Machine learning and LHC event generation

Butter, Anja; Plehn, Tilman; Schumann, Steffen; Badger, Simon; Caron, S.; Cranmer, K.; Bello, F. A. Di; Dreyer, E.; Forte, Stefano; Ganguly, S.; Gonçalves, Dorival; Heimel, Theo; Heinrich, G.; Heinrich, L.; Held, A.; Höche, Stefan; Howard, J.; Ilten, P.; Isaacson, Joshua; Janßen, Timo; Jones, Stephen; Kado, M.; Kagan, M.; Kasieczka, G.; Kling, Felix; Kraml, S.; Krause, Claudius; Krauss, Frank; Kroeninger, K.; Barman, Rahool Kumar; Luchmann, Michel; Magerya, Vitaly; Maître, D.; Malaescu, B.; Maltoni, Fabio; Martini, Till; Mattelaer, Olivier; Nachman, B. P.; Pitz, Sebastian; Rojo, Juan; Schwartz, Matthew D.; Shih, David; Siegert, F.; Stegeman, Roy; Stienen, Bob; Thaler, Jesse; Verheyen, Rob; Whiteson, D.; Winterhalder, Ramon; Zupan, Jure

doi:10.21468/scipostphys.14.4.079

Cited by 32 publications

(13 citation statements)

References 130 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This work significantly impacts high-granularity fast and efficient detector response and collider event simulations. Since they require fine-grained intra-event-correlated data generation, we believe that the Intra-Event Aware GAN (IEA-GAN) offers a robust controllable sampling for all particle physics experiments and simulations, such as detector simulation 11 , 63 and event generation 19 , 20 , 64 , 65 at both Belle II 37 and LHC 66 . In particular, the High-Luminosity Large Hadron Collider (HL-LHC) 25 is expected to surpass the LHC’s design-integrated luminosity by increasing it by a factor of 10.…”

Section: Discussionmentioning

confidence: 99%

Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self-supervised relational reasoning

Hashemi,

Hartmann,

Sharifzadeh

et al. 2024

Nat Commun

View full text Add to dashboard Cite

Simulating high-resolution detector responses is a computationally intensive process that has long been challenging in Particle Physics. Despite the ability of generative models to streamline it, full ultra-high-granularity detector simulation still proves to be difficult as it contains correlated and fine-grained information. To overcome these limitations, we propose Intra-Event Aware Generative Adversarial Network (IEA-GAN). IEA-GAN presents a Transformer-based Relational Reasoning Module that approximates an event in detector simulation, generating contextualized high-resolution full detector responses with a proper relational inductive bias. IEA-GAN also introduces a Self-Supervised intra-event aware loss and Uniformity loss, significantly enhancing sample fidelity and diversity. We demonstrate IEA-GAN’s application in generating sensor-dependent images for the ultra-high-granularity Pixel Vertex Detector (PXD), with more than 7.5 M information channels at the Belle II Experiment. Applications of this work span from Foundation Models for high-granularity detector simulation, such as at the HL-LHC (High Luminosity LHC), to simulation-based inference and fine-grained density estimation.

show abstract

Section: Discussionmentioning

confidence: 99%

Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self-supervised relational reasoning

Hashemi,

Hartmann,

Sharifzadeh

et al. 2024

Nat Commun

View full text Add to dashboard Cite

show abstract

“…It is expected that modern machine learning techniques can improve LHC simulations in other aspects as well [457]. For a more comprehensive overview of the topic, see [458].…”

Section: Machine Learning Techniquesmentioning

confidence: 99%

Event generators for high-energy physics experiments

Campbell,

Diefenthaler,

Hobbs

et al. 2024

SciPost Phys.

Self Cite

View full text Add to dashboard Cite

We provide an overview of the status of Monte-Carlo event generators for high-energy particle physics. Guided by the experimental needs and requirements, we highlight areas of active development, and opportunities for future improvements. Particular emphasis is given to physics models and algorithms that are employed across a variety of experiments. These common themes in event generator development lead to a more comprehensive understanding of physics at the highest energies and intensities, and allow models to be tested against a wealth of data that have been accumulated over the past decades. A cohesive approach to event generator development will allow these models to be further improved and systematic uncertainties to be reduced, directly contributing to future experimental success. Event generators are part of a much larger ecosystem of computational tools. They typically involve a number of unknown model parameters that must be tuned to experimental data, while maintaining the integrity of the underlying physics models. Making both these data, and the analyses with which they have been obtained accessible to future users is an essential aspect of open science and data preservation. It ensures the consistency of physics models across a variety of experiments.

show abstract

“…The experimental and theoretical methods of LHC physics have always been numerical in nature, with the goal to quantitatively, systematically, and comprehensively understand data in terms of fundamental theory. Generative networks are an exciting concept of modern machine learning (ML), combining unsupervised density estimation in an interpretable phase space with fast and flexible sampling and simulations [1]. Currently, the most promising architectures for precision generation are normalizing flows and their invertible network (INN) variants, but we will see that diffusion models and generative transformers might offer an even better balance of precision and expressivity.…”

Section: Introductionmentioning

confidence: 99%

How to understand limitations of generative networks

Das,

Favaro,

Heimel

et al. 2024

SciPost Phys.

Self Cite

View full text Add to dashboard Cite

Well-trained classifiers and their complete weight distributions provide us with a well-motivated and practicable method to test generative networks in particle physics. We illustrate their benefits for distribution-shifted jets, calorimeter showers, and reconstruction-level events. In all cases, the classifier weights make for a powerful test of the generative network, identify potential problems in the density estimation, relate them to the underlying physics, and tie in with a comprehensive precision and uncertainty treatment for generative networks.

show abstract

Machine learning and LHC event generation

Cited by 32 publications

References 130 publications

Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self-supervised relational reasoning

Ultra-high-granularity detector simulation with intra-event aware generative adversarial network and self-supervised relational reasoning

Event generators for high-energy physics experiments

How to understand limitations of generative networks

Contact Info

Product

Resources

About