Parallel Multisplitting Methods with Variable Weighting Matrices for Symmetric Positive Definite Linear System

摘要

In this paper, the new parallel multispltting iterative method with variable weighting matrices is presented for the system of linear equations when the coefficient matrix is a symmetric positive definite matrix. The non-negativity of the weighting matrices, especially, is eliminated. The convergence theories are established for this parallel multisplitting method. Finally, the numerical comparisons are shown.