This repository contains two Python packages, sampler-evaluation and sampler-comparison.

sampler-evaluation has code to take an array of samples of shape [num_chains, num_samples, num_dimensions] and return an estimate of the effective sample size (ESS).

sampler-comparison provides baseline versions of a number of samplers, and depends on sampler-evaluation to collect ESS estimates.

This code is still a draft, not even alpha quality. It is not yet ready for use by others, but is being developed in the open to solicit feedback and contributions.