Hello All,
I am trying to interpolate a finite element solution to use as dirichlet
boundary conditions for an auxiliary problem. I have the following working
for serial runs:
Functions::FEFieldFunction<dim,DoFHandler<dim>, TrilinosWrappers::MPI::
BlockVector> boundaryconditions(dof_handler, curl_vel);
VectorTools::interpolate_boundary_values(dof_handler,
wall,
boundaryconditions,
vorticity_constraints,
fe.component_mask(
vorticity_extractor));
In context, I project the curl of velocity onto a finite element space and
I want to use this as boundary conditions for vorticity.
The problem is when I want to do this in parallel, FEFieldFunction is not
able to handle it directly, as noted in the documentation.
The documentation does give a template on how to use FEFieldFunction with a
distributed vector, see
https://www.dealii.org/9.0.0/doxygen/deal.II/classFunctions_1_1FEFieldFunction.html
What I though might work (using the template given) was something like the
following
class DirichletConditions : public Functions::FEFieldFunction<dim,
DoFHandler<dim>, TrilinosWrappers::MPI::BlockVector>{
public:
DirichletConditions(const DoFHandler<dim>& dofh, const LA::MPI::
BlockVector &src, int n_components):
Functions::FEFieldFunction<dim, DoFHandler<dim>,
TrilinosWrappers::MPI::BlockVector>(dofh, src), n_comp(n_components){};
void vector_value (const Point< dim > &p, Vector< double > &values)
const {
Vector<double> vec_val(n_comp);
bool point_found = true;
try
{
for(int comp = 0; comp < n_comp; ++comp)
vec_val[comp] = this->value(p, comp);
}
catch (...)
{
point_found = false;
}
if (point_found)
values = vec_val;
else
values = 0.0;
}
private:
const int n_comp;
};
DirichletConditions BC(dof_handler, curl_vel, n_components);
VectorTools::interpolate_boundary_values(dof_handler,
wall,
boundaryconditions,
vorticity_constraints,
fe.component_mask(
vorticity_extractor));
This works in serial, but when I switch to MPI I get the following expected
error
An error occurred in line <488> of file
> </home/kyle/lib/dealii-8.5.0/include/deal.II/numerics/fe_field_function.templates.h>
>
> in function
> unsigned int dealii::Functions::FEFieldFunction<dim, DoFHandlerType,
> VectorType>::compute_point_locations(const std::vector<dealii::Point<dim>
> >&, std::vector<typename DoFHandlerType::active_cell_iterator>&,
> std::vector<std::vector<dealii::Point<dim> > >&,
> std::vector<std::vector<unsigned int> >&) const [with int dim = 2;
> DoFHandlerType = dealii::DoFHandler<2>; VectorType =
> dealii::TrilinosWrappers::MPI::BlockVector; typename
> DoFHandlerType::active_cell_iterator =
> dealii::TriaActiveIterator<dealii::DoFCellAccessor<dealii::DoFHandler<2>,
> false> >]
> The violated condition was:
> !my_pair.first->is_artificial()
> Additional information:
> The given point is inside a cell of a
> parallel::distributed::Triangulation that is not locally owned by this
> processor.
>
I'm really not sure if this is the right way to go about what I'm trying to
do.
-Thanks,
Kyle Williams
P.S. I don't think it really matters in this case since this has not been
updated with 9.0.0, but I am using version 8.5.0 with Trilinos and p4est
--
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 dealii+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
