Hello,
I have managed to install Code_Saturne (both versions 4.0.7 and 5.0.3) with Paraview Catalyst support, however, I cannot seem to find anywhere how to make it work. Is there any tutorial or example to help me set up a run with Catalyst?
Thanks!
Code_Saturne and Catalyst
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
Yvan Fournier
- Posts: 4208
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Code_Saturne and Catalyst
Hello,
I assume the install is OK, and you are now looking to use this.
The idea is the following:
1 - run a standard case, using postprocessing with the EnSight format
2 - postprocess the case with ParaView, using the CatalystScriptGeneratorPlugin (or CoprocessorScriptGeneratorPlugin) to generate a Python script (this is basically a graphical "wizard" type tool).
3 - place that script in the case's DATA directory
4 - create a mesh postprocessing "writer" with the same name as the generated Python script (without the ".py" extension, and to which the same meshes are associated as the writer which was used in 1). Renaming the original writer and setting its type to Catalyst works also.
5 - run
There are some caveats:
- with some plugins loaded and/or ParaView builds, some settings in the generated script may match features not available or loaded in your Catalyst build. This will case failures, and the error logs will tell you (through a Python backtrace) which lines cause issues. In most cases, removing the offending lines is enough to solve the issue (this is a pain, and not very elegant, but at least there is a workaround).
- only one Catalyst writer may be used at a time. We have not experimented much yet with multiple inputs.
- I have never tested the "live visualization" option on a cluster. It probably sends everything to the visualization workstation (a more complex setup would be needed with a parallel server running on a visualization cluster, and a client on a workstation or front-end node).
There is also a video generated by an intern a few years ago (with an example of removing incorrect lines) on the web: https://www.youtube.com/watch?v=G-D0SbO ... e=youtu.be.
Best regards,
Yvan
I assume the install is OK, and you are now looking to use this.
The idea is the following:
1 - run a standard case, using postprocessing with the EnSight format
2 - postprocess the case with ParaView, using the CatalystScriptGeneratorPlugin (or CoprocessorScriptGeneratorPlugin) to generate a Python script (this is basically a graphical "wizard" type tool).
3 - place that script in the case's DATA directory
4 - create a mesh postprocessing "writer" with the same name as the generated Python script (without the ".py" extension, and to which the same meshes are associated as the writer which was used in 1). Renaming the original writer and setting its type to Catalyst works also.
5 - run
There are some caveats:
- with some plugins loaded and/or ParaView builds, some settings in the generated script may match features not available or loaded in your Catalyst build. This will case failures, and the error logs will tell you (through a Python backtrace) which lines cause issues. In most cases, removing the offending lines is enough to solve the issue (this is a pain, and not very elegant, but at least there is a workaround).
- only one Catalyst writer may be used at a time. We have not experimented much yet with multiple inputs.
- I have never tested the "live visualization" option on a cluster. It probably sends everything to the visualization workstation (a more complex setup would be needed with a parallel server running on a visualization cluster, and a client on a workstation or front-end node).
There is also a video generated by an intern a few years ago (with an example of removing incorrect lines) on the web: https://www.youtube.com/watch?v=G-D0SbO ... e=youtu.be.
Best regards,
Yvan