Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (20): 30-35.DOI: 10.3778/j.issn.1002-8331.2008.20.010

• 理论研究 • Previous Articles     Next Articles

Smoothing Newton algorithm for solving linear programs problem

SUN Xiu-ping1,ZHENG Pi-e2   

  1. 1.Department of Mathematics,School of Science,Tianjin University,Tianjin 300072,China
    2.School of Management,Tianjin University,Tianjin 300072,China
  • Received:2007-10-09 Revised:2008-01-15 Online:2008-07-11 Published:2008-07-11
  • Contact: SUN Xiu-ping

求解线性规划问题的光滑型牛顿算法

孙秀萍1,郑丕谔2   

  1. 1.天津大学 理学院 数学系,天津 300072
    2.天津大学 管理学院,天津 300072
  • 通讯作者: 孙秀萍

Abstract: We propose a continuation method for solving the Linear Program(LP) by making use of an augmented system of its optimality conditions.The algorithm is shown to be globally convergent without requiring any assumption.It only needs to solve one system of linear equations and to perform one line search at each iteration.To the best of our knowledge this is the first continuation smoothing-type algorithm for solving the LP having the above desired convergence features.

Key words: linear program, smoothing Newton algorithm, global convergence, strictly complementary solution

摘要: 对线性规划的最优性条件,给出一个扩展系统,设计一个连续化的光滑型算法求解该系统。所设计的算法的全局收敛性不需要添加任何假设条件。在每一个迭代点处,只需要解一个线性方程组和做一次线性搜索,比现有求解线性规划问题的连续化方法具有更好的收敛性质。

关键词: 线性规划, 光滑型牛顿算法, 全局收敛, 严格互补解