Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “CERN”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 955 records · Page 53
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Compressing deep neural networks on FPGAs to binary and ternary precision with hls4ml

We present the implementation of binary and ternary neural networks in the hls4ml library, designed to automatically convert deep neural network models to digital circuits with FPGA firmware. Starting from benchmark models trained with floating point precision, we investigate different strategies to reduce the network's resource consumption by reducing the numerical precision of the network parameters to binary or ternary. We discuss the trade-off between model accuracy and resource consumption. In addition, we show how to balance between latency and accuracy by retaining full precision on a selected subset of network components. As an example, we consider two multiclass classification tasks: handwritten digit recognition with the MNIST data set and jet identification with simulated proton-proton collisions at the CERN Large Hadron Collider. The binary and ternary implementation has similar performance to the higher precision implementation while using drastically fewer FPGA resources.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2021

GPU coprocessors as a service for deep learning inference in high energy physics

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

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

Fast convolutional neural networks on FPGAs with hls4ml

We introduce an automated tool for deploying ultra low-latency, low-power deep neural networks with convolutional layers on field-programmable gate arrays (FPGAs). By extending the hls4ml library, we demonstrate an inference latency of 5 µs using convolutional architectures, targeting microsecond latency applications like those at the CERN Large Hadron Collider. Considering benchmark models trained on the Street View House Numbers Dataset, we demonstrate various methods for model compression in order to fit the computational constraints of a typical FPGA device used in trigger and data acquisition systems of particle detectors. In particular, we discuss pruning and quantization-aware training, and demonstrate how resource utilization can be significantly reduced with little to no loss in model accuracy. We show that the FPGA critical resource consumption can be reduced by 97% with zero loss in model accuracy, and by 99% when tolerating a 6% accuracy degradation.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

Lightweight jet reconstruction and identification as an object detection task

We apply object detection techniques based on deep convolutional blocks to end-to-end jet identification and reconstruction tasks encountered at the CERN large hadron collider (LHC). Collision events produced at the LHC and represented as an image composed of calorimeter and tracker cells are given as an input to a Single Shot Detection network. The algorithm, named PFJet-SSD performs simultaneous localization, classification and regression tasks to cluster jets and reconstruct their features. This all-in-one single feed-forward pass gives advantages in terms of execution time and an improved accuracy w.r.t. traditional rule-based methods. A further gain is obtained from network slimming, homogeneous quantization, and optimized runtime for meeting memory and latency constraints of a typical real-time processing environment. We experiment with 8-bit and ternary quantization, benchmarking their accuracy and inference latency against a single-precision floating-point. We show that the ternary network closely matches the performance of its full-precision equivalent and outperforms the state-of-the-art rule-based algorithm. Finally, we report the inference latency on different hardware platforms and discuss future applications.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Ultra-low latency recurrent neural network inference on FPGAs for physics applications with hls4ml

Abstract Recurrent neural networks have been shown to be effective architectures for many tasks in high energy physics, and thus have been widely adopted. Their use in low-latency environments has, however, been limited as a result of the difficulties of implementing recurrent architectures on field-programmable gate arrays (FPGAs). In this paper we present an implementation of two types of recurrent neural network layers—long short-term memory and gated recurrent unit—within the hls4ml framework. We demonstrate that our implementation is capable of producing effective designs for both small and large models, and can be customized to meet specific design requirements for inference latencies and FPGA resources. We show the performance and synthesized designs for multiple neural networks, many of which are trained specifically for jet identification tasks at the CERN Large Hadron Collider.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Differentiable Earth mover’s distance for data compression at the high-luminosity LHC

Abstract The Earth mover’s distance (EMD) is a useful metric for image recognition and classification, but its usual implementations are not differentiable or too slow to be used as a loss function for training other algorithms via gradient descent. In this paper, we train a convolutional neural network (CNN) to learn a differentiable, fast approximation of the EMD and demonstrate that it can be used as a substitute for computing-intensive EMD implementations. We apply this differentiable approximation in the training of an autoencoder-inspired neural network (encoder NN) for data compression at the high-luminosity LHC at CERN The goal of this encoder NN is to compress the data while preserving the information related to the distribution of energy deposits in particle detectors. We demonstrate that the performance of our encoder NN trained using the differentiable EMD CNN surpasses that of training with loss functions based on mean squared error.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Ultrafast jet classification at the HL-LHC

Abstract Three machine learning models are used to perform jet origin classification. These models are optimized for deployment on a field-programmable gate array device. In this context, we demonstrate how latency and resource consumption scale with the input size and choice of algorithm. Moreover, the models proposed here are designed to work on the type of data and under the foreseen conditions at the CERN large hadron collider during its high-luminosity phase. Through quantization-aware training and efficient synthetization for a specific field programmable gate array, we show that O ( 100 ) ns inference of complex architectures such as Deep Sets and Interaction Networks is feasible at a relatively low computational resource cost.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

SymbolNet: neural symbolic regression with adaptive dynamic pruning for compression

Abstract Compact symbolic expressions have been shown to be more efficient than neural network (NN) models in terms of resource consumption and inference speed when implemented on custom hardware such as field-programmable gate arrays (FPGAs), while maintaining comparable accuracy (Tsoi et al 2024 EPJ Web Conf. 295 09036). These capabilities are highly valuable in environments with stringent computational resource constraints, such as high-energy physics experiments at the CERN Large Hadron Collider. However, finding compact expressions for high-dimensional datasets remains challenging due to the inherent limitations of genetic programming (GP), the search algorithm of most symbolic regression (SR) methods. Contrary to GP, the NN approach to SR offers scalability to high-dimensional inputs and leverages gradient methods for faster equation searching. Common ways of constraining expression complexity often involve multistage pruning with fine-tuning, which can result in significant performance loss. In this work, we propose S y m b o l N e t , a NN approach to SR specifically designed as a model compression technique, aimed at enabling low-latency inference for high-dimensional inputs on custom hardware such as FPGAs. This framework allows dynamic pruning of model weights, input features, and mathematical operators in a single training process, where both training loss and expression complexity are optimized simultaneously. We introduce a sparsity regularization term for each pruning type, which can adaptively adjust its strength, leading to convergence at a target sparsity ratio. Unlike most existing SR methods that struggle with datasets containing more than O ( 10 ) inputs, we demonstrate the effectiveness of our model on the LHC jet tagging task (16 inputs), MNIST (784 inputs), and SVHN (3072 inputs).

Tsoi, Ho Fung (ORCID:0000000225502184)↗
DOE OSTI2026

Wasserstein normalized autoencoder for anomaly detection

A novel anomaly detection algorithm is presented. The Wasserstein normalized autoencoder (WNAE) is a normalized probabilistic model that minimizes the Wasserstein distance between the learned probability distribution—a Boltzmann distribution where the energy is the reconstruction error of the autoencoder (AE)—and the distribution of the training data. This algorithm has been developed and applied to the identification of semivisible jets—conical sprays of visible standard model (SM) particles and invisible dark matter states—with the CMS experiment at the CERN LHC. Trained on jets of particles from simulated SM processes, the WNAE is shown to learn the probability distribution of the input data in a fully unsupervised fashion, such that it effectively identifies new physics jets as anomalies. The model exhibits stable, convergent training and recovers strong classification performance for a wide range of signals against the selected background process, for which a standard AE fails because of outlier reconstruction. In addition, the model improves upon standard normalized autoencoders while remaining fully agnostic to the signal. The WNAE directly tackles the problem of outlier reconstruction, a common failure mode of autoencoders in anomaly detection tasks.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗
DOE OSTI2026

Machine-learning techniques for model-independent searches in dijet final states

Anomaly detection methods used in a recent search for new phenomena by CMS at the CERN LHC are presented. The methods use machine learning to detect anomalous jets produced in the decay of new massive particles without depending on a specific theory model. The effectiveness of these approaches in enhancing sensitivity to various simulated signal samples is studied and compared using data collected in proton–proton collisions at a center-of-mass energy of 13 TeV. In an example analysis, the capabilities of anomaly detection methods are further demonstrated by identifying large-radius jets consistent with Lorentz-boosted hadronically decaying top quarks in a model-agnostic framework.

CMS↗
DOE OSTI2026

Machine learning method for enforcing variable independence in background estimation with LHC data: ABCDisCoTEC

A novel solution is presented for the problem of estimating the backgrounds of a signal search using observed data while simultaneously maximizing the sensitivity of the search to the signal. The 'ABCD method' provides a reliable framework for background estimation by partitioning events into one signal-enhanced region (A) and three background-enhanced control regions (B, C, and D) via two smoothly varying, statistically independent variables. In practice, even slight correlations between the two variables can significantly undermine the method's performance. Thus, choosing appropriate variables by hand can present a formidable challenge, especially when background and signal differ only subtly. To address this issue, the ABCD with distance correlation (ABCDisCo) method was developed to construct two learned variables via a neural network trained to provide strong signal-background discrimination with small values of the distance correlation (DisCo) measure between the two learned variables. However, relying solely on minimizing the DisCo can result in learned variables that may not have distributions of background events that are smoothly varying and localized at extreme values, as necessary for the validity of the background estimation. The ABCDisCo training enhanced with closure (ABCDisCoTEC) method is introduced to solve this issue by directly minimizing the nonclosure, expressed as a dedicated differentiable loss term. This extended method is applied to a data set of proton-proton collisions at a center-of-mass energy of 13 TeV recorded by the CMS detector at the CERN Large Hadron Collider. Additionally, given the complexity of the minimization problem with constraints on multiple loss terms, the modified differential method of multipliers is applied and shown to greatly improve the stability and robustness of the ABCDisCoTEC method, compared to grid search hyperparameter optimization procedures.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗
DOE OSTI2019

Proton-driven plasma wakefield acceleration in AWAKE

In this paper, we briefly summarize the experiments performed during the first Run of the Advanced Wakefield Experiment, AWAKE, at CERN (European Organization for Nuclear Research). The final goal of AWAKE Run 1 (2013–2018) was to demonstrate that 10–20 MeV electrons can be accelerated to GeV-energies in a plasma wakefield driven by a highly-relativistic self-modulated proton bunch. We describe the experiment, outline the measurement concept and present first results. Lastly, we outline our plans for the future.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2026

Observation of Long-Range Collective Flow in O + O and Ne + Ne Collisions and Implications for Nuclear Structure Studies

The long-range collective flow of particles produced in oxygen-oxygen ( O + O ) and neon-neon ( Ne + Ne ) collisions is measured with the CMS detector at the CERN LHC. The data samples were collected at a center-of-mass energy per nucleon pair of 5.36 TeV, with integrated luminosities of 7 nb - 1 and 0.8 nb - 1 for O + O and Ne + Ne collisions, respectively. Two- and four-particle azimuthal correlations are measured over nearly five units of pseudorapidity. Significant elliptic ( v 2 ) and triangular ( v 3 ) flow harmonics are observed in both systems. The ratios of v n coefficients between Ne + Ne and O + O collisions suggest sensitivity to the intrinsic nuclear structure of the respective Ne and O ions. Hydrodynamic models with ab initio nuclear structure inputs qualitatively reproduce the collision-overlap dependence of both the v n values and the Ne + Ne to O + O ratios. These measurements provide new constraints when modeling the interplay of nuclear structure and collective dynamics in collisions of O 16 and Ne 20 ions, respectively.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗
DOE OSTI2026

Beam-dump ceiling and its experimental implications: The case of a portable experiment

We generalize the nature of the so-called beam-dump “ceiling” beyond which the improvement on the sensitivity reach in the search for fast-decaying mediators dramatically slows down, and we point out its experimental implications that motivate tabletop-sized beam-dump experiments for the search. Light (bosonic) mediators are well-motivated new-physics particles, as they can appear in dark-sector portal scenarios and models to explain various laboratory-based anomalies. Due to their low mass and feebly interacting nature, beam-dump-type experiments, utilizing high-intensity particle beams, can play a crucial role in probing the parameter space of such visibly decaying mediators—in particular, the “prompt decay” region, where the mediators feature relatively large coupling and mass. We present a general and semianalytic proof that the ceiling effectively arises in the prompt-decay region of an experiment and show its insensitivity to data statistics, background estimates, and systematic uncertainties, considering a concrete example, the search for axion-like particles interacting with ordinary photons at three benchmark beam facilities: PIP-II at FNAL, and SPS and LHC-dump at CERN. We then identify optimal criteria to perform a cost-effective and short-term experiment to reach the ceiling, demonstrating that very short-baseline compact experiments enable access to the parameter space unreachable thus far.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Measurement of 𝑊⁢𝑊⁢𝑍 and 𝑍⁢𝐻 Production Cross Sections at $\sqrt{𝑠}$ = 13 and 13.6 TeV

A measurement is presented of the cross section in proton-proton collisions for the production of two 𝑊 bosons and one 𝑍 boson. It is based on data recorded by the CMS experiment at the CERN LHC at center-of-mass energies $\sqrt{𝑠}$ = 13 and 13.6 TeV, corresponding to an integrated luminosity of 200 fb −1 . Events with four charged leptons (electrons or muons) in the final state are selected. Both nonresonant 𝑊⁢𝑊⁢𝑍 production and 𝑍⁢𝐻 production, with the Higgs boson decaying into two 𝑊 bosons, are reported. For the first time, the two processes are measured separately in a simultaneous fit. Combining the two modes, signal strengths relative to the standard model (SM) predictions of 0.75$^{+0.34}_{−0.29}$ and 1.74$^{+0.71}_{−0.60}$ are measured for $\sqrt{𝑠}$ = 13 and 13.6 TeV, respectively. The observed (expected) significance for the triboson signal is 3.8 (2.5) standard deviations for $\sqrt{𝑠}$ = 13.6 TeV, thus providing the first evidence for triboson production at this center-of-mass energy. Combining the two modes and the two center-of-mass energies, the inclusive signal strength relative to the SM prediction is measured to be 1.03$^{+0.31}_{−0.28}$, with an observed (expected) significance of 4.5 (5.0) standard deviations.

Gauge bosons↗
DOE OSTI2026

Observation of Suppressed Charged-Particle Production in Ultrarelativistic Oxygen-Oxygen Collisions

A hot and dense state of nuclear matter, known as the quark-gluon plasma, is created in collisions of ultrarelativistic heavy nuclei. Highly energetic quarks and gluons, collectively referred to as partons, lose energy as they travel through this matter, leading to suppressed production of particles with large transverse momenta (𝑝 T ). Conversely, high-𝑝 T particle suppression has not been seen in proton-lead collisions, raising questions regarding the minimum system size required to observe parton energy loss. Oxygen-oxygen (OO) collisions examine a region of effective system size that lies between these two extreme cases. The CMS detector at the CERN LHC has been used to quantify charged-particle production in inclusive OO collisions for the first time via measurements of the nuclear modification factor (𝑅 AA ). The 𝑅 AA is derived by comparing particle production to expectations based on proton-proton (𝑝⁢𝑝) data and has a value of unity in the absence of nuclear effects. The data for OO and 𝑝⁢𝑝 collisions at a nucleon-nucleon center-of-mass energy $\sqrt{s_{NN}}$ = 5.36 TeV correspond to integrated luminosities of 6.1 nb −1 and 1.02 pb −1 , respectively. The 𝑅 AA is below unity with a minimum of 0.69 ± 0.04 around 𝑝 T = 6 GeV. The data exhibit better agreement with theoretical models incorporating parton energy loss as compared to baseline models without energy loss.

FOS: Physical sciences↗
DOE OSTI2026

Search for pair production of heavy resonances in final states with a photon and large-radius jets in proton-proton collisions at $\sqrt{s}$ = 13 TeV

A search for the pair production of heavy spin-1/2 or spin-3/2 resonances (𝑡*) in proton-proton collisions at $\sqrt{s}$ =13 TeV is presented. Data collected with the CMS detector at the CERN LHC from 2016 to 2018 corresponding to an integrated luminosity of 138 fb −1 are used. The analysis targets benchmark signal scenarios where one 𝑡* decays into a top quark (𝑡) and a photon (𝛾), and the other into a 𝑡 quark and a gluon (𝑔), i.e., 𝑝⁢𝑝 → $𝑡^*\overline{⁢𝑡^*}$ → 𝑡⁢𝑡⁢𝛾⁢𝑔. All-hadronic final states from the 𝑡 pair decay chain are selected using jet substructure techniques. The signal is probed as a function of the 𝑡* candidate mass, which is reconstructed using the photon and a top quark candidate jet. No significant deviation from the background-only hypothesis is found. Observed (expected) upper limits on the signal cross section at 95% confidence level are set, excluding masses of spin-1/2 𝑡* particles below 930 (930) GeV and spin-3/2 𝑡* particles below 1330 (1390) GeV. This analysis marks the first search for heavy resonances in the $𝑡\bar{⁢𝑡}𝛾⁢𝑔$ channel. Exploiting the high-energy photon to reduce the backgrounds, this search achieves sensitivity competitive with 𝑝⁢𝑝 → $𝑡^*\overline{⁢𝑡^*}$ → $𝑡\bar{⁢𝑡}⁢𝑔⁡𝑔$ searches for spin-1/2 𝑡* despite the small expected 𝑡* → 𝑡⁢𝛾 branching fraction.

hadron colliders↗
DOE OSTI2025

First Search for Axion Dark Matter with a MADMAX Prototype

This Letter presents the first search for dark matter axions with mass in the ranges 76.56 to 76.82 μ⁢eV and 79.31 to 79.53 μ⁢eV using a prototype setup for the MAgnetized Disk and Mirror Axion eXperiment (MADMAX). The experimental setup employs a dielectric haloscope consisting of three sapphire disks and a mirror to resonantly enhance the axion-induced microwave signal within the magnetic dipole field provided by the 1.6 T Morpurgo magnet at CERN. Over 14.5 days of data collection, no axion signal was detected. A 95% CL upper limit on the axion-photon coupling strength down to |𝑔 𝑎⁢𝛾 |∼2×10 −11 GeV −1 is set in the targeted mass ranges, surpassing previous constraints, assuming a local axion dark matter density 𝜌 𝑎 of 0.3 GeV/cm 3 . This study marks the first axion dark matter search using a dielectric haloscope.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗