On 05/09/2018 03:38 AM, [email protected] wrote:
I am using Dealii with the Ifpack_Hypre package in Trilinos. Everything works
well except that I need an Epetra_FEVector vector for which
vector.Map().LinearMap() evaluates to true. However, it appears I am getting a
discontinuous mapping, but I don't understand why based on what I see in the
deal.ii code.
This all relates to the Vector class within the EpetraWrappers namespace. I am
using the reinit function that takes an IndexSet parameter to build
the Epetra_Map. The Vector::reinit function in trilinos_epetra_vector
calls parallel_partitioner.make_trilinos_map(communicator,false). The method
make_trilinos_map in index_set.cc checks if
parallel_partitioner.is_ascending_and_one_to_one(communicator) is true. If
that is true, it appears that a linear map should be created. I checked in my
code whether the IndexSet I pass to reinit is_ascending_and_one_to_one for the
communicator and that returns true.
The only possible complication is that when checking whether there is a linear
map, I am using vector.trilinos_vector().Map().LinearMap(). So perhaps a
linear map is created but then going back and forth between the deal.ii vector
and the underlying Epetra_FEVector causes some issue.
Does anyone familiar with this part of deal.ii or with Epetra vectors have any
ideas that may give me some direction?
Josh -- it's hard to tell without a minimal testcase. Can you try to boil down
your program to the smallest possible thing that shows the error? It doesn't
need to do anything useful other than show the error -- i.e., the matrix can
be all zero, the rhs can be all zero, etc.
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email: [email protected]
www: http://www.math.colostate.edu/~bangerth/
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
