Towards a data-driven model of hadronization using normalizing flows
(2024) In SciPost Physics 17(2).- 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... (More)
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.
(Less)
- author
- Bierlich, Christian LU ; Ilten, Phil ; Menzo, Tony ; Mrenna, Stephen ; Szewc, Manuel ; Wilkinson, Michael K. ; Youssef, Ahmed and Zupan, Jure
- organization
- publishing date
- 2024-08
- type
- Contribution to journal
- publication status
- published
- subject
- in
- SciPost Physics
- volume
- 17
- issue
- 2
- article number
- 045
- publisher
- SciPost
- external identifiers
-
- scopus:85201077333
- ISSN
- 2542-4653
- DOI
- 10.21468/SciPostPhys.17.2.045
- language
- English
- LU publication?
- yes
- id
- 695e61d8-51c5-4db0-a1ca-36685883fcc0
- date added to LUP
- 2024-09-06 14:19:11
- date last changed
- 2024-09-06 14:19:35
@article{695e61d8-51c5-4db0-a1ca-36685883fcc0, abstract = {{<p>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.</p>}}, author = {{Bierlich, Christian and Ilten, Phil and Menzo, Tony and Mrenna, Stephen and Szewc, Manuel and Wilkinson, Michael K. and Youssef, Ahmed and Zupan, Jure}}, issn = {{2542-4653}}, language = {{eng}}, number = {{2}}, publisher = {{SciPost}}, series = {{SciPost Physics}}, title = {{Towards a data-driven model of hadronization using normalizing flows}}, url = {{http://dx.doi.org/10.21468/SciPostPhys.17.2.045}}, doi = {{10.21468/SciPostPhys.17.2.045}}, volume = {{17}}, year = {{2024}}, }