摘要
本文研究了求解无约束优化问题的WYL共轭梯度法.利用修正迭代格式,得到了算法在每步迭代能产生不依赖于搜索条件的充分下降方向.同时,在原算法中关于Wolfe条件中参数去掉的情况下,获得了本文算法是强收敛的.数值实验说明本文算法可以有效求解测试问题.
In this paper, we study the WYL conjugate gradient method for unconstrained optimization problems. By making use of the modified iterative scheme, the sufficient descent conditions are satisfied at each iteration independent of the line search used. Also, by removing the original restriction on the parameter of the Wolfe conditions, we establish the strongly global convergence property for the general function. Numerical results illustrate that our method is efficient for the test problems.
作者
董晓亮
何郁波
孔翔宇
李卫军
DONG Xiao-liang HE Yu-bo KONG Xiang-yu LI Wei-jun(School of Mathematics and Information, Beifang University of Nationalities, Yinchuan, 710021, China Department of Mathematics and Application Mathematics, Huaihua University, Huaihua, 418008, China Network Information Technology Center, Beifang University of Nationalities, Yinchuan, 710021, China)
出处
《数学杂志》
北大核心
2017年第2期231-238,共8页
Journal of Mathematics
基金
Supported by National Natural Science Foundation of China(11601012
11661002)
Ningxia Natural Science Foundation(NZ13095
NZ16093)
Scientiflc Research Foundation of the Higher Education Institutions of Ningxia(NGY2016134)
Beifang University of Nationalities Foundation(2016SXKY06
2014XBZ09
2014XBZ01
2013XYZ028)
关键词
共轭梯度法
充分下降条件
强收敛性
WOLFE搜索
conjugate gradient method
sufficient descent condition
strongly global convergence
Wolfe line search
