Hi James, > Many thanks the answer. You're welcome.
> although I had hoped that I had simply made a mistake. Unfortunetely not. > One thing confused me a bit in the first part of your answer, though: > You wrote that hanging nodes are supported in the official Deal.II > version for anisotropically refined meshes, when the FE is of degree > one, but not for higher degrees. While I would like the option of > using higher order FE also on anisotropically refined meshes, my > attached example (aspect_test.cc) fails for FE degree one. So should > it not work? Sorry, that was my mistake, it does not work for degree one elements (in 3D). I got confused again because dofs_per_quad is zero for degree one elements in 3d, and a face is a quad, but dofs_per_face includes dofs on vertices and lines and is thus not zero. So you cannot use continuous FEs in 3D with certain anisotropic refinement cases if you need to build hanging node constraints. > Second: If the desired behaviour is simply not implemented in the > official Deal.II, is it possible to ask for a copy of your extended > version? I'd have to check with my employer, but I believe that the answer would be yes. > And is it interface compatible with the official version -- > i.e. would I be able to use it without (too many) changes to my code? No, it is not, you'd probably need several changes. > Alternatively: How much effort would it be to add your hanging nodes > handling to the official version - are the changes very extensive? The changes I've done are quite small, but they were never intended to work with different degrees, i.e. with hp on continuous elements. I'm not sure how much effort it is to clarify those points, I'd have to look at the code again, a thing which I could probably do in June. It might be that everything already works fine, but I'm simply not sure about that. It is probably not difficult to get something working for you, but it is probably much more effort to do it in a way that is accepted to be included in the library. > If, for whichever reasons, you do not wish to spend time adding them, > would I be able to do so myself with reasonable effort? As I said, you would probably and without many changes get your problem running. Including it in the library - who knows? > And (to Wolfgang, Guido and Ralf), would this be desirable from the point of > view of the active Deal.II-developers? I can answer on their behalf that they would very much like to include this functionality and probably Wolfgang would also offer his/their support. Best, Tobias _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
