Improved particle-flow event reconstruction with scalable neural networks for current and future particle detectors

Abstract

Efficient and accurate algorithms are necessary to reconstruct particles in the highly granular detectors anticipated at the High-Luminosity Large Hadron Collider and the Future Circular Collider. We study scalable machine learning models for event reconstruction in electron-positron collisions based on a full detector simulation. Particle-flow reconstruction can be formulated as a supervised learning task using tracks and calorimeter clusters. We compare a graph neural network and kernel-based transformer and demonstrate that we can avoid quadratic operations while achieving realistic reconstruction. We show that hyperparameter tuning significantly improves the performance of the models. The best graph neural network model shows improvement in the jet transverse momentum resolution by up to 50% compared to the rule-based algorithm. The resulting model is portable across Nvidia, AMD and Habana hardware. Accurate and fast machine-learning based reconstruction can significantly improve future measurements at colliders.

21 pages, 10 figures

References in corpus (22)

• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• The anti-k_t jet clustering algorithm
• FastJet user manual
• LLaMA: Open and Efficient Foundation Language Models
• Mamba: Linear-Time Sequence Modeling with Selective State Spaces
• Particle-flow reconstruction and global event description with the CMS detector
• Horovod: fast and easy distributed deep learning in TensorFlow
• FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
• Jet reconstruction and performance using particle flow with the ATLAS Detector
• FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning
• LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
• The Pandora Software Development Kit for Pattern Recognition
• MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks
• Performance of Particle Flow Calorimetry at CLIC
• Object condensation: one-stage grid-free multi-object reconstruction in physics detectors, graph and image data
• Reconstructing particles in jets using set transformer and hypergraph prediction networks
• Report of the Topical Group on Higgs Physics for Snowmass 2021: The Case for Precision Higgs Physics
• Machine Learning for Particle Flow Reconstruction at CMS
• Hyperparameter optimization of data-driven AI models on HPC systems
• Learning Tree Structures from Leaves For Particle Decay Reconstruction
• FAIR AI Models in High Energy Physics
• Pre-training strategy using real particle collision data for event classification in collider physics

Cited by in corpus (5)

• Portable acceleration of CMS computing workflows with coprocessors as a service
• One-to-one correspondence reconstruction at the electron-positron Higgs factory
• Parnassus: An Automated Approach to Accurate, Precise, and Fast Detector Simulation and Reconstruction
• Fine-tuning machine-learned particle-flow reconstruction for new detector geometries in future colliders
• Resource-Adaptive Successive Doubling for Hyperparameter Optimization with Large Datasets on High-Performance Computing Systems