rose

rose is a framework to visualize gravitational-wave radiation using ParaView.

It is based on gwpv written by Nils Vu. Check it out!

The name rose comes from the flower shapes of gravitaional emission of compact binary coalescences.

Detrás de los zarzales salvajes de tu pecho
Hay una rosa que deslumbrará todo el jardín

Ruido - La Prohibida

Workflow

1. Create and activate Mamba/Conda environment with the Python version matching the one in ParaView
2. Install the dependencies
3. Start ParaView using rose-pv script to automatically load the plugins
4. Open the waveform files and setup the scene, and export the frames
5. Run the overplotting script to add legends, text, logos, and other data
6. Combine the frames into a video
7. Enjoy the results!

Installation

1. Install MambaForge using the command below:

curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Mambaforge-$(uname)-$(uname -m).sh"
bash Mambaforge-$(uname)-$(uname -m).sh

In case you don't have internet access on the remote: use mitten instead of ssh to pass your internet connection to the remote.

2. Find the exact Python version your ParaView has. Go to ParaView->About ParaView and note down the "Python Library Version". For example, my ParaView 5.13.1 has Python 3.10.13.

3. Create and activate the Mamba environment for rose with the matching Python version and dependencies:

mamba create -y -n rose python=3.10.13 numpy scipy psutil astropy h5py spherical scri spherical_functions
mamba activate rose

4. From the rose directory root, start ParaView via rose-pv script by specifying path to your ParaView binary,

./rose-pv /path/to/paraview

or the application in case of macOS:

./rose-pv /Applications/ParaView-5.13.1.app

That will start the ParaView and load all rose plugins.

5. Now you are ready to open your waveform files using the appropriate reader. For example, rhOverM_Asymptotic_GeometricUnits_CoM.h5 using EnergyFluxVolumeReader. Note that at the momement rose supports only extrapolated waveforms in SXS catalog format, cf. Appendix A.3.1 of Boyle:2019kee.

Running parallel state rendering on an HPC cluster

Running locally on your machine might be too slow or not fitting into RAM for high resolution. For that reason, you might want to run it on a more powerful cluster.

The idea here is to create the scene in ParaView locally, save it to a state file, and then render it on a remote cluster. As the paths to the data files might be different locally and on the cluster, one has to readjust them by editing the state file. As I didn't write yet the script to automatically swap the path, one has has to do it manually.

The below instructions are currently outdated.

1. Copy render/render.begin and your state file (.pvsm) to the cluster.

2. Adjust the number of nodes, path to rose.sif container, and other parameters in render.begin.

3. Begin the rendering job on your state using spanner:

spanner begin -f /path/to/render.begin state.pvsm

4. Check the status of the job:

spanner list

6. Track the process:

spanner logs -f rose_job_name

7. The job will create a directory with the same name as the state file, and output the frames there.

8. Create a video named output.mp4 from the frames in the run directory, adding dark background:

apptainer exec --bind /scratch:/scratch --bind /work:/work ~/apptainers/rose-v2.0.0.sif \
     create_video.py --frames-dir rendered --background-color 0x161616 --output output.mp4

Running parallel frame postprocessing on an HPC cluster

1. Copy render/overplot/overplot.begin to the cluster.

2. Adjust the number of nodes, path to rose.sif container, and other parameters in overplot.begin.

3. Begin the rendering job on your state:

spanner begin -f /path/to/overplot.begin state.pvsm

4. The job will plot legends and a colorbar on top of the frames in the frame directory, and save postprocessed frames into overplotted subdirectory.

5. Create a video from these frames:

apptainer exec --bind /scratch:/scratch --bind /work:/work ~/apptainers/rose-v2.0.0.sif \
     create_video.py --frames-dir overplotted --output output-overplotted.mp4

This is your final movie, enjoy! 💫

Aquí y ahora puede comenzar tu viaje.