Step-32 and step-50 are important for parallelization of TrilinosWrappered
data type.
For someone has to write codes with Trilinos data type, here are necessary
modification
from serial to parallel programs:
1. Change triangulation to parallel::distributed::triangulation rather
than
GridTools::partition_triangulation()
<https://www.dealii.org/current/doxygen/deal.II/namespaceGridTools.html#a99eba8e3b388258eda37a2724579dd1d>
or parallel::shared::Triangulation
<https://www.dealii.org/current/doxygen/deal.II/classparallel_1_1shared_1_1Triangulation.html>
.
2. SparsityPattern and solution/rhs Vector initialization specified for
Trilinos, see Step-42
3. Choose locally owned cells or level cells during assembling.
Though they both allow parallel, step-42's DoFTools::make_sparsity_pattern
<https://www.dealii.org/current/doxygen/deal.II/group__constraints.html#gaf78e864edbfba7e0a7477457bfb96b26>
has
better performance
in matrix LU decomposition (30%-50% less time) than tep-50's.
However another problem pops out. When test a vector-valued problem,
renumber DoFs
component wise make reinit of vectors fail. If
DoFRenumbering::component_wise is called,
1. LinearAlgebra::distributed::Vector<double>
v1(dof_handler.locally_owned_dofs(),
mpi_communicator); v1.reinit(dof_handler.locally_owned_dofs(),
mpi_communicator);
Success
2. LinearAlgebra::distributed::Vector<double> v2;
v2.reinit(dof_handler.locally_owned_dofs(), mpi_communicator);
Fails.
Is that relates to trilinos date type?
在2023年2月23日星期四 UTC+8 22:44:57<Wolfgang Bangerth> 写道：
> On 2/23/23 07:42, 'yy.wayne' via deal.II User Group wrote:
> > So if step-4 doesn't support parallel then what I got is reasonable.
>
> Correct.
> 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].
To view this discussion on the web visit
https://groups.google.com/d/msgid/dealii/da33a84a-5ad3-423a-8377-42eeec16e7f6n%40googlegroups.com.
