Improved smoothing Newton methods for symmetric cone complementarity problems

Improved smoothing Newton methods for symmetric cone complementarity problems

0.00 Avg rating0 Votes
Article ID: iaor20122525
Volume: 6
Issue: 3
Start Page Number: 471
End Page Number: 487
Publication Date: Mar 2012
Journal: Optimization Letters
Authors: , ,
Keywords: programming: nonlinear
Abstract:

There recently has been much interest in smoothing Newton method for solving nonlinear complementarity problems. We extend such method to symmetric cone complementarity problems (SCCP). In this paper, we first investigate a one‐parametric class of smoothing functions in the context of symmetric cones, which contains the Fischer–Burmeister smoothing function and the CHKS smoothing function as special cases. Then we propose a smoothing Newton method for the SCCP based on the one‐parametric class of smoothing functions. For the proposed method, besides the classical step length, we provide a new step length and the global convergence is obtained. Finally, preliminary numerical results are reported, which show the effectiveness of the two step lengthes in the algorithm and provide efficient domains of the parameter for the complementarity problems.

Reviews

Required fields are marked *. Your email address will not be published.