> > FENothing is currently not implemented. I believe that it shouldn't be
> > too complicated to do -- there's no basis functions, no hanging node
> > constraints, etc, but I would anticipate that there are things that can
> > go wrong in various places of the library where *something* is expected.
>
> Great, so different variables can have different discretizations in
> different elements?
Yes, or at least that's the intention. If it doesn't work it's a bug that I'll
be happy to help fixing.
> Thanks. But the question remains: do the native solvers have other
> advantages or features? My comparison point is libMesh, which just
> wraps the PETSc linear and nonlinear solvers.
It's mostly historic: when we started deal.II, PETSc or Trilinos weren't
around so we had to do things ourselves. Performancewise it doesn't make much
of a difference on a single processor (we typically win by a few per cent,
but not by any significant margin). When you go to multiple processors,
deal.II has no support and you need to use the wrappers to PETSc and
Trilinos.
An additional feature of deal.II's linear algebra may be that it's interface
doesn't change. This should be a non-feature (it should in fact be
self-evident) but it becomes one since PETSc changes their interfaces with
every frickin dot version, making it rather hard to keep code compatible
across several PETSc versions. I complain every time I see one of their
developers but it hasn't helped so far ;-)
Best
Wolfgang
-------------------------------------------------------------------------
Wolfgang Bangerth email: [EMAIL PROTECTED]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
