[petsc-users] calling PETSc from C++

[Jed Brown]

On Mon, Mar 12, 2012 at 15:37, Francis Poulin <fpoulin at uwaterloo.ca> wrote:

> I am trying to call Petsc from a C++ program and having difficulties.  I'm
> using v3.2.6 and I can run all the examples so I assumed that everything
> was installed ok.  The body of the function is very simple, see below.  I
> get a segmentation fault.   In my installation I used mpich
>
> I have a colleague who also installed openmp and this works for him on the
> same version of Petsc.  Could it be that Petsc is confused because I have
> two different MPI's installed?  I am hoping that it will use the one that
> it configured but I do have the other MPI in the same path.
>

Yes, the most likely scenario is that your application was linked with a
different MPI than PETSc was compiled with. It is probably a makefile that
uses the wrong MPI. Can you build PETSc examples?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20120312/5b9c48e3/attachment.htm>