INnovative TRiggEr techniques for beyond the standard model PhysIcs Discovery at the LHC

Projektinformationen

INTREPID

ID Finanzhilfevereinbarung: 101115353

DOI

10.3030/101115353

EK-Unterschriftsdatum 18 September 2023

Startdatum 1 Oktober 2023

Enddatum 30 September 2028

Finanziert unter

• European Research Council (ERC)

Gesamtkosten

€ 1 499 375,00

EU-Beitrag

€ 1 499 375,00

Koordiniert durch

UNIVERSIDAD DE OVIEDO
Spain

Periodic Reporting for period 1 - INTREPID (INnovative TRiggEr techniques for beyond the standard model PhysIcs Discovery at the LHC)

Berichtszeitraum: 2023-10-01 bis 2026-03-31

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The INTREPID project aims to transform how particle physics experiments detect rare phenomena that could reveal new physics beyond the Standard Model. At the High-Luminosity Large Hadron Collider (HL-LHC), scientists seek answers to fundamental questions such as the nature of dark matter and the origin of neutrino masses. These searches often involve long-lived particles (LLPs), which produce unusual signatures that traditional detection systems struggle to identify. INTREPID addresses this challenge by developing innovative trigger technologies capable of recognizing these elusive signals in real time. By enhancing the Level-1 Trigger system of the CMS experiment, the project will significantly improve sensitivity to LLPs, paving the way for discoveries that could transform our understanding of the Universe. Beyond particle physics, INTREPID’s advances in hardware acceleration and machine learning have potential applications in other sectors requiring ultra-fast data processing.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

Over the reporting period, INTREPID delivered major technical and scientific milestones across three work packages.

1) Simulation and Benchmarking (WP1): Developed advanced Monte Carlo simulations and a standalone Geant4-based framework to validate new algorithms. These tools enable precise performance studies under HL-LHC conditions.
2) Muon Shower Triggering (WP2): Implemented a clustering algorithm to detect high-multiplicity hit patterns in the CMS muon system, improving efficiency for high-energy muons significantly Firmware demonstrators confirm compliance with strict latency requirements.
3) Graph Neural Networks (WP3): Designed and trained models for reconstructing displaced muons using hit-based pattern recognition. Hardware prototypes using FPGA and ACAP platforms demonstrate feasibility for real-time deployment.

INTREPID also introduced a continuous integration workflow for firmware development and explored FPGA-based acceleration of physics simulations, opening new avenues for high-performance computing. Results have been presented at leading international conferences and published in peer-reviewed journals.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

INTREPID achieved breakthroughs that push the boundaries of current technology. The muon shower algorithm represents a novel approach to recovering high-energy muons that standard triggers at the LHC miss, ensuring accurate momentum reconstruction even at multi-TeV scales. The integration of Graph Neural Networks into real-time trigger systems marks a paradigm shift, enabling sophisticated pattern recognition under stringent latency constraints. Unexpectedly, the project demonstrated that components of event generation software can be mapped to FPGA hardware, suggesting the possibility of real-time physics simulation—a transformative concept for data-intensive research. These innovations not only advance particle physics but also have cross-disciplinary potential in fields such as industrial process control and renewable energy forecasting. Future steps include large-scale validation, firmware optimization for AI models, and exploring commercialization opportunities through industry collaboration.