一族共扼下降算法的全局收敛性  

Global convergence or a class of conjugate descent algorithm

作　　者：王柏岩[1] 欧志英[1] 严克明[1] 

机构地区：[1]甘肃工业大学基础科学系,甘肃兰州730050

出　　处：《甘肃工业大学学报》2001年第4期94-95,共2页Journal of Gansu University of Technology

基　　金：甘肃省科委基金(RS002-B62-087)

摘　　要：给出了无约束问题一族含有一个参数βｋ的共扼下降算法，证明了该共扼下降算法在满足一种非精确线性搜索条件时是下降的，而且是全局收敛的．同时可以看到共扼下降算法是其特例．A class of conjugate descent algorithm with single parameter βｋ for unconstrained minimization problems is presented and it is proved that this algorithm is of descent feature when it satisfies the condi- tion of a sort of inexact linear searching and, in addition, it is globally convergent. It can be found, mean- time, that conventional conjugate descent algorithm is a special case of the algorithm presented.

关 键 词：共扼梯度法 全局收敛 非精确线性搜索 共轭下降算法 无约束最优化 

分 类 号：O224[理学—运筹学与控制论]