摘要
在CD方法和DY方法的基础上对求解无约束优化问题提出了一种混合的CD-DY共轭梯度法.在广义Wolfe线搜索下无需给定下降条件,即可证明混合方法的全局收敛性.初步试验表明新方法的数值效果优于CD方法和DY方法.
In this paper, a mixed CD-DY conjugate gradient method is proposed to solve unconstrained optimiza-tion problems on the basis of CD method and DY method. And, the paper proved the global convergence of the mixed method without the given descent condition, under the general Wolfe line search condition. The numerical experiments show that the mixed method excelled CD method and DY method.
出处
《常熟理工学院学报》
2014年第4期44-47,共4页
Journal of Changshu Institute of Technology
关键词
无约束优化
共轭梯度法
广义Wolfe线搜索
全局收敛性
unconstrained optimization
conjugate gradient method
general Wolfe line search
global conver-gence