Well, they will of course give the same result, as by setting the ith
and jth
degree of freedom to one, you are making a global basis function.
This only works for specific spaces (Lagrange, DG), and not necessarily for Nedelec/RT, as covered in: Assemble with dolfin VS dolfinx - #2 by dokken
Note that the second approach is doing alot more work than the first approach.
It will integrate over the whole domain M**2 times (M=V.dim()), while the first approach does the integration once.
I still do not understand the need for the tabulated values of a basis function here?
You have also not explained what is wrong with the approach in: