[Dune] centerUnitOuterNormal() in UGGrid

[Martin Nolte]

Hi Oli,

should we add this to the FAQ list?

Best,

Martin

On 09/18/2013 11:30 AM, Oliver Sander wrote:
> Hi Eike,
> no worries.  In fact, I should have thought of an orientation problem much
> earlier.
> Also, one may argue that UGGrid should give an error message when an element is
> oriented wrongly, rather than tacitly producing wrong results.  Maybe that's easy
> to do.
> Best,
> Oliver
>
> Am 18.09.2013 11:27, schrieb Eike Mueller:
>> Hi Oliver,
>>
>> thanks for looking into this and confirming that it is not a bug but rather
>> a problem with my grid - several of my elements did indeed have the wrong
>> orientation, i.e. the vertices were not in the
>> correct order. The problem did not show up in ALUGrid as there the grid was
>> 2d instead of 3d.
>> I now fixed my code which generates the .dgf grid files such that it writes
>> out correctly oriented elements, and then centerUnitOuterNormal() shows the
>> expected behaviour.
>>
>> Apologies for the confusion,
>>
>> Eike
>>
>> On 16/09/13 16:24, Oliver Sander wrote:
>>> Hi Eike,
>>> that sounds like a bug.  You are right in expecting a method call
>>> '*OuterNormal'
>>> to return an outer normal.  Can you send a complete test case that reproduces
>>> the problem?
>>> Cheers,
>>> Oliver
>>>
>>> Am 16.09.2013 16:30, schrieb Eike Mueller:
>>>> Dear DUNE-list,
>>>>
>>>> is centerUnitOuterNormal() for the (3d-) UGGrid guaranteed to give the
>>>> OUTER normal in all cases?
>>>> in other word, should the following code, which calculates the scalar
>>>> product of the outer normal and the distance between the center of the
>>>> cell and center of its the intersection with the
>>>> neighbouring cell always print out positive values? [Ok, of course one
>>>> could construct some weird mapping from the reference elements to 3d such
>>>> that this is not the case, but for the grid I'm looking
>>>> at the cells are approximately prisms].
>>>>
>>>> iend = gv.template end<0,Dune::Interior_Partition>()
>>>> for (Iterator it =gv.template
>>>> begin<0,Dune::Interior_Partition>();it!=;++it) {
>>>>      Entity &en = *it;
>>>>      typename Entity::Geometry::GlobalCoordinate    centerEn =
>>>> en.geometry().center();
>>>>      IntersectionIterator isend = gv.iend()
>>>>      for (IntersectionIterator is = gv.ibegin(en);is!=isend; ++is)
>>>>      {
>>>>          typename Entity::Geometry::GlobalCoordinate    normal =
>>>> is.centerUnitOuterNormal();
>>>>          typename Entity::Geometry::GlobalCoordinate    centerIs =
>>>> is.geometry().center();
>>>>          centerIs -= centerEn;
>>>>          std::cout << " Normal * (centerIs - centerEn)" << normal*centerIs
>>>> << std::endl;
>>>>      }
>>>> }
>>>>
>>>> I found that I sometimes get negative values in a cell (and all it's
>>>> children which share the same boundary).
>>>> In my case, I actually need the dot product between the outer normal and
>>>> the distance between the cell and it's neighbours, and I can fix the
>>>> problem by just adding an fabs(), but I was slightly
>>>> surprised by this behaviour which I did not see in ALUGrid.
>>>>
>>>> Thanks a lot,
>>>>
>>>> Eike
>>>>
>>>> _______________________________________________
>>>> Dune mailing list
>>>> Dune at dune-project.org
>>>> http://lists.dune-project.org/mailman/listinfo/dune
>>>
>>>
>>> _______________________________________________
>>> Dune mailing list
>>> Dune at dune-project.org
>>> http://lists.dune-project.org/mailman/listinfo/dune
>>
>
>
> _______________________________________________
> Dune mailing list
> Dune at dune-project.org
> http://lists.dune-project.org/mailman/listinfo/dune

-- 
Dr. Martin Nolte <nolte at mathematik.uni-freiburg.de>

Universität Freiburg                                   phone: +49-761-203-5630
Abteilung für angewandte Mathematik                    fax:   +49-761-203-5632
Hermann-Herder-Straße 10
79104 Freiburg, Germany