节点文献
求解无约束优化问题的一类新的下降算法
A Class of Efficient New Descent Methods
【摘要】 本文对求解无约束优化问题提出了一类新的下降算法,并且给出了HS算法与其相结合的两类杂交算法.在Wolfe线搜索下不需给定下降条件,即证明了它们的全局收敛性.数值实验表明新的算法十分有效,尤其是对求解大规模问题而言.
【Abstract】 In this paper, we propose a class of new descent methods and also we give two hybrids methods based on Hestenes-Stiefel and our new methods. And, we proved their global congenvence in Wolfe line search without descent condition. Numerical experiments shows that our methods are very efficient, especially for large scale problems.
【关键词】 无约束最优化;
下降类算法;
Wolfe线搜索;
全局收敛性;
【Key words】 unconstrained optimization; descent method; wolfe line search; global convergence;
【Key words】 unconstrained optimization; descent method; wolfe line search; global convergence;
【基金】 国家自然科学基金(60472071);北京市教委科研基金(KM200710028001)资助
- 【文献出处】 应用数学学报 ,Acta Mathematicae Applicatae Sinica , 编辑部邮箱 ,2007年01期
- 【分类号】O221.2
- 【被引频次】31
- 【下载频次】372