摘要
谱共轭梯度法是求解无约束优化的一种有效算法.该文首先对JJSL共轭参数[Jiang et al.Computational and Applied Mathematics,2021,40(174)]进行投影修正,再通过选取合适谱参数以保证其搜索方向有下降性,从而得到两个有效的谱共轭梯度法.一般假设下,分别使用常规非精确线搜索计算步长,获得这两个新算法的全局收敛性.数值试验结果以及相应性能图进一步说明其数值有效性.
The spectral conjugate gradient method is an effective algorithm for solving the uncon-strained optimization problems.In this paper,firstly,the projection corrections of JJSL conjugate parameter[Jiang et al.Computational and Applied Mathematics,2021,40(174)]are carried out.Then,two effective spectral conjugate gradient methods are obtained by selecting the appropriate spectral parameters to ensure that the search directions are descend-ing.Under the general assumptions,the global convergence results of two new algorithms are given by using the common inexact line search to calculate the step-size respectively.The numerical test results and the corresponding performance charts further demonstrate the numerical validity of the proposed methods.
作者
刘鹏杰
邵虎
简金宝
宋丹
Liu Pengjie;Shao Hu;Jian Jinbao;Song Dan(School of Mathematics,China University of Mining and Technology,Xuzhou 221116,China;College of Mathematics and Physics,Center for Applied Mathematics of Guangri,Guangri Minzu University,Nanning 530006,China;Department of Public Basic Courses'Teaching,Hezhou University,Hezhou 542899,China)
出处
《计算数学》
CSCD
北大核心
2023年第3期299-308,共10页
Mathematica Numerica Sinica
基金
国家自然科学基金(72071202)
广西科技基地和人才专项(桂科AD23023001)资助。