[dune-fem] Dune::Fem::Function mappings and expression templates

[Claus-Justus Heine]

Just some other silly stuff.

In an attempt to bump Dune::Fem in the context of continuous FEM (read:
NOT dG) to the state of the ALBERTA demo package I undertook the
adventure of trying to define expression templates for GridFunctions
(i.e. promote the evaluate, jacobians and LocalFunction classes and
their respective methods). I stumbled over the fact that any
Dune::Fem::Function inherits the Mapping interface. That also defines
various algebraic operators. I think that this is somehow unfortunate.

Meaning that mayhaps the Mapping interface should possibly only apply to
really discrete functions, which have coefficient vectors as backing
store. The Mapping-operators probably does not make too much sense for
wrapped continuous GridFunctions (which one needs at some point to
define RHS,  boundary data, and coefficients of operators)

Other thoughts about grid-functions and stuff of mine suggest that I
would prefer a jacobian method which generates another GridFunction
(i.e. not the jacobian at a specific point but the jacobian
__function__) Going off-topic.

Best,

Claus

-- 
Dr. Claus-Justus Heine
Institut für Angewandte Analysis und Simulation
* Numerische Mathematik für Höchstleistungsrechner
Universität Stuttgart

Fon: +49 (0) 711 - 685 65558
Fax: +49 (0) 711 - 685 65507