BOOST 2022 is the fourteenth conference of a series of successful joint theory/experiment workshops that bring together the world's leading experts from theory and LHC experiments. Our aim are lively discussions about the latest progress in this field and the exchange of expertise to generate new ideas, in order develop new approaches on the reconstruction and use of boosted decay topologies in particle physics and beyond.
This year is hosted by the University of Hamburg, Germany. We plan for the conference to be in person.
For people with travel restrictions, we offer online participation via zoom. For online participation, please register choosing the option “online-only participation” under “Conference dinner on Wednesday” without fee.
The conference will cover the following topics:
Previous BOOST editions:
Abstract submissions possible until: 15 June 2022
Contribution acceptance notification no later than: 1 July 2022
Early bird registration until: 15 July 2022
Registration deadline: 1 August 2022
Conference dates: 15-19 August 2022
We present results of searches for massive vector-like third-generation quark and lepton partners using proton-proton collision data collected with the CMS detector at the CERN LHC at a center-of-mass energy of 13 TeV. Pair production of vector-like leptons is studied, with decays into final states, containing third generation quarks and leptons. Vector-like quarks are studied in both single and pair production, considering final states, containing top and bottom quarks, electroweak gauge and Higgs bosons. We search using several categories of reconstructed objects, from multi-leptonic to fully hadronic final states. We set exclusion limits on both the vector-like particle mass and cross sections, for combinations of the vector-like particle branching ratios.
A summary of searches for heavy resonances with masses exceeding 1 TeV decaying into pairs or triplets of bosons is presented, performed on data produced by LHC pp collisions at $\sqrt{s}$ = 13 TeV and collected with the CMS detector during 2016, 2017, and 2018. The common feature of these analyses is the boosted topology, namely the decay products of the considered bosons (both electroweak W, Z bosons and the Higgs boson) are expected to be highly energetic and close in angle, leading to a non-trivial identification of the quarks and leptons in the final state. The exploitation of jet substructure techniques allows to increase the sensitivity of the searches where at least one boson decays hadronically. Various background estimation techniques are adopted, based on data-MC hybrid approaches or relying only in control regions in data. Results are interpreted in the context of the Warped Extra Dimension and Heavy Vector Triplet theoretical models, two possible scenarios beyond the standard model.
Many new physics models predict the existence of Higgs-like particles decaying into two bosons (W, Z, photon, or Higgs bosons) making these important signatures in the search for new physics. Searches for Vy, VV, and VH resonances have been performed in various final states. In some of these searches, jet substructure techniques are used to disentangle the hadronic decay products in highly boosted configurations. This talk summarises recent ATLAS searches with Run 2 data collected at the LHC and explains the experimental methods used, including vector- and Higgs-boson-tagging techniques.
Many new-physics signatures at the LHC produce highly boosted particles, leading to close-by objects in the detector and necessitating jet substructure techniques to disentangle the hadronic decay products. This talk will illustrate the use of these techniques in recent ATLAS searches for heavy W' and Z' resonances in top-bottom and di-top final states, as well as in searches for vector-like quarks or dark matter, using the full Run 2 dataset.
The LHC has unlocked a previously unexplored energy regime. Dedicated techniques have been developed to reconstruct and identify boosted top quarks. Measurements in boosted top quark production test the Standard Model in a region with a strongly enhanced sensitivity to high-scale new phenomena. In this contribution, several new measurements of the ATLAS experiment are presented of the differential cross section and asymmetries in this extreme kinematic regime. The measurements are based on the complete 139/fb run-2 data set of proton-proton collisions at 13 TeV collected in run 2 of the LHC. The measurements are interpreted within the Standard Model Effective Field Theory, yielding stringent bounds on the Wilson coefficients of two-light-quark-two-quark operator.
Tensor Networks (TN) are approximations of high-dimensional tensors designed to represent locally entangled quantum many-body systems efficiently. In this talk, we will discuss how to use TN to connect quantum mechanical concepts to machine learning techniques, thereby facilitating the improved interpretability of neural networks. As an application, we will use top jet classification against QCD jets and compare performance against state-of-the-art machine learning applications. Finally, we will discuss how to convert these models into Quantum Circuits to be compiled on a quantum device and show that classical TNs require exponentially large bond dimensions and higher Hilbert-space mapping to perform comparably to their quantum counterparts.
Collider searches face the challenge of defining a representation of high-dimensional data such that (i) physical symmetries are manifest, (ii) the discriminating features are retained, and (iii) the choice of representation is data-driven and new-physics agnostic. We introduce JetCLR (Contrastive Learning of Jet Representations) to solve the mapping from low-level jet constituent data to optimized observables through self-supervised contrastive learning. Using a permutation-invariant transformer-encoder network, physical symmetries such as rotations and translations are encoded as augmentations in a contrastive learning framework. As an example, we construct a data representation for top and QCD jets and visualize its symmetry properties. We benchmark the JetCLR representation against other widely-used jet representations, such as jet images and energy flow polynomials (EFPs).
There is a growing recent interest in endowing the space of collider events with a metric structure calculated directly in the space of its inputs. For quarks and gluons, the recently developed energy mover's distance has allowed for a quantification of what is different between physical events. However, the large number of particles within jets makes using metrics and interpreting these metrics particularly difficult. In this work, we introduce a flexible framework based on neural embedding to embed a manifold from a jet to lower-dimensional spaces using a defined metric. We demonstrate a low distortion and robust embedding can be achieved with Energy movers distance in two dimensions. Furthermore, we show that we can construct a self-organized space that captures the core physical features of a jet, including the splitting angularity and the number of prongs. Using the notion of volume in the embedded space, we propose the volume-adjusted roc-curve to measure the energy mover's volume that a dedicated jet selection has on the total phase space of jets. Finally, we equate the volume to the inclusivity of a jet kinematic selection and show how this approach can quantify the effectiveness of anomaly searches and measurements in performing unbiased, inclusive measurements.
The identification of interesting substructures within jets is an important tool to search for new physics and probe the Standard Model. In this paper, we present \textsc{SHAPER}, a general framework for defining computing shape-based observables, which generalizes the $N$-jettiness from point clusters to any extended shape. This is accomplished by minimizing the $p$-Wasserstein metric between events and parameterized manifolds of energy flows representing idealized shapes, implemented using the dual-potential Sinkhorn approximation for efficient minimization. We show how the geometric language of observables as manifolds can be used to easily define novel event and jet-substructure observables with built-in IRC safety that are useful for physics analyses. We then demonstrate the \textsc{SHAPER} framework by performing an example jet substructure analysis using these new shape-based observables.
The main goal for the upcoming LHC runs is still to discover BSM physics. It will require analyses able to probe regions not linked to specific models but generally identified as beyond the Standard Model. Autoencoders are the ideal analysis tool for this type of search. Energy-based machine learning models have been shown to be flexible and powerful models to describe high-dimensional feature space and combine out-of-distribution searches with density estimation. I will present a Normalized Autoencoder, the first proper and high-performance anomaly search algorithm for LHC jets. I will apply it to jets images and show how the NAE reliably identifies anomalous jets symmetrically in the directions of higher and lower complexity. I will show that NAE works well for top vs QCD jet, as well as for the more challenging dark jet signals.
at Störtebeker Elbphilharmonie GmbH, Platz der Deutschen Einheit 3, 20457 Hamburg
We present the first anti-kT jet spectrum and substructure measurements using the archived ALEPH e+e- data taken in 1994 at a center of mass energy of sqrt(s) = 91.2 GeV. Jets are reconstructed with the anti-kT algorithm with a resolution parameter of 0.4. It is the cleanest test of jets and QCD without the complication of hadronic initial states. The fixed center-of-mass energy also allows the first direct test of pQCD calculation. We present both the inclusive jet energy spectrum and the leading dijet energy spectra, together with a number of substructure observables. They are compared to predictions from PYTHIA6, PYTHIA8, Sherpa, HERWIG, VINCIA, and PYQUEN. None of the models fully reproduce the data. The data are also compared to two perturbative QCD calculations at NLO and with NLL’+R resummation. The results can also serve as reference measurements to compare to results from hadronic colliders. Future directions, including testing jet clustering algorithms designed for future electron-ion collider experiments, will also be discussed.
Energy Correlators (EEC) have recently received great interest both theoretically and experimentally. In particular, the study of EECs in jet substructure has gained deeper understanding with the advent of the light-ray operator product expansion. In this talk, based on this progress, we propose a ratio observable named “celestial non-gaussianity”, which roughly is the ratio between three-point energy correlator and a product of two-point correlators. The underlying motivation for such a construction is to probe how the three-point function deviates from the factorization into a product of two-point functions in the squeezed limit. One salient feature of the “celestial non-gaussianity” is its robustness to hadronization effects. We compare our perturbative prediction with CMS Open Data finding good agreement. We anticipate the celestial non-gaussianity and its possible generalizations will be helpful for future precision measurement of effects like spin correlations and the development of parton showers.
Track functions describe the collective effect of the fragmentation of quarks and gluons into charged hadrons, making them a key ingredient for jet substructure measurements at hadron colliders where track-based measurements offer superior angular resolution. Unlike DGLAP, the evolution of track functions incorporates correlations between final-state hadrons, and is hence non-linear. We derive the NLO renormalization of the track functions in both $\mathcal{N}=4$ SYM and QCD, which are the first example of evolution equations involving the full $1\to 3$ splitting functions. We clarify how our evolution equations can be reduced to those for the single- and multi-hadron fragmentation functions, and can hence be viewed as the most general collinear evolution equations at NLO. We believe that evolution equations incorporating collinear correlations are an important step beyond DGLAP, and will be relevant for the development of higher order partons showers and for paving the way for precision jet substructure at LHC. Finally, we apply these developments to study correlations of energy flows, and have initiated a discussion of NLL resummation for track energy correlators.
communicated via e-mail
Jet angularities are an important class of jet substructure observables
investigated at the LHC. In this talk I will focus the comparison of theoretical
predictions against recent measurements from the CMS experiment [1]. I will
present calculations at NLO+NLL' based on [2, 3], in hadronic dijet and
Z+jet events. Where applicable, the effect of soft drop grooming is included. Those
semi-analytic calculations are also compared to state of the art Monte Carlo
predictions using Sherpa at MC@NLO accuracy for both processes. Non-perturbative
corrections from the underlying event and hadronisation are applied using
parton-to-hadron level transfer matrices extracted from Monte Carlo, as
introduced in [3]. The setup of the calculation allows one to make separate
predictions for quark and gluon jets, and I will highlight the relation to quark
gluon flavour discrimination, see also [4].
[1] JHEP 01 (2022) 188 [arXiv:2109.03340]
[2] JHEP 07 (2021) 076 [arXiv:2104.06920]
[3] JHEP 03 (2022) 131 [arXiv:2112.09545]
[4] Eur.Phys.J.C 81 (2021) 9, 844 [arXiv:2108.10024]
We consider the issue of meaningfully assigning a flavor label to a jet and we show that modern jet substructure techniques can give us new ways to tackle this problem.
On the one hand, we introduce a novel fragmentation-function framework that allows one to connect a flavor definition in the deep UV, where partons live, to an IR definition, where jets live. The IR definition involves the Winner-Take-All axis and it has the advantage that the resulting evolution equations are linear. On the other hand, we consider the issue of interfacing measurements involving flavored jets (typically heavy-flavors) with precision calculations in QCD. We introduce a jet-flavor algorithm, based on Soft Drop grooming, which is at the same time IRC safe through NNLO and easy to implement in experimental analyses.
Quantum chromodynamics is the theory of the strong interaction between quarks and gluons; the coupling strength of the interaction, $\alpha_S$, is the least precisely-known of all interactions in nature. An extraction of the strong coupling from the radiation pattern within jets would provide a complementary approach to conventional extractions from jet production rates and hadronic event shapes, and would be a key achievement of jet substructure at the Large Hadron Collider. Presently, the relative fraction of quark and gluon jets in a sample is the limiting factor in such extractions, as this fraction is degenerate with the value of $\alpha_S$ for the most well-understood observables. To overcome this limitation, we apply recently proposed techniques to statistically demix multiple mixtures of jets and obtain purified quark and gluon distributions based on an operational definiton. We illustrate that studying quark and gluon jet substructure separately can significantly improve the sensitivity of such extractions of the strong coupling. We discuss how using machine learning techniques or infrared- and collinear-unsafe information can improve the demixing performance without the loss of theoretical control.
Identifying the flavour of the experimentally reconstructed hadronic jets is critical to pinpoint specific scattering processes and reject background processes. Jet measurements at the LHC are almost universally performed using the anti-$k_T$ algorithm, however no approach exists to define the jet flavour for this algorithm that is infrared and collinear (IRC) safe. In this talk, we propose a new approach that is IRC safe to all orders in perturbation theory and can be combined with any definition of a jet. We test the algorithm in $e^+e^-$ and $pp$ environments, and as an application we compare our approach to other algorithms for the $Z$+$b$-jet/$Z$+$c$-jet production.
Multiplicity is one of the simplest experimental observables in collider events, whose importance stretches from calibration to advanced tagging techniques. We introduce a new (sub)jet multiplicity, the Lund multiplicity, for lepton and hadron collisions. It probes the full multiple branching structure of QCD and is calculable in perturbation theory. We introduce a formalism allowing us to calculate the average Lund and Cambridge multiplicities to all orders, reaching next-to-next-to double logarithmic (NNDL) accuracy in $e^+e^−$ collisions, an order higher than the existing state-of-the-art, and next-to-double logarithmic accuracy (NDL) in hadronic collisions. Matching our resummed calculation to the NLO result, we find a reduction of theoretical uncertainties by up to 50% compared to the previous state-of-the-art. Adding hadronisation corrections obtained through Monte Carlo simulations, we also show a good agreement with existing Cambridge multiplicity data measured at LEP. Finally we discuss the extension of our formalism to Lund subjet multiplictiy in dijet and Z+jet events at the LHC.
Observables computed on jets groomed with mMDT, or equivalently Soft Drop with $\beta=0$, benefit from reduced sensitivity to pileup, underlying event, and hadronisation, compared to observables computed on un-groomed jets. Observables computed on groomed jets are therefore good candidates for direct comparison between perturbative QCD predictions and measurements.
Focusing on quark initiated jets, we present a method for computing the distributions of groomed observables at next-to-next-to-leading logarithmic (NNLL) accuracy directly in perturbative QCD, in the limit that $z_{\mathrm{cut}}$ is small. This method can be applied to any additive, recursively infra-red and collinear safe (rIRC safe) observable. Predictions for the groomed heavy hemisphere mass, Width, and Les Houches angularity (LHA) are presented, the groomed Width and LHA not having been presented at NNLL before. We then modify our results to include finite $z_{\mathrm{cut}}$ effects at next-to-leading logarithmic (NLL) accuracy as well as next-to-leading order (NLO) matching, this being the first time that finite $z_{\mathrm{cut}}$ effects at NLL accuracy have been included in an NNLL resummation.
Effects of hadronization and underlying event on jet substructure observables must be accurately quantified for precision QCD measurements such as the strong coupling constant and the top quark mass. While these effects have long been studied for ungroomed observables such as the jet mass and jet $p_T$, they are significantly more complicated in groomed observables. In this work we employ a model-independent, field theory-based formalism to systematically quantify hadronization corrections in terms of universal, flavor dependent $O(\Lambda_{\rm QCD})$ constants and moments of groomed jet radius $R_g$. We further extend this framework to describe the impact of the underlying event (UE) by precisely computing the dynamical groomed catchment area as a function of groomed jet mass. We compute these moments of $R_g$ using the soft drop double differential cross section at NLL + $O(\alpha_s)$ accuracy by consistently matching across 2 $\times$ 3 factorization regimes. We test our predictions and fit for hadronization and UE phenomenological parameters by performing detailed comparison with various event generators for quark and gluon initiated jets.
Many New Physics searches and QCD precision measurements at particle colliders involve the study of jet substructure for final state hadrons. While traditionally the state of the art for studying jets at particle colliders have been event shape observables, recently it has been better understood that measuring correlation functions of energy flow operators inside a jet can be a very powerful tool for phenomenology, which naturally stems from first principles of quantum field theory.
In this talk we present a bridge between the huge theoretical progress made to understand the energy correlators from the field theory perspective and their practical implementation into the real world of hadron colliders. I will show a factorization formula we have derived using soft-collinear effective theory (SCET), that together with the fragmenting jet formalism, makes it possible to study such energy correlators inside a jet in the complicated LHC environment.
I will show results for the scaling behavior of the two point energy correlators at NLL and multipoint projected operators at NLL. In addition I will show a comparison of our results with the CMS open data.
State-of-the-art prediction accuracy in jet tagging tasks is currently achieved by modern geometric deep learning architectures incorporating Lorentz group invariance, resulting in computationally costly parameterizations that are moreover complex and thus lack interpretability. To tackle this issue, we propose Boost Invariant Polynomials (BIPs) — a framework to construct highly efficient features that are invariant under permutations, rotations, and boosts in the jet direction. The simplicity of our approach results in a highly flexible and interpretable scheme. We establish the versatility of our method by demonstrating state-of-the-art accuracies in both supervised and unsupervised jet tagging by using several out-of-the-box classifiers with only O(hundreds) of parameters, O(s) training, and O($\mu$s) inference times on CPU.
Motivated by the high computational costs of classical simulations, machine-learned generative models can be extremely useful in particle physics and elsewhere. They become especially attractive when surrogate models can efficiently learn the underlying distribution, such that a generated sample outperforms a training sample of limited size. This kind of GANplification has been observed for simple Gaussian models and large ranges of training sample sizes. In this talk, we extend this histogram based method to show the same effect for a physics simulation, specifically photon showers in an electromagnetic calorimeter.
We propose Classifying Anomalies THrough Outer Density Estimation (CATHODE): A novel, completely data-driven and model-agnostic approach to search for resonant new physics with anomalous jet substructure at the LHC.
Training a conditional normalizing flow on kinematic and substructure variables in a sideband region, we acquire an approximation of their probability densities. We then interpolate our trained background model into the signal region and sample from it, which yields an estimation of the standard model background inside the signal region without relying on simulation. Finally, a classifier is trained to distinguish background and data events in the signal region to find anomalies.
We report an improvement of the nominal statistical significance in a specific example from ~1 𝜎 to as much as ~15 𝜎 using the LHC Olympics R&D dataset as benchmark. Thus, the CATHODE method is able to discover new physics that otherwise would be hidden in data.
The performance of constituent-based jet taggers for boosted top quarks reconstructed from Unified Flow Object jet input is presented. Several taggers which consider all of the information contained in the kinematic information of the jet constituents are tested, and compared to a tagger which relies on high-level summary quantities similar to the taggers used by ATLAS in Runs 1 and 2.
We investigate the pair-production of Right-Handed Neutrinos (RHNs) via a $B-L$ $Z'$ boson and present the sensitivity studies of the active-sterile neutrino mixing ($|V_{\mu N}|$) at the High-Luminosity run of the LHC (HL-LHC) and a future $pp$ collider (FCC-hh). We focus on RHN states with a mass of $10-70$ GeV which naturally results in displaced vertices for small $|V_{\mu N}|$. Being produced through a mass resonance with $M_{Z'} \ge 1$ TeV, the RHNs are heavily boosted, leading to collimated decay products that give rise to fat-jets. We investigate the detection prospect of dedicated signatures in the inner detector and the muon spectrometer, namely a pair of displaced fat-jets and the associated tracks, respectively. We find that both the HL-LHC and FCC-hh can be sensitive to $|V_{\mu N}| > 10^{-6}$ and $|V_{\mu N}| > 10^{-7}$ with the number of events reaching $\mathcal{O}(10)$ and $\mathcal{O}(10^3)$, respectively. This allows probing the generation of light neutrino masses through the Seesaw mechanism.
The large data rates at the LHC make it impossible to store every single observed interaction. Therefore we require an online trigger system to select relevant collisions. We propose an additional approach, where rather than compressing individual events, we compress the entire data set at once. We use a normalizing flow as a deep generative model to learn the probability density of the data online. The events are then represented by the generative neural network and can be inspected offline for anomalies or used for other analysis purposes.
We demonstrate our new approach for a toy model and a correlation-enhanced bump hunt.
Simulation is a key component of modern high energy physics experiments. However, producing simulated data with sufficient detail and in sufficient quantities places a significant strain on the available computing resources. With the increased simulation demands of the upcoming high luminosity phase of the LHC and future colliders expected to contribute to a major bottleneck, computationally efficient solutions are urgently needed.
This contribution presents significant progress in the development of deep generative models for fast shower simulation in highly granular calorimeters. Approaches to simulating both electromagnetic and hadronic showers will be reported, with a focus on the high degree of physical fidelity and computational performance achieved. Additionally, steps taken to overcome the challenges faced when broadening the scope of these simulators, such as those posed by multi-parameter conditioning, will also be presented.
Flavour tagging is a crucial component for the LHC physics program. The performance of the flavour-tagging algorithm is such that the statistical precision of the simulated samples is diluted when flavour tagging is applied in particular to many jets per event. Truth-flavour tagging is based on weighting jets according to their probability of being tagged and is an alternative approach that preserves the statistical power of the simulated samples. This contribution describes a novel implementation of truth-flavour tagging in ATLAS based on graph neural networks. The approach is demonstrated to describe effects due to near-by topologies typical in boosted environments, offering a more elegant solution compared to the traditional strategy based on efficiency histograms.
The Boosted Event Shape Tagger (BEST) is a boosted jet tagger that classifies large radius jets as originating from: Higgs, W, Z, top, bottom, or QCD. In BEST, jet constituents are boosted along the jet axis assuming 7 different mass hypotheses. In each frame, a series of Boosted Event Shape variables are calculated. These variables, along with jet kinematic information, are used as inputs to a fully connected, dense neural network. This relatively simple, physics inspired approach is competitive with advanced convolutional networks. This specialized tagger was developed for the CMS search for pair production of top-like vector-like quarks in an all hadronic final state and can be applied to other searches at CMS. In updating the search from 2016 data to full run 2 data, this tagger underwent many iterations. This poster will show the evolution of BEST with a focus on application to the search for pair production of top-like vector-like quarks.
In high-energy particle physics, complex Monte Carlo simulations are needed to connect the theory to measurable quantities. Often, the significant computational cost of these programs becomes a bottleneck in physics analyses. In this contribution, we evaluate an approach based on a Deep Neural Network to reweight simulations to different models or model parameters, using the full kinematic information in the event. This methodology avoids the need for simulating the detector response multiple times by incorporating the relevant variations in a single sample. We test the method on Monte Carlo simulations of top quark pair production used in CMS, that we reweight to different SM parameter values and to different QCD models.
In this talk, I will present the differential equation solver by using machine learning methods and apply it to solve the evolution equation in QCD as the proof-of-concept, such as the DGLAP equation. Moreover, I will use this method to study the medium-induced radiation loss by solving the time-dependent Schrodinger equation (TDSE) for the light-cone path integral (LCPI) approach, developed by Zakharov.
Reconstructing the type and energy of isolated pions from the ATLAS calorimeters is a key step in the hadronic reconstruction. The baseline methods for local hadronic calibration were optimized early in the lifetime of the ATLAS experiment. Recently, image-based deep learning techniques demonstrated significant improvements over the performance over these traditional techniques. This poster presents an extension of that work using point cloud methods that do not require calorimeter clusters or particle tracks to be projected onto a fixed and regular grid. Instead, transformer, deep sets, and graph neural network architectures are used to process calorimeter clusters and particle tracks as point clouds. This presentation demonstrates the performance of these new approaches as an important step towards a full deep learning-based low-level hadronic reconstruction.
With current and future high-energy collider experiments' vast data collecting capabilities comes an increasing demand for computationally efficient simulations. Generative machine learning models allow fast event generation, yet so far are largely constrained to fixed data and detector geometries.
We introduce a novel generative machine learning setup for generation of permutation invariant point clouds with variable cardinality - a flexible data structure optimal for collider events such as jets. We show that our model scales well to large particle multiplicities and achieves good performance on quark and gluon jet data. To explain the behaviour of the generative model, we perform an analysis of the latent space.
A search for heavy resonances Y decaying into a Standard Model Higgs boson (H) and a new boson (X) is performed with proton-proton collision data with the ATLAS detector at the CERN Large Hadron Collider. The Physics channel where the Higgs decays into bb and the X to light quarks are considered, thus resulting in a fully hadronic final state. A two-dimensional phase space of XH mass versus X mass is scanned for evidence of a signal. Upper limits are set on the production cross-section of the resonance as a function of XH and X masses.
A search is made for a vector-like T quark decaying into a Higgs boson and a top quark in 13 TeV proton-proton collisions using the ATLAS detector at the Large Hadron Collider with a data sample corresponding to an integrated luminosity of 139 fb−1. The all-hadronic decay modes H→b¯b and t→bW→bq¯q′ are reconstructed as large-radius jets and identified using tagging algorithms. Improvements in background estimation, signal discrimination, and a larger data sample, contribute to an improvement in sensitivity over previous all-hadronic searches. No significant excess is observed above the background, so limits are set on the production cross-section of a singlet T quark at 95\% confidence level, depending on the mass, mT, and coupling, κT, of the vector-like T quark to Standard Model particles. This search targets a mass range between 1.0 to 2.3 TeV, and a coupling value between 0.1 to 1.6, expanding the phase space of previous searches. In the considered mass range, the upper limit on the allowed coupling values increases with mT from a minimum value of 0.35 for 1.07 <mT< 1.4 TeV up to 1.6 for mT=2.3 TeV.
The reconstruction and calibration of hadronic final states is an extremely challenging experimental aspect of measurements and searches at the LHC. This talk summarizes the latest results from ATLAS for jet reconstruction and calibration of Anti-kt R=0.4 jets. New approaches to jet inputs better utilize relationships between calorimeter and tracking information to significantly improve the jet performance in preparation for Run-3. Additionally, new pile-up mitigation techniques are shown to provide a better stability with respect to pile-up. Finally, new jet calibration methods are presented, which reduce the uncertainties in the high pileup conditions of the LHC Run 2.
In this talk, I will review my engagement in the field of jet substructure, from work as an undergraduate starting in 2006 to my departure from academia in 2022. I will review what I think the big problems in the field are, from outstanding physics questions to issues with retention and promotion of leaders in this field. I will also discuss what I see as the scope creep of jet substructure as it attempts to maintain relevance into the future.
The ability to differentiate between hadronically decaying massive particles is increasingly important to the LHC physics program. A variety of tagging algorithms for large-radius jets, reconstructed from unified-flow-objects (UFOs), are presented to identify jets containing the hadronic decay of W/Z bosons and top quarks, including both cut-based taggers and machine learning discriminants. The performance of new UFO jet-based taggers will be compared to the taggers deployed in ATLAS during Run-2 for jets reconstructed sole from calorimeter deposits.
We describe a new jet clustering algorithm (SIFT: Scale-Invariant Filter Tree) that does not impose a fixed cone size or associated scale on the event. This construction maintains excellent object discrimination for very collimated partonic systems, tracks accrued mass, and asymptotically recovers favorable behaviors of both the standard KT and anti-KT algorithms. It is intrinsically suitable (without secondary declustering) for the tagging of highly boosted objects, and applicable to the study of jet substructure. Additionally, it is resilient to pileup, via a concurrent filter on soft wide-angle radiation applied within the primary clustering phase. Flexible termination conditions facilitate clustering to a fixed number of objects or identification of the "natural" object count. Linearithmic performance can be achieved through a new neighbor-finding framework based on the KDTree data structure that is compatible with higher-dimensional measures and "sociophobic" coordinates. Clustering histories are visualized in time with video simulation.
Jet tagging is a critical yet challenging classification task in particle physics. While deep learning has transformed jet tagging and significantly improved performance, the lack of a large-scale public dataset impedes further enhancement. In this work, we present JetClass, a new comprehensive dataset for jet tagging. The JetClass dataset consists of 100 M jets, about two orders of magnitude larger than existing public datasets. A total of 10 types of jets are simulated, including several types unexplored for tagging so far. Based on the large dataset, we propose a new Transformer-based architecture for jet tagging, called Particle Transformer (ParT). By incorporating pairwise particle interactions in the attention mechanism, ParT achieves higher tagging performance than a plain Transformer and surpasses the previous state-of-the-art, ParticleNet, by a large margin. The pre-trained ParT models, once fine-tuned, also substantially enhance the performance on two widely adopted jet tagging benchmarks.
https://arxiv.org/abs/2202.03772
The Heavy Object Tagger with Variable R (HOTVR) is an algorithm for the clustering and identification of boosted, hadronically decaying, heavy particles. The central feature of the HOTVR algorithm is a vetoed jet clustering with variable distance parameter R, that decreases with increasing transverse momentum of the jet. In this talk, we present improvements to the HOTVR algorithm, replacing the mass jump with a soft drop veto in the clustering. We study the performance of jet substructure tagging with HOTVR and ungroomed variable R jets, where we use machine learning techniques and energy flow polynomials to analyse the information loss from the soft drop veto. In addition, we show preliminary results of a distance parameter that changes with the jet mass and the transverse momentum, allowing to achieve an optimal value of R for W, Z, H bosons and top quarks simultaneously.
The study of substructure of hadronic jets is key to unlocking further understanding of the physics underlying collisions at the LHC. In the context of precision Standard Model physics, we discuss how substructure variables sensitive to colour flow, such as the Lund Jet Plane, Jet Angularities and Jet Pull projections can be used to develop taggers highly sensitive to the radiation pattern within jets. This can be used, for instance, to build simple but efficient Higgs taggers. Furthermore, despite their relative simplicity, these observables can also be used to construct b-tagging algorithms that can augment standard approaches, mostly based on displaced vertices. Our results are given in terms of tagging performance on simulated data and, when possible, are compared to performance of taggers implemented by experimental collaborations.
Identifying highly boosted resonances, including top quarks, electroweak bosons, and new particles, has become a core topic of research at the LHC. Advances in machine learning have further accelerated interest in boosted resonance identification. However, as machine learning algorithms become more powerful, so too have the correlations of the algorithms with jet kinematics, like mass and momentum. This complicates their use in measurements and searches which, for example, might leverage smoothly falling backgrounds. A number of decorrelation methods have been used previously, including brute-force decorrelation, modifications to ML training loss functions, or modifications to the samples themselves, such as weighted trainings, or trainings with several resonance masses. In this work, we introduce a novel decorrelation method for jet substructure algorithms constructed by modifying sample matrix elements to create "flat" mass and momentum distributed resonances. As the samples have no intrinsic resonance features, ML algorithms are intrinsically decorrelated jet kinematics. This new decorrelation approach is highly versatile and can be applied to boosted object identification, jet mass regression, and jet mass scale calibration. Moreover, this approach allows for robust decorrelation of advanced AI algorithms, where other approaches struggle, such as graph-based interaction networks and transformers. We present this work in the context of tagging Z' resonances decaying to jets and Higgs to tau leptons. Lastly, we show even better performance with modifications to the training architecture, including a KL divergence-weighted loss, and contrastive learning.
As classic WIMP-based signatures for dark matter at the LHC have found no compelling evidence, several phenomenological studies have raised the possibility of accessing a strongly-interacting dark sector through new collider-event topologies. If dark mesons exist, their evolution and hadronization procedure are currently little constrained. They could decay promptly and result in QCD-like jet structures, even though the original decaying particles are dark sector ones; they could behave as semi-visible jets; or they could behave as completely detector-stable hadrons, in which case the final state is just the missing transverse momentum. In this contribution we will introduce a study performed to explore use of jet substructure methods to distinguish dark sector from QCD jets in the first two scenarios, using observables in a IRC-safe linear basis, and discuss ways forward for this approach to dark matter at the LHC, including prospects for estimating modelling uncertainties.
Feature selection algorithms can be an important tool for AI explainability. If the performance of neural networks trained on low-level data can be reproduced by a small set of high-level features, we can hope to understand “what the machine learned”. We present a new algorithm that selects features by ranking their Distance Correlation (DisCo) values with truth labels. We apply this algorithm to the classification of boosted top quarks and use a set of 7,000 Energy Flow Polynomials (EFPs) as our feature space. We show that our method is able to select a small set of high-level features, with a classification performance comparable to the state-of-the-art top taggers.
The production of dark matter particles from confining dark sectors may lead to many novel experimental signatures. Depending on the details of the theory, dark quark production in proton-proton collisions could result in semivisible jets of particles: collimated sprays of dark hadrons of which only some are detectable by particle collider experiments. The experimental signature is characterised by the presence of reconstructed missing momentum collinear with the visible components of the jets. This complex topology is sensitive to detector inefficiencies and mis-reconstruction that generate artificial missing momentum.
We propose a signal-agnostic strategy to reject ordinary jets and identify semivisible jets via anomaly detection techniques. A deep neural autoencoder network with jet substructure variables as input proves highly useful for analyzing anomalous jets. The study focuses on the semivisible jet signature; however, the technique can apply to any new physics model that predicts signatures with anomalous jets from non-SM particles.
Several flavor tagging algorithms exist in ATLAS for jets containing two b-hadrons. These double-b tagger algorithms focus on high transverse-momentum jets, usually above 200 GeV. This work describes the development of a new double-b tagger for jets below 200~GeV. The algorithm relies on large radius track-jets which can be reconstructed at low transverse momenta and implements a neural network architecture based on Deep Sets that uses displaced tracks, secondary vertices, and substructure information to identify the presence of multiple $b$-hadrons. A measurement of the efficiency of the algorithm is performed in ttbar and Z + jets events using the collision data from the Large Hadron Collider at sort{s}= 13 TeV center-of-mass energy recorded with the ATLAS detector between 2015 and 2018, corresponding to an integrated luminosity of 139 fb^-1
Deep learning has transformed jet tagging, in bringing a leap to tagging performance and hence substantially improving the sensitivity of physics searches at the LHC. In seek of further enhancement, recent interests fall in experimenting with more advanced neural network architectures, or injecting physics knowledge into the design of the network. This talk focuses on the latter, with a discussion on how intrinsic physics symmetries play a vital role. We introduce and investigate the LorentzNet, an efficient model with a graph neural network (GNN) backbone that respects the Lorentz group symmetry of a jet. We show how the model improves the tagging performance over the previous state-of-the-art algorithms including ParticleNet, especially when trained only on a few thousand jets. We study how symmetry preservation serves as a strong inductive bias of the jet tagging learning algorithm and hint at its potential role in future tagger development.
This talk is based on https://arxiv.org/abs/2201.08187 and includes new relevant studies on LorentzNet.
at Zum Alten Lotsenhaus, Övelgönne 13, 22605 Hamburg
In this talk, we discuss energy correlators within the context of beauty and charm quark jets to illuminate the effects of the intrinsic mass of the elementary particles of QCD. We extend existing factorization theorems to include the mass of heavy quarks and calculate heavy quark jet functions in order to carry out theoretical calculations to next-to-leading logarithmic accuracy. Using this framework, we then observe a clear transition from the scaling region to the region corresponding to the heavy quark mass scale, manifesting the long-sought-after dead-cone effect and illustrating fundamental effects coming from the intrinsic mass of beauty and charm quarks before they are confined inside hadrons. Our theoretical framework for studying energy correlators using heavy jets has many exciting applications for tuning mass effects for parton shower Monte Carlo event generators, probing medium in heavy-ion collisions, and studying heavy flavor fragmentation functions.
The production of jets at hadron provides stringent tests of perturbative QCD. We present the latest measurements using proton-proton collision data collected by the ATLAS experiment at sqrt(s)=13 TeV. We will discuss the measurement of new event-shape jet observables defined in terms of reference geometries with cylindrical and circular symmetries using the energy mover???s distance. The results are unfolded for detector effects and compared to the state-of-the-art next-to-leading order parton shower generators. If ready, the measurement of strong coupling constant will be presented using the ratio of 3-jet to 2-jet events.
Over the last decades, the theoretical picture of how hadronic jets interact with nuclear matter has been extended to account for the medium’s finite longitudinal length and expansion. However, only recently a first-principle approach has been developed that allows to couple the jet evolution to the medium flow and anisotropic structure in the dilute limit. In this talk, we will show how to extend this approach to the dense regime, where the resummation of multiple in-medium scatterings is necessary. Particularly, we will consider the modifications of the single particle momentum broadening distribution and single gluon production rate in evolving matter. The resummation is performed by either computing the opacity series or starting from the all order BDMPS-Z formalism. We will also discuss the (novel) resulting modifications to jets' substructure.
communicated via e-mail
communicated via e-mail
During heavy-ion collisions, a new phase of matter, the quark-gluon plasma, is believed to have been created. The dense and hot matter interacts with high energy parton leading to the jet quenching effect, which redistributes the energy inside high energy jets. Therefore jets are an important probe to understand the heavy-ion collisions. There have been a lot of progress in recent years in understanding the quark-gluon plasma, and in this talk recent jet substructure measurements from CMS will be summarized.
The energy-flow-operator (EFO) provides an idealised field-theoretic definition of a calorimenter. Recently, the angular correlations between EFOs on the celestial sphere have seen a great deal of interest as a tool for jet substructure. Due to the causal structure of EFO correlators, the angular size of the correlations can be viewed as a time parameter: early time perturbative correlations are imprinted at large angles, whilst later time correlations from hadronisation appear at small angles. In this work we demonstrate that the scales associated with the early time evolution (quenching) of a jet propagating through the QGP are imprinted in the large-angle perturbative structure of the 2-point correlator.
Measuring the jet substructure in heavy-ion collisions provides exciting new opportunities to study detailed aspects of the dynamics of jet quenching in the hot and dense QCD medium created in these collisions. In this talk, we present new ATLAS measurements of jet substructure performed using various jet (de)clustering and grooming techniques. Measurements of inclusive jet suppression (RAA) in heavy-ion collisions are presented for the first time as a function of the jet substructure using both nominal (R=0.4) and large-radius (R=1.0) jets in Pb+Pb and pp collisions at √sNN=5.02 TeV. The jet substructure is characterized using the Soft-Drop grooming procedure in order to identify subjets corresponding to the hardest parton splitting in the jet. The dynamics of jet quenching is measured and presented as a function of the transverse momentum scale (√d12) and the angle of the hardest splitting in the jet. Novel reconstruction methods are utilized to combine and optimize information from the tracker and calorimeter and build jet constituents for the first time in heavy-ion collisions. These new measurements test the sensitivity of jet suppression in the QCD medium to its substructure and the emergence of a critical angle for the onset of color decoherence.
We revisit the picture of jets propagating in the quark-gluon plasma. In addition to vacuum radiation, partons scatter on the medium constituents resulting in induced emissions. We achieve full analytical control of the relevant scales and map out the dominant physical processes in the full phase space for the first time. This covers the whole phase space from early to late times, and from hard splittings to emissions down to the thermal scale. Based on the separation of scales, a space-time picture naturally emerges: at early times, induced emissions start to build from rare scatterings with the medium. At a later stage, induced emissions due to multiple soft scatterings result in a turbulent cascade that rapidly degrades energy down to the thermal scale. Our work serves to improve our understanding of jet quenching from small to large systems and for future upgrades of Monte Carlo generators. Moreover, our factorized picture leads to the possibility for power counting and thus defining accuracy for medium-induced jets for the first time.
References:
J.H.Isaksen, A.Takacs, K.Tywoniuk, arXiv:2206.02811
This measurement is performed using highly boosted top quark pair events resulting in non isolated leptons and overlapping jets. Jet substructure variables are used to identify the boosted top quark and the W boson. The top quark charge asymmetry is measured for events with ttbar invariant mass larger than 750 GeV and corrected for detector and acceptance effects using a binned maximum likelihood fit. The measurement is found to be in good agreement with the standard model prediction at next-to-next-to-leading order in perturbation theory with next-to-leading order electroweak corrections.
Microscopic BSM dynamics can be encoded in a small set of parameters controlling deformations from the SM. I argue that sensitivity to BSM physics can be boosted with observables that discriminate longitudinal and transverse production of EW vectors. This is well studied in leptonic decays, but recently, with the help of energy correlators, studies started to target hadronic decays.
I show that the kinematics of the one- and two-point energy correlators of the hadronic decay of an electroweak vector can discriminate between longitudinal and transverse modes and reveal the interference pattern between different vector polarizations.
Such observables improve the sensitivity to microscopic new physics affecting the production rate of the different helicities. We assess the impact on higher dimensional EFT operators in simple scenarios.
Final states in collider experiments are characterized by correlation functions of the energy flow operator - which plays the roll of an idealised calorimeter. In this talk, I will show that the boosted top quark imprints itself as a peak in the three-point correlator at an angle determined by its mass and transverse momentum. This provides direct access to one of the most important parameters of the Standard Model in one of the simplest field theoretical observables.
The analysis I will present provides a new paradigm for a precise top mass determination that is, for the first time, highly insensitive to soft physics and underlying event contamination.
We present a measurement of the jet mass distribution in fully hadronic decays of boosted top quarks with full Run 2 data. The measurement is performed in the lepton+jets channel of top quark pair production. The top quark decay products of the all-hadronic decay cascade are reconstructed with a single large-radius jet with transverse momentum greater than 400 GeV. The top quark mass is extracted from the normalised differential top quark pair production cross section at the particle level. The uncertainties arising from the calibration of the jet mass scale and modelling of the final state radiation in simulation are improved by dedicated studies of the jet substructure. These studies lead to a significant increase in precision in the top quark mass with respect to an earlier measurement, now reaching a precision below 1 GeV.
In this talk, we study prospects for measuring the strong coupling $\alpha_s$ using the soft-drop jet-mass cross section in hadronic collisions. We compute the normalized cross section for quark and gluon jets to NNLL accuracy, and estimate the nonperturbative hadronization corrections using a model-independent field-theory based formalism involving $\mathcal{O}(\Lambda_{\rm QCD})$ constants for quark and gluon jets and perturbative Wilson coefficients computed to NLL’ accuracy. We clarify the impact of higher logarithmic resummation, that significantly modifies the leading logarithmic shape. Using these calculations, we assess the theoretical uncertainties on an $\alpha_s$-precision measurement due to scale variations, nonperturbative corrections, and quark/gluon fractions. We identify which soft-drop parameters are well-suited for measuring $\alpha_s$, and highlight differences in the $\alpha_s$ sensitivity between quark and gluon jets. This analysis thus paves the way for using soft-drop jet-mass cross section for a measurement of $\alpha_s$ at hadron colliders.
Every bunch crossing at the LHC causes not just one proton-proton interaction, but several. These additional collisions are called "pileup". With the increasing luminosity of the LHC also the number of pileup interactions per bunch crossing increased in the past years and it will reach up to 140 during high-luminosity LHC operation. Removing the pileup from an event is essential, because it does not only affect the jet energy but also other event observables as for example the missing transverse energy, the jet substructure, jet counting and the lepton isolation. In addition, jets as an experimental signature of energetic quarks and gluons produced in high energy processes, need to be calibrated in order to have the correct energy scale. A detailed understanding of both the energy scale and the transverse momentum resolution of jets at the CMS is of crucial importance for many physics analyses. In this talk we present recent developments in terms of jet energy scale and resolution, substructure techniques and pileup mitigation techniques.
Jet and Missing transverse momentum (MET), used to infer the presence of high transverse momentum neutrinos or other weakly interacting neutral particles, are two of the most important quantities to reconstruct at a hadron collider. They are both used by many searches and measurements in ATLAS. New techniques combining calorimeter and tracker measurements, called Particle Flow and Unified Flow, have significantly improved the reconstruction of both transverse momentum and jet substructure observables. The procedure of reconstructing and calibrating ATLAS Anti-kt R=0.4 and R=1.0 jets using in situ techniques is presented. The reconstruction and performance in data and simulation of the MET obtained with different class of jets and different pile-up suppression schemes, including novel machine learning techniques, are also presented.
Jet grooming is an important strategy for analyzing relativistic particle collisions in the presence of contaminating radiation. Most jet grooming techniques introduce hard cutoffs to remove soft radiation, leading to discontinuous behavior and associated experimental and theoretical challenges. In this talk, I introduce Pileup and Infrared Radiation Annihilation (PIRANHA), a paradigm for continuous jet grooming which overcomes the discontinuity and infrared sensitivity of hard cutoff grooming procedures. I motivate PIRANHA from the perspective of optimal transport and introduce a tree-based, computationally inexpensive implementation of PIRANHA called Recursive Subtraction. Finally, I demonstrate the performance of Recursive Subtraction in mitigating sensitivity to soft distortions, such as hadronization and detector effects, and additive contamination from pileup.
The global Feature Extractor (gFEX) is a level one hardware trigger system that is a component of the Phase I ATLAS trigger upgrades. The goal of the gFEX trigger design and implementation is to target global quantities like large radius jets, specifically those large radius jets with interesting substructure. The previous systems are not able to access a large enough area from the calorimeter to effectively calculate and trigger on large radius jets in the available latency. With the gFEX system, this problem is solved by reading in the energy of the entire calorimeter with one single board. These types of jets with interesting substructure are often generated from boosted objects like the Higgs, W, and Z bosons, making the gFEX an invaluable tool for Runs 3 and 4 at the LHC. We will discuss the gFEX algorithms and their expected impact on performance as well as the benefits generated from the novel use of a System on Chip in the real time data path, which introduces the possibility of additional analysis at high rate.
We present the paper "Jets and Jet Substructure at Future Colliders," in which we examine the role of jet substructure on the motivation for and design of future energy frontier colliders. In particular, we discuss the need for a vibrant theory and experimental research and development program to extend jet substructure physics into the new regimes probed by future colliders. Jet substructure has organically evolved with a close connection between theorists and experimentalists and has catalyzed exciting innovations in both communities. We expect such developments will play an important role in the future energy frontier physics program.