Skip to main content
OSTI.GOVU.S. Department of Energy Office of Scientific and Technical Information
U.S. Department of Energy
Office of Scientific and Technical Information
 

Distance-Weighted Graph Neural Networks on FPGAs for Real-Time Particle Reconstruction in High Energy Physics

Journal Article · · Frontiers in Big Data
; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; more »; ; « less
Graph neural networks have been shown to achieve excellent performance for several crucial tasks in particle physics, such as charged particle tracking, jet tagging, and clustering. An important domain for the application of these networks is the FGPA-based first layer of real-time data filtering at the CERN Large Hadron Collider, which has strict latency and resource constraints. We discuss how to design distance-weighted graph networks that can be executed with a latency of less than one μs on an FPGA. To do so, we consider a representative task associated to particle reconstruction and identification in a next-generation calorimeter operating at a particle collider. We use a graph network architecture developed for such purposes, and apply additional simplifications to match the computing constraints of Level-1 trigger systems, including weight quantization. Using the hls4ml library, we convert the compressed models into firmware to be implemented on an FPGA. Performance of the synthesized models is presented both in terms of inference accuracy and resource usage.
Research Organization:
Fermi National Accelerator Laboratory (FNAL), Batavia, IL (United States); Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
Sponsoring Organization:
USDOE Office of Science (SC), High Energy Physics (HEP)
Contributing Organization:
CERN Collaboration
Grant/Contract Number:
AC02-07CH11359
OSTI ID:
1661684
Report Number(s):
FERMILAB-PUB--20-405-E-SCD; arXiv:2008.03601; oai:inspirehep.net:1810883
Journal Information:
Frontiers in Big Data, Journal Name: Frontiers in Big Data Vol. 3; ISSN 2624-909X
Publisher:
FrontiersCopyright Statement
Country of Publication:
United States
Language:
English

References (26)

Probing stop pair production at the LHC with graph neural networks journal August 2019
Object condensation: one-stage grid-free multi-object reconstruction in physics detectors, graph, and image data journal September 2020
Xilinx Vitis Unified Software Platform
  • Kathail, Vinod
  • FPGA '20: The 2020 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, Proceedings of the 2020 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays https://doi.org/10.1145/3373087.3375887
conference February 2020
Deep Learning using Rectified Linear Units (ReLU) preprint January 2018
Learning representations of irregular particle-detector geometry with distance-weighted graph networks text January 2019
A Dynamic Reduction Network for Point Clouds preprint January 2020
Simulation of an imaging calorimeter to demonstrate GarNet on FPGA dataset January 2020
Dataset: Object condensation: one-stage grid-free multi-object reconstruction in physics detectors, graph, and image data dataset January 2020
Probing stop pair production at the LHC with graph neural networks journal August 2019
A Survey on Graph Processing Accelerators: Challenges and Opportunities journal March 2019
Performance of the ALEPH detector at LEP
  • Buskulic, D.; Casper, D.; De Bonis, I.
  • Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Vol. 360, Issue 3 https://doi.org/10.1016/0168-9002(95)00138-7
journal June 1995
Geant4—a simulation toolkit
  • Agostinelli, S.; Allison, J.; Amako, K.
  • Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Vol. 506, Issue 3 https://doi.org/10.1016/S0168-9002(03)01368-8
journal July 2003
Array programming with NumPy journal September 2020
Particle-flow reconstruction and global event description with the CMS detector journal October 2017
Fast inference of deep neural networks in FPGAs for particle physics journal July 2018
Fast inference of Boosted Decision Trees in FPGAs for particle physics journal May 2020
Compressing deep neural networks on FPGAs to binary and ternary precision with hls4ml journal December 2020
Jet tagging via particle clouds journal March 2020
Interaction networks for the identification of boosted H → b b ¯ decays journal July 2020
The NumPy Array: A Structure for Efficient Numerical Computation journal March 2011
Jet reconstruction and performance using particle flow with the ATLAS Detector journal July 2017
Learning representations of irregular particle-detector geometry with distance-weighted graph networks journal July 2019
JEDI-net: a jet identification algorithm based on interaction networks journal January 2020
Pileup mitigation at the Large Hadron Collider with graph neural networks journal July 2019
Energy efficient architecture for graph analytics accelerators journal October 2016
Dynamic Graph CNN for Learning on Point Clouds journal October 2019

Similar Records

Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs
Conference · Sun Nov 29 23:00:00 EST 2020 · OSTI ID:1823535
End-to-end codesign of Hessian-aware quantized neural networks for FPGAs
Journal Article · Mon Jun 24 20:00:00 EDT 2024 · ACM Transactions on Reconfigurable Technology and Systems · OSTI ID:1988435
Graph Neural Networks for Charged Particle Tracking on FPGAs
Journal Article · Tue Mar 22 20:00:00 EDT 2022 · Frontiers in Big Data · OSTI ID:1859670

Related Subjects

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS
deep learning
fast inference
field-programmable gate arrays
graph network
imaging calorimeter