ACCELERATION AND GLOBAL CONVERGENCE OF A FIRST-ORDER PRIMAL-DUAL METHOD FOR NONCONVEX PROBLEMS

被引:14
|
作者
Clason, Christian [1 ]
Mazurenko, Stanislav [2 ,3 ]
Valkonen, Tuomo [2 ,4 ]
机构
[1] Univ Duisburg Essen, Fac Math, D-45117 Essen, Germany
[2] Univ Liverpool, Dept Math Sci, Liverpool L69 7ZL, Merseyside, England
[3] Masaryk Univ, Fac Sci, Kamenice 5,Bld A13, Brno 62500, Czech Republic
[4] Escuela Politec Nacl, Ladron de Guevara E11-253, Quito 170109, Ecuador
基金
英国工程与自然科学研究理事会;
关键词
acceleration; convergence; global; primal-dual; first order; nonconvex; OPTIMIZATION; MINIMIZATION; ALGORITHMS; OPERATORS;
D O I
10.1137/18M1170194
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
The primal-dual hybrid gradient method, modified (PDHGM, also known as the Chambolle-Pock method), has proved very successful for convex optimization problems involving linear operators arising in image processing and inverse problems. In this paper, we analyze an extension to nonconvex problems that arise if the operator is nonlinear. Based on the idea of testing, we derive new step-length parameter conditions for the convergence in infinite-dimensional Hilbert spaces and provide acceleration rules for suitably (locally and/or partially) monotone problems. Importantly, we prove linear convergence rates as well as global convergence in certain cases. We demonstrate the efficacy of these step-length rules for PDE-constrained optimization problems.
引用
收藏
页码:933 / 963
页数:31
相关论文
共 50 条
  • [1] A First-Order Primal-Dual Method for Nonconvex Constrained Optimization Based on the Augmented Lagrangian
    Zhu, Daoli
    Zhao, Lei
    Zhang, Shuzhong
    MATHEMATICS OF OPERATIONS RESEARCH, 2024, 49 (01) : 125 - 150
  • [2] A primal-dual finite element method for first-order transport problems
    Wang, Chunmei
    Wang, Junping
    JOURNAL OF COMPUTATIONAL PHYSICS, 2020, 417
  • [3] On the ergodic convergence rates of a first-order primal-dual algorithm
    Chambolle, Antonin
    Pock, Thomas
    MATHEMATICAL PROGRAMMING, 2016, 159 (1-2) : 253 - 287
  • [4] Unified linear convergence of first-order primal-dual algorithms for saddle point problems
    Fan Jiang
    Zhongming Wu
    Xingju Cai
    Hongchao Zhang
    Optimization Letters, 2022, 16 : 1675 - 1700
  • [5] Unified linear convergence of first-order primal-dual algorithms for saddle point problems
    Jiang, Fan
    Wu, Zhongming
    Cai, Xingju
    Zhang, Hongchao
    OPTIMIZATION LETTERS, 2022, 16 (06) : 1675 - 1700
  • [6] A first-order primal-dual method with adaptivity to local smoothness
    Vladarean, Maria-Luiza
    Malitsky, Yura
    Cevher, Volkan
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021, 34
  • [7] Inexact first-order primal-dual algorithms
    Rasch, Julian
    Chambolle, Antonin
    COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2020, 76 (02) : 381 - 430
  • [8] A FIRST-ORDER PRIMAL-DUAL ALGORITHM WITH LINESEARCH
    Malitsky, Yura
    Pock, Thomas
    SIAM JOURNAL ON OPTIMIZATION, 2018, 28 (01) : 411 - 432
  • [9] A First-Order Primal-Dual Algorithm for Convex Problems with Applications to Imaging
    Antonin Chambolle
    Thomas Pock
    Journal of Mathematical Imaging and Vision, 2011, 40 : 120 - 145
  • [10] A First-Order Primal-Dual Algorithm for Convex Problems with Applications to Imaging
    Chambolle, Antonin
    Pock, Thomas
    JOURNAL OF MATHEMATICAL IMAGING AND VISION, 2011, 40 (01) : 120 - 145