Filippo Leonardi <[email protected]> writes: >> -ksp_chebyshev_estimate_eigenvalues_random > > Can try this but the problem, to me, seems deeper (because inherited from > each > solve afterwards).
Are you changing the operator between solves? Or just solving different right-hand sides? If the latter, the spectral estimates are reused.
signature.asc
Description: PGP signature
