线性约束优化的仿射内点修正梯度路径方法  

INTERIOR AFFINE SCALING MODIFIED GRADIENT PATH ALGORITHM FOR NONLINEAR OPTIMIZATIONS SUBJECT TO LINEAR CONSTRAINTS

作　　者：王云娟[1] 朱德通[2] 

机构地区：[1]上海电机学院文理系,上海200240 [2]上海师范大学数学系,上海200234

出　　处：《高等学校计算数学学报》2005年第S1期93-99,共7页Numerical Mathematics A Journal of Chinese Universities

摘　　要：1 引言本文主要考虑带有线性约束的非线性优化问题: 针对问题(1．1)中具有不等式约束这一特性,本文技巧性的引用仿射变换构造信赖域子问题;同时采用修正梯度路径搜索与线搜索相结合的技术,通过修正梯度路径搜索求得模型的迭代方向,然后沿此方向通过非单调线搜索获得步长因子。In this paper we propose an interior affine scaling modified gradient path algorithm with nonmonotonic interior back-tracking technique for nonlinear optimization subject to linear equality and inequality constraints. Using both modified gradient path search strategy and line search technique, the quadratic model at each iteration generates backtracking step to obtain a new accepted step. The global convergence and fast local convergence rate of the proposed algorithm are established under some reasonable c...

关 键 词：modified gradient path linear inequality constraint interior point method affine scaling nonmonotonic technique. 

分 类 号：O221.2[理学—运筹学与控制论]