摘要
本文提出一种自动确定信赖域半径的新锥模型信赖域算法.该算法在每步迭代中利用以前迭代点的二次信息和水平向量信息自动产生一个信赖域半径.且证明了全局收敛性及超线性收敛性,数值结果验证了新算法的有效性.
In this paper,an adaptive trust-region algorithm based on the new conic model for unconstrained optimization is proposed.On each iteration,the new method will automatically to produce a radius by making full use of the quadratic and the level vector information of the previous points,and the global convergence and superlinear convergence are proved.Numerical experiments show that the new method is efficient.
出处
《应用数学》
CSCD
北大核心
2010年第2期307-312,共6页
Mathematica Applicata
基金
山西省自然科学基金项目(2008011013)
关键词
新锥模型
信赖域
自适应
水平向量
全局收敛性
New conic model
Trust region
Adaptive
Level vector
Global convergence
