Nonmonotone second-order Wolfe's line search method for unconstrained optimization problems

被引：1
|
作者：
Han, Xue [1 ]
Sun, Wenyu [1 ]
Dang, Chuangyin [2 ]
机构：
[1] Nanjing Normal Univ, Sch Math Sci, Nanjing 210046, Peoples R China
[2] City Univ Hong Kong, Dept Mfg Engn & Engn Management, Kowloon, Hong Kong, Peoples R China
来源：
COMPUTERS & MATHEMATICS WITH APPLICATIONS | 2010年 / 60卷 / 09期
基金：
中国国家自然科学基金;
关键词：
Nonmonotone; Second-order line search; Descent pair; Wolfe's line search; Unconstrained optimization; TRUST-REGION METHOD; NEGATIVE CURVATURE; CURVILINEAR PATH; NEWTON METHOD; DIRECTIONS; ALGORITHMS;
D O I：
10.1016/j.camwa.2010.08.002
中图分类号：
O29 [应用数学];
学科分类号：
070104 ;
摘要：
In this paper, we present a new algorithm using the nonmonotone second-order Wolfe's line search. By using the negative curvature information from the Hessian, we prove that the generated sequence converges to the stationary points that satisfy the second-order optimality conditions. We also report numerical results which show the efficiency and robustness of the proposed method. (C) 2010 Elsevier Ltd. All rights reserved.
引用
收藏
页码：2517 / 2525
页数：9
相关论文
共 50 条
12345