摘要
提出了一个求解带等式和不等式约束的Minimax问题的SQP-Filter算法,每步通过求解2个二次规划子问题来得到搜索方向,并沿该方向做线搜索。该算法避免了较难的罚因子的选取,克服了Maratos效应,并在适当的假设条件下,得到了算法的全局收敛性。
In this paper, a SQP - Filter algorithm is proposed for solving minimax problems with equality and inequality constraints. In each step, a quadratic programming sub-problem is solved to get the search direction which is used as line search. This method avoids the difficulty of selecting the penalty factor and overcomes Maratos effects successfully. Its global convergence is attained under some suitable conditions.
出处
《西华大学学报（自然科学版）》
CAS
2011年第6期61-64,共4页
Journal of Xihua University:Natural Science Edition
基金
国家自然科学基金(11071041)
福建省自然科学资金(2009J01002)
福建省教育厅项目(JA11270)
