Iterative methods for linear complementarity problems and large scale quadratic programming problems

Abstract

This paper concerns with iterative methods for large quadratic programming problems and related linear complementarity problems. There have been numerous works done for large scale quadratic programming problems with simple upperbounds. This work extended the available results in two directions. First, an algorithm is suggested to handle quadratic programs with block angular constraints rather than simple upper bounds. Convergence is also investigated. Second part of this work has developed an algorithm which can handle linear complementarity problems of the form obtained from upperbounded quadratic programs, but without symmetry assumption. This algorithm is not entirely new, but its convergence is newly investigated along the line of Ahn``s work. Simple examples along with computational experience is presented to illustrate the proposed method.