Emcee

Emcee: Open-Source Python Library for Bayesian Statistical Modeling and Probabilistic Machine Learning

Emcee is an open-source Python library for Bayesian statistical modeling and probabilistic machine learning. It implements efficient Markov Chain Monte Carlo (MCMC) sampling algorithms that allow users to fit complex models with thousands of parameters.

What is Emcee?

Emcee is an open-source Python library for Bayesian statistical modeling and probabilistic machine learning. It provides an implementation of the affine-invariant ensemble sampler for Markov Chain Monte Carlo (MCMC) proposed by Goodman and Weare.

Some of the key features of Emcee include:

• Implementations of stretch move and snooker update for efficient sampling
• Support for Bayesian model fitting with thousands of parameters
• Parallelization on multi-core machines and GPUs
• Easy to use API for defining models and fitting parameters
• Integrations with common Python data science stacks like NumPy, SciPy and Matplotlib

Emcee allows users to easily set up Bayesian models with prior distributions and likelihoods. The MCMC sampler will then efficiently explore the parameter space to maximize the posterior probability. This makes Emcee useful for problems like parameter estimation, uncertainty quantification, density estimation and experimental design.

Since it is optimized for problems with large numbers of parameters, Emcee is popular in fields like astrophysics, geoscience, biology and physics where complex probabilistic models are common.

Emcee Features

Official Links