摘要
本文证明了求解无约束最优化的广义拟牛顿算法在Goldstein非精确线搜索下对一般目标函数的全局收敛性 ,并在一定条件下证明了算法的局部超线性收敛性 .
In this paper,we develop the Generalized Quasi Newton methods for unconstrained optimization which was formed in paper,and we use inexact line searches (Goldstein rule).These methods are globally convergent when applied to a general objective function under the weak condition,and are locally super linearly convergent when applied to a uniformly convex function whoes Hessian matrix G(x) is Lipschitz continuous in the neighborhood of the optimal solution point.So we develop the results of paper and .
出处
《应用数学》
CSCD
北大核心
2002年第3期69-75,共7页
Mathematica Applicata
基金
北京市教委科研基金资助项目 (99KJ10 )
