Intra-block amalgamation in sparse hypermatrix Cholesky factorization

jo s e pr,jua njo @ a c.up c.e du Abstract- In this paper we present an im-prove m e nt to o ur s e que nti al i n- co re i m pl e-mentation of a sparse Cholesky factorization based on a hypermatrix storage structure. We allow the inclusion of additional zeros in data submatrices to create larger blocks and in this way use more efficient routines for matrix multiplication. Since matrix multi-plication takes about 90 % of the total fac-torization time this is an important point to optimize.