Local convergence of an inexact Newton-type method involving optimization model on subproblems
Abstract
The present paper deals with inexact Newton-type scheme for solving generalized equation governed by set-valued mappings defined on finitely dimensional spaces. We propose a new dynamical updating strategy by adapting in a mathematical program modeling based on the linearization of the single-valued part at each step. We investigate the local convergence behavior of the proposed framework and applied it to design a structural algorithm for solving complementarity problems. Implementation of several numerical tests was also considered to illustrate the feasibility of such framework.