| 一类新的超记忆多步曲线搜索方法及其全局收敛性 |
| |
| 引用本文： | 孙敏. 一类新的超记忆多步曲线搜索方法及其全局收敛性[J]. 青岛化工学院学报(自然科学版), 2008, 0(5): 464-466 |
| |
| 作者姓名： | 孙敏 |
| |
| 作者单位： | 枣庄学院数学与信息科学系,山东枣庄277160 |
| |
| 摘 要： | 提出一种求解无约束最优化问题的超记忆多步曲线搜索方法,此方法具有如下特点：（1）每次迭代目标函数f（x）下降量更大;（2）充分利用前m步的迭代信息;（3）每次迭代同时确定下降方向和步长;（4）步长一致有正下界。在较弱的条件下,证明了此方法的收敛性。
|
| 关 键 词： | 无约束最优化 多步法 曲线搜索法 收敛性 |
A New Super-memory Multi-step Curve Search Method and its Global Convergence |
| |
| SUN Min. A New Super-memory Multi-step Curve Search Method and its Global Convergence[J]. Journal of Qingdao Institute of Chemical Technology(Natural Science Edition), 2008, 0(5): 464-466 |
| |
| Authors: | SUN Min |
| |
| Affiliation: | SUN Min (Department of Mathematics and Information Science, Zaozhuang University, Zaozhuang 277160,China) |
| |
| Abstract: | A new super-memory multi-step curve search method is proposed for unconstrained optimization in this paper, the following four properties are obtained： （1） the value of f（x） decreases more at each iteration; （2） it uses the previous m-step iterative information sufficiently; （3） the descent direction and the step-size are determined at the same time; （4） the step-sizes have a positive uniform bound from below. Under mild conditions, the convergence of the new method is proved. |
| |
| Keywords: | unconstrained optimization multi-step method curve search method convergence. |
| 本文献已被 维普 等数据库收录！ |
