Cargando…

Refining fast simulation using machine learning

At the CMS experiment, a growing reliance on the fast Monte Carlo application (FastSim) will accompany the high luminosity and detector granularity expected in Phase 2. The FastSim chain is roughly 10 times faster than the application based on the GEANT4 detector simulation and full reconstruction r...

Descripción completa

Detalles Bibliográficos
Autores principales: Bein, Samuel, Connor, Patrick, Pedro, Kevin, Schleper, Peter, Wolf, Moritz
Lenguaje:eng
Publicado: 2023
Materias:
Acceso en línea:http://cds.cern.ch/record/2872274
_version_ 1780978597190696960
author Bein, Samuel
Connor, Patrick
Pedro, Kevin
Schleper, Peter
Wolf, Moritz
author_facet Bein, Samuel
Connor, Patrick
Pedro, Kevin
Schleper, Peter
Wolf, Moritz
author_sort Bein, Samuel
collection CERN
description At the CMS experiment, a growing reliance on the fast Monte Carlo application (FastSim) will accompany the high luminosity and detector granularity expected in Phase 2. The FastSim chain is roughly 10 times faster than the application based on the GEANT4 detector simulation and full reconstruction referred to as FullSim. However, this advantage comes at the price of decreased accuracy in some of the final analysis observables. In this contribution, a machine learning-based technique to refine those observables is presented. We employ a regression neural network trained with a sophisticated combination of multiple loss functions to provide post-hoc corrections to samples produced by the FastSim chain. The results show considerably improved agreement with the FullSim output and an improvement in correlations among output observables and external parameters. This technique is a promising replacement for existing correction factors, providing higher accuracy and thus contributing to the wider usage of FastSim.
id cern-2872274
institution Organización Europea para la Investigación Nuclear
language eng
publishDate 2023
record_format invenio
spelling cern-28722742023-10-21T02:28:54Zhttp://cds.cern.ch/record/2872274engBein, SamuelConnor, PatrickPedro, KevinSchleper, PeterWolf, MoritzRefining fast simulation using machine learningDetectors and Experimental TechniquesAt the CMS experiment, a growing reliance on the fast Monte Carlo application (FastSim) will accompany the high luminosity and detector granularity expected in Phase 2. The FastSim chain is roughly 10 times faster than the application based on the GEANT4 detector simulation and full reconstruction referred to as FullSim. However, this advantage comes at the price of decreased accuracy in some of the final analysis observables. In this contribution, a machine learning-based technique to refine those observables is presented. We employ a regression neural network trained with a sophisticated combination of multiple loss functions to provide post-hoc corrections to samples produced by the FastSim chain. The results show considerably improved agreement with the FullSim output and an improvement in correlations among output observables and external parameters. This technique is a promising replacement for existing correction factors, providing higher accuracy and thus contributing to the wider usage of FastSim.At the CMS experiment, a growing reliance on the fast Monte Carlo application (FastSim) will accompany the high luminosity and detector granularity expected in Phase 2. The FastSim chain is roughly 10 times faster than the application based on the GEANT4 detector simulation and full reconstruction referred to as FullSim. However, this advantage comes at the price of decreased accuracy in some of the final analysis observables. In this contribution, a machine learning-based technique to refine those observables is presented. We employ a regression neural network trained with a sophisticated combination of multiple loss functions to provide post-hoc corrections to samples produced by the FastSim chain. The results show considerably improved agreement with the FullSim output and an improvement in correlations among output observables and external parameters. This technique is a promising replacement for existing correction factors, providing higher accuracy and thus contributing to the wider usage of FastSim.arXiv:2309.12919CMS CR-2023/128FERMILAB-CONF-23-537-CMS-CSAID-PPDoai:cds.cern.ch:28722742023-08-25
spellingShingle Detectors and Experimental Techniques
Bein, Samuel
Connor, Patrick
Pedro, Kevin
Schleper, Peter
Wolf, Moritz
Refining fast simulation using machine learning
title Refining fast simulation using machine learning
title_full Refining fast simulation using machine learning
title_fullStr Refining fast simulation using machine learning
title_full_unstemmed Refining fast simulation using machine learning
title_short Refining fast simulation using machine learning
title_sort refining fast simulation using machine learning
topic Detectors and Experimental Techniques
url http://cds.cern.ch/record/2872274
work_keys_str_mv AT beinsamuel refiningfastsimulationusingmachinelearning
AT connorpatrick refiningfastsimulationusingmachinelearning
AT pedrokevin refiningfastsimulationusingmachinelearning
AT schleperpeter refiningfastsimulationusingmachinelearning
AT wolfmoritz refiningfastsimulationusingmachinelearning