CERN Accelerating science

ATLAS Slides
Report number	ATL-SOFT-SLIDE-2023-169
Title	ATLAS data analysis using a parallelized workflow on distributed cloud-based services with GPUs
Author(s)	Sandesara, Jay Ajitbhai (Amherst College (US)) ; Coelho Lopes De Sa, Rafael (University of Massachusetts (US)) ; Martinez Outschoorn, Verena Ingrid (University of Massachusetts (US)) ; Barreiro Megino, Fernando Harald (University of Texas at Arlington (US)) ; Elmsheuser, Johannes (Brookhaven National Laboratory (US)) ; Klimentov, Alexei (Brookhaven National Laboratory (US))
Corporate author(s)	The ATLAS collaboration
Collaboration	ATLAS Collaboration
Submitted to	26th International Conference on Computing in High Energy & Nuclear Physics, Norfolk, Virginia, Us, 8 - 12 May 2023
Submitted by	[email protected] on 08 May 2023
Subject category	Particle Physics - Experiment
Accelerator/Facility, Experiment	CERN LHC ; ATLAS
Abstract	We present a new implementation of simulation-based inference using data collected by the ATLAS experiment at the LHC. The method relies on large ensembles of deep neural networks to approximate the exact likelihood. Additional neural networks are introduced to model systematic uncertainties in the measurement. Training of the large number of deep neural networks is automated using a parallelized workflow with distributed computing infrastructure integrated with cloud-based services. We will show an example workflow using the ATLAS PanDA framework integrated with GPU infrastructure from Google Cloud Platform. Numerical analysis of the neural networks is optimized with JAX and JIT. The novel machine-learning method and cloud-based parallel workflow can be used to improve the sensitivity of several other analyses of LHC data.