Thomas Klijnsma

Pattern recognition problems in high energy physics are notably different from traditional machine learning applications in computer vision. Reconstruction algorithms identify and measure the kinematic properties of particles produced in high energy collisions and recorded with complex detector systems. Two critical applications are the reconstruction of charged particle trajectories in tracking detectors and the reconstruction of particle showers in calorimeters. These two problems have unique challenges and characteristics, but both have high dimensionality, high degree of sparsity, and complex geometric layouts. Graph Neural Networks (GNNs) are a relatively new class of deep learning architectures which can deal with such data effectively, allowing scientists to incorporate domain knowledge in a graph structure and learn powerful representations leveraging that structure to identify patterns of interest. In this work we demonstrate the applicability of GNNs to these two diverse particle reconstruction problems.

In this community review report, we discuss applications and techniques for fast machine learning (ML) in science-the concept of integrating powerful ML methods into the real-time experimental data processing loop to accelerate scientific discovery. The material for the report builds on two workshops held by the Fast ML for Science community and covers three main areas: applications for fast ML across a number of scientific domains; techniques for training and implementing performant and resource-efficient ML algorithms; and computing architectures, platforms, and technologies for deploying these algorithms. We also present overlapping challenges across the multiple scientific domains where common solutions can be found. This community report is intended to give plenty of examples and inspiration for scientific discovery through integrated and accelerated ML solutions. This is followed by a high-level overview and organization of technical advances, including an abundance of pointers to source material, which can enable these breakthroughs.

The Exa.TrkX project has applied geometric learning concepts such as metric learning and graph neural networks to HEP particle tracking. Exa.TrkX's tracking pipeline groups detector measurements to form track candidates and filters them. The pipeline, originally developed using the TrackML dataset (a simulation of an LHC-inspired tracking detector), has been demonstrated on other detectors, including DUNE Liquid Argon TPC and CMS High-Granularity Calorimeter. This paper documents new developments needed to study the physics and computing performance of the Exa.TrkX pipeline on the full TrackML dataset, a first step towards validating the pipeline using ATLAS and CMS data. The pipeline achieves tracking efficiency and purity similar to production tracking algorithms. Crucially for future HEP applications, the pipeline benefits significantly from GPU acceleration, and its computational requirements scale close to linearly with the number of particles in the event.

Abstract Recently, graph neural networks (GNNs) have been successfully used for a variety of particle reconstruction problems in high energy physics, including particle tracking. The Exa.TrkX pipeline based on GNNs demonstrated promising performance in reconstructing particle tracks in dense environments. It includes five discrete steps: data encoding, graph building, edge filtering, GNN, and track labeling. All steps were written in Python and run on both GPUs and CPUs. In this work, we accelerate the Python implementation of the pipeline through customized and commercial GPU-enabled software libraries, and develop a C++ implementation for inferencing the pipeline. The implementation features an improved, CUDA-enabled fixed-radius nearest neighbor search for graph building and a weakly connected component graph algorithm for track labeling. GNNs and other trained deep learning models are converted to ONNX and inferenced via the ONNX Runtime C++ API. The complete C++ implementation of the pipeline allows integration with existing tracking software. We report the memory usage and average event latency tracking performance of our implementation applied to the TrackML benchmark dataset.

In the next decade, the demands for computing in large scientific experiments are expected to grow tremendously. During the same time period, CPU performance increases will be limited. At the CERN Large Hadron Collider (LHC), these two issues will confront one another as the collider is upgraded for high luminosity running. Alternative processors such as graphics processing units (GPUs) can resolve this confrontation provided that algorithms can be sufficiently accelerated. In many cases, algorithmic speedups are found to be largest through the adoption of deep learning algorithms. We present a comprehensive exploration of the use of GPU-based hardware acceleration for deep learning inference within the data reconstruction workflow of high energy physics. We present several realistic examples and discuss a strategy for the seamless integration of coprocessors so that the LHC can maintain, if not exceed, its current performance throughout its running.

We present a determination of the strong coupling constant αsmZ using inclusive top-quark pair production cross section measurements performed at the LHC and at the Tevatron. Following a procedure first applied by the CMS Collaboration, we extract individual values of αsmZ from measurements by different experiments at several centre-of-mass energies, using QCD predictions complete in NNLO perturbation theory, supplemented with NNLL approximations to all orders, and suitable sets of parton distribution functions. The determinations are then combined using a likelihood-based approach, where special emphasis is put on a consistent treatment of theoretical uncertainties and of correlations between various sources of systematic uncertainties. Our final combined result is αsmZ=0.1177-0.0036+0.0034 .

Abstract We present the current stage of research progress towards a one-pass, completely Machine Learning (ML) based imaging calorimeter reconstruction. The model used is based on Graph Neural Networks (GNNs) and directly analyzes the hits in each HGCAL endcap. The ML algorithm is trained to predict clusters of hits originating from the same incident particle by labeling the hits with the same cluster index. We impose simple criteria to assess whether the hits associated as a cluster by the prediction are matched to those hits resulting from any particular individual incident particles. The algorithm is studied by simulating two tau leptons in each of the two HGCAL endcaps, where each tau may decay according to its measured standard model branching probabilities. The simulation includes the material interaction of the tau decay products which may create additional particles incident upon the calorimeter. Using this varied multiparticle environment we can investigate the application of this reconstruction technique and begin to characterize energy containment and performance.

This document provides a writeup of all contributions to the workshop on "High precision measurements of $α_s$: From LHC to FCC-ee" held at CERN, Oct. 12--13, 2015. The workshop explored in depth the latest developments on the determination of the QCD coupling $α_s$ from 15 methods where high precision measurements are (or will be) available. Those include low-energy observables: (i) lattice QCD, (ii) pion decay factor, (iii) quarkonia and (iv) $τ$ decays, (v) soft parton-to-hadron fragmentation functions, as well as high-energy observables: (vi) global fits of parton distribution functions, (vii) hard parton-to-hadron fragmentation functions, (viii) jets in $e^\pm$p DIS and $γ$-p photoproduction, (ix) photon structure function in $γ$-$γ$, (x) event shapes and (xi) jet cross sections in $e^+e^-$ collisions, (xii) W boson and (xiii) Z boson decays, and (xiv) jets and (xv) top-quark cross sections in proton-(anti)proton collisions. The current status of the theoretical and experimental uncertainties associated to each extraction method, the improvements expected from LHC data in the coming years, and future perspectives achievable in $e^+e^-$ collisions at the Future Circular Collider (FCC-ee) with $\cal{O}$(1--100 ab$^{-1}$) integrated luminosities yielding 10$^{12}$ Z bosons and jets, and 10$^{8}$ W bosons and $τ$ leptons, are thoroughly reviewed. The current uncertainty of the (preliminary) 2015 strong coupling world-average value, $α_s(m_Z)$ = 0.1177 $\pm$ 0.0013, is about 1\%. Some participants believed this may be reduced by a factor of three in the near future by including novel high-precision observables, although this opinion was not universally shared. At the FCC-ee facility, a factor of ten reduction in the $α_s$ uncertainty should be possible, mostly thanks to the huge Z and W data samples available.

The observation of Higgs boson production in association with a top quark-antiquark pair is reported, based on a combined analysis of proton-proton collision data at center-of-mass energies of √s = 7, 8, and 13 TeV, corresponding to integrated luminosities of up to 5.1, 19.7, and 35.9  fb^(-1), respectively. The data were collected with the CMS detector at the CERN LHC. The results of statistically independent searches for Higgs bosons produced in conjunction with a top quark-antiquark pair and decaying to pairs of W bosons, Z bosons, photons, τ leptons, or bottom quark jets are combined to maximize sensitivity. An excess of events is observed, with a significance of 5.2 standard deviations, over the expectation from the background-only hypothesis. The corresponding expected significance from the standard model for a Higgs boson mass of 125.09 GeV is 4.2 standard deviations. The combined best fit signal strength normalized to the standard model prediction is 1.26^(+0.31)_(−0.26).

Motivated by the success of the flavour physics programme carried out over the last decade at the Large Hadron Collider (LHC), we characterize in detail the physics potential of its High-Luminosity and High-Energy upgrades in this domain of physics. We document the extraordinary breadth of the HL/HE-LHC programme enabled by a putative Upgrade II of the dedicated flavour physics experiment LHCb and the evolution of the established flavour physics role of the ATLAS and CMS general purpose experiments. We connect the dedicated flavour physics programme to studies of the top quark, Higgs boson, and direct high-$p_T$ searches for new particles and force carriers. We discuss the complementarity of their discovery potential for physics beyond the Standard Model, affirming the necessity to fully exploit the LHC's flavour physics potential throughout its upgrade eras.

Computing needs for high energy physics are already intensive and are expected to increase drastically in the coming years. In this context, heterogeneous computing, specifically as-a-service computing, has the potential for significant gains over traditional computing models. Although previous studies and packages in the field of heterogeneous computing have focused on GPUs as accelerators, FPGAs are an extremely promising option as well. A series of workflows are developed to establish the performance capabilities of FPGAs as a service. Multiple different devices and a range of algorithms for use in high energy physics are studied. For a small, dense network, the throughput can be improved by an order of magnitude with respect to GPUs as a service. For large convolutional networks, the throughput is found to be comparable to GPUs as a service. This work represents the first open-source FPGAs-as-a-service toolkit.

Abstract The Exa.TrkX project presents a graph neural network (GNN) technique for low-level reconstruction of neutrino interactions in a Liquid Argon Time Projection Chamber (LArTPC). GNNs are still a relatively novel technique, and have shown great promise for similar reconstruction tasks in the Large Hadron Collider (LHC). Graphs describing particle interactions are formed by treating each detector hit as a node, with edges describing the relationships between hits. We utilise a multi-head attention message passing network which performs graph convolutions in order to label each node with a particle type. We present an updated variant of our GNN architecture, with several improvements. After testing the model on more realistic simulation with regions of unresponsive wires, the target was modified from edge classification to node classification in order to increase robustness. Removing edges as a classification target opens up a broader possibility space for edge-forming techniques; we explore the model’s performance across a variety of approaches, such as Delaunay triangulation, kNN, and radius-based methods. We also extend this model to the 3D context, sharing information between detector views. By using reconstructed 3D spacepoints to map detector hits from each wire plane, the model naively constructs 2D representations that are independent yet fully consistent.

Particle tracking is a challenging pattern recognition task at the Large Hadron Collider (LHC) and the High Luminosity-LHC. Conventional algorithms, such as those based on the Kalman Filter, achieve excellent performance in reconstructing the prompt tracks from the collision points. However, they require dedicated configuration and additional computing time to efficiently reconstruct the large radius tracks created away from the collision points. We developed an end-to-end machine learning-based track finding algorithm for the HL-LHC, the Exa.TrkX pipeline. The pipeline is designed so as to be agnostic about global track positions. In this work, we study the performance of the Exa.TrkX pipeline for finding large radius tracks. Trained with all tracks in the event, the pipeline simultaneously reconstructs prompt tracks and large radius tracks with high efficiencies. This new capability offered by the Exa.TrkX pipeline may enable us to search for new physics in real time.

This paper presents a graph neural network (GNN) technique for low-level reconstruction of neutrino interactions in a Liquid Argon Time Projection Chamber (LArTPC). GNNs are still a relatively novel technique, and have shown great promise for similar reconstruction tasks in the LHC. In this paper, a multihead attention message passing network is used to classify the relationship between detector hits by labelling graph edges, determining whether hits were produced by the same underlying particle, and if so, the particle type. The trained model is 84% accurate overall, and performs best on the EM shower and muon track classes. The model’s strengths and weaknesses are discussed, and plans for developing this technique further are summarised.

To address the unprecedented scale of HL-LHC data, the Exa.TrkX project is investigating a variety of machine learning approaches to particle track reconstruction. The most promising of these solutions, graph neural networks (GNN), process the event as a graph that connects track measurements (detector hits corresponding to nodes) with candidate line segments between the hits (corresponding to edges). Detector information can be associated with nodes and edges, enabling a GNN to propagate the embedded parameters around the graph and predict node-, edge- and graph-level observables. Previously, message-passing GNNs have shown success in predicting doublet likelihood, and we here report updates on the state-of-the-art architectures for this task. In addition, the Exa.TrkX project has investigated innovations in both graph construction, and embedded representations, in an effort to achieve fully learned end-to-end track finding. Hence, we present a suite of extensions to the original model, with encouraging results for hitgraph classification. In addition, we explore increased performance by constructing graphs from learned representations which contain non-linear metric structure, allowing for efficient clustering and neighborhood queries of data points. We demonstrate how this framework fits in with both traditional clustering pipelines, and GNN approaches. The embedded graphs feed into high-accuracy doublet and triplet classifiers, or can be used as an end-to-end track classifier by clustering in an embedded space. A set of post-processing methods improve performance with knowledge of the detector physics. Finally, we present numerical results on the TrackML particle tracking challenge dataset, where our framework shows favorable results in both seeding and track finding.

Author(s): Choma, Nicholas; Murnane, Daniel; Ju, Xiangyang; Calafiura, Paolo; Conlon, Sean; Farrell, Steven; Prabhat; Cerati, Giuseppe; Gray, Lindsey; Klijnsma, Thomas; Kowalkowski, Jim; Spentzouris, Panagiotis; Vlimant, Jean-Roch; Spiropulu, Maria; Aurisano, Adam; Hewes, Jeremy; Tsaris, Aristeidis; Terao, Kazuhiro; Usher, Tracy | Abstract: To address the unprecedented scale of HL-LHC data, the Exa.TrkX project is investigating a variety of machine learning approaches to particle track reconstruction. The most promising of these solutions, graph neural networks (GNN), process the event as a graph that connects track measurements (detector hits corresponding to nodes) with candidate line segments between the hits (corresponding to edges). Detector information can be associated with nodes and edges, enabling a GNN to propagate the embedded parameters around the graph and predict node-, edge- and graph-level observables. Previously, message-passing GNNs have shown success in predicting doublet likelihood, and we here report updates on the state-of-the-art architectures for this task. In addition, the Exa.TrkX project has investigated innovations in both graph construction, and embedded representations, in an effort to achieve fully learned end-to-end track finding. Hence, we present a suite of extensions to the original model, with encouraging results for hitgraph classification. In addition, we explore increased performance by constructing graphs from learned representations which contain non-linear metric structure, allowing for efficient clustering and neighborhood queries of data points. We demonstrate how this framework fits in with both traditional clustering pipelines, and GNN approaches. The embedded graphs feed into high-accuracy doublet and triplet classifiers, or can be used as an end-to-end track classifier by clustering in an embedded space. A set of post-processing methods improve performance with knowledge of the detector physics. Finally, we present numerical results on the TrackML particle tracking challenge dataset, where our framework shows favorable results in both seeding and track finding.

[This corrects the article DOI: 10.3389/fdata.2022.787421.].

The Exa.TrkX project has applied geometric learning concepts such as metric learning and graph neural networks to HEP particle tracking. The Exa.TrkX tracking pipeline clusters detector measurements to form track candidates and filters them. The pipeline, originally developed using the TrackML dataset (a simulation of an LHC-like tracking detector), has been demonstrated on various detectors, including the DUNE LArTPC and the CMS High-Granularity Calorimeter. This paper documents new developments needed to study the physics and computing performance of the Exa.TrkX pipeline on the full TrackML dataset, a first step towards validating the pipeline using ATLAS and CMS data. The pipeline achieves tracking efficiency and purity similar to production tracking algorithms. Crucially for future HEP applications, the pipeline benefits significantly from GPU acceleration, and its computational requirements scale close to linearly with the number of particles in the event.

Classifying whole images is a classic problem in machine learning, and graph neural networks are a powerful methodology to learn highly irregular geometries. It is often the case that certain parts of a point cloud are more important than others when determining overall classification. On graph structures this started by pooling information at the end of convolutional filters, and has evolved to a variety of staged pooling techniques on static graphs. In this paper, a dynamic graph formulation of pooling is introduced that removes the need for predetermined graph structure. It achieves this by dynamically learning the most important relationships between data via an intermediate clustering. The network architecture yields interesting results considering representation size and efficiency. It also adapts easily to a large number of tasks from image classification to energy regression in high energy particle physics.

Classifying whole images is a classic problem in machine learning, and graph neural networks are a powerful methodology to learn highly irregular geometries. It is often the case that certain parts of a point cloud are more important than others when determining overall classification. On graph structures this started by pooling information at the end of convolutional filters, and has evolved to a variety of staged pooling techniques on static graphs. In this paper, a dynamic graph formulation of pooling is introduced that removes the need for predetermined graph structure. It achieves this by dynamically learning the most important relationships between data via an intermediate clustering. The network architecture yields interesting results considering representation size and efficiency. It also adapts easily to a large number of tasks from image classification to energy regression in high energy particle physics.

Computing needs for high energy physics are already intensive and are expected to increase drastically in the coming years. In this context, heterogeneous computing, specifically as-a-service computing, has the potential for significant gains over traditional computing models. Although previous studies and packages in the field of heterogeneous computing have focused on GPUs as accelerators, FPGAs are an extremely promising option as well. A series of workflows are developed to establish the performance capabilities of FPGAs as a service. Multiple different devices and a range of algorithms for use in high energy physics are studied. For a small, dense network, the throughput can be improved by an order of magnitude with respect to GPUs as a service. For large convolutional networks, the throughput is found to be comparable to GPUs as a service. This work represents the first open-source FPGAs-as-a-service toolkit.

This report describes the measurements performed on Bi1:5Sb0:5Te1:8Se1:2 (BSTS). The goal of this research is to determine the qualities of BSTS as a topological insulator. A topological insulator is a material that is insulating in the bulk, but conducting on the surface. For BSTS, the magnetoresistance and Hall resistance are researched as a function of temperature and magnetic field, and various other properties as a function of temperature. By characterizing the weak antilocalization in BSTS under a variety of angles and the use of very thin flakes, it is likely that surface states are observed.

Five measurements of the top quark pair production cross section from hadron colliders are compared with the theoretical prediction (up the next-to-next-to-leading order in QCD) in order to determine the strong coupling constant αs. The determination is performed for several PDF sets. The individual determinations of αs are combined to form a single best estimate using the Best Linear Unbiased Estimate (BLUE) procedure. Preliminary combined results indicate a higher value of αs than that determined in a previous extraction of αs from the top quark pair production cross section.

Recently, graph neural networks (GNNs) have been successfully used for a variety of particle reconstruction problems in high energy physics, including particle tracking. The Exa.TrkX pipeline based on GNNs demonstrated promising performance in reconstructing particle tracks in dense environments. It includes five discrete steps: data encoding, graph building, edge filtering, GNN, and track labeling. All steps were written in Python and run on both GPUs and CPUs. In this work, we accelerate the Python implementation of the pipeline through customized and commercial GPU-enabled software libraries, and develop a C++ implementation for inferencing the pipeline. The implementation features an improved, CUDA-enabled fixed-radius nearest neighbor search for graph building and a weakly connected component graph algorithm for track labeling. GNNs and other trained deep learning models are converted to ONNX and inferenced via the ONNX Runtime C++ API. The complete C++ implementation of the pipeline allows integration with existing tracking software. We report the memory usage and average event latency tracking performance of our implementation applied to the TrackML benchmark dataset.

We present the current stage of research progress towards a one-pass, completely Machine Learning (ML) based imaging calorimeter reconstruction. The model used is based on Graph Neural Networks (GNNs) and directly analyzes the hits in each HGCAL endcap. The ML algorithm is trained to predict clusters of hits originating from the same incident particle by labeling the hits with the same cluster index. We impose simple criteria to assess whether the hits associated as a cluster by the prediction are matched to those hits resulting from any particular individual incident particles. The algorithm is studied by simulating two tau leptons in each of the two HGCAL endcaps, where each tau may decay according to its measured standard model branching probabilities. The simulation includes the material interaction of the tau decay products which may create additional particles incident upon the calorimeter. Using this varied multiparticle environment we can investigate the application of this reconstruction technique and begin to characterize energy containment and performance.

Accelerated Machine Learning as a Service for Particle Physics Computing: • Amount and complexity of high-energy physics data increases dramatically from 2025 onward • Traditional algorithms will require too much CPU time • Machine learning can solve combinatorially-scaling problems in constant time, but must be fast enough

The Exa.TrkX project has applied geometric learning concepts such as metric learning and graph neural networks to HEP particle tracking. Exa.TrkX's tracking pipeline groups detector measurements to form track candidates and filters them. The pipeline, originally developed using the TrackML dataset (a simulation of an LHC-inspired tracking detector), has been demonstrated on other detectors, including DUNE Liquid Argon TPC and CMS High-Granularity Calorimeter. This paper documents new developments needed to study the physics and computing performance of the Exa.TrkX pipeline on the full TrackML dataset, a first step towards validating the pipeline using ATLAS and CMS data. The pipeline achieves tracking efficiency and purity similar to production tracking algorithms. Crucially for future HEP applications, the pipeline benefits significantly from GPU acceleration, and its computational requirements scale close to linearly with the number of particles in the event.

In this community review report, we discuss applications and techniques for fast machine learning (ML) in science -- the concept of integrating power ML methods into the real-time experimental data processing loop to accelerate scientific discovery. The material for the report builds on two workshops held by the Fast ML for Science community and covers three main areas: applications for fast ML across a number of scientific domains; techniques for training and implementing performant and resource-efficient ML algorithms; and computing architectures, platforms, and technologies for deploying these algorithms. We also present overlapping challenges across the multiple scientific domains where common solutions can be found. This community report is intended to give plenty of examples and inspiration for scientific discovery through integrated and accelerated ML solutions. This is followed by a high-level overview and organization of technical advances, including an abundance of pointers to source material, which can enable these breakthroughs.