[Paraview] Interfacing with FE simulation using Catalyst

Wed Nov 26 12:06:13 EST 2014

Hi,

My guess is that the templating isn't working the way you think it is. I
couldn't see where vtkout_type was being set but in general you probably
don't want to template on the VTK data sets since the API where the work is
done is different for each of them. Also, input->GetClassName() should be
returning the name of the actual class you want to use, e.g. vtkPolyData,
vtkUnstructuredGrid, etc.

Another thing I noticed is that you probably want to get rid of the
SetForceOutput(true) in:
if(inSituProcessor->RequestDataDescription(dataDescription.GetPointer()) !=
0)  {  dataDescription->GetInputDescriptionByName("input")->SetGrid(out);
dataDescription->SetForceOutput(true);  std::cout << "CoProcess " <<
inSituProcessor->CoProcess(dataDescription.GetPointer())<< std::endl;  }
This will ensure that all output from Catalyst is generated when possibly
only a single output was requested in the pipelines.

Regards,
Andy

On Wed, Nov 26, 2014 at 11:33 AM, Alexandre Ancel <
alexandre.ancel at cemosis.fr> wrote:

> Hello,
>
> An update on this issue that I'm still having:
> I tried to pinpoint the source of the problem. I went through Paraview and
> VTK code, and found that there is a check that fails to pass apparently,
> resulting in the errors previously mentionned. This check fails on the
> Paraview side.
> In VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx (from which
> vtkPVPostFilterExecutive inherits), the InputTypeIsValid (l.763) function
> seems to check for data type, and fails to find a matching type when I try
> to visualize the data extracted from the PVTrivialProducer.
> The recorded type (with input->GetClassName()) is vtkDataObject and the
> IsA function is unable to match it with any of the type in the list
> iterated over.
>
> Did I went to far in the code ? Could the information about the
> Unstructured grid nature be lost when I extract the data in Paraview ?
>
> By the way, this list is probably not the right place to talk about it,
> since it is related to VTK, but isn't the error printed when no match is
> found a bit confusing for the user ?
> I mean if the function does not find a valid type, it will by default
> choose the first type of the list, resulting in the kind of errors I had
> (vtkPolyData in my case):
> vtkPVPostFilterExecutive (0x3efc890): Input for connection index 0 on
> input port index 0 for algorithm vtkPVPostFilter(0x3efbe70) is of type
> vtkDataObject, but a vtkPolyData is required.
>
> Whereas the type checked was not matched with any type, thus not a
> vtkPolyData.
>
> By the way, if you want to have a look at the C++ code used in our
> library, it is available in:
>
> https://github.com/feelpp/feelpp/blob/develop/feel/feelfilters/exporterVTK_impl.hpp
> With:
> - the initialization part starting at line 115
> - the coprocessing part starting at line 406
>
> Thanks again for the help,
> Best regards,
> Alexandre Ancel
>
>
> On Mon, Nov 24, 2014 at 4:56 PM, Alexandre Ancel <
> alexandre.ancel at cemosis.fr> wrote:
>
>> Hello,
>>
>> I tested writing data from the python script, and the data is apparently
>> written correctly
>>  when opening the files with Paraview. The issue is only present I try to
>> extract the data in Paraview.
>> I'll continue to investigate both in my code and Paraview. If you have
>> other pieces of advice as to where I could be looking, I would be
>> interested.
>>
>> I will also test if I can reproduce the issue with one of the example.
>>
>> Regards,
>> Alexandre Ancel
>>
>> On Fri, Nov 21, 2014 at 10:57 PM, Andy Bauer <andy.bauer at kitware.com>
>> wrote:
>>
>>> Hi Alexandre,
>>>
>>> If you could reproduce the errors that you're getting in one of the
>>> examples, we could try to track that down. I tried using your script and
>>> didn't have any issues with the CxxFullExample in the Catalyst example code
>>> and PV 4.2.
>>>
>>>
>>> Regards,
>>> Andy
>>>
>>> On Fri, Nov 21, 2014 at 12:14 PM, Dan Lipsa <dan.lipsa at kitware.com>
>>> wrote:
>>>
>>>>
>>>>> When I click on the icon located near the producer to visualize the
>>>>> data, I get an Extract component added to the pipeline. Trying to visualize
>>>>> this, then throws me the following errors:
>>>>>
>>>>> ---
>>>>> ERROR: In
>>>>> /home/ancel/Downloads/ParaView-v4.2.0-source/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
>>>>> line 810
>>>>> vtkPVPostFilterExecutive (0x3efc890): Input for connection index 0 on
>>>>> input port index 0 for algorithm vtkPVPostFilter(0x3efbe70) is of type
>>>>> vtkDataObject, but a vtkPolyData is required.
>>>>>
>>>>>
>>>>> ERROR: In
>>>>> /home/ancel/Downloads/ParaView-v4.2.0-source/VTK/Common/ExecutionModel/vtkDemandDrivenPipeline.cxx,
>>>>> line 810
>>>>> vtkPVPostFilterExecutive (0x3efc890): Input for connection index 0 on
>>>>> input port index 0 for algorithm vtkPVPostFilter(0x3efbe70) is of type
>>>>> vtkDataObject, but a vtkPolyData is required.
>>>>>
>>>>>
>>>>> ERROR: In
>>>>> /home/ancel/Downloads/ParaView-v4.2.0-source/ParaViewCore/ServerManager/Rendering/vtkSMParaViewPipelineControllerWithRendering.cxx,
>>>>> line 504
>>>>> vtkSMParaViewPipelineControllerWithRendering (0x38bbc00): Data cannot
>>>>> be shown in the defaulted render view!!
>>>>> ---
>>>>>
>>>>
>>>> I have not run an example for an unstructured grid. It should work
>>>> similarly though.
>>>> I recently wrote a Catalyst adapter for an unstructured grid. I will
>>>> run Catalyst Live with that to see how that works.
>>>>
>>>> You could try however to save images or entire data from the pipeline
>>>> script as it is done in the Catalyst examples.
>>>>
>>>>
>>>>
>>>>> A few more questions come in my mind now:
>>>>> - If I understand correctly and if the user wants to pause the
>>>>> simulation, for example when he has only 1 timestep, he would have to do
>>>>> the following steps:
>>>>> * Connect with catalyst in Paraview
>>>>> * Set a breakpoint for time step 0
>>>>> * launch the simulation
>>>>> Is that alright ?
>>>>> Would it be possible to do this with the play/next frame paraview
>>>>> buttons ?
>>>>>
>>>>
>>>> The communication between catalyst and the simulation happens only at
>>>> time steps. Because of that, Catalyst could only stop the simulation at
>>>> time step 2.
>>>>
>>>>
>>>> - Looking at the last lines of the blog post concerning pvserver. Is
>>>>> pvserver catalyst enabled by default ?
>>>>>
>>>> Yes.
>>>>
>>>>
>>>>> Does it mean that I can build the following pipeline ?
>>>>> Paraview (local computer) <-> pvserver (cluster frontal node) <->
>>>>> catalyst enabled simulation (cluster nodes)
>>>>>
>>>>
>>>>  You need to enable MPI to be able to run this. pvserver is the
>>>> paraview server that does all the rendering, only images are sent to
>>>> ParaView (GUI).
>>>>
>>>> Dan
>>>>
>>>>
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>>
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>>
>>>> Please keep messages on-topic and check the ParaView Wiki at:
>>>> http://paraview.org/Wiki/ParaView
>>>>
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://public.kitware.com/mailman/listinfo/paraview
>>>>
>>>>
>>>
>>
>>
>> --
>> Alexandre Ancel
>> Docteur, Ingénieur de recherche / Phd, Research Engineer
>> Cemosis <http://www.cemosis.fr> - alexandre.ancel at cemosis.fr
>> Tel: +33 (0)3 68 8*5 02 06*
>> IRMA - 7, rue René Descartes
>> 67 000 Strasbourg, France
>>
>
>
>
> --
> Alexandre Ancel
> Docteur, Ingénieur de recherche / Phd, Research Engineer
> Cemosis <http://www.cemosis.fr> - alexandre.ancel at cemosis.fr
> Tel: +33 (0)3 68 8*5 02 06*
> IRMA - 7, rue René Descartes
> 67 000 Strasbourg, France
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20141126/b37f19ea/attachment-0001.html>