---

Instructions

Install necessary package using pipenv:

cd /path/to/git/repo/root/
pipenv install .

or

pip install -r requirement.txt

To run the full set of experiments:

for i in $(seq 0 3839)
do 
 python experiments.py $i
done

Alternatively, see slurmsweep.sh or slurmsweep_cpu.sh SLURM scripts for running in HPC environment.

To collect results into dataframes (as pickled .pkl files)

python experiments_to_pandas.py --path /path/to/outputdir/

To generate plots from the pickled files:

python plot_utils.py --output_dirpath "/path/to/output/directory" --datafilepaths /path/to/data1.pkl /path/to/data2.pkl

---

Other content

• experiment_output contains outputs of running experiments.py and experiments_to_pandas.py as instructed above. This is the dataset used to generate all figures and results in the paper.
• demo.ipynb is a standalone Jupyter notebook with a small demonstration.