University)Open access

An Accelerated Inexact Dampened Augmented Lagrangian Method for Linearly-Constrained Nonconvex Composite Optimization Problems

Weiwei Kong, Renato D. C. Monteiro

Open full text 1 citations

Abstract

This paper proposes and analyzes an accelerated inexact dampened augmented Lagrangian (AIDAL) method for solving linearly-constrained nonconvex composite optimization problems. Each iteration of the AIDAL method consists of: (i) inexactly solving a dampened proximal augmented Lagrangian (AL) subproblem by calling an accelerated composite gradient (ACG) subroutine; (ii) applying a dampened and under-relaxed Lagrange multiplier update; and (iii) using a novel test to check whether the penalty parameter of the AL function should be increased. Under several mild assumptions involving the dampening factor and the under-relaxation constant, it is shown that the AIDAL method generates an approximate stationary point of the constrained problem in ${\cal O}(\varepsilon^{-5/2}\log\varepsilon^{-1})$ iterations of the ACG subroutine, for a given tolerance $\varepsilon>0$. Numerical experiments are also given to show the computational efficiency of the proposed method.

Open-access reader

About this research paper

What this paper is about

This paper proposes and analyzes an accelerated inexact dampened augmented Lagrangian (AIDAL) method for solving linearly-constrained nonconvex composite optimization problems. Each iteration of the AIDAL method consists of: (i) inexactly solving a dampened proximal augmented Lagrangian (AL) subproblem by calling an accelerated composite gradient (ACG) subroutine; (ii) applying a dampened and under-relaxed Lagrange multiplier update; and (iii) using a novel test to check whether the penalty parameter of the AL function should be increased. Under several mild assumptions involving the dampening factor and the under-relaxation constant, it is shown that the AIDAL method generates an approximate stationary point of the constrained problem in ${\cal O}(\varepsilon^{-5/2}\log\varepsilon^{-1})$ iterations of the ACG subroutine, for a given tolerance $\varepsilon>0$. Numerical experiments are also given to show the computational efficiency of the proposed method.

Why it matters

OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

This paper proposes and analyzes an accelerated inexact dampened augmented Lagrangian (AIDAL) method for solving linearly-constrained nonconvex composite optimization problems. Each iteration of the AIDAL method consists of: (i) inexactly solving a dampened proximal augmented Lagrangian (AL) subproblem by calling an accelerated composite gradient (ACG) subroutine; (ii) applying a dampened and under-relaxed Lagrange multiplier update; and (iii) using a novel test to check whether the penalty parameter of the AL function should be increased. Under several mild assumptions involving the dampening factor and the under-relaxation constant, it is shown that the AIDAL method generates an approximate stationary point of the constrained problem in ${\cal O}(\varepsilon^{-5/2}\log\varepsilon^{-1})$ iterations of the ACG subroutine, for a given tolerance $\varepsilon>0$. Numerical experiments are also given to show the computational efficiency of the proposed method.

Key concepts: Augmented Lagrangian method, Lagrange multiplier, Subroutine, Lagrangian relaxation, Mathematics, Mathematical optimization, Lagrangian, Relaxation (psychology)

Related papers

Back to paper searchBrowse research topicsOriginal source
An Accelerated Inexact Dampened Augmented Lagrangian Method for Linearly-Constrained Nonconvex Composite Optimization Problems — Research Paper | ScholarLens