Re: [Paraview] Slice not appearing while coprocessing
Hi Tim, Everything seems to be correct. Maybe on the "input" name/identifier isn't matching? This is an unstructured grid output, correct? Maybe try writing out the full grid to see if Catalyst is getting the proper input information. Cheers, Andy On Fri, Jan 29, 2016 at 4:18 PM, Tim Gallagher wrote: > Hi everybody, > > I was working with a colleague to get his simulation running with > coprocessing and I am totally stumped on the problem. I have been able to > set up and render some very complex things with our code, but for some > reason this simulation just won't work as expected. > > We generated the coprocessing script with version 4.3 and we are running > it on a cluster with the same version, using off-screen rendering. I have > attached the script, it really just takes a slice through the center of the > domain. > > It is producing images of the colorbar and nothing else. I tried to save > the slice data to an xdmf file from the coprocessing script and it all > reports dimensions of zero. So the slice doesn't seem to be actually doing > anything. > > We took the exact lines from the Pipeline class and moved them to their > own python file. We then removed the coprocessor.RegisterView and changed > the coprocessor.CreateProducer line so it loads a datafile off disk > instead. We then ran it through the python shell in the paraview GUI and we > get exactly the slice we expected to get. > > So, for some reason, this renders fine in the GUI but outputs the > background with just the colorbar when we are using coprocessing. > > Any ideas or glaring problems in the coprocessing script that we are > missing? > > Thanks, > > Tim > ___ > 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 > > Search the list archives at: http://markmail.org/search/?q=ParaView > > Follow this link to subscribe/unsubscribe: > http://public.kitware.com/mailman/listinfo/paraview > > ___ 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 Search the list archives at: http://markmail.org/search/?q=ParaView Follow this link to subscribe/unsubscribe: http://public.kitware.com/mailman/listinfo/paraview
Re: [Paraview] Slice not appearing while coprocessing
Hi Andy,
Thanks for the reply! The data is a vtkMultiBlockDataSet with structured grids.
I forgot about the grid writing debugging technique, I found the issue.
It turns out that all of my runs/testing in the past 4 years I've had
coprocessing had only 1 block per processor and this simulation we were trying
had multiple blocks per processor. Buried deep in my interface adapter code
where I set the pointers for the grid arrays, I was pointing to the first
pointer in the array of pointers instead of incrementing through the array for
each block. So all the blocks besides the first on the processor were using the
wrong coordinates and it just so happened that meant we didn't have any grid
points where the slice was. So changing my ptr = x[0] to ptr = x[i] was all it
took... kind of embarrassing.
At any rate, one of the things I tried to do while debugging was to write out
the grid using Xdmf instead of VTK. But that didn't work. I had:
writer = coprocessor.CreateWriter( XdmfWriter, "filename_%t.xmf", 1 )
and I got:
...
File "/data4/pv4.4Build/lib/site-packages/paraview/coprocessing.py", line 319,
in WriterParametersProxy
proxy.GetProperty("Input").SetInputConnection(
AttributeError: 'NoneType' object has no attribute 'GetProperty'
ERROR: In
/data3/ParaView/ParaViewCore/ServerImplementation/Core/vtkSIProxyDefinitionManager.cxx,
line 565
vtkSIProxyDefinitionManager (0x3854340): No proxy that matches:
group=insitu_writer_parameters and proxy=XdmfWriter were found.
Is it possible to write out a multiblock dataset from coprocessing using XDMF?
Thanks again,
Tim
- Original Message -
From: "Andy Bauer"
To: "tim gallagher"
Cc: "paraview"
Sent: Saturday, January 30, 2016 7:46:01 AM
Subject: Re: [Paraview] Slice not appearing while coprocessing
Hi Tim,
Everything seems to be correct. Maybe on the "input" name/identifier isn't
matching? This is an unstructured grid output, correct?
Maybe try writing out the full grid to see if Catalyst is getting the proper
input information.
Cheers,
Andy
On Fri, Jan 29, 2016 at 4:18 PM, Tim Gallagher < tim.gallag...@gatech.edu >
wrote:
Hi everybody,
I was working with a colleague to get his simulation running with coprocessing
and I am totally stumped on the problem. I have been able to set up and render
some very complex things with our code, but for some reason this simulation
just won't work as expected.
We generated the coprocessing script with version 4.3 and we are running it on
a cluster with the same version, using off-screen rendering. I have attached
the script, it really just takes a slice through the center of the domain.
It is producing images of the colorbar and nothing else. I tried to save the
slice data to an xdmf file from the coprocessing script and it all reports
dimensions of zero. So the slice doesn't seem to be actually doing anything.
We took the exact lines from the Pipeline class and moved them to their own
python file. We then removed the coprocessor.RegisterView and changed the
coprocessor.CreateProducer line so it loads a datafile off disk instead. We
then ran it through the python shell in the paraview GUI and we get exactly the
slice we expected to get.
So, for some reason, this renders fine in the GUI but outputs the background
with just the colorbar when we are using coprocessing.
Any ideas or glaring problems in the coprocessing script that we are missing?
Thanks,
Tim
___
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
Search the list archives at: http://markmail.org/search/?q=ParaView
Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview
___
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
Search the list archives at: http://markmail.org/search/?q=ParaView
Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/paraview
Re: [Paraview] Calculate Edge Lengths in Unstructured Grid
Hey Tim, you're absolutely right, thanks for pointing this out! I just wanted to mention, especially for other users searching through all the mails, that this actually provides the minimum, average, and maximum of the quality metric calculated, rather then the minimum, average, and maximum element edge length, which is what I'd need, but thanks anyway! Best, Tom From: Tim Gallagher Sent: Wednesday, January 27, 2016 12:27 AM To: Fastl, Thomas Cc: paraview@paraview.org Subject: Re: [Paraview] Calculate Edge Lengths in Unstructured Grid Hi Thomas, Sorry for the delay responding, I got side tracked doing some other stuff and didn't know the answer off the top of my head! But, it looks the filter provides what you are looking for. Based on the documentation of the vktMeshQuality filter (http://www.vtk.org/doc/nightly/html/classvtkMeshQuality.html#details), in addition to the per-cell value in the Quality scalar, there is also a Field Data set that is produced called Mesh Quality. In your case, you are looking for the Mesh Tetrahedron Quality field. The field has 5 components. The first 4 are the ones you are looking for, and they are: > The minimum, average, maximum, and unbiased variance of quality for each type > of cell is stored in the output mesh's FieldData. When you visualize the Mesh Tetrahedron Quality variable, it will default to the Magnitude. If you change it to show component 0 you will get the minimum for each cell, 1 is the average, etc.. The 5th field (component 4) is used internally and doesn't matter. Hope that helps, Tim From: "Thomas Fastl" To: "tim gallagher" Cc: paraview@paraview.org Sent: Monday, January 25, 2016 5:07:29 PM Subject: Re: [Paraview] Calculate Edge Lengths in Unstructured Grid Hey Tim, thanks for your reply! The Mesh Quality filter indeed provides the Edge Ratio for all tetrahedral elements, a single value calculated for every cell using the values I actually need. There must be a way to access them specifically (min/max/mean edge length) or a workaround to obtain this information. Does anybody know how I could do that? Thank you for your help! Best, Tom From: Tim Gallagher Sent: Monday, January 25, 2016 8:13 PM To: Fastl, Thomas Cc: paraview@paraview.org Subject: Re: [Paraview] Calculate Edge Lengths in Unstructured Grid Thomas, You can use the MeshQuality filter to compute the Edge Ratio for tetrahedral cells. This will give you a field called Quality in your dataset. You can then compute the statistics on it using other filters as needed. That should do what you are looking for. The definitions used in the MeshQuality filter can be found at: http://www.vtk.org/Wiki/images/6/6b/VerdictManual-revA.pdf Tim From: "Thomas Fastl" To: paraview@paraview.org Sent: Monday, January 25, 2016 1:02:35 PM Subject: [Paraview] Calculate Edge Lengths in Unstructured Grid Hello Community, I'm trying to calculate the min/max/mean edge length (or even an edge length distribution) in a tetrahedral element mesh specified as an unstructured grid in Paraview. I've played with the 'Extract Edges' filter, however, wasn't successful so far. Does anybody know how I could do that? Thank you for your input! Best, Thomas Fastl ___ 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 Search the list archives at: http://markmail.org/search/?q=ParaView Follow this link to subscribe/unsubscribe: http://public.kitware.com/mailman/listinfo/paraview ___ 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 Search the list archives at: http://markmail.org/search/?q=ParaView Follow this link to subscribe/unsubscribe: http://public.kitware.com/mailman/listinfo/paraview
