基于共轭梯度的锥模型信赖域算法  

A CONIC MODEL TRUST-REGION ALGORITHM BASED ON CONJUGATE GRADIENT

作　　者：陆晓平[1] 倪勤[2] 

机构地区：[1]中国人民解放军93704部队 [2]南京航空航天大学理学院

出　　处：《高等学校计算数学学报》2011年第2期122-137,共16页Numerical Mathematics A Journal of Chinese Universities

基　　金：国家自然科学基金(11071117)资助项目

摘　　要：1引言 本文考虑求解无约束优化问题 minf（x），0∈R^n，A conic model trust-region algorithm based on conjugate gradient is described in this paper we derive a conjugate gradient algorithm for solving new conic model trust-region subproblems, discuss the monotonicity of the conic func- tion in the conic conjugate directions, and prove the descent properties. According to the theory we develop a conic model trust-region algorithm for solving uncon- strained optimization problem. Then we present the global convergence analysis of this trust-region algorithm. The numerical experiment shows that this algorithm may be efficient.

关 键 词：信赖域算法 共轭梯度 锥模型 无约束优化问题 R^N 求解 

分 类 号：O221.2[理学—运筹学与控制论]