ARMS: an algebraic recursive multilevel solver for general sparse linear systems

From MaRDI portal





The paper describes new recursive multilevel method for preconditioning of general sparse linear systems. This strategy is used in the new solver (ARMS) that generalize previous authors' codes BILUM and BILUTM. All these methods are based on a block incomplete LU factorization. The ARMS is fully recursive and employs the nested dissection reordering and inner-level iterations. Assumptions, under which the new preconditioning is exact, are given together with the proof that eigenvalues of the preconditioned matrix are close to 1. Extensive numerical tests are presented and cover various features of the method. They show that the solver ARMS is more robust, saves memory, but performs slower than ILUT and ILUTP (incomplete LU factorization with threshold and with threshold and pivoting).




Cited in
(48)


Describes a project that uses

Uses Software






This page was built for publication: ARMS: an algebraic recursive multilevel solver for general sparse linear systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5317851)