Information Flow Analysis

IFA is a simple and fast library for information theory research and information flow analysis. It's a Python module written C++, Cython.

Installation

Dependencies:

• numpy

If you have Cython some cpp files will get regenerated during installation

    pip install ifa

Or if you want the developmen version:

    git clone https://github.com/janekolszak/ifa.git;
    cd ifa;
    sudo make install;

Usage

Computing Jensen–Shannon divergence:

    from ifa.distribution import Distribution
    from ifa.divergence import jsd

    from numpy.testing import assert_allclose

    p = Distribution(["A", "B"], [0.5, 0.5])
    q = Distribution(["A", "C"], [0.5, 0.5])

    assert_allclose(jsd(p, 0.5, q, 0.5), [0.5])

What's inside:

• Distribution class with some basic operations
• Divergences:
• Jensen–Shannon divergence
• Kullback–Leibler divergence
• Functions to compute information flow between distributions