Chainsail

sampling multimodal distributions made easy

Chainsail helps to sample multimodal distributions by implementing an automatically scaling Replica Exchange MCMC algorithm. All you need to provide is a Python module defining your probability density and its log-probability gradient.

Chainsail is currently not deployed under this URL. If you would like to see a demo, please contact the Chainsail team at support@chainsail.io.

Chainsail is open-source and you can find the source code here. Please feel free to create an issue or contribute via PRs!

The Chainsail resources repository contains some background on the algorithms used in Chainsail and a complete walk-through. Also see our blog posts for details on Chainsail and the algorithms at work:

Chainsail announcement blog post Inferring Gaussian mixture parameters with Chainsail Replica Exchange Hamiltonian Monte Carlo