您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(理学版)》

J4 ›› 2011, Vol. 46 ›› Issue (7): 101-105.

• 数学 • 上一篇    下一篇

一种充分下降的DY共轭梯度法及其收敛性

李敏,陈宇,屈爱平   

  1. 怀化学院数学系, 湖南 怀化 418008
  • 收稿日期:2010-01-13 出版日期:2011-07-20 发布日期:2011-09-08
  • 作者简介:李敏(1980- ),男,讲师,硕士研究生,研究方向为最优化理论与算法.Email:minjiecow@163.com
  • 基金资助:

    怀化学院资金资助项目(HHUQ200901)

A sufficient descent DY conjugate gradient method and its global convergence

LI Min, CHEN Yu, QU Ai-ping   

  1. Department of Mathematics and Applied mathematics, Huaihua University, Huaihua 418008, Hunan, China
  • Received:2010-01-13 Online:2011-07-20 Published:2011-09-08

摘要:

基于已有的DY方法和HZ方法,提出了一种修正的DY共轭梯度法(MDY算法)。该算法产生的搜索方向为充分下降方向,且这一性质与所采用的线搜索方法无关。在一定的条件下证明了保守MDY算法(CMDY算法)基于Armijo线搜索和Wolfe线搜索求解非凸优化问题的全局收敛性。相关的数值试验结果验证了该方法的有效性。

关键词: 充分下降;DY方法;Armijo线搜索;Wolfe线搜索;全局收敛

Abstract:

 A modified DY conjugate gradient method(MDY)is proposed based on DY method and HZ method. This method can generate sufficient descent directions for the objective functions, and this property is independent of the line search method used. Under mild conditions, it is proved that the conservative MDY method with Armijo line search or Wolfe line search converges globally even if the minimization function is nonconvex. Numerical results show that the proposed method is efficient.

Key words:  sufficient descent; DY method; armijo line search; wolfe line search; global convergence

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!