[dune-pdelab] ImplementationsTag for different grid functions

[Christian Engwer]

OK,

I realized, that the transformations don't actually look at the
NodeTag and thus the ImplementationTag has to contain both
information.

Besides this, I think it would be convient to have two additional
features (I'm up for discussion about how to implement it).

A) when implementing the interpolate for the new dune-functions like
   functions interface, I relized, that you might want to directly use
   a function (without wrapping it) as a leaf node. As you don't wrap
   it, it will usually not contain a LeafTag. Perhaps the NodeTag
   should be deduced by a typetrait and usually gets the information
   from the typedef, but can be specialized (like anything that
   implements the functions interface and doesn't contain a NodeTag
   typedef is implicitly a LeafNode!?)

B) the functions don't necessarily contain an ImplementationTag. If I
   would be able to choose between an implicit ImplementationTag,
   deduced by the transformation, and an explicit ImplementationTag, I
   could specify the ImplementationTag in cases where the class
   doesn't contain one, or where I know better for some reason. This
   second idea however would require us to change the lookup strategy
   such, that ImplementationTag really only describes the concept of
   the input tree, whereas the type of tree node is described by the
   NodeTag.

Christian
   

On Tue, Mar 31, 2015 at 11:26:43PM +0200, Christian Engwer wrote:
> Hi Steffen,
> 
> why do we use different ImplementationTags for composite-, power- and
> "plain" gridfunctions?
> 
> As fas as I understood the tags, the ImplementationTag describes a
> concept of functionality, while the NodeTag specifies the type of node
> in the tree. Thus I'd expect that we only need GridFunctionTag for all
> node type, as they differ in the definition of NodeTag?! Did I miss
> something? At least I'd expect that CompositeGridFunctionTag and
> PowerGridFunctionTag inherit GridFunctionTag.
> 
> Ciao
> Christian
> 
> _______________________________________________
> dune-pdelab mailing list
> dune-pdelab at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-pdelab
> 

-- 
Prof. Dr. Christian Engwer 
Institut für Numerische und Angewandte Mathematik
Fachbereich Mathematik und Informatik der Universität Münster
Einsteinstrasse 62
48149 Münster

E-Mail	christian.engwer at uni-muenster.de
Telefon	+49 251 83-35067
FAX		+49 251 83-32729