Explainable machine learning of the underlying physics of high-energy particle collisions

Abstract

We present an implementation of an explainable and physics-aware machine learning model capable of inferring the underlying physics of high-energy particle collisions using the information encoded in the energy-momentum four-vectors of the final state particles. We demonstrate the proof-of-concept of our White Box AI approach using a Generative Adversarial Network (GAN) which learns from a DGLAP-based parton shower Monte Carlo event generator. We show, for the first time, that our approach leads to a network that is able to learn not only the final distribution of particles, but also the underlying parton branching mechanism, i.e. the Altarelli-Parisi splitting function, the ordering variable of the shower, and the scaling behavior. While the current work is focused on perturbative physics of the parton shower, we foresee a broad range of applications of our framework to areas that are currently difficult to address from first principles in QCD. Examples include nonperturbative and collective effects, factorization breaking and the modification of the parton shower in heavy-ion, and electron-nucleus collisions.

11 pages, 4 figures

References in corpus (20)

• Event generation with SHERPA 1.1
• Herwig++ Physics and Manual
• Calculating the Jet Quenching Parameter from AdS/CFT
• Classification without labels: Learning from mixed samples in high energy physics
• Parton Fragmentation Functions
• Equivariant flow-based sampling for lattice gauge theory
• Jet physics in heavy-ion collisions
• Parton showers beyond leading logarithmic accuracy
• Pileup Mitigation with Machine Learning (PUMML)
• Simulation of electron-proton scattering events by a Feature-Augmented and Transformed Generative Adversarial Network (FAT-GAN)
• Lorentz Group Equivariant Neural Network for Particle Physics
• Mapping Machine-Learned Physics into a Human-Readable Space
• Factorization of jet cross sections in heavy-ion collisions
• Equivariant Energy Flow Networks for Jet Tagging
• LHC analysis-specific datasets with Generative Adversarial Networks
• Interpretable deep learning for nuclear deformation in heavy ion collisions
• Jet grooming through reinforcement learning
• The JETSCAPE framework
• Covariantizing Phase Space
• Calculating the energy loss of leading jets

Cited by in corpus (13)

• Machine Learning in Nuclear Physics
• Machine Learning and LHC Event Generation
• CaloFlow: Fast and Accurate Generation of Calorimeter Showers with Normalizing Flows
• Precision Studies of QCD in the Low Energy Domain of the EIC
• E Pluribus Unum Ex Machina: Learning from Many Collider Events at Once
• Energy-weighted Message Passing: an infra-red and collinear safe graph neural network algorithm
• Towards a Deep Learning Model for Hadronization
• A Detailed Study of Interpretability of Deep Neural Network based Top Taggers
• FAIR AI Models in High Energy Physics
• Systematic Quark/Gluon Identification with Ratios of Likelihoods
• Latent Space Refinement for Deep Generative Models
• Applicability Evaluation of Selected xAI Methods for Machine Learning Algorithms for Signal Parameters Extraction
• Explaining machine-learned particle-flow reconstruction