Lucas:
Cutting to the chase: my problem involves using traceless, symmetric 3x3
tensor fields, and I represent these as 5-component vector finite element
fields, where each vector-component is a degree of freedom of the tensor. My
domain is a cylinder, and I'm trying to set up periodic boundary conditions on
the ends, with one end rotated pi/2 with respect to the other. However, the
`make_periodicity_constraints` function assumes an actual vector finite
element field (in the sense of having the same dimension as space-dim) when
applying the rotation matrix, and so I am unable to use periodic constraints
in a straightforward way for my application.
Yes, tough case :-(
Two questions on this: 1) how difficult would it be to accommodate this
particular issue from outside the library? I've been looking through the code
and seeing "this is a hack" and "we have to be careful" so I'm a bit wary to
try to re-implement `make_periodicity_constraints` on my own. 2) Is it
possible (in the future) to be able to apply arbitrary linear constraints
between periodically-joined fields?
It's definitely not an easy case, in particular because the way you discretize
your fields is just so non-standard (but not unreasonable). I think the way
one would generalize the existing function would be if the library identifies
matching sets of degrees of freedom (say from corresponding vertices on
periodic faces) and then instead of setting them equal as it does now call
back a user-defined function that then does what needs to be done. The
existing function already does the first part; it would then take an argument
of the kind
std::function<void (const std::vector<types::global_dof_index> &dof_indices_1,
const std::vector<types::global_dof_index> &dof_indices_2,
AffineConstraints<double> &constraints)
which for now is simply a loop adding constraints
U[dof_indices_1[i]]=U[dof_indices_2[i]] but which could do more complicated
things like in your case.
(The current function can actually do more complex things such as rotate
vectors as well, but you get the idea of the user-provided function object. I
think you should be able to generalize the existing function in this direction.)
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email: [email protected]
www: http://www.math.colostate.edu/~bangerth/
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/dealii/fbc85bd3-8020-4d30-9339-09014c109407%40colostate.edu.
