hello, I find one kind of treatment in
dealii/tests/matrix_free/stokes_computation.cc(line 884-919):
use FEEvaluation<...>::begin_dof_values() to loop over the dofs_per_cell
rather than dofs_per_component
FEEvaluation velocity(data, 0);
for (unsigned int cell = cell_range.first; cell <
Pai,
can you come up with a small, self-contained testcase? This indeed looks like
a bug somewhere.
I looked at the code you are calling, but it seems correct to me -- it may be
that the error is in Trilinos, but it's hard to tell without a code we can
play with.
Best
W.
On 10/28/19
Mauro,
apologies for not getting back to this earlier. Have you found a solution?
I think what you find is something that we had to learn over time and have
learned since step-35 was written: That it's typically easiest to just put all
solution variables into one large finite element system,
Dear Luca,
Thank you very much for your help. I tried to use your recommendation, but
it does not seem to work correctly. My code is:
CylindricalManifold<2> cylman(0);
triangulation.set_manifold(numbers::flat_manifold_id, cylman);
QGauss<2> quadrature_formula(fe.degree + 1);