Instructions

• Create Coka cluster account (login from INFN or axpfe)

• Create Conda environment able to run the script --> attached txt file with packages, to create env type : 'conda create -n myenv --file package-list.txt'

• Script to run here: '/eos/user/e/efranzos/SWAN_projects/analysis_attempt/yields_fit/fit_templates.py'

• Copy the script in a folder in coka (also utils.py) and run with the conda env created (before running, create a folder called 'img' in the same path of the 'fit_templates.py' script to save the plots)

• Check of you can access the path to the trained NN in coka: '/home/efranzos/ML_pytorch/MODELS'

• The dataset to study needs to be changed in the script and it will be loaded with uproot (hint: can't be too large so create a small root files with only the variables required by the NN as done in '/eos/user/e/efranzos/SWAN_projects/analysis_attempt/save_clean_dataset.ipynb' and then copy it in coka)

• To submit the job run 'sbatch my_job_file.slurm' (must be located in the same folder of 'fit_templates.py'

-- EdoardoFranzoso - 2022-03-17