丁老师 <[email protected]> writes: > i think bandwidth minimization will improve the convergence of the kspsolver, > am i right?
Bandwidth is *irrelevant* for Krylov methods. It can affect preconditioning, but it's really not the right metric. Choose a partitioning that respects the coupling strength in your problem, then choose a local ordering that has the properties you like. (Low bandwidth is just one possibility.)
pgpCCCulds9_1.pgp
Description: PGP signature
