This repo contains code and demos of estimation procedures for posterior distributions, conditional entropy, and mutual information between random variables X and Y.
To reproduce any of the figures, navigate to the corresponding directory, and run the Jupyter notebook.
cd figs/fig1
jupyter nbconvert --to notebook --inplace --execute figure-1.ipynb --ExecutePreprocessor.timeout=-1
Commands are similar for Figures 2 and 3. The application and hypothesis test code can be found in the figs/application director. The above commands convert the notebook to a Python file and produces the figures as PDFs. An alternate option is to open to the notebook, and select "Restart and Run All".
UF requires only a standard computer with enough RAM to support the in-memory operations.
The code has been tested on the following systems:
- macOS: Mojave (10.14.1)
- Linux: Ubuntu 16.04
The code mainly depends on the Python scientific stack.
numpy
scipy
scikit-learn
joblib
matplotlib
seaborn
tqdm