MCMC fitting with emcee

Speaker: Marcelo Ojeda • Date: 10-10-2025

When / Where

Friday 10th, October at 11:30 am • Sala Seminario

Abstract

Too many datapoints? Models with too many parameters? Bored of the old and inconsistent curvefit outputs? MCMC is here to save you. Markov Chain Monte Carlo (MCMC) methods are powerful tools for exploring complex parameter spaces and estimating uncertainties in model fitting. In this meeting, I will introduce the Python package emcee, a widely used, flexible MCMC sampler. We will cover the basics of setting up a likelihood function, initializing walkers, and interpreting the resulting posterior distributions. I will demonstrate how emcee can be applied to fit models to astrophysical data, providing robust parameter estimates and uncertainties.

Files

• Slides: slides.pdf

• Notebook: emcee_example.ipynb

• Video: video