Estudo de técnicas de otimização para reconstrução de energia de jatos no primeiro nível de seleção de eventos do experimento ATLAS

Abstract

The LHC is the biggest and most energetic particle collider, performing proton-proton collisions with 7 TeV on center of mass at 50 ns rate. Several detectors are placed along the LHC in order to measure the collisions results, envisaging the validation or rejection of the new particle physics theories. The ATLAS experiment is the biggest detector at LHC, operating very well since 2009 and providing large amount of data for the physics analysis envisaging the search for the Higgs Particle. The online event selection system (trigger) is one of the major reponsibles for the quality of the acquired data in ATLAS, it was designed to reject the huge amount of background noise generated at LHC. However, the LHC upgrade for high luminosity (10' to 1035) in the following ten years will increase the amount of events per collison, increasing the event pileup probability. In order to cope with the new luminosity requirements, the ATLAS and its trigger system will also upgrade its components and algorithms. This work presents the studies about the use of new algorithms for jet energy estimation in the first level of the ATLAS trigger system. Envisaging the reduction of the pileup effect, a bidimensional window is applied on the region where the event was identified (RoI). Additionally, the longitudinal information from the ATLAS hadronic calorimeter (TileCal) is used in order to improve the jet energy estimation, using linear (Least Square) and nonlinear (Artificial Neural Networks) optimization techniques. The achieved results, through Monte Carlo simulations, show that the TileCal logitudinal information reduces the energy estimation error by 23%, while the bidimensional window slightly improves the jet turn on curve indicating the reduction of the pileup effects.