arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.01876hep-excs.LGphysics.ins-det
收藏

面向HL-LHC的基于FPGA的超快喷注分类

Ultrafast jet classification on FPGAs for the HL-LHC

ETH Zürich(苏黎世联邦理工学院);Imperial College London(伦敦帝国学院);University of California San Diego(加州大学圣迭戈分校);Universität Hamburg(汉堡大学);Massachusetts Institute of Technology(麻省理工学院);Institute of Physics Belgrade(贝尔格莱德物理研究所);Fermi National Accelerator Laboratory(费米国家加速器实验室);European Organization for Nuclear Research (CERN)(欧洲核子研究组织(CERN));Universität Göttingen(哥廷根大学);Uppsala Universitet(乌普萨拉大学);Universidade do Estado do Rio de Janeiro (UERJ)(里约热内卢州立大学);University of Southern California(南加州大学)
Patrick Odagiu, Zhiqiang Que, Javier Duarte, Johannes Haller, Gregor Kasieczka, Artur Lobanov, Vladimir Loncar, Wayne Luk, Jennifer Ngadiuba, Maurizio Pierini, … 展开作者

Patrick Odagiu, Zhiqiang Que, Javier Duarte, Johannes Haller, Gregor Kasieczka, Artur Lobanov, Vladimir Loncar, Wayne Luk, Jennifer Ngadiuba, Maurizio Pierini, Philipp Rincke, Arpita Seksaria, Sioni Summers, Andre Sznajder, Alexander Tapper, Thea K. Aarrestad

AI总结:

本文针对HL-LHC提出三种面向FPGA部署的喷注起源分类机器学习模型,通过量化感知训练和高效综合,以较低资源成本实现Deep Sets和Interaction Networks等复杂架构约百纳秒级推理。

摘要:

使用三种机器学习模型进行喷注起源分类。这些模型经过优化,可部署在现场可编程门阵列(FPGA)设备上。在此背景下,我们展示了延迟和资源消耗如何随输入规模和算法选择而变化。此外,本文提出的模型旨在处理欧洲核子研究中心大型强子对撞机(CERN LHC)在高亮度阶段所预期的那类数据和条件。通过量化感知训练和针对特定现场可编程门阵列的高效综合,我们证明,以相对较低的计算资源成本,即可实现诸如Deep Sets和Interaction Networks等复杂架构的O(100) ns量级推理。

英文摘要:

Three machine learning models are used to perform jet origin classification. These models are optimized for deployment on a field-programmable gate array device. In this context, we demonstrate how latency and resource consumption scale with the input size and choice of algorithm. Moreover, the models proposed here are designed to work on the type of data and under the foreseen conditions at the CERN LHC during its high-luminosity phase. Through quantization-aware training and efficient synthetization for a specific field programmable gate array, we show that $O(100)$ ns inference of complex architectures such as Deep Sets and Interaction Networks is feasible at a relatively low computational resource cost.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑