Kumar,
for triangles and tetrahedra, mappings are generally implemented by
relying in MappingFE. As a consequence, if there is a Lagrange element
of polynomial degree p, there is a corresponding mapping of polynomial
degree p. Right now, FE_SimplexP is available up to polynomial degree
p=2, so quadratic elements should be ok.
Whether or not GridIn can read such meshes is a different question. If
you get an error, then the answer is apparently no. But that shouldn't
be too hard to fix, and we would be quite happy to accept a patch (and
point you in the right direction regarding what needs to be
implemented). My suspicion is that it would not take more then 100 lines
of code to read these meshes.
The bigger problem with quadratic meshes (which also exists for the case
of quadratic quadrilaterals/hexahedra) is that we do not currently have
a good way to translate the information encoded by mid-edge nodes into a
mapping. If we can read such meshes for quadrilaterals, I suspect that
we just ignore the mid-edge nodes and rely on the user's ability to
provide a mapping that restores this kind of information.
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email:
bang...@colostate.edu
www:
http://www.math.colostate.edu/~bangerth/