PETSc CFD advice

Matthew Knepley knepley at
Wed Oct 18 07:26:31 CDT 2006

On 10/18/06, Nils Erik Svangård <nilserik at> wrote:
> Hello!
> Last year I did my master thesis which consisted of converting and old
> fortran based CFD application using block structured explicit solver
> to use the PETSc framework.
> Now my mentor/supervisor has contacted me again, they are thinking
> about doing a unstructured solver and are looking at PETSc.
> I said that I think using PETSc is a good idea, and if they are going
> to write new code I think they should look at using C++ instead of
> Fortran (because I think C++ is better suported by PETSc).
> Was my advice any good? Any other advices they should look at before
> beginning?

While we do fully support Fortran, we have new more advanced features in
C++, so
I think yor advice was good as long as they can be comfortable in C++ :)
There is new
support for unstructured meshes in PETSc. It is still in alpha release, so
it is not in the
same category as the other PETSc structures, but I would encourage them to
look at
it. I recently gave a tutorial at Columbia (which I will post on the web
tonight), which
demonstrates some of this.



best regards
> /nisse
> --
> Nils-Erik Svangård
> E-Mail: nilserik at
> MSN: nilserik at
> Skype: schweingaard
> Mobil: +46-(0)70-3612178

"Failure has a thousand explanations. Success doesn't need one" -- Sir Alec
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the petsc-users mailing list