Thanks Barbel,
That's really useful. I will let you know how things go implementing it
for hp::DoFHandler.
John
John Chapman
www.maths.dur.ac.uk/~tphj28
On 08/09/10 10:07, Bärbel Janssen wrote:
Hi John,
the function is there and can be used, but not for hp::DoFHandler yet.
Maybe you can try to implement it for hp::DoFHandler yourself. Feel
free to ask if you need some help. I would be happy to help.
Bärbel
--
Bärbel Janssen
Institut für Angewandte Mathematik
Universität Heidelberg
Im Neuenheimer Feld 293, Raum 213
Telefon: +49 6221 54-5449
On Wed, 8 Sep 2010, Bärbel Janssen wrote:
Hi John,
yes, Martin and I we finished the make_flux_sparsity_pattern
function. Please give me a sec and I try to check it in. I will let
you know then.
Bärbel
--
Bärbel Janssen
Institut für Angewandte Mathematik
Universität Heidelberg
Im Neuenheimer Feld 293, Raum 213
Telefon: +49 6221 54-5449
On Wed, 8 Sep 2010, John Chapman wrote:
Hi Martin,
You asked two questions, which I'll answer in turn. There was an
exception thrown, namely
An error occurred in line <466> of file
</extra/tmp/dealii_6.3.1/deal.II/lac/include/lac/sparse_matrix.templates.h>
in function
void dealii::SparseMatrix<number>::add(unsigned int, unsigned
int, const unsigned int*, const number2*, bool, bool) [with number2
= double, number = double]
The violated condition was:
counter < cols->row_length(row)
The name and call sequence of the exception was:
ExcInternalError()
Additional Information:
(none)
Secondly I think that your comment about the sparsity pattern is
true. I am creating the sparsity pattern with
constraints.clear ();
DoFTools::make_hanging_node_constraints (dof_handler,constraints);
constraints.close ();
CompressedSetSparsityPattern csp
(dof_handler.n_dofs(),dof_handler.n_dofs());
DoFTools::make_sparsity_pattern (dof_handler, csp, constraints,
false);
sparsity_pattern.copy_from (csp);
system_matrix.reinit (sparsity_pattern);
but of course I need DoFTools::make_flux_sparsity_pattern, but this
doesn't take constraints (as typically a dG method doesn't have
any). However my application will be introducing some constraints
so changing this section would only be a temporary solution.
I know Barbel was also looking at a
make_flux_sparsity_pattern(...,constraints,...) function, so perhaps
she can shed some light on the problem.
John
On 08/09/10 09:31, Martin Kronbichler wrote:
Dear John,
Assuming that I want to continue to use
constraints.distribute_local_to_global, how do I treat the assembly
terms ui_ve_matrix, ue_vi_matrix i.e. those that contain integrals
involving cell and neighbor? The two cells may not have the same
number
of degrees of freedom, so the matrices may not be square, so I
tried to use
void ConstraintMatrix::distribute_local_to_global ( const
FullMatrix< double> & local_matrix,
const std::vector< unsigned int> & row_indices,
const std::vector< unsigned int> & col_indices,
MatrixType& global_matrix
) const
but the local_dof_indices and neighbor_dof_indices are not the
correct
variables to pass - how do I get row_indices and col_indices from the
dof_indices?
What do you mean by "not the correct variables to pass"? To me your
assembly routines seems correct. Your matrix ui_ve_matrix has
local_dof_indices.size() rows and neighbor_dof_indices.size() columns,
which is exactly the dimension the distribute_local_to_global function
expects. So I can't really see any flaw with this problem. Is some
kind
of exception thrown (there might still be a bug in that
distribute_l_t_g
function, we haven't been using it so much yet)? How is the sparsity
pattern of your matrix obtained - did you use
make_flux_sparsity_pattern?
Best,
Martin
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
