Pointwise evaluation of a vector

Hi @kinnala ,

Is it possible to do pointwise evalulation of InteriorBasis on a set of input points for ElementHex1 ? I can see an interpolator method, but not sure of what it does. I am trying:

import numpy as np
from skfem import *
mesh  = MeshHex()
elem = ElementHex1()
velem = ElementVectorH1(elem)
basis = InteriorBasis(mesh, velem,  MappingIsoparametric(mesh, elem), intorder=1)
eval = basis.interpolator(np.random.random(basis.N))

but this raises:

Traceback (most recent call last):
  File ".\generateStructuredMesh.py", line 110, in <module>
    eval = basis.interpolator(np.random.random(basis.N))
  File "C:\Users\bshri\AppData\Local\Programs\Python\Python37\lib\site-packages\skfem\assembly\basis\interior_basis.py", line 141, in interpolator
    finder = self.mesh.element_finder()
  File "C:\Users\bshri\AppData\Local\Programs\Python\Python37\lib\site-packages\skfem\mesh\mesh.py", line 342, in element_finder
    raise NotImplementedError("element_finder not implemented "
NotImplementedError: element_finder not implemented for the given Mesh type.

Is this something that you put there for future implementation?

https://github.com/kinnala/scikit-fem/blob/884a427043577233584ea4257dea6f9c2f599d64/skfem/mesh/mesh.py#L339-L343

[kinnala]

There is now a preliminary implementation for tets and 1D meshes available for testing in #487 .

[kinnala]

An additional remark: inverse mapping of MappingIsoparametric will clip the results to the unit cube for robustness so the above approach might require some tuning.