摘要
本文对无约束优化问题提出了一类基于锥模型的非单调信赖域算法.二次模型非单调信赖域算法是新算法的特例.在适当的条件下,证明了算法的全局收敛性及Q-二次收敛性.
This paper present a new class of nonmonotone trust region algorithms based on conic model for unconstrained optimization.The class of nonmonotone strust region algorithms based on quadratic model is a special example of the new algorithms.Global and Q-quadratic convergence are proved under certain conditions.
出处
《应用数学》
CSCD
北大核心
2005年第S1期13-17,共5页
Mathematica Applicata
关键词
无约束优化
锥模型
非单调信赖域算法
全局收敛性
Q-二次收敛性
Unconstrained optimization
Conic model
Nonmonotone trust region algorithms
Global convergence
Q-quadratic convergence
