Analysis-ready Generative Unfolding

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Butter, Anja, Huetsch, Nathan, Mikuni, Vinicius, Nachman, Benjamin, Schweitzer, Sofia Palacios
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909768012005376
author Butter, Anja
Huetsch, Nathan
Mikuni, Vinicius
Nachman, Benjamin
Schweitzer, Sofia Palacios
author_facet Butter, Anja
Huetsch, Nathan
Mikuni, Vinicius
Nachman, Benjamin
Schweitzer, Sofia Palacios
contents Machine Learning (ML)-based unfolding methods have enabled high-dimensional and unbinned differential cross section measurements. While a suite of such methods has been proposed, most focus exclusively on the challenge of statistically removing resolution effects. In practice, unfolding methods must also account for impurities and finite acceptance and efficiency effects. In this paper, we extend a class of unfolding methods based on generative ML to include the full suite of effects relevant for cross section measurements. Our new methods include fully generative solutions as well as generative-discriminative hybrid approaches (GenFoldG and GenFoldC). We demonstrate these new techniques in both Gaussian and simulated LHC examples. Overall, we find that both methods are able to accommodate all effects, thus adding a complementary and analysis-ready method to the unfolding toolkit.
format Preprint
id arxiv_https___arxiv_org_abs_2509_02708
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Analysis-ready Generative Unfolding
Butter, Anja
Huetsch, Nathan
Mikuni, Vinicius
Nachman, Benjamin
Schweitzer, Sofia Palacios
High Energy Physics - Phenomenology
High Energy Physics - Experiment
Machine Learning (ML)-based unfolding methods have enabled high-dimensional and unbinned differential cross section measurements. While a suite of such methods has been proposed, most focus exclusively on the challenge of statistically removing resolution effects. In practice, unfolding methods must also account for impurities and finite acceptance and efficiency effects. In this paper, we extend a class of unfolding methods based on generative ML to include the full suite of effects relevant for cross section measurements. Our new methods include fully generative solutions as well as generative-discriminative hybrid approaches (GenFoldG and GenFoldC). We demonstrate these new techniques in both Gaussian and simulated LHC examples. Overall, we find that both methods are able to accommodate all effects, thus adding a complementary and analysis-ready method to the unfolding toolkit.
title Analysis-ready Generative Unfolding
topic High Energy Physics - Phenomenology
High Energy Physics - Experiment
url https://arxiv.org/abs/2509.02708