Mpi for heat eqn #215
Mpi for heat eqn #215
Conversation
There was a problem hiding this comment.
@WasimNiyazMunshi Here are some comments. I think the rest looks good. If you know step-26 and step-40, it should not be difficult to understand the program.
|
@WasimNiyazMunshi -- ping |
|
@WasimNiyazMunshi Can you squash the commits? This way, we don't have one commit that adds the executable and another that removes it again. That'd be a waste of disk space for everyone who clones the repository. |
WasimNiyazMunshi
force-pushed
the
MPI_for_Heat_Eqn
branch
3 times, most recently
from
69c3ce7 to
3259e03
Compare
WasimNiyazMunshi
force-pushed
the
MPI_for_Heat_Eqn
branch
from
3259e03 to
359ef19
Compare
| @@ -0,0 +1 @@ | |||
| DEAL_II_WITH_TRILINOS DEAL_II_WITH_MPI | |||
There was a problem hiding this comment.
This does not match the requirements stated in CMakeLists.txt.
| DynamicSparsityPattern dsp(locally_relevant_dofs); | ||
| DoFTools::make_sparsity_pattern(dof_handler, dsp, constraints, true); | ||
| SparsityTools::distribute_sparsity_pattern( | ||
| dsp, dof_handler.locally_owned_dofs(), mpi_communicator, |
There was a problem hiding this comment.
| dsp, dof_handler.locally_owned_dofs(), mpi_communicator, | |
| dsp, locally_owned_dofs, mpi_communicator, |
| system_rhs.reinit(locally_owned_dofs, mpi_communicator); | ||
| DynamicSparsityPattern dsp(locally_relevant_dofs); | ||
| DoFTools::make_sparsity_pattern(dof_handler, dsp, constraints, true); | ||
| SparsityTools::distribute_sparsity_pattern( | ||
| dsp, dof_handler.locally_owned_dofs(), mpi_communicator, | ||
| locally_relevant_dofs); | ||
| system_matrix.reinit(locally_owned_dofs, locally_owned_dofs, dsp, | ||
| mpi_communicator); |
There was a problem hiding this comment.
| system_rhs.reinit(locally_owned_dofs, mpi_communicator); | |
| DynamicSparsityPattern dsp(locally_relevant_dofs); | |
| DoFTools::make_sparsity_pattern(dof_handler, dsp, constraints, true); | |
| SparsityTools::distribute_sparsity_pattern( | |
| dsp, dof_handler.locally_owned_dofs(), mpi_communicator, | |
| locally_relevant_dofs); | |
| system_matrix.reinit(locally_owned_dofs, locally_owned_dofs, dsp, | |
| mpi_communicator); |
Is there a reason why you need to reinitialize your data structures here?
You already call setup_system at the appropriate places, which does exactly the preparation of data structures, so I don't think you will need to reinit most structures again here.
If needed, I would move the setup_system call after the goto hook.
Mpi solver for heat eqn