Applied Mathematics and Mechanics (English Edition) ›› 2003, Vol. 24 ›› Issue (6): 738-746.

• 论文 • 上一篇    

GLOBAL LINEAR AND QUADRATIC ONE-STEP SMOOTHING NEWTON METHOD FOR VERTICAL LINEAR COMPLEMENTARITY PROBLEMS

张立平1, 高自友2   

  1. 1. Department of Methematical Sciences, Tsinghua University, Beijing 100084, P. R. China;
    2. School of Traffic and Transportation, Northern Jiaotong University, Beijing 100044, P. R. China
  • 收稿日期:2002-01-29 修回日期:2003-03-15 出版日期:2003-06-18 发布日期:2003-06-18
  • 基金资助:
    the National Natural Science Foundation of China(10201001);the National Outstanding Young Investigator Grant(70225005)

GLOBAL LINEAR AND QUADRATIC ONE-STEP SMOOTHING NEWTON METHOD FOR VERTICAL LINEAR COMPLEMENTARITY PROBLEMS

ZHANG Li-ping1, GAO Zi-you2   

  1. 1. Department of Methematical Sciences, Tsinghua University, Beijing 100084, P. R. China;
    2. School of Traffic and Transportation, Northern Jiaotong University, Beijing 100044, P. R. China
  • Received:2002-01-29 Revised:2003-03-15 Online:2003-06-18 Published:2003-06-18
  • Supported by:
    the National Natural Science Foundation of China(10201001);the National Outstanding Young Investigator Grant(70225005)

摘要: A one-step smoothing Newton method is proposed for solving the vertical linear complementarity problem based on the so-called aggregation function. The proposed algorithm has the following good features: (ⅰ) It solves only one linear system of equations and does only one line search at each iteration;(ⅱ) It is well-defined for the vertical linear complementarity problem with vertical block P0 matrix and any accumulation point of iteration sequence is its solution.Moreover, the iteration sequence is bounded for the vertical linear complementarity problem with vertical block P0+R0 matrix;(ⅲ) It has both global linear and local quadratic convergence without strict complementarity. Many existing smoothing Newton methods do not have the property (ⅲ).

Abstract: A one-step smoothing Newton method is proposed for solving the vertical linear complementarity problem based on the so-called aggregation function. The proposed algorithm has the following good features: (ⅰ) It solves only one linear system of equations and does only one line search at each iteration;(ⅱ) It is well-defined for the vertical linear complementarity problem with vertical block P0 matrix and any accumulation point of iteration sequence is its solution.Moreover, the iteration sequence is bounded for the vertical linear complementarity problem with vertical block P0+R0 matrix;(ⅲ) It has both global linear and local quadratic convergence without strict complementarity. Many existing smoothing Newton methods do not have the property (ⅲ).

中图分类号: 

APS Journals | CSTAM Journals | AMS Journals | EMS Journals | ASME Journals