基于非单调技术的ODE型算法  

Non-monotone ODE-type Trust Region Method

作　　者：张军[1] 王冠舒[1] 

机构地区：[1]海南大学信息科学技术学院,海南海口570228

出　　处：《海南大学学报（自然科学版）》2012年第1期16-19,共4页Natural Science Journal of Hainan University

基　　金：海南省自然科学基金项目(111001)

摘　　要：将非单调技术与信赖域ODE算法相结合,提出了一种求解无约束优化的新算法,从而减少了迭代次数以及信赖域子问题的计算次数.并给出在一定条件下算法的整体收敛性,数值试验表明算法有效.Non-monotone technology were combined with traditional trust region ODE-algorithm,and a new algorithm for unconstrained optimization problem were proposed,and which can reduce the number of iterations and trust region sub-problems number of calculation.Under certain assumptions,this paper also proved algorithm＇s global convergence.Numerical results indicated that the proposed algorithm is effective and feasible.

关 键 词：非单调技术 信赖域ODE算法 整体收敛 无约束优化 

分 类 号：O221[理学—运筹学与控制论]