Thank you, Daniel. That confirms my intuition about that. The push forward from the ref cell to physical cells is, I found after thinking about it again, for RaviartThomas, BDM, Nedelec elements etc not the same as a Piola transform (which eliminates scaling but not rotation). My code works now without having to rescale.
Thanks again :-) Konrad On Monday, March 18, 2019 at 11:17:40 AM UTC+1, Daniel Arndt wrote: > > Konrad, > > I implemented a class derived from the Function<dim> class that evaluates > a scalar or vector shape function at a given (set of) point(s) in a > physical cell. If I check the output graphically I see that the vector > shape functions on the physical cell, for example for lowest order > Raviart-Thomas elements has a magnitude that scales with the inverse edge > length. But on any physical cell the magnitude should be one I believe. > > For the Raviart-Thomas the degrees of freedom for the lowest order > element should be the face integrals. Hence, I would expect them to scale > with the (inverse of the) size of the face. > > > I assume that has something to do with the mapping used by the FEValues > class to map to the physical cell, right? Does the FEValues class initiated > with any vector element use the Piola transform? If not how can I influence > this? > > Have a look at > https://www.dealii.org/current/doxygen/deal.II/group__mapping.html. The > transformation type used in Mapping::transform() is normally set by the > given FiniteElement class. > > Best, > Daniel > -- 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 dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.