Global convergence enhancement of classical linesearch interior point methods for MCPs

作者:

Highlights:

摘要

Recent works have shown that a wide class of globally convergent interior point methods may manifest a weakness of convergence. Failures can be ascribed to the procedure of linesearch along the Newton step. In this paper, we introduce a globally convergent interior point method which performs backtracking along a piecewise linear path. Theoretical and computational results show the effectiveness of our proposal.

论文关键词:65K05,65H10,65H20,90C33,Complementarity problems,Interior point methods,Backtracking strategy

论文评审过程:Received 23 November 2001, Revised 8 July 2002, Available online 6 December 2002.

论文官网地址:https://doi.org/10.1016/S0377-0427(02)00745-8