Armijo型线搜索下的新共轭梯度法的全局收敛性

GLOBAL CONVERGENCE OF A NEW CONJUGATE GRADIENT METHOD WITH ARMIJO LINEAR SEARCH

共轭梯度法是解决无约束非线性最优化问题的重要的方法之一.基于FR方法好的收敛性并考虑到dk的下降性,提出了一类新的共轭梯度法,并在两种Armijo型搜索下,研究了新方法的全局收敛性.数据实验表明新方法是有效的.

Conjugate gradient methods are very important for solving nonlinear optimization problems. On the base of the good convergence of FR, and in view of the descend property of the paper presents a new conjugate gradient method, and studies the global convergence with two kinds of Armijo-type linear search. The numerical results show that new CG methods are very perfect.