15–19 Sept 2025
CERN
Europe/Zurich timezone

End-to-end Fast Detector Simulation and Reconstruction

15 Sept 2025, 12:35
5m
500/1-001 - Main Auditorium (CERN)

500/1-001 - Main Auditorium

CERN

400
Show room on map
7. Experimental Technologies Cutting Edge AI for Offline Data Processing

Speakers

Michael Duehrssen-Debling (CERN) Nedaa Alexandra Asbah (CERN)

Description

The standard Monte Carlo pipeline separates generation, detector simulation and reconstruction. This project advances an end-to-end generative approach that maps truth-level particles directly to reconstructed objects, reducing per-event runtime to ≪ 1 s by bypassing detailed detector simulation and algorithmic reconstruction. For correctly identified objects, the simulation of kinematic properties and efficiencies are essentially a solved problem with existing HEP tools; the emphasis here is on mis-identified particles and fakes that arise from rare combinations of input kinematics and unusual detector interactions. The model is conditioned on pile-up, detector conditions and trigger selections to preserve kinematic correlations. Confusion-aware generation (explicit modelling of mis-ID channels), imbalance-robust training and domain adaptation are used to capture rare effects, while simulation-based inference and uncertainty calibration propagate errors to analysis-level observables. An extensive validation programme benchmarks physics performance against the standard chain across key analyses and systematic scans. The goal is a very fast (≪ 1 s) simulation that remains sufficiently accurate for a wide range of LHC and future-collider studies—particularly searches and systematic-uncertainty evaluations that require large, independent Monte Carlo samples.

CERN group/ Experiment

CERN ATLAS TEam

Working area Area 7: Experimental Technologies
Project goals Establish very fast simulation of Monte Carlo events suitable to be used for a sizeable fraction of ATLAS and future detector analyses needs
Timeline Year 1: Investigation of AI model architectures to model each source of mis-identified particles independently Year 2: Integration of these AI models for mis-identified particles into existing approaches to end-to-end simulation and reconstruction Year 3: Validation against ATLAS Geant4 Monte Carlo and tuning to the ATLAS detector simulation and reconstruction
Available person power 0.2 FTE
Additional person power request 36 GRAP months, 36 DOCT monts
Is this an already ongoing activity? No
Indicative hardware resources needs Hardware: current GPU training resources through CERN IT sufficient as long as they stay available at current occupation

Authors

Presentation materials

There are no materials yet.