Paralell sparse matrix computations by