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.
出处
《海南大学学报(自然科学版)》
CAS
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)
关键词
非单调技术
信赖域ODE算法
整体收敛
无约束优化
non-monotonic technique
trust region ODE-algorithm
global convergence
unconstrained optimization