This repository contains the supporting codes for the article Robust inference of causality in high-dimensional dynamical processes from the Information Imbalance of distance ranks (arXiv, PNAS).
The functions employed to compute the Imbalance Gain in the paper are in the Python modules utilities.py and imbalance_gain.py, and their use is illustrated in the notebook tutorial.ipynb. The Information Imbalance for the causality tests can also be computed using the DADApy package, as shown in the same tutorial.
The subdirectory dynamical-systems contains the scripts to generate the trajectories of the dynamical systems analyzed in the paper.
The codes require installing the packages NumPy (1.24.2), Matplotlib (3.7.0), SciPy (1.10.1), scikit-learn (1.2.1), Joblib (1.2.0) and DADApy (0.2.0).