摘要
基于最速下降法把无约束优化问题同约束优化问题结合起来,形成一种能避免最速下降法的重要缺陷,而具有全局下降性的新算法,同时给出了其总体收敛性定理和简单数值实验.应用此算法,分析了某些无约束多峰函数最优化问题.
Based on the steepest decent method and combining unconstrained optimization problems with (constrained) optimization ones,the paper aims to develop a new method and algorithm with a global decreasing property that can deal with the crucial flow of the steepest decent method.Furthermore,two theorems on the global convergence of the algorithm and one numerical experiment help to understand it.An exploration is (proposed) to extend the method to multiextremal global optimization in some situations.
出处
《纺织高校基础科学学报》
CAS
2004年第1期40-42,共3页
Basic Sciences Journal of Textile Universities
基金
陕西省教育厅专项科研基金资助项目(03jk065)
西安建筑科技大学基础研究基金资助项目(DB12006)
关键词
混合
整体下降性
梯度
非线性规划
搜索点
hybrid method
global decreasing property
search point
gradient method
nonlinear programming