Thanks to all. I found my error. I didn't take into account that the
contribution of the faces of neighbouring cells cancel. So I was a bit confused
why in step-7 only the boundaries of the domains are considered.
If I get this right and I want to apply dirichlet boundaries on two boundaries
of the domain, I can discard the boundary term at these faces, right? And on
the other ones I will do like in step-7 or the code of Martin.
Best regards,
Klaus
> Hi Klaus,
>
> you can do assembling as usual for cells, just using FEFaceFalues, (snippet
> from my code):
>
> FEValues<dim> fe_values (dof_handler.get_fe(), quadrature,
> //setup FE values
> UpdateFlags(update_values |
> update_gradients |
> update_q_points |
> update_JxW_values));
>
> FEFaceValues<dim> fe_face_values (dof_handler.get_fe(), face_quadrature,
> //and FE face values
> UpdateFlags(update_values |
> update_gradients |
> update_q_points |
> update_normal_vectors |
> update_JxW_values));
>
> typename DoFHandler<dim>::active_cell_iterator
> cell = dof_handler.begin_active(),
> endc = dof_handler.end();
> for (; cell!=endc; ++cell)
> if ((cell->subdomain_id() == subdomain_id) || (subdomain_id ==
> types::invalid_subdomain_id))
> {
> cell_rhs = 0;
>
> fe_values.reinit (cell);
>
> rhs_functions->body_value_list (fe_values.get_quadrature_points(),
> //get the body force value
> body_rhs_values);
>
> for (unsigned int q_point=0; q_point<n_q_points; ++q_point)
> for (unsigned int i=0; i<dofs_per_cell; ++i)
> {
> const unsigned int component_i =
> dof_handler.get_fe().system_to_component_index(i).first;
>
> cell_rhs(i) += (body_rhs_values[q_point](component_i) *
> //and integrate, this is (f,v)_cell
> fe_values.shape_value(i,q_point) *
> fe_values.JxW(q_point));
> };
>
> for (unsigned int face=0; face<GeometryInfo<dim>::faces_per_cell;
> ++face) //loop over cell faces
> if (cell->face(face)->at_boundary())
> //if we are on the neumann boundary
> if (cell->face(face)->boundary_indicator() !=
> BoundaryIndicators::dirichlet_boundary)
> {
> fe_face_values.reinit (cell,face);
>
> rhs_functions->boundary_value_list
> (fe_face_values.get_quadrature_points(), //get the boundary values
>
> fe_face_values.get_normal_vectors(),
>
> boundary_rhs_values);
>
> for (unsigned int q_point=0; q_point<n_face_q_points;
> ++q_point)
> for (unsigned int i=0; i<dofs_per_cell; ++i)
> {
> const unsigned int component_i =
> dof_handler.get_fe().system_to_component_index(i).first;
>
> cell_rhs(i) +=
> boundary_rhs_values[q_point](component_i) * //and integrate, this is
> (g,v)_face
> fe_face_values.shape_value(i, q_point) *
> fe_face_values.JxW(q_point);
> };
> };
>
> cell->get_dof_indices (local_dof_indices);
> constraints.distribute_local_to_global (cell_rhs,
> local_dof_indices,
> rhs);
> };
>
>
> May this helps,
>
> Best,
> Martin
>
------------------------------------------------------------
> Von: Klaus Fischer <[email protected]>
> An: [email protected]
> Gesendet: Dienstag, den 5. April 2011, 14:24:10 Uhr
> Betreff: Re: [deal.II] Handle boundary terms
>
> Hi,
>
> I think we're talking about different things here. I'm totally aware of the
> boundary conditions of my system and found examples how to implement these
> constraints using deal.II. What I meant was looking for example at the
> "vector-valued > problems" module description, one has the final form (v,u) -
> (div v,p) - (q, div u) = (q, f) + (nv, p)_\Omega.
>
> In the implementation however the last term in the above form, the boundary
> term, is neglected. I'm now wondering how would this look like if I don't
> neglect it. Somehow I then have to be able to evaluate the shape functions
> and their gradients on the surface of each cell. How can this be done?
>
> Best regards,
>
> Klaus
> ___________________________________________________________
> Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
> Toolbar eingebaut! http://produkte.web.de/go/toolbar
> _______________________________________________
> dealii mailing list
> http://poisson.dealii.org/mailman/listinfo/dealii[http://poisson.dealii.org/mailman/listinfo/dealii]>>
___________________________________________________________
Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
Toolbar eingebaut! http://produkte.web.de/go/toolbar
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
