The ATLAS Collaboration

\AtlasOrcid · 2025

Read The ATLAS Collaboration by \AtlasOrcid (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-19689.txt.

Introduction

The ATLAS Collaboration presents, in 2025 work, a new transformer-based tool for telling which kind of quark a spray of particles came from. Such tagging is a core need at the Large Hadron Collider, where sprays from heavy quarks mark key events. Past taggers read built traits; the new one reads raw track data end to end.

Extra training goals, drawn from physics, keep the tool's choices readable. Checked on fake and real crash data, it shuns charm sprays three and a half times better than the old tool. Light sprays are shunned near twice better, at the same catch rate for bottom sprays. Gains feed hunts for paired Higgs births and for quark-Higgs links.

Catalog record

CREATOR\AtlasOrcid
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-19689

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The ATLAS Collaboration \AtlasOrcid Abstract \AtlasTitle Transforming jet flavour tagging at ATLAS \AtlasAbstract Jet flavour tagging enables the identification of jets originating from heavy-flavour quarks in proton–proton collisions at the Large Hadron Collider, playing a critical role in its physics programmes. This paper presents GN2, a novel transformer-based flavour tagging algorithm deployed by the ATLAS Collaboration that represents a paradigm shift from previous approaches. Designed to classify jets based on the flavour of their constituent particles, GN2 processes low-level tracking information in an end-to-end architecture and incorporates physics-informed auxiliary training objectives to enhance both interpretability and performance. Its performance is validated in both simulation and collision data. The GN2 algorithm provides substantial benefits for physics analyses involving heavy-flavour jets, such as measurements of Higgs boson pair production and the couplings of bottom and charm quarks to the Higgs boson, and demonstrates the impact of advanced machine learning methods in experimental particle physics. \AtlasRefCode FTAG-2023-05 \PreprintIdNumber CERN-EP-2025-103 \AtlasJournal Nature Communications The Large Hadron Collider (LHC) [ 1 ] is the world’s most powerful particle collider. It is used to extend the boundaries of our understanding of fundamental particles and their interactions. It offers a unique opportunity to test the Standard Model (SM) of particle physics, as well as search for new phenomena beyond the Standard Model (BSM). The demanding experimental conditions at the LHC necessitate continuous innovation by the main experiments, pushing them to apply cutting-edge technologies to efficiently identify physics processes of interest within the largest proton–proton ( pp ) collision dataset ever recorded. Hadronic jets are the most abundant physics objects in pp collision events, and their characteristics are widely utilised in data analyses. The flavour of a hadronic jet is determined by the types of hadrons or leptons it contains. Flavour tagging concerns the classification of hadronic jets into those containing b\text{-hadrons} ( b\text{-jets} ), c\text{-hadrons} ( c\text{-jets} ), hadronic \tau\text{-lepton} decays ( \tau\text{-jets} ), and neither of the above (light-jets), using algorithms sensitive to the distinctive properties of the respective classes. Since the beginning of Run 1 of the LHC (2009–2013), the ATLAS experiment [ 2 , 3 ] has achieved continuous improvement in the performance of these algorithms. The progress has mostly been driven by the integration of machine-learning techniques, including boosted decision trees and neural networks. The state-of-the-art algorithm used thus far to analyse the data at \sqrt{s}=$13\text{\,}\mathrm{TeV}$ from Run 2 of the LHC (2015–2018) [ 4 , 5 ] led to very impactful physics results such as the observations of the Higgs boson decaying to bottom quarks [ 6 ] and its production in association with a pair of top quarks [ 7 ] . Flavour tagging plays an essential role in the comprehensive research programme of ATLAS, which includes precision measurements of the Higgs boson [ 8 ] , top quark [ 9 ] and other SM processes [ 10 ] , as well as the searches for supersymmetry [ 11 ] and other BSM phenomena [ 12 ] . This work describes a novel flavour tagging algorithm developed by the ATLAS Collaboration for the analysis of data from pp collisions recorded during Run 2 (2015–2018) and Run 3 (2022–2026) of the LHC at centre-of-mass energies of \sqrt{s}=$13\text{\,}\mathrm{TeV}$ and \sqrt{s}=$13.6\text{\,}\mathrm{TeV}$ . Flavour-tagging techniques rely on the long lifetime, high mass, high decay multiplicity and characteristic decay modes of b - and c\text{-hadrons} , and the properties of heavy-quark fragmentation [ 13 ] . The typical lifetime of the order of \tau\approx$1.5\text{\,}\mathrm{ps}$ [ 13 , 14 , 15 ] for b\text{-hadrons} in jets with transverse momenta in the range from tens to hundreds of \mathrm{G}\mathrm{eV} results in them travelling a mean flight length \langle l\rangle=\beta\gamma c\tau in the range from few millimetres to centimetres before decaying, which often leads to a secondary vertex significantly displaced from the collision point. The majority of b\text{-jets} also contain a tertiary vertex from the decay of the c\text{-hadron} produced in the b\text{-hadron} decay. The traditional flavour-tagging algorithms developed by the ATLAS Collaboration are based on a two-stage approach [ 16 , 4 , 5 ] . In the first step, specialised low-level algorithms employ complementary approaches to extract information from the trajectories of the charged-particle constituents ( ‘tracks’ ) associated with the jet. These specialised algorithms either rely on the properties of individual tracks or leverage their correlations with properties of other tracks to explicitly reconstruct displaced vertices. In the second step, the output of low-level algorithms are subsequently combined in a high-level multivariate classifier to maximise performance. The most recent algorithm employed by the ATLAS Collaboration, following this paradigm, is a deep neural network ( ‘DL1d’ ) that leverages a low-level track-based algorithm ( ‘DIPS’ ) [ 17 ] based on Deep Sets [ 18 ] . DL1d has already substantially improved the performance relative to the algorithm used in most published Run-2 physics analyses [ 5 ] . GN2 presents a novel approach, illustrated in Figure 1 , which breaks the design paradigm of the two-staged algorithms. A single Transformer model [ 19 ] that directly processes track and jet information is trained with target labels extracted from the Monte Carlo (MC) simulation. In addition to the primary training target, jet flavour prediction, auxiliary training objectives are introduced to reconstruct the internal structure of a jet by grouping tracks originating from a common vertex and predicting the underlying physics process from which a track originated. Such physics domain knowledge is embedded in a combined loss function that enables a simultaneous optimisation, instead of relying on manually optimised low-level algorithms. This flexible structure allows the swift re-tuning of GN2 to suit alternative experimental conditions or physics goals. Details of the algorithm architectures are summarised in the Methods section, together with descriptions of the ATLAS detector, simulation samples, physics objects, and analysis strategies. Figure 1 : Illustration of the GN2 algorithm with jet and track input variables, discriminating between jet flavours by exploiting secondary vertices and other properties stemming from the displaced decays of b -hadrons. The jet features are copied for each track associated with the jet. The combined vectors are then fed into a per-track initialisation network, followed by a transformer encoder and a global representation of the jet. n_{\mathrm{jf}} ( n_{\mathrm{tf}} ) corresponds to the number of jet (track) features. The pooled jet representation and output track embeddings are provided as inputs to the three task-specific networks. Details of the GN2 architecture are summarised in the Methods section. GN2 achieves a remarkable performance boost compared with the DL1d algorithm, with improvements by a factor of 1.5–4 observed in its major experimental applications. The deployment of GN2 should greatly enhance the physics reach of ATLAS in flagship analyses, such as the search for Higgs pair production and the c -quark Yukawa coupling measurement, for which the projected sensitivity at the High Luminosity LHC is improved by up to 30% [ 20 ] . These improvements do not come with a strong dependence on the choice and configuration of the MC event generator, and are confirmed by measured performance in recorded collisions. The innovative auxiliary training objectives bring excellent interpretability and opens up new avenues for future applications. To facilitate future developments and strengthen the connections between

Read the complete text