Stationary splitting iterative methods for the matrix equation AXB=C

Abstract Stationary splitting iterative methods for solving A X B = C are considered in this paper. The main tool to derive our new method is the induced splitting of a given nonsingular matrix A = M − N by a matrix H such that ( I − H ) − 1 exists. Convergence properties of the proposed method are discussed and numerical experiments are presented to illustrate its computational efficiency and the effectiveness of some preconditioned variants. In particular, for certain surface fitting applications our method is much more efficient than the progressive iterative approximation (PIA), a conventional iterative method often used in computer aided geometric design (CAGD).