非线性极大极小问题的可行信赖域算法
A Feasible Trust Algorithm of Nonlinear Max-Min Problems
-
摘要: 针对一类非线性约束极大极小问题,利用极大熵方法将转化为带不等式约束的非线性规划问题,给出了一种可行信赖域算法,解决了不等式约束的非线性大系统优化问题,并证明了该算法的全局收敛性。初步的数值试验表明,对于该类极大极小问题,本算法有良好的数值表现。Abstract: A feasible trust algorithm which used maxim-entropy methods, a class of max-min problems with nonlinear constraints turn into nonlinear programming problems with inequality constraints is proposed. The algorithm which resolve general max-min problems is global convergent. Preliminary numerical experiments show that the proposed algorithm is effective.