Re: [deal.II] Compute the face embedding matrix for FE_NedelecSZ

2021-09-22 Thread Wolfgang Bangerth
On 9/22/21 2:37 AM, Sebastian Kinnewig wrote: When I implement the computation of the embedding matrix , as done in FE_Nedelec, in FE_NedelecSZ the program compiles and runs without any warnings or errors (even in debug mode). But from the results it appears that there still is a problem with

Re: [deal.II] Compute the face embedding matrix for FE_NedelecSZ

2021-09-22 Thread Alexander
Hi Sebastian FE_NedelecSZ does not support hanging nodes, i.e. the mesh must be conforming. Implementing handing-node constraints for this element is possible, but would require a special approach since constraints are not static. If you feel like implementing this, I would be glad to

Re: [deal.II] Compute the face embedding matrix for FE_NedelecSZ

2021-09-21 Thread Wolfgang Bangerth
On 9/21/21 8:35 AM, Sebastian Kinnewig wrote: To make the switch to FE_NedelecSZ I need to compute the face embedding. I tried to compute the face embedding via the function FETools::compute_face_embedding_matricies() and pass the resulting face embedding matrix to the interface_constraints

[deal.II] Compute the face embedding matrix for FE_NedelecSZ

2021-09-21 Thread Sebastian Kinnewig
Hello all, I am currently working on an adaptive solver for the time-harmonic Maxwell equation. At the moment I am using the Nédélec elements provided by the class FE_Nedelec. Since FE_Nedelec requires a consistent orientation of the faces, I would like to switch to FE_NedelecSZ (therefore I