利用Armijo型线性搜索HZ共轭梯度法的全局收敛性(英文)  

Global Convergence of HZ's Conjugate Gradient Method with Armijo-type Line Search

作　　者：魏敬广[1] 张建军[1] 

机构地区：[1]上海大学数学系,上海200444

出　　处：《运筹学学报》2011年第1期85-94,共10页Operations Research Transactions

基　　金：supported by the Innovation Foundation of Shanghai University(Grant No.A.16-0101-10-031);Shanghai Leading Academic Discipline project(Grant No.J5101)

摘　　要：由William W.Hager和张洪超提出的一种新的共轭梯度法(简称HZ方法),已被证明是一种有效的方法.本文证明了HZ共轭梯度法在Armijo型线性搜索下的全局收敛性.数值实验显示,在Armijo型线性搜索下的HZ共轭梯度法比在Wolfe线性搜索下更有效.HZ＇s conjugate gradient method（proposed by William W.Hager and Hongchao Zhang） has been proved to be an efficient method.In this paper, we prove the global convergence of HZ＇s method with Armijo-type line search. Our numerical experiments show that the new algorithm are more efficient and competitive with HZ＇s method with Wolfe line search in most cases.

关 键 词：运筹学 共轭梯度法 全局收敛性 线性搜索 无约束最优化 

分 类 号：O224[理学—运筹学与控制论]