Wolfe线搜索充分下降的修正DY共轭梯度法  

A Modified DY conjugate gradient method with sufficient descent property for wolfe line search

作　　者：黄海[1] 

机构地区：[1]广西民族师范学院数学与计算机科学系,广西崇左532200

出　　处：《河南理工大学学报（自然科学版）》2013年第3期368-372,共5页Journal of Henan Polytechnic University（Natural Science）

基　　金：国家自然科学基金资助项目(10761001);广西教育厅科研项目(201012MS215)

摘　　要：基于DY和DL共轭梯度法,给出一个新的βk公式,在精确线搜索下该公式等价于βDkY.基于新参数公式建立了采用Wolfe线搜索的共轭梯度算法,证明了算法满足充分下降性和全局收敛性,初步的数值试验结果表明该方法是有效的,适合于求解非线性无约束优化问题.Based on the DY and DL conjugate gradient methods,a new parameter formula βk was presented,which is the same as,βDYk when the line search is exact.Based on the new formula,a new conjugate gradient algorithm with the Wolfe line search was proposed,and the sufficient descent property and the global convergence of the algorithm were proved.Preliminary numerical results show that the new method was effective and it suitable for solving nonlinear unconstrained optimization problems.

关 键 词：无约束优化 共轭梯度法 充分下降 全局收敛 

分 类 号：O221.2[理学—运筹学与控制论]