Accelerating MPGP-type Methods Through Preconditioning
For researchers solving quadratic programming problems, this work provides a practical acceleration technique for MPGP-type methods.
This paper accelerates MPGP-type algorithms for quadratic programming by introducing an approximate variant of preconditioning in face that computes the inner preconditioner only once, achieving large speedups in numerical experiments.
This work investigates the acceleration of MPGP-type algorithms using preconditioning for the solution of quadratic programming problems. The preconditioning needs to be done only on the free set so as not to change the constraints. A variant of preconditioning restricted to the free set is the preconditioning in face. The inner preconditioner in preconditioning in face needs to be recomputed or updated every time the free set changes. Here, we investigate an approximate variant of preconditioning in face that computes the inner preconditioner only once. We analyze the error of the approximate variant, give a sharp bound on the condition number of the preconditioned operator, and provide numerical experiments demonstrating that very large speedups can be achieved by the approximate variant.