The Rheinfall parallel algorithm for Gaussian Elimination

The "Rheinfall" algorithm is a way of re-arranging the operations in Gaussian Elimination, which has a natural parallel and distributed-memory formulation but degrades gracefully to sequential execution. It is suitable for elimination of general sparse matrices when exact arithmetic computations are used. I will introduce the algorithm and the motivation for its development, and present some performance results over selected matrices of the SIMC collection, both for the sequential and the parallel execution modes.