Proof of feasibility of an integration of AdePT in Gaussino simulations for improving performance

dc.contributor.advisorSandoval Usme, Carlos Eduardo
dc.contributor.authorBenavides Rubio, Juan Bernardo
dc.contributor.orcidBenavides Rubio, Juan Bernardo [0009-0006-5300-8497]
dc.contributor.researchgroup
dc.date.accessioned2026-02-09T19:58:41Z
dc.date.available2026-02-09T19:58:41Z
dc.date.issued2025
dc.descriptionIlustraciones, diagramas, fotografías, gráficosspa
dc.description.abstractThe Large Hadron Collider beauty (LHCb) experiment at CERN faces a significant computational challenge as increasing data collection rates demand proportionally larger sets of simulated data. The simulation of the electromagnetic calorimeter (ECAL) has been identified as the primary performance bottleneck, consuming over 60% of the total simulation time and limiting the statistical precision of key physics analyses. This thesis explores a novel solution to this bottleneck by integrating AdePT (Accelerated demonstrator of electromagnetic Particle Transport), a library designed for GPU-accelerated simulation of electromagnetic showers, into Gaussino, the LHCb experiment-agnostic simulation framework. A proof-of-concept integration was developed to offload the computationally intensive ECAL simulation from the CPU to the GPU. The feasibility of this approach was evaluated through a series of benchmark tests comparing the performance and physics fidelity of the AdePT-integrated simulations against the standard Geant4 baseline. The results demonstrate significant performance improvements, with speed-up factors of up to 4.12x in calorimeter-centric benchmarks. The performance was found to scale effectively with event complexity, highlighting the effciency of GPU parallelization for high-multiplicity events. Physics validation studies confirmed a high degree of fidelity, with key observables such as energy deposition and longitudinal shower profiles showing strong agreement with the baseline. While some discrepancies were identified, they are under active investigation. This work successfully demonstrates that integrating AdePT into Gaussino is a viable strategy to alleviate the simulation bottleneck, paving the way for a more effcient use of heterogeneous computing resources and enhancing the physics potential of the LHCb experiment.eng
dc.description.abstractEl experimento Large Hadron Collider beauty (LHCb) del CERN se enfrenta a un importante reto computacional, ya que el aumento de las tasas de recopilación de datos exige conjuntos de datos simulados proporcionalmente mayores. La simulación del calorímetro electromagnético (ECAL) se ha identificado como el principal cuello de botella en cuanto al rendimiento, ya que consume más del 60% del tiempo total de simulación y limita la precisión estadística de los análisis físicos clave. Esta tesis explora una solución novedosa a este cuello de botella mediante la integración de AdePT (demostrador acelerado de transporte de partículas electromagnéticas), una librería diseñada para la simulación acelerada por GPU de cascadas electromagnéticas, en Gaussino, el framework de simulación del LHCb. Se desarrolló una prueba de concepto de la integración para descargar la simulación ECAL, que requiere un gran esfuerzo computacional, de la CPU a la GPU. La viabil-idad de este enfoque se evaluó mediante una serie de pruebas en las que se comparó el rendimiento y la fidelidad física de las simulaciones usando AdePT con la referencia estándar Geant4. Los resultados demuestran mejoras significativas en el rendimiento, con factores de aceleración de hasta 4,12 veces en pruebas de referencia centradas en el calorímetro. Se comprobó que el rendimiento se adapta eficazmente a la complejidad de los even-tos, lo que pone de relieve la eficiencia de la paralelización de la GPU para eventos de alta multiplicidad. Los estudios de validación física confirmaron un alto grado de fidelidad, con observables clave como la deposición de energía y los perfiles de cascada longitudinales mostrando una fuerte concordancia con la referencia. Si bien se iden-tificaron algunas discrepancias, estas se están investigando activamente. Este trabajo demuestra con éxito que la integración de AdePT en Gaussino es una estrategia viable para aliviar el cuello de botella de la simulación, allanando el camino para un uso más eficiente de los recursos computacionales heterogéneos y mejorando el potencial físico del experimento LHCb. (Texto tomado de la fuente)spa
dc.description.degreelevelMaestría
dc.description.degreenameMaestría en Ingeniería de Sistemas y Computación
dc.format.extentvii, 70 páginas
dc.format.mimetypeapplication/pdf
dc.identifier.instnameUniversidad Nacional de Colombiaspa
dc.identifier.reponameRepositorio Institucional Universidad Nacional de Colombiaspa
dc.identifier.repourlhttps://repositorio.unal.edu.co/spa
dc.identifier.urihttps://repositorio.unal.edu.co/handle/unal/89429
dc.language.isoeng
dc.publisherUniversidad Nacional de Colombia
dc.publisher.branchUniversidad Nacional de Colombia - Sede Bogotá
dc.publisher.facultyFacultad de Ingeniería
dc.publisher.placeBogotá, Colombia
dc.publisher.programBogotá - Ingeniería - Maestría en Ingeniería - Ingeniería de Sistemas y Computación
dc.relation.referencesS. F. Novaes, “Standard model: An introduction.” https://arxiv. org/abs/hep-ph/0001283, 2000.
dc.relation.referencesL. Evans and P. Bryant, “LHC Machine,” Journal of Instrumentation, vol. 3, p. S08001, aug 2008.
dc.relation.referencesThe LHCb collaboration, “The LHCb Detector at the LHC,” Journal of Instrumentation, vol. 3, p. S08005, aug 2008.
dc.relation.referencesThe LHCb Collaboration, “Performance of the fast simulation interface in Gauss-on-Gaussino,” tech. rep., CERN, 2022.
dc.relation.referencesThe LHCb collaboration, “LHCb detector performance,” Int. J. Mod. Phys., vol. A30, p. 1530022, 2015.
dc.relation.referencesThe LHCb collaboration, “The LHCb Upgrade I,” JINST, vol. 19, p. P05065, 2024.
dc.relation.referencesN. Srimanobhas, “Introduction to Monte Carlo for Particle Physics Study.” https://indico.cern.ch/event/92209/contributions/ 2114409/attachments/1098701/1567290/CST2010-MC.pdf, 2010.
dc.relation.referencesM. Clemencic et al., “The LHCb Simulation Application, Gauss: De- sign, Evolution and Experience,” Journal of Physics: Conference Se- ries, vol. 331, p. 032023, 12 2011.
dc.relation.referencesB. G. Siddi and D. Muller, “Gaussino - a gaudi-based core simula- tion framework,” 2019 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), pp. 1–4, 10 2019.
dc.relation.referencesM. Mazurek, M. Clemencic, and G. Corti, “Gauss and Gaussino: the LHCb simulation software and its new experiment agnostic core frame- work,” Proceedings of Science, vol. 414, 2022.
dc.relation.referencesG. Barrand and others., “GAUDI - A software architecture and framework for building HEP data processing applications,” Computer Physics Communications, vol. 140, pp. 45–55, 10 2001.
dc.relation.referencesM. Clemencic, H. Degaudenzi, P. Mato, S. Binet, W. Lavrijsen, C. Leggett, and I. Belyaev, “Recent developments in the LHCb software framework gaudi,” Journal of Physics: Conference Series, vol. 219, 2010.
dc.relation.referencesT. Sj¨ostrand, S. Mrenna, and P. Skands, “A brief introduction to PYTHIA 8.1,” Computer Physics Communications, vol. 178, pp. 852– 867, 2008.
dc.relation.referencesD. Lange, “The EvtGen particle decay simulation package,” Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 462, pp. 152– 155, 2001.
dc.relation.referencesS. Agostinelli et al., “GEANT4 - A simulation toolkit,” Nuclear In- struments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 506, pp. 250– 303, 7 2003.
dc.relation.referencesJ. Allison et al., “Geant4 developments and applications,” IEEE Trans- actions on Nuclear Science, vol. 53, pp. 270–278, 2006.
dc.relation.referencesJ. Allison et al., “Recent developments in Geant4,” Nuclear Instru- ments and Methods in Physics Research Section A: Accelerators, Spec- trometers, Detectors and Associated Equipment, vol. 835, pp. 186–225, 11 2016.
dc.relation.referencesThe Geant4 Collaboration, Geant4 User’s Guide for Application De- velopers. CERN, 2024. Accessed: 2025-08-23.
dc.relation.referencesThe LHCb Collaboration, “LHCb CPU Usage Forecast,” tech. rep., CERN, 2019.
dc.relation.referencesThe LHCb Collaboration, “Computing and software for LHCb Upgrade II,” ArXiv, 3 2025.
dc.relation.referencesS. Roiser and C. Bozzi, “The LHCb Software and Computing Upgrade towards LHC Run 3,” Journal of Physics: Conference Series, vol. 1085, 10 2018.
dc.relation.referencesC. Bozzi, “Utilization of LHCb O!ine Computing Resources in 2024,” tech. rep., CERN, Geneva, 2025.
dc.relation.referencesM. Whitehead, “A palette of fast simulations in LHCb,” Proceedings of Science, vol. 340, 2019.
dc.relation.referencesM. Barbetti, “Lamarr: LHCb ultra-fast simulation based on machine learning models deployed within Gauss,” Journal of Physics: Confer- ence Series (ACAT 2022), vol. 414, 3 2023.
dc.relation.referencesM. Mazurek, G. Corti, and D. M¨uller, “New Simulation Software Tech- nologies at the LHCb Experiment at CERN,” Computing and Infor- matics, vol. 40, pp. 815–832, 12 2021.
dc.relation.referencesG. Corti et al., “New software technologies in the LHCb Simulation,” Journal of Physics: Conference Series, vol. 2438, 2023.
dc.relation.referencesN. V. Canudas and Others, “Graph Clustering: a graph-based cluster- ing algorithm for the electromagnetic calorimeter in LHCb,” European Physical Journal C, vol. 83, 2023.
dc.relation.referencesM. Rovere and Others, “CLUE: A Fast Parallel Clustering Algorithm for High Granularity Calorimeters in High-Energy Physics,” Frontiers in Big Data, vol. 3, 11 2020.
dc.relation.referencesE. Buhmann, S. Diefenbacher, E. Eren, F. Gaede, G. Kasieczka, A. Ko- rol, and K. Kr¨uger, “Getting High: High Fidelity Simulation of High Granularity Calorimeters with High Speed,” Computing and Software for Big Science, vol. 5, 2021.
dc.relation.referencesThe Atlas Collaboration, “AtlFast3: The Next Generation of Fast Sim- ulation in ATLAS,” Computing and Software for Big Science, vol. 6, 2022.
dc.relation.referencesJ. Apostolakis and Others, “Detector simulation challenges for future accelerator experiments,” Frontiers in Physics, vol. 10, 2022.
dc.relation.referencesC. Krause and D. Shih, “Fast and accurate simulations of calorimeter showers with normalizing flows,” Physical Review D, vol. 107, 2023.
dc.relation.referencesF. Sergeev and Others, “Fast simulation of the LHCb electromagnetic calorimeter response using VAEs and GANs,” Journal of Physics: Con- ference Series, vol. 1740, 1 2021.
dc.relation.referencesD. Lancierini, P. Owen, and N. Serra, “Simulating the LHCb hadron calorimeter with generative adversarial networks,” Nuovo Cimento della Societa Italiana di Fisica C, vol. 42, 2019.
dc.relation.referencesA. Rogachev and F. Ratnikov, “GAN with an Auxiliary Regressor for the Fast Simulation of the Electromagnetic Calorimeter Response,” Journal of Physics: Conference Series, vol. 2438, 2023.
dc.relation.referencesC. Krause et al., “CaloChallenge 2022: A Community Challenge for Fast Calorimeter Simulation,” 2024.
dc.relation.referencesG. Amadio et al., “Offoading electromagnetic shower transport to GPUs,” Journal of Physics: Conference Series, vol. 2438, p. 012055, 2 2023.
dc.relation.referencesJ. Apostolakis et al., “AdePT-Enabling GPU electromagnetic transport with Geant4,” tech. rep., CERN, 2024.
dc.relation.referencesG4HepEM, “G4HepEm.” https://github.com/mnovak42/g4hepem.
dc.relation.referencesJ. Apostolakis and others., “Towards a high performance geometry li- brary for particle-detector simulations,” Journal of Physics: Confer- ence Series, vol. 608, p. 012023, 4 2015.
dc.relation.referencesThe LHCb Collaboration, “Allen: A High-Level Trigger on GPUs for LHCb,” Computing and Software for Big Science, vol. 4, pp. 1–11, 12 2020.
dc.relation.referencesThe LHCb Collaboration, “A Comparison of CPU and GPU Imple- mentations for the LHCb Experiment Run 3 Trigger,” Computing and Software for Big Science, vol. 6, 12 2022.
dc.relation.referencesV. W. Berger and Y. Zhou, Kolmogorov–Smirnov Test: Overview. John Wiley & Sons, Ltd, 2014.
dc.rights.accessrightsinfo:eu-repo/semantics/openAccess
dc.rights.licenseReconocimiento 4.0 Internacional
dc.rights.urihttp://creativecommons.org/licenses/by/4.0/
dc.subject.blaaProcesadores gráficosspa
dc.subject.blaaSimulación por computadorspa
dc.subject.ddc530 - Física::539 - Física moderna
dc.subject.ddc000 - Ciencias de la computación, información y obras generales::005 - Programación, programas, datos de computación
dc.subject.lembPartículas (física)spa
dc.subject.lembParticles (Nuclear physics)eng
dc.subject.lembCalorimetríaspa
dc.subject.lembCalorimeters and calorimetryeng
dc.subject.proposalLHCbspa
dc.subject.proposalGPU accelerationeng
dc.subject.proposalAceleración por GPUspa
dc.subject.proposalParticle transport simulationeng
dc.subject.proposalSimulación de transporte de partículasspa
dc.subject.proposalGaussinoeng
dc.subject.proposalAdePTeng
dc.subject.proposalComputación heterogéneaspa
dc.subject.proposalHeterogeneous computingeng
dc.subject.proposalElectromagnetic calorimetereng
dc.subject.proposalCalorimetro electromagnéticospa
dc.subject.wikidataLHCbspa
dc.titleProof of feasibility of an integration of AdePT in Gaussino simulations for improving performanceeng
dc.title.translatedPrueba de viabilidad de una integración de AdePT en las simulaciones de Gaussino para mejorar el desempeñospa
dc.typeTrabajo de grado - Maestría
dc.type.coarhttp://purl.org/coar/resource_type/c_bdcc
dc.type.coarversionhttp://purl.org/coar/version/c_ab4af688f83e57aa
dc.type.contentText
dc.type.driverinfo:eu-repo/semantics/masterThesis
dc.type.redcolhttp://purl.org/redcol/resource_type/TM
dc.type.versioninfo:eu-repo/semantics/acceptedVersion
dcterms.audience.professionaldevelopmentInvestigadores
dcterms.audience.professionaldevelopmentEstudiantes
oaire.accessrightshttp://purl.org/coar/access_right/c_abf2

Archivos

Bloque original

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
Trabajo_Final_Maestría.pdf
Tamaño:
3.22 MB
Formato:
Adobe Portable Document Format

Bloque de licencias

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
license.txt
Tamaño:
5.74 KB
Formato:
Item-specific license agreed upon to submission
Descripción: