Summary.
In this paper, tangential frequency filtering decompositions (TFFD) for unsymmetric matrices are introduced. Different algorithms for the construction of unsymmetric tangential frequency filtering decompositions are presented. These algorithms yield for a specified class of matrices equivalent decompositions. The convergence rates of an iterative scheme, which uses a sequence of TFFDs as preconditioners, are independent of the number of unknowns for this class of matrices. Several numerical experiments verify the efficiency of these methods for the solution of linear systems of equations which arise from the discretisation of convection-diffusion differential equations.
Similar content being viewed by others
Author information
Authors and Affiliations
Additional information
Received April 1, 1996 / Revised version received July 4, 1996
Rights and permissions
About this article
Cite this article
Wagner, C. Tangential frequency filtering decompositions for unsymmetric matrices. Numer. Math. 78, 143–163 (1997). https://doi.org/10.1007/s002110050308
Issue Date:
DOI: https://doi.org/10.1007/s002110050308