[petsc-users] Use of PCApplyBAorAB with PC_SYMMETRIC side

[Jed Brown]

Sylvain Mercier <sylvainmercier85 at gmail.com> writes:

> Hello everyone,
>
> I want to use PCApplyBAorAB symmetrically with a preconditioner from an ICC
> factorization.
> However, I don't understand why the result is not the same if I use a
> natural or a RCM ordering during the factorization.

Uh, ICC is not the same in RCM versus natural ordering.  Incomplete
factorization is _highly_ order-dependent.

> This problem does not occur if I use this routine with PC_LEFT or PC_RIGHT

What do you mean "does not occur"?  You should expect that changing
factorization ordering will always change the preconditioner, though
some cases could cancel.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140217/faadd62b/attachment.pgp>