[Dune-devel] Issues in dune-localfunctions

[Christian Engwer]

Dear all,

I updated the tests in dune-localfe.hh to check the state of
diffOrder. Currently the test fails for _all_ implementations, at
least if I check for the semantics as described in interface.hh

Christian

On Wed, Mar 26, 2014 at 05:03:01PM +0100, Christian Engwer wrote:
> Dear all,
> 
> I'm currently working on the new functions interface in pdelab and
> encountered some issues with dune-localfunctions.
> 
> A) what exactly is the definition of the diffOrder in the
>    basis-traits?
>    Currently it defaults to 0 and most implementaitons will never
>    have specified the diffOrder
> B) the C1LocalBasisTraits set diffOrder to 1, but we currently have
>    no (globally) C1 functions in dune-localfunctions, so what does the
>    C0,C1,Ck here refer to?
> C) it is a bit surprising (for the user) that in case of diffOrder=1 I
>    can use evaluateJacobian(...), but not evaluate(...) with the
>    appropriate direction information.
> D) diffOrder sounds to me like something mathematically implied and
>    locally most local functions are C-infinity. So in these cases the
>    appropriate methematical information would be more something like
>    differentiable, but non-trivial. On the other hand we usually use
>    this entry to tell how man derivatives are _implemented_, but in
>    this case the name is imho missleading.
> 
> Besides these point there are a few further topics we might want to
> discuss seperately.
> 
> I hope we can sort out some of these issues.
> 
> Ciao
> Christian
> 
> _______________________________________________
> Dune-devel mailing list
> Dune-devel at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune-devel
> 

-- 
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