A decomposition algorithm for solving large-scale quadratic programming problems

In this paper an algorithm for solving large-scale programming is proposed. We decompose a large-scale quadratic programming into a serial of small-scale ones and then approximate the solution of the large-scale quadratic programming via the solutions of these small-scale ones. It is proved that the accumulation point of the iterates generated by the algorithm is a global minimum point of the quadratic programming. The algorithm has performed very well in numerical testing. It is a new way of solving large-scale quadratic programming problems.