nessai.utils.information

Utilities for computing information and entropy

Functions

differential_entropy(→ float)

Approximate the differential entropy from samples.

Module Contents

nessai.utils.information.differential_entropy(log_p: numpy.ndarray) → float

Approximate the differential entropy from samples.

Parameters:

log_pnumpy.ndarray

Array of log-probabilities.

Returns:

float

The differential entropy

Notes

Assumes samples are drawn from \(p(x)\) such that

\[h(x) = -\int p(x) \log p(x) dx\]

can be approximated via Monte Carlo integration.