reshmaannathomas / thornyflat_dollysods

Steps to run heimdall+fetch on WVU's Thorny Flats or Dolly Sods.

Copy and paste the submit_script.sh, edit the lines and submit with sbatch submit_script.sh. May you find lots of FRBs!

Submit_script.sh will submit a slurm job to a GPU node.

There are several singularity containers available on Thorny Flats and Dolly Sods. On Dolly Sods you also have /shared/containers/tpp/tpp_20231212.sif container.

Both these containers have your package and Fetch.

To run your or Fetch on the command line:

On TF, do qsub -I -q comm_gpu_inter -l nodes=1:ppn=1:gpus=3 (change the no: of gPUs and node numbers as per your wish):

On DS, ssh <somenode>,:

Then run: singualrity shell --nv <path_to_your_container>

Now you are in the shell. Run your commands now!

Step-by-step procedure to run Heimdall+Fetch pipeline using your

1. your_writer.py can convert files from one format to another. This is optional. your_rfimask.py will give a mask of bad channels which you can use in the subsequent steps.
2. your_heimdall.py runs heimdall on the input file
3. candcsvmaker.py converts the heimdall .cand files to a csv file
4. your_candmaker.py will take this .csv file as input and convert them to h5 files
5. predict.py in FETCH will run on the h5 files and creates a results.csv file
6. your_h5plotter.py can either plot all the h5s files (freq vs time, DM vs time, time profile etc) or only the positives in files