A NEW TRUST-REGION ALGORITHM FOR FINITE MINIMAX PROBLEM
A NEW TRUST-REGION ALGORITHM FOR FINITE MINIMAX PROBLEM
复制标题
一种解决有限极小极大问题的新信任域算法
DOI：
10.4208/jcm.1109-m3567
复制
发表时间：
2012-05-01
影响因子：
0.9
通讯作者：
Wang, Li
中科院分区：
文献类型：
--
作者：
Wang, Fusheng;Wang, Chuanlong;Wang, Li
本文针对极大极小优化问题提出了一种新的信赖域算法，该算法在每次迭代时基于一种新的近似模型仅求解一个二次子问题。这种方法不同于传统算法，传统算法通常需要求解两个二次子问题。此外，为避免马拉托斯效应，采用了非单调策略。分析表明，在标准条件下，该算法具有全局收敛性和超线性收敛性。进行了初步的数值实验以展示新方法的有效性。
In this paper, a new trust region algorithm for minimax optimization problems is proposed, which solves only one quadratic subproblem based on a new approximation model at each iteration. The approach is different with the traditional algorithms that usually require to solve two quadratic subproblems. Moreover, to avoid Maratos effect, the nonmonotone strategy is employed. The analysis shows that, under standard conditions, the algorithm has global and superlinear convergence. Preliminary numerical experiments are conducted to show the effiency of the new method.