Hi Pearl,

> I am not using parallel::distributed::Triangulation. I used to create
> PetscWrappers::MPI::Vectors with ghost elements. by using the
> functions to extract the indices for locally_owned_dofs and
> locally_relevant_dofs.

Are you sure? Before the distributed::Triangulation existed, you would use
 DoFRenumbering::subdomain_wise (dof_handler);
 DoFTools::count_dofs_with_subdomain_association()
etc, like in step-17 or not?

> However, these functions seemed to have changed
> drastically.

When did that change happen?

> Now they just return all the dofs for the dof_handler
> when you are not using p4est.

No, they always worked like this. I think we could generate these
IndexSets for a non-distributed::Triangulation by looking at the
subdomain. Would that be a good idea? (I never liked the fact, that
the ownership was only implicitly defined and returned by the
DoFTools::count_dofs_with_subdomain_association() which is rather hard
to find)

-- 
Timo Heister
http://www.math.tamu.edu/~heister/
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii

Reply via email to