Gregor Kasieczka

Results on two-particle angular correlations for charged particles emitted in proton-proton collisions at center-of-mass energies of 0.9, 2.36, and 7 TeV are presented, using data collected with the CMS detector over a broad range of pseudorapidity (η) and azimuthal angle (ϕ). Short-range correlations in Δη, which are studied in minimum bias events, are characterized using a simple “independent cluster” parametrization in order to quantify their strength (cluster size) and their extent in η (cluster decay width). Long-range azimuthal correlations are studied differentially as a function of charged particle multiplicity and particle transverse momentum using a 980 nb−1 data set at 7 TeV. In high multiplicity events, a pronounced structure emerges in the two-dimensional correlation function for particle pairs with intermediate p T of 1–3 GeV/c, 2.0 < |Δη| < 4.8 and Δϕ ≈ 0. This is the first observation of such a long-range, near-side feature in two-particle correlation functions in pp or $$ p\overline p $$ collisions.

Jet production in PbPb collisions at a nucleon-nucleon center-of-mass energy of 2.76 TeV was studied with the CMS detector at the LHC, using a data sample corresponding to an integrated luminosity of 6.7 inverse microbarns. Jets are reconstructed using the energy deposited in the CMS calorimeters and studied as a function of collision centrality. With increasing collision centrality, a striking imbalance in dijet transverse momentum is observed, consistent with jet quenching. The observed effect extends from the lower cut-off used in this study (jet transverse momentum = 120 GeV/c) up to the statistical limit of the available data sample (jet transverse momentum approximately 210 GeV/c). Correlations of charged particle tracks with jets indicate that the momentum imbalance is accompanied by a softening of the fragmentation pattern of the second most energetic, away-side jet. The dijet momentum balance is recovered when integrating low transverse momentum particles distributed over a wide angular range relative to the direction of the away-side jet.

Charged-hadron transverse-momentum and pseudorapidity distributions in proton-proton collisions at sqrt(s) = 7 TeV are measured with the inner tracking system of the CMS detector at the LHC. The charged-hadron yield is obtained by counting the number of reconstructed hits, hit-pairs, and fully reconstructed charged-particle tracks. The combination of the three methods gives a charged-particle multiplicity per unit of pseudorapidity, dN(charged)/d(eta), for |eta| &lt; 0.5, of 5.78 +/- 0.01 (stat) +/- 0.23 (syst) for non-single-diffractive events, higher than predicted by commonly used models. The relative increase in charged-particle multiplicity from sqrt(s) = 0.9 to 7 TeV is 66.1% +/- 1.0% (stat) +/- 4.2% (syst). The mean transverse momentum is measured to be 0.545 +/- 0.005 (stat) +/- 0.015 (syst) GeV/c. The results are compared with similar measurements at lower energies.

Machine learning based on convolutional neural networks can be used to study jet images from the LHC. Top tagging in fat jets offers a well-defined framework to establish our DeepTop approach and compare its performance to QCD-based top taggers. We first optimize a network architecture to identify top quarks in Monte Carlo simulations of the Standard Model production channel. Using standard fat jets we then compare its performance to a multivariate QCD-based top tagger. We find that both approaches lead to comparable performance, establishing convolutional networks as a promising new approach for multivariate hypothesis-based top tagging.

Autoencoder networks, trained only on QCD jets, can be used to search for anomalies in jet-substructure. We show how, based either on images or on 4-vectors, they identify jets from decays of arbitrary heavy resonances. To control the backgrounds and the underlying systematics we can de-correlate the jet mass using an adversarial network. Such an adversarial autoencoder allows for a general and at the same time easily controllable search for new physics. Ideally, it can be trained and applied to data in the same phase space region, allowing us to efficiently search for new physics using un-supervised learning.

We introduce a new and highly efficient tagger for hadronically decaying top quarks, based on a deep neural network working with Lorentz vectors and the Minkowski metric.With its novel machine learning setup and architecture it allows us to identify boosted top quarks not only from calorimeter towers, but also including tracking information.We show how the performance of our tagger compares with QCD-inspired and image-recognition approaches and find that it significantly increases the performance for strongly boosted top quarks.

Based on the established task of identifying boosted, hadronically decaying top quarks, we compare a wide range of modern machine learning approaches. Unlike most established methods they rely on low-level input, for instance calorimeter output. While their network architectures are vastly different, their performance is comparatively similar. In general, we find that these new approaches are extremely powerful and great fun.

A new paradigm for data-driven, model-agnostic new physics searches at colliders is emerging, and aims to leverage recent breakthroughs in anomaly detection and machine learning. In order to develop and benchmark new anomaly detection methods within this framework, it is essential to have standard datasets. To this end, we have created the LHC Olympics 2020, a community challenge accompanied by a set of simulated collider events. Participants in these Olympics have developed their methods using an R&D dataset and then tested them on black boxes: datasets with an unknown anomaly (or not). Methods made use of modern machine learning tools and were based on unsupervised learning (autoencoders, generative adversarial networks, normalizing flows), weakly supervised learning, and semi-supervised learning. This paper will review the LHC Olympics 2020 challenge, including an overview of the competition, a description of methods deployed in the competition, lessons learned from the experience, and implications for data analyses with future datasets as well as future colliders.

First-principle simulations are at the heart of the high-energy physics research program. They link the vast data output of multi-purpose detectors with fundamental theory predictions and interpretation. This review illustrates a wide range of applications of modern machine learning to event generation and simulation-based inference, including conceptional developments driven by the specific requirements of particle physics. New ideas and tools developed at the interface of particle physics and machine learning will improve the speed and precision of forward simulations, handle the complexity of collision data, and enhance inference as an inverse simulation problem.

The spectra of strange hadrons are measured in proton-proton collisions, recorded by the CMS experiment at the CERN LHC, at centre-of-mass energies of 0.9 and 7 TeV. The $ {\text{K}}_{\text{S}}^0 $ , Λ, and Ξ− particles and their antiparticles are reconstructed from their decay topologies and the production rates are measured as functions of rapidity and transverse momentum, p T. The results are compared to other experiments and to predictions of the Pythia Monte Carlo program. The p T distributions are found to differ substantially from the Pythia results and the production rates exceed the predictions by up to a factor of three.

This report of the BOOST2012 workshop presents the results of four working groups that studied key aspects of jet substructure. We discuss the potential of first-principle QCD calculations to yield a precise description of the substructure of jets and study the accuracy of state-of-the-art Monte Carlo tools. Limitations of the experiments' ability to resolve substructure are evaluated, with a focus on the impact of additional (pile-up) proton proton collisions on jet substructure performance in future LHC operating scenarios. A final section summarizes the lessons learnt from jet substructure analyses in searches for new physics in the production of boosted top quarks.

Measurements of inclusive W and Z boson production cross sections in pp collisions at $ \sqrt {s} = 7 $ TeV are presented, based on 2.9 pb−1 of data recorded by the CMS detector at the LHC. The measurements, performed in the electron and muon decay channels, are combined to give $ \sigma \left( {{\text{pp}} \to {\text{W}}X} \right) \times \mathcal{B}\left( {{\text{W}} \to \ell \nu } \right) = 9.95\pm 0.07\left( {{\text{stat}}{.}} \right)\pm 0.28\left( {{\text{syst}}{.}} \right)\pm 1.09 $ (lumi.) nb and $ \sigma \left( {{\text{pp}} \to {\text{Z}}X} \right) \times \mathcal{B}\left( {Z \to {\ell^{+} }{\ell^{-} }} \right) = 0.931\pm 0.026\left( {{\text{stat}}{.}} \right)\pm 0.023\left( {{\text{syst}}{.}} \right)\pm 0.102 $ (lumi.) nb, where ℓ stands for either e or μ. Theoretical predictions, calculated at the next-to-next-to-leading order in QCD using recent parton distribution functions, are in agreement with the measured cross sections. Ratios of cross sections, which incur an experimental systematic uncertainty of less than 4%, are also reported.

A search for narrow resonances in the dijet mass spectrum is performed using data corresponding to an integrated luminosity of $2.9\text{ }\text{ }{\mathrm{pb}}^{\ensuremath{-}1}$ collected by the CMS experiment at the Large Hadron Collider. Upper limits at the 95% confidence level are presented on the product of the resonance cross section, branching fraction into dijets, and acceptance, separately for decays into quark-quark, quark-gluon, or gluon-gluon pairs. The data exclude new particles predicted in the following models at the 95% confidence level: string resonances, with mass less than 2.50 TeV, excited quarks, with mass less than 1.58 TeV, and axigluons, colorons, and ${E}_{6}$ diquarks, in specific mass intervals. This extends previously published limits on these models.

The first LHC pp collisions at centre-of-mass energies of 0.9 and 2.36 TeV were recorded by the CMS detector in December 2009. The trajectories of charged particles produced in the collisions were reconstructed using the all-silicon Tracker and their momenta were measured in the 3.8 T axial magnetic field. Results from the Tracker commissioning are presented including studies of timing, efficiency, signal-to-noise, resolution, and ionization energy. Reconstructed tracks are used to benchmark the performance in terms of track and vertex resolutions, reconstruction of decays, estimation of ionization energy loss, as well as identification of photon conversions, nuclear interactions, and heavy-flavour decays.

Measurements of primary charged hadron multiplicity distributions are presented for non-single-diffractive events in proton-proton collisions at centre-of-mass energies of $ \sqrt {s} = 0.9 $ , 2.36, and 7 TeV, in five pseudorapidity ranges from |η| < 0.5 to |η| < 2.4. The data were collected with the minimum-bias trigger of the CMS experiment during the LHC commissioning runs in 2009 and the 7 TeV run in 2010. The multiplicity distribution at $ \sqrt {s} = 0.9\;{\text{TeV}} $ is in agreement with previous measurements. At higher energies the increase of the mean multiplicity with $ \sqrt {s} $ is underestimated by most event generators. The average transverse momentum as a function of the multiplicity is also presented. The measurement of higher-order moments of the multiplicity distribution confirms the violation of Koba-Nielsen-Olesen scaling that has been observed at lower energies.

Over the past decade, a large number of jet substructure observables have been proposed in the literature, and explored at the LHC experiments. Such observables attempt to utilize the internal structure of jets in order to distinguish those initiated by quarks, gluons, or by boosted heavy objects, such as top quarks and W bosons. This report, originating from and motivated by the BOOST2013 workshop, presents original particle-level studies that aim to improve our understanding of the relationships between jet substructure observables, their complementarity, and their dependence on the underlying jet properties, particularly the jet radius and jet transverse momentum. This is explored in the context of quark/gluon discrimination, boosted W boson tagging and boosted top quark tagging.

The first measurement of the cross section for top-quark pair production in pp collisions at the LHC at center-of-mass energy sqrt(s)= 7 TeV has been performed using 3.1 {\pm} 0.3 inverse pb of data recorded by the CMS detector. This result utilizes the final state with two isolated, highly energetic charged leptons, large missing transverse energy, and two or more jets. Backgrounds from Drell-Yan and non-W/Z boson production are estimated from data. Eleven events are observed in the data with 2.1 {\pm} 1.0 events expected from background. The measured cross section is 194 {\pm} 72 (stat.) {\pm} 24 (syst.) {\pm} 21 (lumi.) pb, consistent with next-to-leading order predictions.

LHC analyses directly comparing data and simulated events bear the danger of using first-principle predictions only as a black-box part of event simulation. We show how simulations, for instance, of detector effects can instead be inverted using generative networks. This allows us to reconstruct parton level information from measured events. Our results illustrate how, in general, fully conditional generative networks can statistically invert Monte Carlo simulations. As a technical by-product we show how a maximum mean discrepancy loss can be staggered or cooled.

A critical question concerning generative networks applied to event generation in particle physics is if the generated events add statistical precision beyond the training sample. We show for a simple example with increasing dimensionality how generative networks indeed amplify the training statistics. We quantify their impact through an amplification factor or equivalent numbers of sampled events.

For simulations where the forward and the inverse directions have a physics meaning, invertible neural networks are especially useful. A conditional INN can invert a detector simulation in terms of high-level observables, specifically for ZW production at the LHC. It allows for a per-event statistical interpretation. Next, we allow for a variable number of QCD jets. We unfold detector effects and QCD radiation to a pre-defined hard process, again with a per-event probabilistic interpretation over parton-level phase space.

Abstract Accurate simulation of physical processes is crucial for the success of modern particle physics. However, simulating the development and interaction of particle showers with calorimeter detectors is a time consuming process and drives the computing needs of large experiments at the LHC and future colliders. Recently, generative machine learning models based on deep neural networks have shown promise in speeding up this task by several orders of magnitude. We investigate the use of a new architecture—the Bounded Information Bottleneck Autoencoder—for modelling electromagnetic showers in the central region of the Silicon-Tungsten calorimeter of the proposed International Large Detector. Combined with a novel second post-processing network, this approach achieves an accurate simulation of differential distributions including for the first time the shape of the minimum-ionizing-particle peak compared to a full Geant4 simulation for a high-granularity calorimeter with 27k simulated channels. The results are validated by comparing to established architectures. Our results further strengthen the case of using generative networks for fast simulation and demonstrate that physically relevant differential distributions can be described with high accuracy.

Bayesian neural networks allow us to keep track of uncertainties, for example in top tagging, by learning a tagger output together with an error band. We illustrate the main features of Bayesian versions of established deep-learning taggers. We show how they capture statistical uncertainties from finite training samples, systematics related to the jet energy scale, and stability issues through pile-up. Altogether, Bayesian networks offer many new handles to understand and control deep learning at the LHC without introducing a visible prior effect and without compromising the network performance.

We propose a new model-agnostic search strategy for physics beyond the standard model (BSM) at the LHC, based on a novel application of neural density estimation to anomaly detection. Our approach, which we call Classifying Anomalies THrough Outer Density Estimation (CATHODE), assumes the BSM signal is localized in a signal region (defined e.g. using invariant mass). By training a conditional density estimator on a collection of additional features outside the signal region, interpolating it into the signal region, and sampling from it, we produce a collection of events that follow the background model. We can then train a classifier to distinguish the data from the events sampled from the background model, thereby approaching the optimal anomaly detector. Using the LHC Olympics R&D dataset, we demonstrate that CATHODE nearly saturates the best possible performance, and significantly outperforms other approaches that aim to enhance the bump hunt (CWoLa Hunting and ANODE). Finally, we demonstrate that CATHODE is very robust against correlations between the features and maintains nearly-optimal performance even in this more challenging setting.

Compelling experimental evidence suggests the existence of new physics beyond the well-established and tested standard model of particle physics. Various current and upcoming experiments are searching for signatures of new physics. Despite the variety of approaches and theoretical models tested in these experiments, what they all have in common is the very large volume of complex data that they produce. This data challenge calls for powerful statistical methods. Machine learning has been in use in high-energy particle physics for well over a decade, but the rise of deep learning in the early 2010s has yielded a qualitative shift in terms of the scope and ambition of research. These modern machine learning developments are the focus of the present Review, which discusses methods and applications for new physics searches in the context of terrestrial high-energy physics experiments, including the Large Hadron Collider, rare event searches and neutrino experiments. Owing to the growing volumes of data from high-energy physics experiments, modern deep learning methods are playing an increasingly important role in all aspects of data taking and analysis. This Review provides an overview of key developments, with a focus on the search for physics beyond the standard model.

With the vast data-collecting capabilities of current and future high-energy collider experiments, there is an increasing demand for computationally efficient simulations. Generative machine learning models enable fast event generation, yet so far these approaches are largely constrained to fixed data structures and rigid detector geometries. In this paper, we introduce EPiC-GAN - equivariant point cloud generative adversarial network - which can produce point clouds of variable multiplicity. This flexible framework is based on deep sets and is well suited for simulating sprays of particles called jets. The generator and discriminator utilize multiple EPiC layers with an interpretable global latent vector. Crucially, the EPiC layers do not rely on pairwise information sharing between particles, which leads to a significant speed-up over graph- and transformer-based approaches with more complex relation diagrams. We demonstrate that EPiC-GAN scales well to large particle multiplicities and achieves high generation fidelity on benchmark jet generation tasks.

Abstract We explore the use of normalizing flows to emulate Monte Carlo detector simulations of photon showers in a high-granularity electromagnetic calorimeter prototype for the International Large Detector (ILD). Our proposed method — which we refer to as “Layer-to-Layer Flows” ( L2LFlows ) — is an evolution of the CaloFlow architecture adapted to a higher-dimensional setting (30 layers of 10× 10 voxels each). The main innovation of L2LFlows consists of introducing 30 separate normalizing flows, one for each layer of the calorimeter, where each flow is conditioned on the previous five layers in order to learn the layer-to-layer correlations. We compare our results to the BIB-AE, a state-of-the-art generative network trained on the same dataset and find our model has a significantly improved fidelity.

We introduce a new technique named Latent CATHODE (LaCATHODE) for performing "enhanced bump hunts", a type of resonant anomaly search that combines conventional one-dimensional bump hunts with a model-agnostic anomaly score in an auxiliary feature space where potential signals could also be localized. The main advantage of LaCATHODE over existing methods is that it provides an anomaly score that is well behaved when evaluating it beyond the signal region, which is essential to prevent the sculpting of background distributions in the bump hunt. LaCATHODE accomplishes this by constructing the anomaly score directly in the latent space learned by a conditional normalizing flow trained on sideband regions. We demonstrate the superior stability and comparable performance of LaCATHODE for enhanced bump hunting in an illustrative toy example as well as on the LHC Olympics R&D dataset.

Physics beyond the Standard Model that is resonant in one or more dimensions has been a longstanding focus of countless searches at colliders and beyond. Recently, many new strategies for resonant anomaly detection have been developed, where sideband information can be used in conjunction with modern machine learning, in order to generate synthetic datasets representing the Standard Model background. Until now, this approach was only able to accommodate a relatively small number of dimensions, limiting the breadth of the search sensitivity. Using recent innovations in point cloud generative models, we show that this strategy can also be applied to the full phase space, using all relevant particles for the anomaly detection. As a proof of principle, we show that the signal from the R dataset from the LHC Olympics is findable with this method, opening up the door to future studies that explore the interplay between depth and breadth in the representation of the data for anomaly detection.

Weakly supervised methods have emerged as a powerful tool for model-agnostic anomaly detection at the Large Hadron Collider (LHC). While these methods have shown remarkable performance on specific signatures such as dijet resonances, their application in a more model-agnostic manner requires dealing with a larger number of potentially noisy input features. In this paper, we show that using boosted decision trees as classifiers in weakly supervised anomaly detection gives superior performance compared to deep neural networks. Boosted decision trees are well known for their effectiveness in tabular data analysis. Our results show that they not only offer significantly faster training and evaluation times, but they are also robust to a large number of noisy input features. By using advanced gradient boosted decision trees in combination with ensembling techniques and an extended set of features, we significantly improve the performance of weakly supervised methods for anomaly detection at the LHC. This advance is a crucial step toward a more model-agnostic search for new physics.

Abstract Machine learning-based anomaly detection (AD) methods are promising tools for extending the coverage of searches for physics beyond the Standard Model (BSM). One class of AD methods that has received significant attention is resonant anomaly detection, where the BSM physics is assumed to be localized in at least one known variable. While there have been many methods proposed to identify such a BSM signal that make use of simulated or detected data in different ways, there has not yet been a study of the methods’ complementarity. To this end, we address two questions. First, in the absence of any signal, do different methods pick the same events as signal-like? If not, then we can significantly reduce the false-positive rate by comparing different methods on the same dataset. Second, if there is a signal, are different methods fully correlated? Even if their maximum performance is the same, since we do not know how much signal is present, it may be beneficial to combine approaches. Using the Large Hadron Collider (LHC) Olympics dataset, we provide quantitative answers to these questions. We find that there are significant gains possible by combining multiple methods, which will strengthen the search program at the LHC and beyond.

A search for neutral minimal supersymmetric standard model (MSSM) Higgs bosons in pp collisions at the LHC at a center-of-mass energy of 7 TeV is presented. The results are based on a data sample corresponding to an integrated luminosity of 36 pb(-1) recorded by the CMS experiment. The search uses decays of the Higgs bosons to tau pairs. No excess is observed in the tau-pair invariant-mass spectrum. The resulting upper limits on the Higgs boson production cross section times branching fraction to tau pairs, as a function of the pseudoscalar Higgs boson mass, yield stringent new bounds in the MSSM parameter space.

A search for Z bosons in the μ(+)μ(-) decay channel has been performed in PbPb collisions at √S(NN)=2.76 TeV with the CMS detector at the LHC, in a 7.2 μb(-1) data sample. The number of opposite-sign muon pairs observed in the 60-120 GeV/c(2) invariant mass range is 39, corresponding to a yield per unit of rapidity (y) and per minimum bias event of [33.8±5.5(stat)±4.4(syst)]×10(-8), in the |y|<2.0 range. Rapidity, transverse momentum, and centrality dependencies are also measured. The results agree with next-to-leading order QCD calculations, scaled by the number of incoherent nucleon-nucleon collisions.

A search for microscopic black hole production and decay in pp collisions at a center-of-mass energy of 7 TeV has been conducted by the CMS Collaboration at the LHC, using a data sample corresponding to an integrated luminosity of 35 pb−1. Events with large total transverse energy are analyzed for the presence of multiple high-energy jets, leptons, and photons, typical of a signal expected from a microscopic black hole. Good agreement with the standard model backgrounds, dominated by QCD multijet production, is observed for various final-state multiplicities and model-independent limits on new physics in these final states are set. Using simple semi-classical approximation, limits on the minimum black hole mass are derived as well, in the range 3.5–4.5 TeV. These are the first direct limits on black hole production at a particle accelerator.

Measurements of the total and differential cross sections with respect to transverse momentum and rapidity for B+ mesons produced in pp collisions at sqrt(s) = 7 TeV are presented. The data correspond to an integrated luminosity of 5.8 inverse picobarns collected by the CMS experiment operating at the LHC. The exclusive decay B+ to J/psi K+, with the J/psi decaying to an oppositely charged muon pair, is used to detect B+ mesons and to measure the production cross section as a function of the transverse momentum and rapidity of the B. The total cross section for p_t(B) > 5 GeV and |y(B)| < 2.4 is measured to be 28.1 +/- 2.4 +/- 2.0 +/- 3.1 microbarns, where the first uncertainty is statistical, the second is systematic, and the last is from the luminosity measurement.

The differential cross section for the inclusive production of isolated prompt photons has been measured as a function of the photon transverse energy E(T)(γ) in pp collisions at √s=7 TeV using data recorded by the CMS detector at the LHC. The data sample corresponds to an integrated luminosity of 2.9 pb(-1). Photons are required to have a pseudorapidity |η(γ)|<1.45 and E(T)(γ)>21 GeV, covering the kinematic region 0.006<x(T)<0.086. The measured cross section is found to be in agreement with next-to-leading-order perturbative QCD calculations.

The performance of top taggers, for example in resonance searches, can be significantly enhanced through an increased set of variables, with a special focus on final-state radiation. We study the production and the decay of a heavy gauge boson in the upcoming LHC run. For constant signal efficiency, the multivariate analysis achieves an increased background rejection by up to a factor 30 compared to our previous tagger. Based on this study and the documentation in the appendix we release a new HEPTopTagger2 for the upcoming LHC run. It now includes an optimal choice of the size of the fat jet, N-subjettiness, and different modes of Qjets.

Distinguishing quarks from gluons based on low-level detector output is one of the most challenging applications of multi-variate and machine learning techniques at the LHC. We first show the performance of our 4-vector-based LoLa tagger without and after considering detector effects. We then discuss two benchmark applications, mono-jet searches with a gluon-rich signal and di-jet resonances with a quark-rich signal. In both cases an immediate benefit compared to the standard event-level analysis exists.

Significant advances in deep learning have led to more widely used and precise neural network-based generative models such as Generative Adversarial Networks (GANs). We introduce a post-hoc correction to deep generative models to further improve their fidelity, based on the Deep neural networks using the Classification for Tuning and Reweighting (DCTR) protocol. The correction takes the form of a reweighting function that can be applied to generated examples when making predictions from the simulation. We illustrate this approach using GANs trained on standard multimodal probability densities as well as calorimeter simulations from high energy physics. We show that the weighted GAN examples significantly improve the accuracy of the generated samples without a large loss in statistical power. This approach could be applied to any generative model and is a promising refinement method for high energy physics applications and beyond.

While deep learning has proven to be extremely successful at supervised classification tasks at the LHC and beyond, for practical applications, raw classification accuracy is often not the only consideration. One crucial issue is the stability of network predictions, either versus changes of individual features of the input data, or against systematic perturbations. We present a new method based on a novel application of "distance correlation" (DisCo), a measure quantifying non-linear correlations, that achieves equal performance to state-of-the-art adversarial decorrelation networks but is much simpler and more stable to train. To demonstrate the effectiveness of our method, we carefully recast a recent ATLAS study of decorrelation methods as applied to boosted, hadronic W-tagging. We also show the feasibility of DisCo regularization for more powerful convolutional neural networks, as well as for the problem of hadronic top tagging.

The ABCD method is one of the most widely used data-driven background estimation techniques in high energy physics. Cuts on two statistically independent classifiers separate signal and background into four regions, so that background in the signal region can be estimated simply using the other three control regions. Typically, the independent classifiers are chosen ``by hand'' to be intuitive and physically motivated variables. Here, we explore the possibility of automating the design of one or both of these classifiers using machine learning. We show how to use state-of-the-art decorrelation methods to construct powerful yet independent discriminators. Along the way, we uncover a previously unappreciated aspect of the ABCD method: its accuracy hinges on having low signal contamination in control regions not just overall, but relative to the signal fraction in the signal region. We demonstrate the method with three examples: a simple model consisting of three-dimensional Gaussians; boosted hadronic top jet tagging; and a recasted search for paired dijet resonances. In all cases, automating the ABCD method with machine learning significantly improves performance in terms of ABCD closure, background rejection, and signal contamination.

Given the increasing data collection capabilities and limited computing resources of future collider experiments, interest in using generative neural networks for the fast simulation of collider events is growing. In our previous study, the Bounded Information Bottleneck Autoencoder (BIB-AE) architecture for generating photon showers in a high-granularity calorimeter showed a high accuracy modeling of various global differential shower distributions. In this work, we investigate how the BIB-AE encodes this physics information in its latent space. Our understanding of this encoding allows us to propose methods to optimize the generation performance further, for example, by altering latent space sampling or by suggesting specific changes to hyperparameters. In particular, we improve the modeling of the shower shape along the particle incident axis.

Abstract Motivated by the computational limitations of simulating interactions of particles in highly-granular detectors, there exists a concerted effort to build fast and exact machine-learning-based shower simulators. This work reports progress on two important fronts. First, the previously investigated Wasserstein generative adversarial network and bounded information bottleneck autoencoder generative models are improved and successful learning of hadronic showers initiated by charged pions in a segment of the hadronic calorimeter of the International Large Detector is demonstrated for the first time. Second, we consider how state-of-the-art reconstruction software applied to generated shower energies affects the obtainable energy response and resolution. While many challenges remain, these results constitute an important milestone in using generative models in a realistic setting.

Collider searches face the challenge of defining a representation of high-dimensional data such that physical symmetries are manifest, the discriminating features are retained, and the choice of representation is new-physics agnostic. We introduce JetCLR to solve the mapping from low-level data to optimized observables through self-supervised contrastive learning. As an example, we construct a data representation for top and QCD jets using a permutation-invariant transformer-encoder network and visualize its symmetry properties. We compare the JetCLR representation with alternative representations using linear classifier tests and find it to work quite well.

Abstract The demands placed on computational resources by the simulation requirements of high energy physics experiments motivate the development of novel simulation tools. Machine learning based generative models offer a solution that is both fast and accurate. In this work we extend the Bounded Information Bottleneck Autoencoder (BIB-AE) architecture, designed for the simulation of particle showers in highly granular calorimeters, in two key directions. First, we generalise the model to a multi-parameter conditioning scenario, while retaining a high degree of physics fidelity. In a second step, we perform a detailed study of the effect of applying a state-of-the-art particle flow-based reconstruction procedure to the generated showers. We demonstrate that the performance of the model remains high after reconstruction. These results are an important step towards creating a more general simulation tool, where maintaining physics performance after reconstruction is the ultimate target.

Abstract Fast simulation of the energy depositions in high-granular detectors is needed for future collider experiments at ever-increasing luminosities. Generative machine learning (ML) models have been shown to speed up and augment the traditional simulation chain in physics analysis. However, the majority of previous efforts were limited to models relying on fixed, regular detector readout geometries. A major advancement is the recently introduced CaloClouds model, a geometry-independent diffusion model, which generates calorimeter showers as point clouds for the electromagnetic calorimeter of the envisioned International Large Detector (ILD). In this work, we introduce CaloClouds II which features a number of key improvements. This includes continuous time score-based modelling, which allows for a 25-step sampling with comparable fidelity to CaloClouds while yielding a 6× speed-up over Geant4 on a single CPU (5× over CaloClouds ). We further distill the diffusion model into a consistency model allowing for accurate sampling in a single step and resulting in a 46× speed-up over Geant4 (37× over CaloClouds ). This constitutes the first application of consistency distillation for the generation of calorimeter showers.

Measurements of dijet azimuthal decorrelations in pp collisions at √s=7 TeV using the CMS detector at the CERN LHC are presented. The analysis is based on an inclusive dijet event sample corresponding to an integrated luminosity of 2.9 pb⁻¹. The results are compared to predictions from perturbative QCD calculations and various Monte Carlo event generators. The dijet azimuthal distributions are found to be sensitive to initial-state gluon radiation.

Measurements of the differential production cross sections dσ/dpTB and dσ/dyB for B0 mesons produced in pp collisions at sqrt[s] = 7 TeV are presented. The data set used was collected by the CMS experiment at the LHC and corresponds to an integrated luminosity of 40 pb-1. The production cross section is measured from B0 meson decays reconstructed in the exclusive final state J/ψKS0, with the subsequent decays J/ψ → μ + μ - and KS0 → π+}π-. The total cross section for pTB>5 GeV and |yB|<2.2 is measured to be 33.2 ± 2.5 ± 3.5 μb, where the first uncertainty is statistical and the second is systematic.

Dijet angular distributions are measured over a wide range of dijet invariant masses in pp collisions at √s = 7 TeV, at the CERN LHC. The event sample, recorded with the CMS detector, corresponds to an integrated luminosity of 36 pb⁻¹. The data are found to be in good agreement with the predictions of perturbative QCD, and yield no evidence of quark compositeness. With a modified frequentist approach, a lower limit on the contact interaction scale for left-handed quarks of Λ⁺ = 5.6 TeV (Λ⁻ = 6.7 TeV) for destructive (constructive) interference is obtained at the 95% confidence level.

A measurement of the double-differential inclusive dijet production cross section in proton–proton collisions at s=7TeV is presented as a function of the dijet invariant mass and jet rapidity. The data correspond to an integrated luminosity of 36pb−1, recorded with the CMS detector at the LHC. The measurement covers the dijet mass range 0.2TeV to 3.5TeV and jet rapidities up to |y|=2.5. It is found to be in good agreement with next-to-leading-order QCD predictions.

Beam test results of the radiation tolerance study of chemical vapour deposition (CVD) diamond against different particle species and energies is presented. We also present beam test results on the independence of signal size on incident particle rate in charged particle detectors based on un-irradiated and irradiated poly-crystalline CVD diamond over a range of particle fluxes from 2 kHz/cm2 to 10 MHz/cm2. The pulse height of the sensors was measured with readout electronics with a peaking time of 6 ns. In addition functionality of poly-crystalline CVD diamond 3D devices was demonstrated in beam tests and 3D diamond detectors are shown to be a promising technology for applications in future high luminosity experiments.

Next-generation radio surveys will yield an unprecedented amount of data, warranting analysis by use of machine learning techniques. Convolutional neural networks are the deep learning technique that has proven to be the most successful in classifying image data. Capsule networks are a more recently developed technique that use capsules comprised of groups of neurons, that describe properties of an image including the relative spatial locations of features. The current work explores the performance of different capsule network architectures against simpler convolutional neural network architectures, in reproducing the classifications into the classes of unresolved, FRI and FRII morphologies. We utilise images from a LOFAR survey which is the deepest, wide-area radio survey to date, revealing more complex radio-source structures compared to previous surveys, presenting further challenges for machine learning algorithms. The 4- and 8-layer convolutional networks attain an average precision of 93.3% and 94.3% respectively, compared to 89.7% obtained with the capsule network, when training on original and augmented images. Implementing transfer learning achieves a precision of 94.4%, that is within the confidence interval of the 8-layer convolutional network. The convolutional networks always outperform any variation of the capsule network, as they prove to be more robust to the presence of noise in images. The use of pooling appears to allow more freedom for the intra-class variability of radio galaxy morphologies, as well as reducing the impact of noise.

The past few years have seen a rapid development of machine-learning algorithms. While surely augmenting performance, these complex tools are often treated as black-boxes and may impair our understanding of the physical processes under study. The aim of this paper is to move a first step into the direction of applying expert-knowledge in particle physics to calculate the optimal decision function and test whether it is achieved by standard training, thus making the aforementioned black-box more transparent. In particular, we consider the binary classification problem of discriminating quark-initiated jets from gluon-initiated ones. We construct a new version of the widely used N-subjettiness, which features a simpler theoretical behaviour than the original one, while maintaining, if not exceeding, the discrimination power. We input these new observables to the simplest possible neural network, i.e. the one made by a single neuron, or perceptron, and we analytically study the network behaviour at leading logarithmic accuracy. We are able to determine under which circumstances the perceptron achieves optimal performance. We also compare our analytic findings to an actual implementation of a perceptron and to a more realistic neural network and find very good agreement.

Abstract 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. We show the same effect for a physics simulation, specifically photon showers in an electromagnetic calorimeter.

There is a growing need for machine-learning-based anomaly detection strategies to broaden the search for beyond-the-Standard-Model physics at the Large Hadron Collider (LHC) and elsewhere. The first step of any anomaly detection approach is to specify observables and then use them to decide on a set of anomalous events. One common choice is to select events that have low probability density. It is a well-known fact that probability densities are not invariant under coordinate transformations, so the sensitivity can depend on the initial choice of coordinates. The broader machine learning community has recently connected coordinate sensitivity with anomaly detection and our goal is to bring awareness of this issue to the growing high-energy physics literature on anomaly detection. In addition to analytical explanations, we provide numerical examples from simple random variables and from the LHC Olympics dataset that show how using probability density as an anomaly score can lead to events being classified as anomalous or not depending on the coordinate frame.

A search for narrow resonances at high mass in the dimuon and dielectron channels has been performed by the CMS experiment at the CERN LHC, using pp collision data recorded at $ \sqrt {s} = 7 $ TeV. The event samples correspond to integrated luminosities of 40 pb−1 in the dimuon channel and 35 pb−1 in the dielectron channel. Heavy dilepton resonances are predicted in theoretical models with extra gauge bosons (Z′) or as Kaluza-Klein graviton excitations (GKK) in the Randall-Sundrum model. Upper limits on the inclusive cross section of Z′(GKK) → ℓ + ℓ − relative to Z → ℓ + ℓ − are presented. These limits exclude at 95% confidence level a Z′ with standard-model-like couplings below 1140GeV, the superstring-inspired Z ψ ′ below 887 GeV, and, for values of the coupling parameter $ {{k} \left/ {{{{\overline M }_{\text{Pl}}}}} \right.} $ of 0.05 (0.1), Kaluza-Klein gravitons below 855 (1079) GeV.

Bose-Einstein correlations have been measured using samples of proton-proton collisions at 0.9 and 2.36 TeV center-of-mass energies, recorded by the CMS experiment at the CERN Large Hadron Collider. The signal is observed in the form of an enhancement of pairs of same-sign charged particles with small relative four-momentum. The size of the correlated particle emission region is seen to increase significantly with the particle multiplicity of the event.

A search for supersymmetry in the context of general gauge-mediated breaking with the lightest neutralino as the next-to-lightest supersymmetric particle and the gravitino as the lightest is presented. The data sample corresponds to an integrated luminosity of 36 pb(-1) recorded by the CMS experiment at the LHC. The search is performed by using events containing two or more isolated photons, at least one hadronic jet, and significant missing transverse energy. No excess of events at high missing transverse energy is observed. Upper limits on the signal cross section for general gauge-mediated supersymmetry between 0.3 and 1.1 pb at the 95% confidence level are determined for a range of squark, gluino, and neutralino masses, excluding supersymmetry parameter space that was inaccessible to previous experiments.

A measurement of Wγ and Zγ production in proton–proton collisions at s=7TeV is presented. Results are based on a data sample recorded by the CMS experiment at the LHC, corresponding to an integrated luminosity of 36pb−1. The electron and muon decay channels of the W and Z are used. The total cross sections are measured for photon transverse energy ETγ>10GeV and spatial separation from charged leptons in the plane of pseudorapidity and azimuthal angle ΔR(ℓ,γ)>0.7, and with an additional dilepton invariant mass requirement of Mℓℓ>50GeV for the Zγ process. The following cross section times branching fraction values are found: σ(pp→Wγ+X)×B(W→ℓν)=56.3±5.0(stat.)±5.0(syst.)±2.3(lumi.)pb and σ(pp→Zγ+X)×B(Z→ℓℓ)=9.4±1.0(stat.)±0.6(syst.)±0.4(lumi.)pb. These measurements are in agreement with standard model predictions. The first limits on anomalous WWγ, ZZγ, and Zγγ trilinear gauge couplings at s=7TeV are set.

A measurement of the angular correlations between beauty and anti-beauty hadrons ( $ {\text{B}}\overline {\text{B}} $ ) produced in pp collisions at a centre-of-mass energy of 7 TeV at the CERN LHC is presented, probing for the first time the region of small angular separation. The B hadrons are identified by the presence of displaced secondary vertices from their decays. The B hadron angular separation is reconstructed from the decay vertices and the primary-interaction vertex. The differential $ {\text{B}}\overline {\text{B}} $ production cross section, measured from a data sample collected by CMS and corresponding to an integrated luminosity of 3.1 pb−1, shows that a sizable fraction of the $ {\text{B}}\overline {\text{B}} $ pairs are produced with small opening angles. These studies provide a test of QCD and further insight into the dynamics of $ {\text{b}}\overline {\text{b}} $ production.

The results of the first search for long-lived gluinos produced in 7 TeV $pp$ collisions at the CERN Large Hadron Collider are presented. The search looks for evidence of long-lived particles that stop in the CMS detector and decay in the quiescent periods between beam crossings. In a dataset with a peak instantaneous luminosity of $1\ifmmode\times\else\texttimes\fi{}{10}^{32}\text{ }\text{ }{\mathrm{cm}}^{\ensuremath{-}2}\text{ }{\mathrm{s}}^{\ensuremath{-}1}$, an integrated luminosity of $10\text{ }\text{ }{\mathrm{pb}}^{\ensuremath{-}1}$, and a search interval corresponding to 62 hours of LHC operation, no significant excess above background was observed. Limits at the 95% confidence level on gluino pair production over 13 orders of magnitude of gluino lifetime are set. For a mass difference ${m}_{\stackrel{\texttildelow{}}{g}}\ensuremath{-}{m}_{{\stackrel{\texttildelow{}}{\ensuremath{\chi}}}_{1}^{0}}&gt;100\text{ }\text{ }\mathrm{GeV}/{c}^{2}$, and assuming $\mathrm{BR}(\stackrel{\texttildelow{}}{g}\ensuremath{\rightarrow}g{\stackrel{\texttildelow{}}{\ensuremath{\chi}}}_{1}^{0})=100%$, ${m}_{\stackrel{\texttildelow{}}{g}}&lt;370\text{ }\text{ }\mathrm{GeV}/{c}^{2}$ are excluded for lifetimes from $10\text{ }\text{ }\ensuremath{\mu}\mathrm{s}$ to 1000 s.

We show how to treat systematic uncertainties using Bayesian deep networks for regression. First, we analyze how these networks separately trace statistical and systematic uncertainties on the momenta of boosted top quarks forming fat jets. Next, we propose a novel calibration procedure by training on labels and their error bars. Again, the network cleanly separates the different uncertainties. As a technical side effect, we show how Bayesian networks can be extended to describe non-Gaussian features.

Abstract We introduce a Python package that provides simple and unified access to a collection of datasets from fundamental physics research—including particle physics, astroparticle physics, and hadron- and nuclear physics—for supervised machine learning studies. The datasets contain hadronic top quarks, cosmic-ray-induced air showers, phase transitions in hadronic matter, and generator-level histories. While public datasets from multiple fundamental physics disciplines already exist, the common interface and provided reference models simplify future work on cross-disciplinary machine learning and transfer learning in fundamental physics. We discuss the design and structure and line out how additional datasets can be submitted for inclusion. As showcase application, we present a simple yet flexible graph-based neural network architecture that can easily be applied to a wide range of supervised learning tasks. We show that our approach reaches performance close to dedicated methods on all datasets. To simplify adaptation for various problems, we provide easy-to-follow instructions on how graph-based representations of data structures, relevant for fundamental physics, can be constructed and provide code implementations for several of them. Implementations are also provided for our proposed method and all reference algorithms.

With the vast data-collecting capabilities of current and future high-energy collider experiments, there is an increasing demand for computationally efficient simulations. Generative machine learning models enable fast event generation, yet so far these approaches are largely constrained to fixed data structures and rigid detector geometries. In this paper, we introduce EPiC-GAN - equivariant point cloud generative adversarial network - which can produce point clouds of variable multiplicity. This flexible framework is based on deep sets and is well suited for simulating sprays of particles called jets. The generator and discriminator utilize multiple EPiC layers with an interpretable global latent vector. Crucially, the EPiC layers do not rely on pairwise information sharing between particles, which leads to a significant speed-up over graph- and transformer-based approaches with more complex relation diagrams. We demonstrate that EPiC-GAN scales well to large particle multiplicities and achieves high generation fidelity on benchmark jet generation tasks.

A measurement of the underlying activity in scattering processes with transverse momentum scale in the GeV region is performed in proton-proton collisions at sqrt(s) = 0.9 TeV, using data collected by the CMS experiment at the LHC. Charged hadron production is studied with reference to the direction of a leading object, either a charged particle or a set of charged particles forming a jet. Predictions of several QCD-inspired models as implemented in PYTHIA are compared, after full detector simulation, to the data. The models generally predict too little production of charged hadrons with pseudorapidity eta < 2, p_T > 0.5 GeV/c, and azimuthal direction transverse to that of the leading object.

A search for quark compositeness in the form of quark contact interactions, based on hadronic jet pairs (dijets) produced in proton-proton collisions at √s=7 TeV, is described. The data sample of the study corresponds to an integrated luminosity of 2.9 pb(-1) collected with the CMS detector at the LHC. The dijet centrality ratio, which quantifies the angular distribution of the dijets, is measured as a function of the invariant mass of the dijet system and is found to agree with the predictions of the standard model. A statistical analysis of the data provides a lower limit on the energy scale of quark contact interactions. The sensitivity of the analysis is such that the expected limit is 2.9 TeV; because the observed value of the centrality ratio at high invariant mass is below the expectation, the observed limit is 4.0 TeV at the 95% confidence level.

We present a measurement of the ratio of positive to negative muon fluxes from cosmic ray interactions in the atmosphere, using data collected by the CMS detector both at ground level and in the underground experimental cavern at the CERN LHC. Muons were detected in the momentum range from 5 GeV/c to 1 TeV/c. The surface flux ratio is measured to be 1.2766±0.0032(stat.)±0.0032(syst.), independent of the muon momentum, below 100 GeV/c. This is the most precise measurement to date. At higher momenta the data are consistent with an increase of the charge ratio, in agreement with cosmic ray shower models and compatible with previous measurements by deep-underground experiments.

Bose-Einstein correlations between identical particles are measured in samples of proton-proton collisions at 0.9 and 7 TeV centre-of-mass energies, recorded by the CMS experiment at the LHC. The signal is observed in the form of an enhancement of number of pairs of same-sign charged particles with small relative momentum. The dependence of this enhancement on kinematic and topological features of the event is studied. Anticorrelations between same-sign charged particles are observed in the region of relative momenta higher than those in the signal region.

First-principle simulations are at the heart of the high-energy physics research program. They link the vast data output of multi-purpose detectors with fundamental theory predictions and interpretation. This review illustrates a wide range of applications of modern machine learning to event generation and simulation-based inference, including conceptional developments driven by the specific requirements of particle physics. New ideas and tools developed at the interface of particle physics and machine learning will improve the speed and precision of forward simulations, handle the complexity of collision data, and enhance inference as an inverse simulation problem.

A measurement of the b-hadron production cross section in proton-proton collisions at $ \sqrt {s} = 7\;{\text{TeV}} $ is presented. The dataset, corresponding to 85 nb−1, was recorded with the CMS experiment at the LHC using a low-threshold single-muon trigger. Events are selected by the presence of a muon with transverse momentum $ p_T^\mu > 6\;{\text{GeV}} $ with respect to the beam direction and pseudorapidity |η μ | < 2.1. The transverse momentum of the muon with respect to the closest jet discriminates events containing b hadrons from background. The inclusive b-hadron production cross section is presented as a function of muon transverse momentum and pseudorapidity. The measured total cross section in the kinematic acceptance is σ(pp → b + X → μ + X′) = 1.32 ± 0.01(stat) ± 0.30(syst) ± 0.15(lumi)μb.

A search for pair-produced bottom-like quarks in pp collisions at s=7TeV is conducted with the CMS experiment at the LHC. The decay b′→tW is considered in this search. The b′b¯′→tW−t¯W+ process can be identified by the distinctive signature of trileptons and same-sign dileptons. With a data sample corresponding to an integrated luminosity of 34 pb−1, no excess above the standard model background predictions is observed and a b′ quark with a mass between 255 and 361 GeV/c2 is excluded at the 95% confidence level.

A search for a heavy gauge boson W' has been conducted by the CMS experiment at the LHC in the decay channel with an electron and large transverse energy imbalance, using proton-proton collision data corresponding to an integrated luminosity of 36 inverse picobarns. No excess above standard model expectations is seen in the transverse mass distribution of the electron-(missing E_T) system. Assuming standard-model-like couplings and decay branching fractions, a W' boson with a mass less than 1.36 TeV/c^2 is excluded at 95% confidence level.

Capsule networks are ideal tools to combine event-level and subjet information at the LHC. After benchmarking our capsule network against standard convolutional networks, we show how multi-class capsules extract a resonance decaying to top quarks from both, QCD di-jet and the top continuum backgrounds. We then show how its results can be easily interpreted. Finally, we use associated top-Higgs production to demonstrate that capsule networks can work on overlaying images to go beyond calorimeter information.

A new paradigm for data-driven, model-agnostic new physics searches at colliders is emerging, and aims to leverage recent breakthroughs in anomaly detection and machine learning. In order to develop and benchmark new anomaly detection methods within this framework, it is essential to have standard datasets. To this end, we have created the LHC Olympics 2020, a community challenge accompanied by a set of simulated collider events. Participants in these Olympics have developed their methods using an R&D dataset and then tested them on black boxes: datasets with an unknown anomaly (or not). This paper will review the LHC Olympics 2020 challenge, including an overview of the competition, a description of methods deployed in the competition, lessons learned from the experience, and implications for data analyses with future datasets as well as future colliders.

A new heavy gauge boson, W′, decaying to a muon and a neutrino, is searched for in pp collisions at a centre-of-mass energy of 7 TeV. The data, collected with the CMS detector at the LHC, correspond to an integrated luminosity of 36 pb−1. No significant excess of events above the standard model expectation is found in the transverse mass distribution of the muon–neutrino system. Masses below 1.40 TeV are excluded at the 95% confidence level for a sequential standard-model-like W′. The W′ mass lower limit increases to 1.58 TeV when the present analysis is combined with the CMS result for the electron channel.

Top taggers are established analysis tools used to reconstruct boosted hadronically decaying top quarks, for example, in searches for heavy resonances. We first present a dedicated study of signal efficiency versus background rejection, allowing for an improved choice of working points. Next, we determine to what degree our mass drop selection can be improved by systematically including angular correlations between subjets or $N$-subjettiness. Finally, we extend the reach of the top tagger to transverse momenta below the top mass. This momentum range will be crucial in searches for the associated production of a Higgs boson with top quarks.

Confining dark sectors with pseudo-conformal dynamics produce SUEP, or Soft Unclustered Energy Patterns, at colliders: isotropic dark hadrons with soft and democratic energies. We target the experimental nightmare scenario, SUEPs in exotic Higgs decays, where all dark hadrons decay promptly to SM hadrons. First, we identify three promising observables, the charged particle multiplicity, the event ring isotropy, and the matrix of geometric distances between charged tracks. Their patterns can be exploited through a cut-and-count search, supervised machine learning, or an unsupervised autoencoder. We find that the HL-LHC will probe exotic Higgs branching ratios at the per-cent level, even without a detailed knowledge of the signal features. Our techniques can be applied to other SUEP searches, especially the unsupervised strategy, which is independent of overly specific model assumptions and the corresponding precision simulations.

QCD-jets at the LHC are described by simple physics principles. We show how super-resolution generative networks can learn the underlying structures and use them to improve the resolution of jet images. We test this approach on massless QCD-jets and on fat top-jets and find that the network reproduces their main features even without training on pure samples. In addition, we show how a slim network architecture can be constructed once we have control of the full network performance.

We explore the use of normalizing flows to emulate Monte Carlo detector simulations of photon showers in a high-granularity electromagnetic calorimeter prototype for the International Large Detector (ILD). Our proposed method -- which we refer to as "Layer-to-Layer-Flows" (L$2$LFlows) -- is an evolution of the CaloFlow architecture adapted to a higher-dimensional setting (30 layers of $10\times 10$ voxels each). The main innovation of L$2$LFlows consists of introducing $30$ separate normalizing flows, one for each layer of the calorimeter, where each flow is conditioned on the previous five layers in order to learn the layer-to-layer correlations. We compare our results to the BIB-AE, a state-of-the-art generative network trained on the same dataset and find our model has a significantly improved fidelity.

Machine learning techniques that perform morphological classification of astronomical sources often suffer from a scarcity of labelled training data. Here, we focus on the case of supervised deep learning models for the morphological classification of radio galaxies, which is particularly topical for the forthcoming large radio surveys. We demonstrate the use of generative models, specifically Wasserstein GANs (wGANs), to generate data for different classes of radio galaxies. Further, we study the impact of augmenting the training data with images from our wGAN on three different classification architectures. We find that this technique makes it possible to improve models for the morphological classification of radio galaxies. A simple Fully Connected Neural Network (FCN) benefits most from including generated images into the training set, with a considerable improvement of its classification accuracy. In addition, we find it is more difficult to improve complex classifiers. The classification performance of a Convolutional Neural Network (CNN) can be improved slightly. However, this is not the case for a Vision Transformer (ViT).

We study various machine learning based algorithms for performing accurate jet flavor classification on field-programmable gate arrays and demonstrate how latency and resource consumption scale with the input size and choice of algorithm. These architectures provide an initial design for models that could be used for tagging at the CERN LHC during its high-luminosity phase. The high-luminosity upgrade will lead to a five-fold increase in its instantaneous luminosity for proton-proton collisions and, in turn, higher data volume and complexity, such as the availability of jet constituents. Through quantization-aware training and efficient hardware implementations, we show that O(100) ns inference of complex architectures such as deep sets and interaction networks is feasible at a low computational resource cost.

In recent years, neural network-based classification has been used to improve data analysis at collider experiments. While this strategy proves to be hugely successful, the underlying models are not commonly shared with the public and they rely on experiment-internal data as well as full detector simulations. We propose a new strategy, so-called classifier surrogates, to be trained inside the experiments, that only utilise publicly accessible features and truth information. These surrogates approximate the original classifier distribution, and can be shared with the public. Subsequently, such a model can be evaluated by sampling the classification output from high-level information without requiring a sophisticated detector simulation. Technically, we show that Continuous Normalizing Flows are a suitable generative architecture that can be efficiently trained to sample classification results using Conditional Flow Matching. We further demonstrate that these models can be easily extended by Bayesian uncertainties to indicate their degree of validity when confronted with unknown inputs to the user. For a concrete example of tagging jets from hadronically decaying top quarks, we demonstrate the application of flows in combination with uncertainty estimation through either inference of a mean-field Gaussian weight posterior, or Monte Carlo sampling network weights.

Choosing which properties of the data to use as input to multivariate decision algorithms---also known as feature selection---is an important step in solving any problem with machine learning. While there is a clear trend towards training sophisticated deep networks on large numbers of relatively unprocessed inputs (so-called automated feature engineering), for many tasks in physics, sets of theoretically well-motivated and well-understood features already exist. Working with such features can bring many benefits, including greater interpretability, reduced training and run time, and enhanced stability and robustness. We develop a new feature selection method based on distance correlation, and demonstrate its effectiveness on the tasks of boosted top- and $W$-tagging. Using our method to select features from a set of over 7,000 energy flow polynomials, we show that we can match the performance of much deeper architectures, by using only ten features and two orders-of-magnitude fewer model parameters.

A neural network for software compensation was developed for the highly granular CALICE Analogue Hadronic Calorimeter (AHCAL). The neural network uses spatial and temporal event information from the AHCAL and energy information, which is expected to improve sensitivity to shower development and the neutron fraction of the hadron shower. The neural network method produced a depth-dependent energy weighting and a time-dependent threshold for enhancing energy deposits consistent with the timescale of evaporation neutrons. Additionally, it was observed to learn an energy-weighting indicative of longitudinal leakage correction. In addition, the method produced a linear detector response and outperformed a published control method regarding resolution for every particle energy studied.

Foundation models are multi-dataset and multi-task machine learning methods that once pre-trained can be fine-tuned for a large variety of downstream applications. The successful development of such general-purpose models for physics data would be a major breakthrough as they could improve the achievable physics performance while at the same time drastically reduce the required amount of training time and data. We report significant progress on this challenge on several fronts. First, a comprehensive set of evaluation methods is introduced to judge the quality of an encoding from physics data into a representation suitable for the autoregressive generation of particle jets with transformer architectures (the common backbone of foundation models). These measures motivate the choice of a higher-fidelity tokenization compared to previous works. Finally, we demonstrate transfer learning between an unsupervised problem (jet generation) and a classic supervised task (jet tagging) with our new OmniJet-$\alpha$ model. This is the first successful transfer between two different and actively studied classes of tasks and constitutes a major step in the building of foundation models for particle physics.

We propose the first-ever complete, model-agnostic search strategy based on the optimal anomaly score, for new physics on the tails of distributions. Signal sensitivity is achieved via a classifier trained on auxiliary features in a weakly-supervised fashion, and backgrounds are predicted using the ABCD method in the classifier output and the primary tail feature. The independence between the classifier output and the tail feature required for ABCD is achieved by first training a conditional normalizing flow that yields a decorrelated version of the auxiliary features; the classifier is then trained on these features. Both the signal sensitivity and background prediction require a sample of events accurately approximating the SM background; we assume this can be furnished by closely related control processes in the data or by accurate simulations, as is the case in countless conventional analyses. The viability of our approach is demonstrated for signatures consisting of (mono)jets and missing transverse energy, where the main SM background is $Z(\nu \nu) +\text{jets}$, and the data-driven control process is $\gamma+\text{jets}$.

A search for pair production of first-generation scalar leptoquarks is performed in the final state containing two electrons and two jets using proton-proton collision data at √s=7 TeV. The data sample used corresponds to an integrated luminosity of 33 pb−1 collected with the CMS detector at the CERN LHC. The number of observed events is in good agreement with the predictions for the standard model background processes, and an upper limit is set on the leptoquark pair production cross section times β2 as a function of the leptoquark mass, where β is the branching fraction of the leptoquark decay to an electron and a quark. A 95% confidence level lower limit is set on the mass of a first-generation scalar leptoquark at 384 GeV for β=1, which is the most stringent direct limit to date.Received 17 December 2010DOI:https://doi.org/10.1103/PhysRevLett.106.201802This article is available under the terms of the Creative Commons Attribution 3.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.© 2011 CERN, for the CMS Collaboration

A measurement of the ratio of the inclusive 3-jet to 2-jet cross sections as a function of the total jet transverse momentum, HT, in the range 0.2 < HT < 2.5 TeV is presented. The data have been collected at a proton-proton centre-of-mass energy of 7 TeV with the CMS detector at the LHC, and correspond to an integrated luminosity of 36 inverse picobarns. Comparisons are made between the data and the predictions of different QCD-based Monte Carlo models for multijet production. All models considered in this study are consistent with the data for HT > 0.5 TeV. This measurement extends to an HT range that has not been explored before.

Abstract We have measured the radiation tolerance of poly-crystalline and single-crystalline diamonds grown by the chemical vapor deposition (CVD) process by measuring the charge collected before and after irradiation in a 50 m pitch strip detector fabricated on each diamond sample. We irradiated one group of sensors with 800 MeV protons, and a second group of sensors with 24 GeV protons, in steps, to protons cm −2 and protons cm −2 respectively. We observe the sum of mean drift paths for electrons and holes for both poly-crystalline CVD diamond and single-crystalline CVD diamond decreases with irradiation fluence from its initial value according to a simple damage curve characterized by a damage constant for each irradiation energy and the irradiation fluence. We find for each irradiation energy the damage constant, for poly-crystalline CVD diamond to be the same within statistical errors as the damage constant for single-crystalline CVD diamond. We find the damage constant for diamond irradiated with 24 GeV protons to be and the damage constant for diamond irradiated with 800 MeV protons to be . Moreover, we observe the pulse height decreases with fluence for poly-crystalline CVD material and within statistical errors does not change with fluence for single-crystalline CVD material for both 24 GeV proton irradiation and 800 MeV proton irradiation. Finally, we have measured the uniformity of each sample as a function of fluence and observed that for poly-crystalline CVD diamond the samples become more uniform with fluence while for single-crystalline CVD diamond the uniformity does not change with fluence.

A set of MC simulated training/testing events for the evaluation of top quark tagging architectures. In total 1.2M training events, 400k validation events and 400k test events. Use “train” for training, “val” for validation during the training and “test” for final testing and reporting results. <strong>Description</strong> 14 TeV, hadronic tops for signal, qcd diets background, Delphes ATLAS detector card with Pythia8 No MPI/pile-up included Clustering of particle-flow entries (produced by Delphes E-flow) into anti-kT 0.8 jets in the pT range [550,650] GeV All top jets are matched to a parton-level top within ∆R = 0.8, and to all top decay partons within 0.8 Jets are required to have |eta| &lt; 2 The leading 200 jet constituent four-momenta are stored, with zero-padding for jets with fewer than 200 Constituents are sorted by pT, with the highest pT one first The truth top four-momentum is stored as truth_px etc. A flag (1 for top, 0 for QCD) is kept for each jet. It is called is_signal_new The variable "ttv" (= test/train/validation) is kept for each jet. It indicates to which dataset the jet belongs. It is redundant as the different sets are already distributed as different files.

Collider searches face the challenge of defining a representation of high-dimensional data such that physical symmetries are manifest, the discriminating features are retained, and the choice of representation is new-physics agnostic. We introduce JetCLR to solve the mapping from low-level data to optimized observables though self-supervised contrastive learning. As an example, we construct a data representation for top and QCD jets using a permutation-invariant transformer-encoder network and visualize its symmetry properties. We compare the JetCLR representation with alternative representations using linear classifier tests and find it to work quite well.

The computational cost for high energy physics detector simulation in future experimental facilities is going to exceed the current available resources. To overcome this challenge, new ideas on surrogate models using machine learning methods are being explored to replace computationally expensive components. Additionally, differentiable programming has been proposed as a complementary approach, providing controllable and scalable simulation routines. In this document, new and ongoing efforts for surrogate models and differential programming applied to detector simulation are discussed in the context of the 2021 Particle Physics Community Planning Exercise (`Snowmass').

A search for pair production of second-generation scalar leptoquarks in the final state with two muons and two jets is performed using proton-proton collision data at √s=7 TeV collected by the CMS detector at the LHC. The data sample used corresponds to an integrated luminosity of 34 pb−1. The number of observed events is in good agreement with the predictions from the standard model processes. An upper limit is set on the second-generation leptoquark cross section times β2 as a function of the leptoquark mass, and leptoquarks with masses below 394 GeV are excluded at a 95% confidence level for β=1, where β is the leptoquark branching fraction into a muon and a quark. These limits are the most stringent to date.Received 17 December 2010DOI:https://doi.org/10.1103/PhysRevLett.106.201803This article is available under the terms of the Creative Commons Attribution 3.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.© 2011 CERN, for the CMS Collaboration

Diamond devices have now become ubiquitous in the LHC experiments, finding applications in beam background monitoring and luminosity measuring systems. This sensor material is now maturing to the point that the large pads in existing diamond detectors are being replaced by highly granular tracking devices, in both pixel and strip configurations, for detector systems that will be used in Run II at the LHC and beyond. The RD42 collaboration has continued to seek out additional diamond manufacturers and quantify the limits of the radiation tolerance of this material. The ATLAS experiment has recently installed, and is now commissioning a fully-fledged pixel tracking detector system based on diamond sensors. Finally, RD42 has recently demonstrated the viability of 3D biased diamond sensors that can be operated at very low voltages with full charge collection. These proceedings describe all of these advances.

Detectors based on Chemical Vapor Deposition (CVD) diamond have been used extensively and successfully in beam conditions/beam loss monitors as the innermost detectors in the highest radiation areas of Large Hadron Collider (LHC) experiments. The startup of the LHC in 2015 brought a new milestone where the first polycrystalline CVD (pCVD) diamond pixel modules were installed in an LHC experiment and successfully began operation. The RD42 collaboration at CERN is leading the effort to develop polycrystalline CVD diamond as a material for tracking detectors operating in extreme radiation environments. The status of the RD42 project with emphasis on recent beam test results is presented.

The large data rates at the LHC require an online trigger system to select relevant collisions. Rather than compressing individual events, we propose to compress an 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.

This upload contains the datasets used in arXiv:2302.11594. The file <em>g4-showers_950k_10x10_train_val_test.pt</em> contains the <strong>760k training</strong>, <strong>95k validation</strong> and <strong>95k test</strong> showers as well as their incident energies. It should be loaded as follows: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - <em>import torch </em> <em>list_tensors = torch.load(args.file_path)</em> <em>for (idx, tensor) in enumerate(list_tensors):</em> <em> [showers_train, showers_val, showers_test, inc_energies_train, inc_energies_val, inc_energies_test] = list_tensors</em> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - The file <em>g4-showers_665k_10x10_test.pt</em> contains 665k additional showers that were used for the classifier scaling studies, in addition to the 95k test showers from the file <em>g4-showers_950k_10x10_train_val_test.pt</em>. It should be loaded as follows: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - <em>import torch</em> <em>list_tensors = torch.load("g4-showers_950k_10x10_train_val_test.pt") </em> <em>[showers_geant, inc_energies_geant] = list_tensors</em> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - A detailed description of how the datasets were simulated can be found in the paper.

A search for narrow resonances in the dijet mass spectrum is performed using data corresponding to an integrated luminosity of 2.9 inverse pb collected by the CMS experiment at the LHC. Upper limits at the 95% confidence level (CL) are presented on the product of the resonance cross section, branching fraction into dijets, and acceptance, separately for decays into quark-quark, quark-gluon, or gluon-gluon pairs. The data exclude new particles predicted in the following models at the 95% CL: string resonances, with mass less than 2.50 TeV, excited quarks, with mass less than 1.58 TeV, and axigluons, colorons, and E_6 diquarks, in specific mass intervals. This extends previously published limits on these models.

In sight of the luminosity increase of the High Luminosity-LHC (HL-LHC), most experiments at the CERN Large Hadron Collider (LHC) are planning upgrades for their innermost layers in the next 5–10 years. These upgrades will require more radiation tolerant technologies than exist today. Usage of Chemical Vapor Deposition (CVD) diamond as detector material is one of the potentially interesting technologies for the upgrade. CVD diamond has been used extensively in the beam condition monitors of BaBar, Belle, CDF and all LHC experiments. Measurements of the radiation tolerance of the highest quality polycrystalline CVD material for a range of proton energies, pions and neutrons obtained with this material are presented. In addition, new results on the evolution of various semiconductor parameters as a function of the dose rate are described.

In the present study, results towards the development of a 3D diamond sensor are presented. Conductive channels are produced inside the sensor bulk using a femtosecond laser. This electrode geometry allows full charge collection even for low quality diamond sensors. Results from testbeam show that charge is collected by these electrodes. In order to understand the channel growth parameters, with the goal of producing low resistivity channels, the conductive channels produced with a different laser setup are evaluated by Raman spectroscopy.