The upcoming upgrades at the LHC have fueled increasing interest in alternative highly-parallel and GPU friendly algorithms for tracking and reconstruction. The PV-Finder project is developing a novel prototype algorithm for vertexing in high density collisions using a Convolutional Neural Network (CNN).

The PV-Finder algorithm uses a custom kernel to transform the sparse 3D space of hits and tracks into a dense 1D dataset, and then Deep Learning techniques are used to find PV locations. By training networks on our kernels using several CNN layers, we have achieved better than 90% efficiency with no more than 0.2 False Positives (FPs) per event. Beyond its physics performance, this algorithm also provides a rich collection of possibilities for visualization and study of 1D convolutional networks.

The current version of PV-Finder is based on a toy simulation of the LHCb detector in Run 3 conditions. We are breaking out the kernel generation, to allow the algorithm to be run on different inputs, such as the official LHCb framework, ATLAS or ACTS, and CMS track output.

The code currently lives at gitlab.cern.ch/LHCb-Reco-Dev/pv-finder.

Team

• Henry Schreiner
• Mike Sokoloff
• Mike Williams
• Simon Akar
• Marian Stahl
• Rui Fang
• Gowtham Atluri
• Kendrick Li

Presentations

• 5 Apr 2024 - "New Deep Learning based approach to Primary Vertex finding in ATLAS experiment", Rocky Bala Garg, APS April Meeting, 2024
• 1 Feb 2024 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC (poster)", Simon Akar, 6th Inter-experiment Machine Learning Workshop
• 30 Jan 2024 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Simon Akar, 6th Inter-experiment Machine Learning Workshop
• 30 Jan 2024 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Rocky Bala Garg, 6th Inter-experiment Machine Learning Workshop
• 22 Sep 2023 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Mike Sokoloff, seminar at Ramon LLull University, LaSalle Campus
• 9 May 2023 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Rocky Bala Garg, CHEP 2023
• 9 May 2023 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Mike Sokoloff, CHEP-2023
• 16 Apr 2023 - "PV-Finder Approval: A Deep Learning based Approach to finding Primary Vertices in ATLAS data", Rocky Bala Garg, Inner Tracking CP Plenary
• 9 Mar 2023 - "Update on PV-Finder Study: A Deep Learning based Approach to finding Primary Vertices in ATLAS data", Rocky Bala Garg, Tracking and Vertexing for Prompt and Displaced Particles Meeting, ATLAS
• 4 Nov 2022 - "Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC", Mike Sokoloff, ML4Jets2022
• 25 Oct 2022 - "Comparing and improving hybrid deep learning algorithms for identifying and locating primary vertices (poster)", Simon Akar, ACAT 2022 (21st International Workshop on Advanced Computing and Analysis Techniques in Physics Research)
• 13 Oct 2021 - "Implementation of Deep Learning Algorithm for identifying and locating primary vertices in ATLAS data", Rocky Bala Garg, DIANA/HEP Fellow Presentations
• 19 May 2021 - "Progress in developing a hybrid deep learning algorithm for identifying and locating primary vertices", Simon Akar, CHEP 2021 (25th International Conference on Computing in High-Energy and Nuclear Physics)
• 10 May 2021 - "Progress in developing a hybrid deep learning algorithm for identifying and locating primary vertices, Performances evaluation & comparison", Simon Akar, Internal LHCb RTA-DPA meeting
• 29 Mar 2021 - "Progress in developing a hybrid deep learning algorithm for identifying and locating primary vertices, Learning KDEs and PVs from track parameters", Simon Akar, Internal LHCb RTA-DPA meeting
• 20 Apr 2020 - "An updated hybrid deep learning algorithm for identifying and locating primary vertices", Marian Stahl, Connecting The Dots 2020
• 29 Jan 2020 - "LHCb HLT performance- and regression-tests A hybrid deep learning approach to vertexing", Marian Stahl, IRIS-HEP Topical Meeting
• 17 Apr 2019 - "A hybrid deep learning approach to vertexing", Henry Schreiner, 3rd IML Machine Learning Workshop
• 3 Apr 2019 - "A hybrid deep learning approach to vertexing", Henry Schreiner, Connecting The Dots and Workshop on Intelligent Trackers 2019
• 20 Mar 2019 - "Machine Learning for the Primary Vertex reconstruction", Henry Schreiner, 2019 Joint HSF/OSG/WLCG Workshop
• 11 Mar 2019 - "A hybrid deep learning approach to vertexing", Henry Schreiner, 19th International Workshop on Advanced Computing and Analysis Techniques in Physics Research

Publications

• Advances in developing deep neural networks for finding primary vertices in proton-proton collisions at the LHC, S. Akar, M. Elashri, R. Garg, E. Kauffman, M. Peters, H. Schreiner, M. Sokoloff, W. Tepe and L. Tompkins, arXiv 2309.12417 (21 Sep 2023).
• Comparing and improving hybrid deep learning algorithms for identifying and locating primary vertices, S. Akar, M. Peters, H. Schreiner, M. Sokoloff and W. Tepe, arXiv 2304.02423 (05 Apr 2023) [1 citation].
• Progress in developing a hybrid deep learning algorithm for identifying and locating primary vertices, S. Akar, G. Atluri, T. Boettcher, M. Peters, H. Schreiner, M. Sokoloff, M. Stahl, W. Tepe, C. Weisser and M. Williams, EPJ Web Conf. 251 04012 (2021) (08 Mar 2021) [3 citations] [NSF PAR].
• An updated hybrid deep learning algorithm for identifying and locating primary vertices, S. Akar, T. J. Boettcher, S. Carl, H. F. Schreiner, M. D. Sokoloff, M. Stahl, C. Weisser, M. Williams, arXiv:2007.01023 [physics.ins-det] (Submitted to CTD2020) (02 Jul 2020).
• A hybrid deep learning approach to vertexing, R. Fang, H. Schreiner, M. Sokoloff, C. Weisser and M. Williams, J.Phys.Conf.Ser. 1525 012079 (2020) (19 Jun 2019) [5 citations] [NSF PAR].

Recent recordings