I think my question is slightly different to Thomas's. I have the same
fields on either side of the interface but they are discontinuous over
the interface but continuous on either side. So I would like the
interface to be DG and away from that I would use CG.
I think you can use the same trick, which allows you to introduce double
dofs on an internal interface. What you do with those dofs is up to you
and your problem.
If you don't add boundary conditions (BC) on them (both dofs on the
interface) you have a discontinuity with a "do nothing" condition.
I give the example of Laplace on both sides:
without BC you have a discontinuity if the "diffusion" coefficients are
different in the two domains, and on the interface on both sides you have
a normal derivative equal to zero (but of course a gap between them);
you have the the same if you impose nonhomogeneous Neumann BC, but the
normal derivatives on the interface is imposed by your fluxes;
you can use Robin BC and impose in this way a discontinuity, that depends
on the fluxes;
or you can impose Dirichlet BC on one or both sides, fixing the
discontinuity in the latter case.
Best
Thomas
@Wolfgang : yesterday I've sent a mail explaining where occurs the error
with FE_Nothing, but I think we had problems with the mail server and was
delivered late.
_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
