摘要
本文对BFGS公式进行修正,形成了MBFGS公式,并结合Wolfe-Powell型非精确线性搜索准则设计出了MBFGS算法,通过对目标函数合理的假设证明了该算法具有全局收敛性,又运用Matlab编写程序实现了该算法,初步的数值例子表明了该算法是有效的,并且有收敛速度快的特点.
This paper modified the BFGS-type formula form a new formula,which is the MBFGS-type formula. Furthermore,combined with the Wolfe-Powell inexact linear search rule,a MBFGS algorithm is designed. In addition,the global convergence of the algorithm is proved on conditions of assuming the objective function. Finally,preliminary numerical experiments are carried out,which suggests that the algorithm is validity and is of the faster convergence speed characteristic.
出处
《科技信息》
2008年第26期209-210,共2页
Science & Technology Information
关键词
无约束最优化
MBFGS算法
全局收敛性
Wolfe-Powell型线性搜索
unconstrained optimization
Modified Broyden-Fletcher-Goldfard-Shanna algorithm
global con-vergence property
Wolfe-Powell line search.
