标准Wolfe线搜索下改进的HS共轭梯度法  

Improved HS Conjugate Gradient Method with Standard Wolfe Line Search

作　　者：王森森 郑宗剑 韩信 WANG Sensen;ZHENG Zongjian;HAN Xin(Mathematics and Information School of HeTian Normal College,Hetian Xinjiang 84800;Mathematics School of Sichuan University of Arts and Sciences,Dazhou Sichuan 635000;Electronics and Information Engineering School of Southwest University,Chongqing 400715,China)

机构地区：[1]和田师范专科学校数学与信息学院,新疆和田848000 [2]四川文理学院数学学院,四川达州635000 [3]西南大学电子信息工程学院,重庆400715

出　　处：《四川文理学院学报》2024年第2期50-55,共6页Sichuan University of Arts and Science Journal

基　　金：达州市社科联重点研究基地数学与金融研究中心一般项目(SCMF202206)。

摘　　要：通过对现有的HS共轭梯度法进行修正,提出一个具有下降性质的改进型HS共轭梯度法,该算法的下降性质得到论证.在标准Wolfe线搜索条件下,证明了改进的HS算法具有全局收敛性.最后,通过数值实验结果的对比,发现新算法数值效果是优异的.By modifying the existing HS conjugate gradient method,an improved HS conjugate gradient method with descent property has been proposed,and the descent property of this algorithm has been demonstrated.Under the standard Wolfe line search conditions,it has been proven that the improved HS algorithm possesses global convergence.Finally,through the comparison of numerical experimental results,it was found that the new algorithm enjoys excellent numerical performance.

关 键 词：无约束优化 共轭梯度法 标准Wolfe线搜索 全局收敛性 

分 类 号：O224[理学—运筹学与控制论]