摘要
提出一类新的求解无约束优化问题的记忆梯度法,在较弱条件下证明了其全局收敛性.新算法有如下特点:(1)采用曲线搜索方法,在每步迭代时同时确定下降方向和步长;(2)利用当前和前面迭代点的信息产生下降方向,无需计算和存储矩阵,适于求解大型优化问题.
A new memory gradient method for unconstrained optimization problems was presented and its global convergence under some mild conditions was proved. The new method has the following properties： （ 1 ） at each iteration the descent direction and step size were determined at same time by using curve search method. （2）it is suitable in solving large scale optimization problems by using the current and previous iterative information to generate the descent di- rection and avoiding the computation and storage of some matrices.
出处
《信阳师范学院学报（自然科学版）》
CAS
2009年第2期179-182,共4页
Journal of Xinyang Normal University(Natural Science Edition)
基金
河南省教育厅自然科学基金项目(2008A110016)
关键词
无约束优化
记忆梯度法
曲线搜索
unconstrained optimization
memory gradient method
curve search
