On 09/05/2017 07:29 AM, Dante De Santis wrote:
This is working for me, except that the solution on the boundary nodes
of each partition is always zero while
for the interior nodes is OK.
I am using a parallel distributed triangulation and PETSc vector and
here is the extracted piece of the code:
/*
Save the solution
where sol_u is a PETScWrappers::MPI::Vector
*/
parallel::distributed::SolutionTransfer<dim,
PETScWrappers::MPI::Vector>system_trans (dof_handler);
system_trans.prepare_serialization (sol_u);
triangulation.save ( filename.c_str() );
//Reload the solution
parallel::distributed::SolutionTransfer<dim,
PETScWrappers::MPI::Vector>system_trans (dof_handler);
system_trans.deserialize(sol_u);
I know that there must be some issues with distributed vector/dofs but I
cannot figure it out.
I can't right away see what the issue may be, but to help debug this:
* Change your code in such a way that just before serialization, you set
all entries of the solution vector to one. This way, you know exactly
what you have. What you get back should match this exactly.
* Check the following: Is sol_u a vector that has ghost elements before
serialization? Is sol_u a vector that has ghost elements after
serialization? Because if these vectors don't have ghost elements, then
what you see visually may not be what you actually have.
As always, you should run your program in debug mode to make sure as
many bugs are found automatically as possible.
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.
