共 28 条
Smoothing Newton method for NCP with the identification of degenerate indices
被引:4
作者:
Yu, Haodong
[1
]
Pu, Dingguo
[1
]
机构:
[1] Tongji Univ, Dept Math, Shanghai 200092, Peoples R China
基金:
中国国家自然科学基金;
关键词:
Nonlinear complementarity problems;
Degenerate indices;
Smoothing methods;
Global convergence;
Superlinear convergence;
NONLINEAR COMPLEMENTARITY-PROBLEMS;
LEVENBERG-MARQUARDT METHOD;
SEMISMOOTH REFORMULATION;
ALGORITHM;
D O I:
10.1016/j.cam.2010.05.004
中图分类号:
O29 [应用数学];
学科分类号:
070104 ;
摘要:
We present a new smoothing Newton method for nonlinear complementarity problems (NCP(F)) by using an NCP function to reformulate the problem to its equivalent form. Compared with most current smoothing methods, our method contains an estimating technique based on the active-set strategy. This technique focuses on the identification of the degenerate set for a solution x* of the NCP(F). The proposed method has the global convergence, each accumulation point is a solution of the problem. The introduction of the active-set strategy effectively reduces the scale of the problem. Under some regularity assumption, the degenerate set can be identified correctly near the solution and local superlinear convergence is obtained as well. (C) 2010 Elsevier B.V. All rights reserved.
引用
收藏
页码:3424 / 3435
页数:12
相关论文