非凸无约束优化问题的广义拟牛顿法的全局收敛性  

On the Global Convergence of the Genralized Quasi-Newton Method for Nonconvex Unconstrained Optimization Problems

作　　者：陈兰平[1] 焦宝聪[1] 

机构地区：[1]首都师范大学数学系,北京100037

出　　处：《应用数学》2005年第4期573-579,共7页Mathematica Applicata

基　　金：国家自然科学基金(60472071);北京市教委科研基金(KM200510028019)资助项目

摘　　要：本文对无约束优化问题提出一类新的广义拟牛顿法,并采用一类非精确线搜索证明了算法对一般非凸目标函数极小化问题的全局收敛性.This paper is concerned with open problem of whether the quasi-Newton methods with inexact line search converges globally when applied to nonconvex unconstrained optimization problems. We propose a genralized quasi-Newton methods update and prove that methods with Goldstein type line search converges globally if the function to be minimized has Lipschitz continuous gradients.

关 键 词：广义拟牛顿算法 无约束最优化 全局收敛性 

分 类 号：O174.13[理学—数学]