A globally convergent BFGS method for nonconvex minimization without line searches

被引:10
|
作者
Zhang, L [1 ]
机构
[1] Hunan Univ, Dept Appl Math, Changsha 410082, Peoples R China
来源
OPTIMIZATION METHODS & SOFTWARE | 2005年 / 20卷 / 06期
关键词
nonconvex minimization; BFGS method; global convergence; superlinear convergence;
D O I
10.1080/10556780500065499
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
In this paper, by using a so-called fixed steplength strategy, we propose a BFGS method without use of line searches for unconstrained optimization. Under mild conditions, we show the global and superlinear convergence of the proposed method even when the objective function is nonconvex.
引用
收藏
页码:737 / 747
页数:11
相关论文
共 50 条