REFINEMENT OF A PRECONDITIONED ACCELERATED OVERRELAXATION (AOR) ITERATIVE METHOD

Abstract

Iterative methods for solution of linear algebraic systems abound literature. Of primary concern to any acceptable iteration method is its efficiency in terms of rate of convergence; the faster the convergence the better. This present paper is an attempt to investigate the effect of refinement on a fast-converging preconditioned AOR iteration, in order to further improve on its convergence. The resulting iteration, the refinement of preconditioned AOR (RPAOR) method is subjected to validation through numerical experiments, and it was revealed that the RPAOR attains convergence rates that surpass those of the classical AOR, the preconditioned AOR and the refinement of AOR methods, in all cases.