Iterative methods for \(X-AXB=C\) (Q1378994)

The paper describes and analyzes several variants of alternating directions implicit iterative methods for the solution of matrix equations of the form \(X-AXB=C\). The convergence analysis is based on potential theory and shows that iterating in one direction more than the other can give faster convergence than strict alternation of directions. Numerical examples suggest that these methods, when applied to image restoration, are competitive with the conjugate gradient algorithm.