Guerrero De La Cruz, Harold David

Profile Picture

Filters

20221
Reset filters

Settings

Citations

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Optimizing GEANT4 for event generation in Monte Carlo simulation at the HL-LHC
    (2022-05-20) Guerrero De La Cruz, Harold David; Malik, Sudhir; College of Arts and Sciences - Sciences; Marrero-Soto, Pablo J.; Santana, Samuel A.; Department of Physics; Rodríguez, Daniel
    Detector simulation is an essential tool to design, build and commission the highly sophisticated detectors utilised in modern particle physics experiments like the CMS Experiment at the Large Hadron Collider (LHC) at CERN. It is also a fundamental tool for data analysis and interpretation of the physics measurements. CMS detector simulation software is based on the GEANT4 simulation toolkit. It is a software package that provides the tools to describe the detector geometry and materials, and incorporates a number of models to simulate electromagnetic and hadronic interactions of particles with matter. The CMS detector will undergo sophisticated upgrades with more granular parts required to cope with the conditions of high luminosity run of the LHC, after 2027, referred to as the HL-LHC. The running conditions will be an instantaneous luminosity of 5–7.5x1034cm-2s-1, more radiation, more pileup and and data of up to 300 inverse femtobarns per year. The increasing complexity in the detector accompanied by an increase in data size adds to the very high computational cost with no change in CPU clock speeds. The data reconstruction time scales up with the pileup and with more complex algorithms to run, leads to larger fraction of total CPU usage. More accuracy is expected despite more complicated geometry and with a smaller fraction of total CPU usage compared to the past. This poses a considerable challenge for the already optimized GEANT4-based CMS detector simulation. One avenue being explored is to modify the simulation settings to process events even faster, but with less precision. Machine learning algorithms can then be applied to the reduced-precision output for a high-quality end result. In the work described in this thesis we investigated the impact of varying parameters, such as the energy limits below which particles can be discarded, on running time and physical performance of the detector simulation . We made modifications in the parameters individually and in combination.