Communicating Likelihoods with Normalising Flows
This work addresses the problem of likelihood communication for high-energy physicists, providing an incremental solution.
The authors tackled the problem of modeling unbinned likelihoods from samples, achieving reliable communication of experimental and phenomenological likelihoods through their workflow. They demonstrated its effectiveness in three high-energy physics case studies.
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 of the joint distribution, 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.