[dune-pdelab] Problem with Dune::PDELab::NonOverlappingEntitySet - UGGrid with non overlapping parallel solvers.

[Dodwell, Timothy]

Hi Steffen, That's great! Sorry to ask you nasty question :) Tim

________________________________________
From: Steffen Müthing <steffen.muething at iwr.uni-heidelberg.de>
Sent: 13 October 2015 15:09
To: Dodwell, Timothy
Cc: dune-pdelab mailing list
Subject: Re: [dune-pdelab] Problem with Dune::PDELab::NonOverlappingEntitySet - UGGrid with non overlapping parallel solvers.

Hey Tim,


> Am 09.10.2015 um 18:06 schrieb Dodwell, Timothy <T.Dodwell at exeter.ac.uk>:
>
> Hey Steffen,
>
> So I checked out feature/fix-nonoverlapping-mode, and ran into this error!
>
> /Users/tjd20/git_workspace/dune/dune-pdelab/dune/pdelab/common/intersectiontype.hh:35:37: error: no match for 'operator==' (operand types are 'Dune::PDELab::PartitionViewEntitySet<Dune::GridView<Dune::UGGridLeafGridViewTraits<const Dune::UGGrid<3>, (Dune::PartitionIteratorType)4u> >, Dune::PartitionSet<31u> >::Partitions {aka Dune::PartitionSet<31u>}' and 'const All {aka const Dune::PartitionSet<31u>}')
>         if (entity_set.partitions() == Partitions::all)
>
> Cheers

I’m working on the nonoverlapping support right now - the feature branch has landed in master this afternoon. That updated version
should also resolve your error.

PDELab should stabilize over the next few days, I hope that we have a 2.4 release branch before the end of the week.

Steffen

>
> Tim
> ________________________________________
> From: Steffen Müthing <steffen.muething at iwr.uni-heidelberg.de>
> Sent: 09 October 2015 13:27
> To: Dodwell, Timothy
> Cc: Eike Mueller; dune-pdelab mailing list
> Subject: Re: [dune-pdelab] Problem with Dune::PDELab::NonOverlappingEntitySet - UGGrid with non overlapping parallel solvers.
>
> Hi Tim,
>
> you’ve run into a bleeding edge problem… I’ve fixed the nonoverlapping computations just last week, but the
> code hasn’t landed in the master of dune-pdelab yet. You can try switching PDELab to the branch
> feature/fix-nonoverlapping-mode (I think that’s what is called, I’m traveling right now). That branch contains
> the EntitySet.
>
> Cheers,
> Steffen
>
>> Am 09.10.2015 um 15:08 schrieb Dodwell, Timothy <T.Dodwell at exeter.ac.uk>:
>>
>> Hi Eike (Hope things are well in Bath),
>>
>> Sorry I have confused issues mentioning  PkLocalFiniteElementMap, what you spotted was a bug in my testing branch, when checking out my master I called FEM fem(gv); So forget the second half of my question.
>>
>> The issue is not being able to find 'Dune::PDELab::NonOverlappingEntitySet'
>>
>> I need to use this ( from what I understand) as UG doesn't support overlapping domain decomposition.
>>
>> Cheers
>>
>> Tim
>>
>> From: Eike Mueller <E.Mueller at bath.ac.uk>
>> Sent: 09 October 2015 10:54
>> To: Dodwell, Timothy
>> Cc: dune-pdelab at dune-project.org
>> Subject: Re: [dune-pdelab] Problem with Dune::PDELab::NonOverlappingEntitySet - UGGrid with non overlapping parallel solvers.
>>
>> Hi Tim,
>>
>> which version of DUNE is this? Looking at finiteelementmap/pkfem.hh in the master which I updated not too long ago, the constructor of the class PkLocalFiniteElementMap seems to require the gridview as an argument, so maybe this deletes the default constructor (which gets called by “FEM fem")? Have you tried
>>
>> FEM fem(gv);
>>
>> But it’s odd that it works in the sequenetial version...
>>
>> Eike
>>
>>> On 9 Oct 2015, at 10:28, Dodwell, Timothy <T.Dodwell at exeter.ac.uk> wrote:
>>>
>>> PkLocalFiniteElementMap
>>
>> _______________________________________________
>> dune-pdelab mailing list
>> dune-pdelab at dune-project.org
>> http://lists.dune-project.org/mailman/listinfo/dune-pdelab