ATLAS Fast Simulation - from classical to deep learning

by Hasib Ahmed (The University of Edinburgh (GB))

Wednesday 16 Feb 2022, 11:00 → 12:00 Europe/Zurich

ATLAS, one of the largest experiments at the Large Hadron Collider, has a broad physics program, ranging from precision measurements to the discovery of new interactions. Completing that program requires gargantuan amounts of simulated Monte Carlo events. Detailed detector simulation with Geant4 provides good agreement to data, but, due to the complexity of the detector, the CPU resources required are extraordinary. For more than 10 years, ATLAS has developed and utilised tools that replace the slowest part of the simulation - the calorimeter shower simulation - by faster alternatives. AtlFast3, or AF3, is the latest generation of high precision fast simulation in ATLAS. AF3 combines Geant4 with a parametrization-based Fast Calorimeter Simulation and a new deep learning-based Fast Calorimeter Simulation. AF3 has achieved the speed up required to meet the computing challenges and Monte Carlo needs for Run 3.  With unprecedented precision and the ability to model jet substructure, AF3 can be used to simulate almost all physics processes. For high luminosity LHC, further improvement in physics modelling along with a fast simulation for the inner detector is expected.

 

Hasib Ahmed is a Research Associate at the University of Edinburgh and a member of the ATLAS collaboration at CERN. He received his BSc from the University of Dhaka, Bangladesh and PhD from the University of Oklahoma, USA. His research interests are in experimental particle physics especially in Higgs physics and in novel techniques to model and simulate physics processes. 

ATLAS_FastSim_Hasib_Ahmed.pdf

video1945304383.mp4