[petsc-dev] strangness in Chebyshev estimate of eigenvalues

[Jed Brown]

Mark Adams <mfadams at lbl.gov> writes:
> The point is that it is possible to have a process invariant test, as
> opposed to impossible. PETSc's MIS is not invariant now anyway.  

If we want something invariant that is algorithmically defensible, we
have to use a quality hash function.  It's entirely feasible, we just
have to decide to do it.

> I don't think about DMDA; the problems that I have experience with
> have a global numbering that can be preserved if one wishes for
> debugging.

That implies a "1D partition" of a fixed ordering.  I suppose that's
okay when people have already chosen SFC orderings.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20150829/c94214b5/attachment.sig>