International Linear Collider
Not a member yet
509047 research outputs found
Sort by
Jet tagging with more-interaction particle transformer
No full textIn this paper, we introduce the More-Interaction Particle Transformer (MIParT), a novel deep-learning neural network designed for jet tagging. This framework incorporates our own design, the More-Interaction Attention (MIA) mechanism, which increases the dimensionality of particle interaction embeddings. We tested MIParT using the top tagging and quark-gluon datasets. Our results show that MIParT not only matches the accuracy and AUC of LorentzNet and a series of Lorentz-equivariant methods, but also significantly outperforms the ParT model in background rejection. Specifically, it improves background rejection by approximately 25% with a signal efficiency of 30% on the top tagging dataset and by 3% on the quark-gluon dataset. Additionally, MIParT requires only 30% of the parameters and 53% of the computational complexity needed by ParT, proving that high performance can be achieved with reduced model complexity. For very large datasets, we double the dimension of particle embeddings, referring to this variant as MIParT-Large (MIParT-L). We found that MIParT-L can further capitalize on the knowledge from large datasets. From a model pre-trained on the 100M JetClass dataset, the background rejection performance of fine-tuned MIParT-L improves by 39% on the top tagging dataset and by 6% on the quark-gluon dataset, surpassing that of fine-tuned ParT. Specifically, the background rejection of fine-tuned MIParT-L improves by an additional 2% compared to that of fine-tuned ParT. These results suggest that MIParT has the potential to increase the efficiency of benchmarks for jet tagging and event identification in particle physics.In this study, we introduce the More-Interaction Particle Transformer (MIParT), a novel deep learning neural network designed for jet tagging. This framework incorporates our own design, the More-Interaction Attention (MIA) mechanism, which increases the dimensionality of particle interaction embeddings. We tested MIParT using the top tagging and quark-gluon datasets. Our results show that MIParT not only matches the accuracy and AUC of LorentzNet and a series of Lorentz-equivariant methods, but also significantly outperforms the ParT model in background rejection. Specifically, it improves background rejection by approximately 25% at a 30% signal efficiency on the top tagging dataset and by 3% on the quark-gluon dataset. Additionally, MIParT requires only 30% of the parameters and 53% of the computational complexity needed by ParT, proving that high performance can be achieved with reduced model complexity. For very large datasets, we double the dimension of particle embeddings, referring to this variant as MIParT-Large (MIParT-L). We find that MIParT-L can further capitalize on the knowledge from large datasets. From a model pre-trained on the 100M JetClass dataset, the background rejection performance of the fine-tuned MIParT-L improved by 39% on the top tagging dataset and by 6% on the quark-gluon dataset, surpassing that of the fine-tuned ParT. Specifically, the background rejection of fine-tuned MIParT-L improved by an additional 2% compared to the fine-tuned ParT. The results suggest that MIParT has the potential to advance efficiency benchmarks for jet tagging and event identification in particle physics. The code is available at the following GitHub repository: https://github.com/USST-HEP/MIPar
Building CERN’s Future Circular Collider—An Estimation of Its Impact on Value Added and Employment
No full textThis chapter explores the potential economic and employment impacts of constructing the Future Circular Collider (FCC), a next-generation particle accelerator being developed by CERN. The FCC project aims to build upon the existing accelerator complex near Geneva, extending into the Haute-Savoie region and introducing an unparalleled research facility for the global scientific community. By integrating a high-intensity electron-positron collider and a high-energy hadron collider, the FCC is designed to push the boundaries of particle physics throughout the twenty-first century. Beyond its scientific aspirations, the project has the potential to create significant economic value through direct and indirect employment, technology transfer, and innovation spillovers across sectors. The analysis presented in this chapter examines the anticipated impacts on regional and international economies, highlighting the benefits of such a large-scale infrastructure in advancing scientific frontiers while also delivering tangible contributions to society, innovation, and employment growth. Through advanced modelling and projections, the chapter estimates the FCC’s potential to act as a catalyst for economic development, further solidifying Europe’s leadership in high-energy physics research
High-Voltage studies in the new GE1/1 GEM station at CMS experiment
No full textPart of the muon upgrade of the CMS Experiment for the High-Luminosity Large Hadron Collider consists of three new GEM stations, GE1/1, GE2/1 and ME0. The purpose of these stations is to increase the redundancy of the CMS muon spectrometer in the forward endcap regions and to extend the acceptance of the detector up to a pseudorapidity |η| ∼ 2.8 with the ME0 station. To achieve this result, these detectors must be able to withstand the background radiation of the installation environment, with a rate capability of up to 150 kHz/cm.The first station, GE1/1, was installed during the Long Shutdown 2. Since the start of Run 3 in 2022, GE1/1 has been active in CMS operations and data acquisition; this will be the focus of this work. The presence of high radiation in the area where the detector operates is a challenge for the High-Voltage system, since a large charge has to be handled without reducing the effective gain of the detector. This phenomenon leads to a drop in the effective voltage applied and therefore voltage compensation must be applied to have a stable gain. We quantify this effect by measuring the currents flowing in the High-Voltage system of GE1/1 detectors and comparing them with the observed particle hit-rate measured in the same chambers, as a function of LHC beam luminosity. This study, based on data provided by detectors already installed in CMS, aims to test the performance of detectors in terms of rate capability in order to help the development of the GE1/1 station and to support the development and to highlight the operational needs of the next two stations to be built, GE2/1 and ME0
Real-time Gravitational Wave data analysis with Machine Learning
No full textIn recent years, deep learning algorithms have excelled in various domains, including Astronomy. Despite this success, few deep learning models are planned for online deployment in the O4 data collection run of the LIGO-Virgo-KAGRA collaboration. This is partly due to a lack of standardized software tools for quick implementation and deployment of novel ideas with confidence in production performance. Our team addressed this gap by developing ml4gw and hermes libraries. We’ll discuss how these libraries enhanced efficiency and model robustness in several applications: Aframe, a low-latency machine learning pipeline for compact binary sources of gravitational waves, and a deep learning-based denoising scheme for astrophysical gravitational waves, covering Binary Neutron Stars (BNS), Neutron Star-Black Hole (NSBH), and Binary Black Hole (BBH) events. We'll explore the potential of machine learning for real-time detection and end-to-end searches for gravitational-wave transients. We also introduce anomaly detection techniques using deep recurrent autoencoders and a semi-supervised strategy called Gravitational Wave Anomalous Knowledge (GWAK) to identify binaries, detector glitches, and hypothesized astrophysical sources emitting GWs in the LIGO-Virgo-KAGRA frequency band. We discuss how in the future these developments can lead to rapid deployment of next-generation deep learning technology for fast gravitational wave detection.
Bio: Katya Govorkova completed her PhD at Nikhef with LHCb experiment, where she conducted high-precision physics measurements and contributed to the development of a fully software-based trigger system. Later, as a CERN Fellow, she worked on the CMS experiment, developing and deploying machine learning techniques for anomaly detection in the hardware trigger. Currently, she is a Postdoctoral Associate at MIT, where she develops anomaly detection techniques for the LIGO trigger system, enabling advancements in Multi-messenger Astronomy. Katya recently returned to LHCb, focusing on FPGA-based solutions for real-time data processing as part of the LHCb Upgrade II.
Coffee will be served at 10:30.</p
Evaluation of pixel sensors produced with a commercial 150nm CMOS process for the CMS Phase-2 Upgrade
No full textThe CMS experiment will undergo a
major upgrade to prepare for the High-Luminosity phase of the LHC.
Within the context of this upgrade, studies on a novel passive sensor
production technique for hybrid pixel detectors were performed.
The sensors were produced using a commercial CMOS process
with a feature size of that enables the use of stitching to produce large sensors out of different sub-reticles of . This provides the
possibility to produce sensors larger than the size of a reticle of
while retaining the small feature sizes enabled
through projection lithography. Additionally, the use of commercial
production lines enables higher throughput and is potentially less
expensive with the possibility to process larger wafers than with contact lithography techniques, which is currently commonly used in High-Energy physics.
To evaluate this novel sensor production process, two large prototyping
campaigns were performed for the CMS Phase-2 Inner Tracker. This
includes the production of large sensors with a size of up to and a pixel pitch of . The sensors
were bump-bonded to CROC read-out chips and irradiated to a non-ionizing radiation dose of up to . Both, before and after the irradiation
campaign, the sensors were tested in test beam environments.
The prototyping campaigns have shown, that sensors produced with a commercial LFoundry CMOS process fulfill the performance requirements for CMS Phase-2 Inner Tracker and are a promising candidate, for current and future detectors. In this report, a summary of the prototyping campaign is presented, including the yield of the sensor production, the performance of the sensors and a comparison of their detection efficiency before and after irradiation
NA62e+: dark sector searches with high intensity positron beams in ECN3
No full textDark sector models present a rich phenomenology that requires high-intensity beams and precision detectors for thorough exploration. The NA62 experiment has already published several constraints on dark sector candidates, leveraging proton beam dump and meson decay techniques. This proposal seeks to significantly enhance NA62's discovery potential for dark sector candidates by using the positron-on-target technique. High intensity and high-energy positron beams, reaching up to 150 GeV energy, have already been produced at the SPS extracted beam lines. If a positron beam with an intensity in the range of 2 positrons on target per year is delivered to the present K12 beam line, the NA62 detector would be ideal for searches of dark sector particles in both visible and invisible decay channels. Additionally, this approach would enable precision measurement of key standard model observable, including a detailed scan of ) and at the di-pion and di-muon production threshold, with discovery potential for the True Muonium () bound state