摘要
在本文中,我们给出一个求解无约束优化问题的秩一适定方法,该方法具有下述较好性质:校正矩阵是对称正定的;在适当条件下,对非凸函数拥有全局收敛性.我们还给出数值检验结果.
We propose a rank-one fitting method for solving unconstrained optimzation problems. This method possesses some better properties.. （a） which can ensure that the undated matrices is symmetric and positive definite; （b） the global convergence for non-convex function is established under suitable conditions. The numerical results are reported.
出处
《应用数学》
CSCD
北大核心
2009年第1期118-122,共5页
Mathematica Applicata
基金
Supported by China NSF(10761001)
the Scientific Research Foundation of Guangxi University (X081082)
关键词
无约束优化
秩一校正
全局收敛
Unconstrained optimization
Rank-one update
Global convergence
