engleski

Improved Symmetric Gauss-Seidel Method for Solving Sparse Linear Systems Appear in CFD

Two new variants of iterative solvers for sparse linear systems of algebraic equations that appear in Computational Fluid Dynamics (CFD) are proposed. The proposed methods are compared with selected Krylov space methods in terms of CPU time required to achieve prescribed accuracy for selected test problems. When the coefficient matrix is diagonally dominant there are no significant differences between efficiency of methods and all of them are quite applicable. For the most demanding problems corresponding to the pressure equation in a incompressible fluid flow solver, the new proposed methods show a great advantage over the popular GMRES(5) method and they are comparable with PTFQMR method with the simple ILU preconditioner (corresponding the Symmetric Gauss-Seidel method).

Sparse linear systems; Iterative method; Symmetric Gauss-Seidel method

nije evidentirano