Added hp::Refinement::copy_future_fe_indices(). #14173
Conversation
marcfehling
force-pushed
the
copy_future
branch
from
3d02dc1
to
2c908f9
Compare
|
Found a typo. Now fixed. |
|
|
||
| Assert(cell_destination->is_locally_owned(), ExcInternalError()); |
There was a problem hiding this comment.
| Assert(cell_destination->is_locally_owned(), ExcInternalError()); |
There was a problem hiding this comment.
This assertion is also in set_future_fe_index, so we really do not need it.
| ExcMessage( | ||
| "Both DoFHandler objects must work on the same Triangulation!")); | ||
|
|
||
| typename DoFHandler<dim, spacedim>::active_cell_iterator cell_destination; |
There was a problem hiding this comment.
| typename DoFHandler<dim, spacedim>::active_cell_iterator cell_destination; |
| if (cell_source->is_locally_owned() && | ||
| cell_source->future_fe_index_set()) | ||
| { | ||
| cell_destination = |
There was a problem hiding this comment.
| cell_destination = | |
| auto cell_destination = |
| Assert(dof_handler_source.get_triangulation().n_active_cells() == | ||
| dof_handler_destination.get_triangulation().n_active_cells(), | ||
| ExcMessage( | ||
| "Both DoFHandler objects must work on the same Triangulation!")); |
There was a problem hiding this comment.
| Assert(dof_handler_source.get_triangulation().n_active_cells() == | |
| dof_handler_destination.get_triangulation().n_active_cells(), | |
| ExcMessage( | |
| "Both DoFHandler objects must work on the same Triangulation!")); | |
| Assert(&dof_handler_source.get_triangulation() == | |
| &dof_handler_destination.get_triangulation(), | |
| ExcMessage( | |
| "Both DoFHandler objects must work on the same Triangulation!")); |
There was a problem hiding this comment.
Ah yes, comparing addresses is smarter :)
There was a problem hiding this comment.
One question: don't we need something like this also for actve_fe_index? What do you the first time?
The workflow is usually that you only set active FE indices once in the beginning and manually. So if you have multiple DoFHandlers, you simply set the same indices on both. An additional function wouldn't be too useful I think.
The idea behind this function is that after you set future FE indices on one DoFHandler with one of the hp::Refinement functions, you can easily transfer the result to other DoFHandlers. This only makes sense for future FE indices in my opinion.
As an alternative, we could write {get|set}_future_fe_indices functions similar to DoFHandler::{get|set}_active_fe_indices. Then the user has to juggle with a container with all future FE indices. This might be a more versatile approach, and would form a pendant to the active FE indices functions.
I would be happy to transform this PR accordingly if you agree!
| Assert(dof_handler_source.get_triangulation().n_active_cells() == | ||
| dof_handler_destination.get_triangulation().n_active_cells(), | ||
| ExcMessage( | ||
| "Both DoFHandler objects must work on the same Triangulation!")); |
There was a problem hiding this comment.
Ah yes, comparing addresses is smarter :)
|
|
||
| Assert(cell_destination->is_locally_owned(), ExcInternalError()); |
There was a problem hiding this comment.
This assertion is also in set_future_fe_index, so we really do not need it.
marcfehling
force-pushed
the
copy_future
branch
from
2c908f9
to
6725f0e
Compare
Since there is already Personally, I like this suggestion 👍 |
|
Superseded by #14208. |
Functions that copies future FE indices from one DoFHandler to another. Useful when dealing with multiple DoFHandler objects for the same problem, for example, if you have one for Stokes and one for the temperature as in step-32.