High performance FPGA embedded system for machine learning based tracking and trigger in sPhenix and EIC
Abstract
We present a comprehensive end-to-end pipeline to classify triggers versus background events in this paper. This pipeline makes online decisions to select signal data and enables the intelligent trigger system for efficient data collection in the Data Acquisition System (DAQ) of the upcoming sPHENIX and future EIC (Electron-Ion Collider) experiments. Starting from the coordinates of pixel hits that are lightened by passing particles in the detector, the pipeline applies three-stage of event processing (hits clustering, track reconstruction, and trigger detection) and labels all processed events with the binary tag of trigger versus background events. The pipeline consists of deterministic algorithms such as clustering pixels to reduce event size, tracking reconstruction to predict candidate edges, and advanced graph neural network-based models for recognizing the entire jet pattern. In particular, we apply the message-passing graph neural network to predict links between hits and reconstruct tracks and a hierarchical pooling algorithm (DiffPool) to make the graph-level trigger detection. We obtain an impressive performance (≥70% accuracy) for trigger detection with only 3200 neuron weights in the end-to-end pipeline. We deploy the end-to-end pipeline into a field-programmable gate array (FPGA) and accelerate the three stages with speedup factors of 1152, 280, and 21, respectively.
- Authors:
- Publication Date:
- Research Org.:
- Sunrise Technology, Inc., Stonybrook, NY (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC)
- OSTI Identifier:
- 1979437
- DOE Contract Number:
- SC0019518
- Resource Type:
- Journal Article
- Journal Name:
- Journal of Instrumentation
- Additional Journal Information:
- Journal Volume: 17; Journal Issue: 07; Journal ID: ISSN 1748-0221
- Publisher:
- Institute of Physics (IOP)
- Country of Publication:
- United States
- Language:
- English
- Subject:
- Instruments & Instrumentation
Citation Formats
Xuan, T., Durao, F., and Sun, Y. High performance FPGA embedded system for machine learning based tracking and trigger in sPhenix and EIC. United States: N. p., 2022.
Web. doi:10.1088/1748-0221/17/07/c07003.
Xuan, T., Durao, F., & Sun, Y. High performance FPGA embedded system for machine learning based tracking and trigger in sPhenix and EIC. United States. https://doi.org/10.1088/1748-0221/17/07/c07003
Xuan, T., Durao, F., and Sun, Y. 2022.
"High performance FPGA embedded system for machine learning based tracking and trigger in sPhenix and EIC". United States. https://doi.org/10.1088/1748-0221/17/07/c07003.
@article{osti_1979437,
title = {High performance FPGA embedded system for machine learning based tracking and trigger in sPhenix and EIC},
author = {Xuan, T. and Durao, F. and Sun, Y.},
abstractNote = {We present a comprehensive end-to-end pipeline to classify triggers versus background events in this paper. This pipeline makes online decisions to select signal data and enables the intelligent trigger system for efficient data collection in the Data Acquisition System (DAQ) of the upcoming sPHENIX and future EIC (Electron-Ion Collider) experiments. Starting from the coordinates of pixel hits that are lightened by passing particles in the detector, the pipeline applies three-stage of event processing (hits clustering, track reconstruction, and trigger detection) and labels all processed events with the binary tag of trigger versus background events. The pipeline consists of deterministic algorithms such as clustering pixels to reduce event size, tracking reconstruction to predict candidate edges, and advanced graph neural network-based models for recognizing the entire jet pattern. In particular, we apply the message-passing graph neural network to predict links between hits and reconstruct tracks and a hierarchical pooling algorithm (DiffPool) to make the graph-level trigger detection. We obtain an impressive performance (≥70% accuracy) for trigger detection with only 3200 neuron weights in the end-to-end pipeline. We deploy the end-to-end pipeline into a field-programmable gate array (FPGA) and accelerate the three stages with speedup factors of 1152, 280, and 21, respectively.},
doi = {10.1088/1748-0221/17/07/c07003},
url = {https://www.osti.gov/biblio/1979437},
journal = {Journal of Instrumentation},
issn = {1748-0221},
number = 07,
volume = 17,
place = {United States},
year = {Fri Jul 01 00:00:00 EDT 2022},
month = {Fri Jul 01 00:00:00 EDT 2022}
}
Works referenced in this record:
Performance of a geometric deep learning pipeline for HL-LHC particle tracking
journal, October 2021
- Ju, Xiangyang; Murnane, Daniel; Calafiura, Paolo
- The European Physical Journal C, Vol. 81, Issue 10
Searching for exotic particles in high-energy physics with deep learning
journal, July 2014
- Baldi, P.; Sadowski, P.; Whiteson, D.
- Nature Communications, Vol. 5, Issue 1
Boosted Decision Trees in the Level-1 Muon Endcap Trigger at CMS
journal, September 2018
- Acosta, Darin; Brinkerhoff, Andrew; Busch, Elena
- Journal of Physics: Conference Series, Vol. 1085
Charged Particle Tracking via Edge-Classifying Interaction Networks
journal, November 2021
- DeZoort, Gage; Thais, Savannah; Duarte, Javier
- Computing and Software for Big Science, Vol. 5, Issue 1
Deep Learning and Its Application to LHC Physics
journal, October 2018
- Guest, Dan; Cranmer, Kyle; Whiteson, Daniel
- Annual Review of Nuclear and Particle Science, Vol. 68, Issue 1
Operation of the ATLAS trigger system in Run 2
journal, October 2020
- collaboration, The ATLAS
- Journal of Instrumentation, Vol. 15, Issue 10
Fast inference of deep neural networks in FPGAs for particle physics
journal, July 2018
- Duarte, J.; Han, S.; Harris, P.
- Journal of Instrumentation, Vol. 13, Issue 07
Distance-Weighted Graph Neural Networks on FPGAs for Real-Time Particle Reconstruction in High Energy Physics
journal, January 2021
- Iiyama, Yutaro; Cerminara, Gianluca; Gupta, Abhijay
- Frontiers in Big Data, Vol. 3