A necessary and sufficient condition for semiconvergence and optimal parameter of the SSOR method for solving the rank deficient linear least squares problem (Q858764)

For finding the minimum norm least squares solution of a general linear system, one can apply the symmetric successive overrelaxation (SSOR) iteration method to the corresponding augmented system. In this paper, the authors establish a necessary and sufficient condition for semiconvergence of the SSOR method, and give the optimal values of its parameter \(\omega\). In addition, some related problems concerning the successive overrelaxation (SOR) and accelerated overrelaxation (AOR) are studied as well.