Data Science and Bayesian methods

Modern physics places demanding requirements on data analysis and statistical inference. Data can range from a handful of rare events to very large datasets and complex simulations. Extracting physics knowledge from such data requires powerful statistical methods, efficient algorithms, and modern data-science tools.

A particular focus of our group are Bayesian methods. The Bayesian approach provides a consistent framework for confronting complex models with measurements, determining model parameters, and quantifying uncertainties. At MPP, new methods and software for these tasks are developed and applied to physics analyses. This includes the Bayesian Analysis Toolkit (BAT), a powerful open-source framework for Bayesian inference.