A parallel algorithm for sparse symbolic LU factorization without pivoting on out—of—core matrices
Michel Cosnard, Laura Grigori · 2001
Finding the nonzero structures of the lower and upper triangular factors of an unsymmetric sparse matrix A is an important problem in the field of sparse matrix computations. Complementing previous research on sequential algorithms, we develop a parallel algorithm by appropriately intertwining a fully concurrent algorithm with an experimentally proved efficient algorithm (both algorithms are previous art). The resulting algorithm, intended to use with out-of-core matrices, leads to sensitive improvements in memory usage.