Estrategias de segunda ordem para problemas de complementaridade / Second order strategies for complementarity problems

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

In this work we reformulate the generalized nonlinear complementarity problem (GNCP) in polyhedral cones as a nonlinear system with nonnegativity in some variables and propose the resolution of such reformulation through interior-point methods. In particular we define two algorithms and prove the local convergence of these algorithms under standard assumptions. The first algorithm is based on Newton s method and the second, on the Chebyshev s tensorial method. The algorithm based on Chebyshev s method may be considered a predictor-corrector one. Such algorithm, when applied to problems for which the functions are affine, and the parameters are properly chosen, turns into the well-known Mehrotra s predictor corrector algorithm. We also present numerical results that illustrate the competitiveness of both proposals

ASSUNTO(S)

complementaridade metodo preditor-corretor metodo de predictor-corrector method newton singular systems polyhedra cones interior-points methods sistemas singulares complementary newton method metodos de pontos interiores cones poliedrais

Documentos Relacionados