[Dune] GhostEntity

Andreas Dedner dedner at mathematik.uni-freiburg.de
Thu Jan 11 12:03:17 CET 2007


>> This definition is already rather contradictory. It says
>>* the intersection iterator only stops at the border of the domain.
>>  If we have an inner processor, it has no border edjes at all, 

Why does it not have border edges? It must have border edges (not
boundary edges) on all sides if it is an inner processor - or have
I understood something wrong?


>>  that would mean the ghostelement intersection iterator would not stop
>>  all?!
>>* the example says it stops only on the intersection with entities in
>>  the interior region, but these intersection are per definition not
>>  border intersections (if we assume that we don't have inner
>>  boundaries, which applies to all existing implementations)
>>
>> 
Ok, I see the problem: it should read:
the IIter stops only at intersections with interior or overlap codim
     zero entities.
Alternativly there is something defined as gamma^f in the paper so that
we stop at intersections on gamma^f (what does f stand for?)

>>
>>Does this apply for all grids?
>>* Are all subentites (and especially their ids) accessible?
> 
I though that this is the case since we need the data also on all
verticies for example.
> 
>>* Are all geometries accessible?
> 
I think that this should be the case for subentities - I have not
thought about father/children entities...

Andreas






More information about the Dune mailing list