Hi all,
I am not a finite element specialist and just started using Fenics. Is it possible to implement a PDE solver such that taking the gradient of the solution gives me back the exact Neumann boundary data (after multiplying with the normal vector)? At least in the case where I assume that the order of my data corresponds to the order of my elements in some proper sense? I hoped that it would work with order one less than the element order for example. I was playing around a bit and it seemed like that in comparison to Dirichlet conditions this is not straight forward. It is probably not important for an answer to this question, but I was using modifications of ft01_poisson.py, one of the scripts in the tutorial book. (Chapter 2.2 and modifications in 4.1.3)
Thanks,
Sebastian
