A NEW SQP-FILTER METHOD PROGRAMMING FOR SOLVING NONLINEAR PROBLEMS

作     者：Duoquan Li 

作者机构：LSEC ICMSEC Academy of Mathematics and System Sciences Chinese Academy of Sciences Beijing 100080 China Graduate School of Chinese Academy of Sciences Beijing 100080 China) 

出 版 物：《Journal of Computational Mathematics》 (计算数学（英文）)

年 卷 期：2006年第24卷第5期

页      面：609-634页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 07[理学] 070105[理学-运筹学与控制论] 0701[理学-数学] 

基　　金：Partly supported by Chinese 

主　　题：Nonlinear programming Sequential quadratic programming Filter Restoration phase Maratos affects Global convergence Multi-objective optimization Quadratic programming subproblem. 

摘      要：In [4], Fletcher and Leyffer present a new method that solves nonlinear programming problems without a penalty function by SQP-Filter algorithm. It has attracted much attention due to its good numerical results. In this paper we propose a new SQP-Filter method which can overcome Maratos effect more effectively. We give stricter acceptant criteria when the iterative points are far from the optimal points and looser ones vice-versa. About this new method, the proof of global convergence is also presented under standard assumptions. Numerical results show that our method is efficient.