English

运筹学学报 >

2017 , Vol. 21 >Issue 1: 111 - 117

DOI: https://doi.org/10.15960/j.cnki.issn.1007-6093.2017.01.011

运筹学

从常步长梯度方法的视角看不可微凸优化增广Lagrange方法的收敛性

展开
  • 1. 华侨大学数学科学学院, 福建泉州 362021

收稿日期: 2016-05-11

  网络出版日期: 2017-03-15

基金资助

国家自然科学基金(Nos. 91330206, 11571059), 福建省中青年教师教育科研项目(No. JAT160024)

收起

A look at the convergence of the augmented  Lagrange method for nondifferentiable convex programming from  the view of a gradient method with constant stepsize

Expand
  • 1. School of Mathematical Sciences, Huaqiao University, Quanzhou 362021, Fujian, China

Received date: 2016-05-11

  Online published: 2017-03-15

Fold

摘要

增广Lagrange方法是求解非线性规划的一种有效方法. 从一新的角度证明不等式约束非线性非光滑凸优化问题的增广Lagrange方法的收敛性. 用常步长梯度法的收敛性定理证明基于增广Lagrange函数的对偶问题的常步长梯度方法的收敛性, 由此得到增广Lagrange方法乘子迭代的全局收敛性.

关键词： 梯度法; Moreau包络; 增广Lagrange对偶; 全局收敛

本文引用格式

田朝薇, 张立卫 . 从常步长梯度方法的视角看不可微凸优化增广Lagrange方法的收敛性[J]. 运筹学学报, 2017 , 21(1) : 111 -117 . DOI: 10.15960/j.cnki.issn.1007-6093.2017.01.011

Abstract

The augmented Lagrange method is an effective method for solving nonlinear optimization problems. This paper, from a new pointview, studies the convergence of the augmented Lagrange method for the nonlinear nonsmooth convex programming problem with inequality constraints. The convergence of the gradient method with constant stepsize for the dual problem, based on the augmented Lagrange function, is demonstrated by using a convergence theorem of a gradient method with constant stepsize, from which the global convergence of the multiplier iteration of augmented Lagrange method is obtained.

Key words： gradient method; Moreau envelope; augmented Lagrangian dual; global convergence

参考文献

[1] Hestenes M R. Multiplier and gradient methods [J]. Journal of Optimization Theory and Applications, 1969, 4: 303-320.
[2] Powell M J D.  A method for nonlinear constraints in minimization problems [C]//Optimization. London: Academic Press, 1969, 283-298.
[3] Rockafellar R T. Lagrange multipliers and optimality [J]. SIAM Review, 1993, 35: 183-238.
[4] Rockafellar R T. Augmented Lagrange multiplier functions and duality in nonconvex programming [J]. SIAM Journal on Control, 1974, 12: 268-285..
[5] Conn A, Gould N, Toint P. A globally convergent augmented Lagrangian algorithm for optimization with general constraints and simple bounds [J]. SIAM Journal on Numerical Analysis, 1991, 28: 545-572.
[6] Wu H X, Luo H Z. Saddle points of general augmented Lagrangian for constrained nonconvex optimization [J]. Journal of Global Optimization, 2012, 53: 683-687.
[7] Liu Y J, Zhang L W. Convergence analysis of the augmented Lagrangian method for nonlinear second-order cone optimization problem [J]. Nonlinear Analysis: Theory, Methods and Applications, 2007, 67: 1359-1373.
[8] 张立卫, 吴佳, 张艺. 变分分析与优化 [M]. 北京: 科学出版社, 2013.
[9] Meng F W, Zhao G Y. On second-order properties of the Moreau-Yosida regularization for constrainted nonsmooth convex programs [J]. Numerical Functional Analysis and Optimization, 2004, 25: 515-529.
[10] Rockafellar R T. A dual approach to solving nonlinear programming problems by unconstrained optimization [J]. Mathematical Programming, 1973, 5: 354-373.
Options
文章导航

/

〈 〉