Communicating likelihoods with normalising flows
Abstract
We present a machine-learning-based workflow to model an unbinned likelihood from its samples. A key advancement over existing approaches is the validation of the learned likelihood using rigorous statistical tests, such as the Kolmogorov-Smirnov test of the joint distribution. Our method enables the reliable communication of experimental and phenomenological likelihoods for subsequent analyses. We demonstrate its effectiveness through three case studies in high-energy physics. To support broader adoption, we provide an open-source reference implementation, nabu.
Fields of Study & Subjects
Computational Physics and Python Applications
Scientific Computing and Data Management
Data Analysis with R
Publication details
- Journal / source
- The European Physical Journal C
- Publisher
- Springer Science and Business Media LLC
- Volume
- 86
- Issue
- 7
- Pages
- 805–805
- Publication date
- 13 July 2026
- Language
- EN