An improved LQP-based method for solving nonlinear complementarity problems

Abstract

The well-known logarithmic-quadratic proximal (LQP) method has motivated a number of efficient numerical algorithms for solving nonlinear complementarity problems (NCPs). In this paper, we aim at improving one of them, i.e., the LQP-based interior prediction-correction method proposed in [He, Liao and Yuan, J. Comp. Math., 2006, 24(1): 33-44], via identifying more appropriate step-sizes in the correction steps. Preliminary numerical results for solving some NCPs arising in traffic equilibrium problems are reported to verify the theoretical assertions.