Currently only SuiteSparse/CHOLMOD benefits from the block AMD ordering. This
can have a significant performance impact on the cholesky factorization
routines.
Both SPARSE_SCHUR and SPARSE_NORMAL_CHOLESKY will benefit from this enhancement.
The work involves refactoring some of the block matrix ordering code from
suitesparse.cc
and combining it with the internals of cs_schol.c, where cs_amd is explicitly
called and the resulting block permutation is used to order the scalar matrix.
Original issue reported on code.google.com by sameerag...@google.com on 20 Apr 2013 at 11:07
Original issue reported on code.google.com by
sameerag...@google.com
on 20 Apr 2013 at 11:07