We should provide a general mechanism, maybe in the form of an additional argument, to enable Hermitian transpose versus real transpose for all these operations.
It pains me to tell someone on petsc-maint that they should use MatHermitianTranspose() for a PtAP-like operation. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20121012/48598342/attachment.html>
