Monte Carlo simulations for the Pierre Auger Observatory using the VO auger grid resources

Abstract

The Pierre Auger Observatory, located near Malargüe, Argentina, is the world’s largest cosmic-ray detector. It comprises a 3000 km2 surface detector and 27 fluorescence telescopes, which measure the lateral and longitudinal distributions of the many millions of air-shower particles produced in the interactions initiated by a cosmic ray in the Earth’s atmosphere. The determination of the nature of cosmic rays and studies of the detector performances rely on extensive Monte Carlo simulations describing the physics processes occurring in extensive air showers and the detector responses. The aim of the Monte Carlo simulations task is to produce and provide the Auger Collaboration with reference libraries used in a wide variety of analyses. All multipurpose detector simulations are currently produced in local clusters using Slurm and HTCondor. The bulk of the shower simulations are produced on the grid, via the Virtual Organization auger, using the DIRAC middleware. The job submission is made via python scripts using the DIRAC-API. The Auger site is undergoing a major upgrade, which includes the installation of new types of detectors, demanding increased simulation resources. The novel detection of the radio component of extensive air showers is the most challenging endeavor, requiring dedicated shower simulations with very long computation times, not optimized for the grid production. For data redundancy, the simulations are stored on the Lyon server and the grid Disk Pool Manager and are accessible to the Auger members via iRODS and DIRAC, respectively. The CERN VM-File System is used for software distribution where, soon, the Auger Offline software will also be made available.