Matrix Krylov subspace methods for linear systems with multiple right-hand sides (Q2583219)

The authors discuss global Krylov methods for solving systems of linear equations with multiple right-hand sides. At first a relation between the global methods for such systems of equations and the corresponding classical methods for systems of equations with a single right-hand side is given. Then the global generalized Hessenberg (GGH) process as well as the global Galerkin and minimal residual norm methods are discussed. The GGH process constructs a basis of a matrix Krylov subspace and makes use of the Frobenius scalar product. A weighting technique proposed by \textit{A. Essai} [Numer.~Algorithms 18, No. 3--4, 277--292 (1998; Zbl 0926.65036)] is applied to accelerate the convergence of the proposed global methods. Complexity comparisons as well as comparisons on the numerical behaviour between the global methods, the block GMRES, and the classical Krylov methods are presented. Finally, numerical experiments are reported. These experiments show that the weighted methods converge faster than the original methods.