Towards a data-driven model of hadronization using normalizing flows

TL;DR

This paper presents a novel invertible neural network model for hadronization, employing a new training method called MAGIC, and introduces a Bayesian extension for uncertainty analysis, advancing machine learning applications in particle physics.

Contribution

It introduces a data-driven hadronization model using normalizing flows, a new training method MAGIC, and a Bayesian framework for uncertainty quantification.

Findings

Faithful reproduction of simplified Lund string model

Improved agreement with experimental data using MAGIC

Bayesian extension provides uncertainty estimates

Abstract

We introduce a model of hadronization based on invertible neural networks that faithfully reproduces a simplified version of the Lund string model for meson hadronization. Additionally, we introduce a new training method for normalizing flows, termed MAGIC, that improves the agreement between simulated and experimental distributions of high-level (macroscopic) observables by adjusting single-emission (microscopic) dynamics. Our results constitute an important step toward realizing a machine-learning based model of hadronization that utilizes experimental data during training. Finally, we demonstrate how a Bayesian extension to this normalizing-flow architecture can be used to provide analysis of statistical and modeling uncertainties on the generated observable distributions.