不等式约束Minimax问题的一个可行序列线性方程组算法  

A Feasible Sequential Systems of Linear Equations Algorithm for Inequality Constrained Minimax Problem

作　　者：赵媛璐 王福胜[1] 

机构地区：[1]太原师范学院数学系,山西晋中030619

出　　处：《太原师范学院学报（自然科学版）》2017年第4期6-11,共6页Journal of Taiyuan Normal University:Natural Science Edition

基　　金：山西省留学回国基金(2017-104)

摘　　要：提出了求解不等式约束minimax问题的一个可行序列线性方程组算法.在每次迭代中可行下降方向通过求解两个具有相同系数矩阵的线性方程组产生,系数矩阵具有较好的稀疏性,计算量少.在适当的假设条件下,算法具有全局收敛性,初步数值实验验证了算法的有效性.A feasible sequential system of linear equations algorithm for inequality con- strained minimax problems is proposed. At each iteration,the proposed algorithm solves only two systems of linear equations with a same coefficient matrix obtain the feasible descent direction. Furthermore,the sparsity of the coefficient matrix is good. It is proved that under appropriate as- sumptions, the algorithm is globally convergent. Finally, several numerical examples are reported to verify effectiveness of the algorithm.

关 键 词：不等式约束minimax问题 序列线性方程组 全局收敛性 

分 类 号：O221.2[理学—运筹学与控制论]