标题:
一个修改的LS非线性共轭梯度算法
A Modified LS Nonlinear Conjugate Gradient Algorithms
作者:
陈海
关键字:
共轭梯度法, 下降性, 全局收敛性
Conjugate Gradient Method; Descent Property; Global Convergence
期刊名称:
《Advances in Applied Mathematics》, Vol.2 No.1, 2013-02-26
摘要:
我们给出一个修改的LS共轭梯度公式,此公式能保证参数βk非负且搜索方向在不需要任何线搜索下具有充分下降性。在适当条件下,证明该方法对一般函数具有全局收敛性,同时给出数值检验结果。In this paper, a modified LS conjugate gradient formula is proposed. This formula can ensure that the scalar holds and the search direction possesses the sufficiently descent property without any line search. The global convergence will be established for general functions under suitable conditions and numerical results are reported.