The scalability and robustness of a class of non-overlapping domain decomposition preconditioners using 2-way nested dissection reordering is studied. In particular, three methods are considered: a nested symmetric successive over-relaxation (NSSOR), a nested version of modified ILU with rowsum constraint (NMILUR), and nested filtering factorization (NFF). The NMILUR preconditioner satisfies the rowsum property i.e., a right filtering condition on the vector (1,...,1)T . The NFF method is more general in the sense that it satisfies right filtering condition on any given vector. There is a subtle difference between NMILUR and NFF, but NFF is much more robust and converges faster than NSSOR and NMILUR. The test cases consist of a Poisson problem and convection-diffusion problems with jumping coefficients. © 2013 Springer-Verlag.
CITATION STYLE
Kumar, P., Meerbergen, K., & Roose, D. (2013). Multi-threaded nested filtering factorization preconditioner. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7782 LNCS, pp. 220–234). https://doi.org/10.1007/978-3-642-36803-5_16
Mendeley helps you to discover research relevant for your work.