NNPT Numerical Experiments

This repo contains small scripts to reproduce the figures/experiments in the project.

Setup

Create an environment with the required Python packages (at minimum: numpy, matplotlib, seaborn, tqdm). Make sure the local modules (lpt.py, utility.py, etc.) are on your PYTHONPATH (running from the repo root is simplest).

Reproducing the examples

Step 1: Run simulations

example1.py, example2.py, and example3.py run Monte Carlo simulations and save p-value arrays to data/.

python example1.py
python example2.py
python example3.py

To run only a subset of simulations within a script, use --sims:

python example1.py --sims nnpt
python example2.py --sims sizing validity
python example3.py --sims 2d_sizing 3d_fixed

Step 2: Make plots

Once the .npy files are in data/, generate figures with make_plots.py:

python make_plots.py

To make only specific plots:

python make_plots.py --plots example1_nnpt example2_sizing example3_bins

Available plot names:

Plot name	Source
example1_no_clt	example1
example1_nnpt	example1
example2_sizing	example2
example2_fixed	example2
example2_sizing + example2_fixed	example2 (combined plot, fixed as dashed lines)
example2_sizing_validity	example2
example2_fixed_validity	example2
example3_2d_sizing	example3
example3_2d_fixed	example3
example3_2d_sizing_validity	example3
example3_2d_fixed_validity	example3
example3_3d_sizing	example3
example3_3d_fixed	example3
example3_3d_sizing_validity	example3
example3_3d_fixed_validity	example3
example3_bins	example3 (bin visualisations, no .npy needed)

Figures are saved to figures/.

Notes

• The data/ and figures/ directories are created automatically if needed.
• Simulation settings (sample sizes, bin exponents, number of MC repetitions) are defined near the top of main() in each script.