Wolfgang,
> > One dof_handler that I reinitialize. Is that likely to complcate things?
> > Do you think I should be using two dof_handlers?
>
> You will have to. The DoFHandler object stores which degree of freedom lives
> where, and so makes the association between the elements of your solution
> vector and which shape function/node they belong to. If you re-initialize
> your DoFHandler object with another finite element, you forget this
> connection, and in effect you end up with a solution vector from the first
> step for which you don't know any more what vector elements correspond to the
> shape functions on your current cell.
Now this suddenly makes a whole hyper-sphere of sense!
...and that is exactly what dealii is ccorrectly complaining about. I've
lost the relation between cell and solution when I reinialized the
dof_handler - I did not realise this happened at that point, probably I
should dig a little deeper. I get it now due to your extra explanation.
Many thanks. I'll let you know how things turn out with this, we are
getting some pretty darn good results already...
Quick follow-up question:
If I really need to separate these two problems in that way (yes, I do!),
then I should probably be writing two separate classes:
(1) ElasticProblem
(2) ElectrostaticProblem
and let these two clases share only the mesh and let
ElectrostaticProblem have read-only access to the solution vector from
ElasticProblem. Does that sound like a good start?
Best,
Toby
-----
Toby D. Young
Assistant Professor
Philosophy-Physics
Polish Academy of Sciences
Warszawa, Polska
www: http://www.ippt.gov.pl/~tyoung
skype: stenografia
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
