dolfin team mailing list archive

[Kristian Oelgaard <K.Oelgaard@xxxxxxxxxxxxxxx>]

Hi,
I have some difficulties with the subfunction facility in Dolfin. I'm interested
in saving the normal and shear components of a stress function (in 3D) for post
processing purposes.

the stress function is initialised in this way:

	stress.init(mesh, a_strain.trial());

and the trial element 'a_strain.trial()' is defined as:

	elementB = FiniteElement("Discontinuous vector Lagrange", "tetrahedron", 0, 3)
	elementC = FiniteElement("Discontinuous vector Lagrange", "tetrahedron", 0, 3)

	strain_element = elementB + elementC

	(wn,ws) = TrialFunctions(strain_element)

Now, when doing:

	Function n_stress = stress[0];
	Function s_stress = stress[1];

 	File file_normal_stress("normal_stress.pvd");
 	File file_shear_stress("shear_stress.pvd");

	while( t < T)
	{

    nonlinear_solver.solve(nonlinear_problem, x);

		n_stress = stress[0];
		s_stress = stress[1];

	  file_normal_stress << n_stress;
	  file_shear_stress  << s_stress;

  }

the values of the two files are exactly the same as if only the first half of
stress.vector() is used as a basis for both subfunctions.

Funny thing is that a similar approach works fine in the elasticity demo. Only
difference is that in this case the strains are computed in FFC, where my
stresses are computed in the non-linear problem, on element level, and then
written into the stress.vector(). So, what am I doing wrong?

Regards,

Kristian