A theoretical overview of Krylov subspace methods (Q1917418): Difference between revisions

This excellent paper contains a wide spectrum of iterative methods for solving linear systems. The author presents a survey of Krylov subspace methods for solving nonsymmetric and not positive definite linear systems. To provide a framework for these methods, the outline of an orthogonalization method is presented. Estimates of a norm of the residual and error vector for generalized minimum error methods and conjugate Krylov subspace methods, including BCG, BICO, QMR, TFQMR, the rank-3 update CKS method, the residual or error-minimizing CG, and the energy norm-minimizing CG are given. Discussion and practical experiences with described methods conclude this paper.