Improving the modified Gauss-Seidel method for \(Z\)-matrices (Q1373313)

The authors investigate the Gauss-Seidel method for solving the linear system \(Ax=b\) with a preconditioning matrix \(I + S(\alpha )\), \(\alpha\) a given parameter vector. The proposed method is a generalization of the modified Gauss-Seidel method with \(I + S\) as preconditioner. It is shown that if \(A\) is an irreducibly diagonally dominant \(Z\)-matrix, then \((I+S(\alpha))A\) is a strictly diagonally dominant \(Z\)-matrix. A practical technique for the determination of an optimal parameter \(\alpha\) is proposed. Numerical experiments demonstrate the improved convergence rate of the new method. It is also shown that the spectral radius of the iteration matrix of the new method is smaller than that of the successive overrelaxation algorithm.