Generative Unfolding with Distribution Mapping

TL;DR

This paper extends generative unfolding techniques using Schr"odinger Bridges and Direct Diffusion to improve the accuracy of distribution mapping in high-dimensional particle physics data analysis.

Contribution

It introduces extensions of two morphing methods to ensure correct conditional probabilities in generative unfolding, matching state-of-the-art accuracy.

Findings

Achieved high accuracy on single jet substructure dataset

Successfully applied methods to 22-dimensional Z + 2-jets phase space

Demonstrated improved distribution mapping in complex high-dimensional data

Abstract

Machine learning enables unbinned, highly-differential cross section measurements. A recent idea uses generative models to morph a starting simulation into the unfolded data. We show how to extend two morphing techniques, Schr\"odinger Bridges and Direct Diffusion, in order to ensure that the models learn the correct conditional probabilities. This brings distribution mapping to a similar level of accuracy as the state-of-the-art conditional generative unfolding methods. Numerical results are presented with a standard benchmark dataset of single jet substructure as well as for a new dataset describing a 22-dimensional phase space of Z + 2-jets.