A PROXIMAL ALGORITHM WITH VARIABLE METRIC FOR THE P0 COMPLEMENTARITY PROBLEM

We consider a regularization proximal method with variable metric to solve the nonlinear complementarity problem (NCP) for P0-functions. We establish global convergence properties when the solution set is non empty and bounded. Furthermore, we prove, without boundedness of the solution set, that the sequence generated by the algorithm is a minimizing sequence for the implicit Lagrangian function, as defined by Mangasarian and Solodov.Those results are tronger than the presented in a previous paper.

Citation

TR ES 683/05, PESC/COPPE/UFRJ, Rio de Janeiro, BR, 05/2005

Article

Download

View PDF