Trajectory-following methods for large-scale degenerate convex quadratic programming (Q2392932): Difference between revisions

This paper considers an infeasible interior-point method for solving the general convex quadratic programming problem. Global convergence of the method is established. Additional trial steps for the choice of idealised solution trajectory are described to accelerate the convergence of the method on both degenerate and nondegenerate problems. The implementation of the convex quadratic programming package is described. Some experimental results on 178 convex examples are presented to illustrate the performance of the package.