Fusão de informação com o NeuralRinger do Trigger do ATLAS para classificação de elétrons Boosted

Abstract

Online filtering systems typically have high speed design requirements and limited memory use, making difficult to develop new improvements to the systems. Moreover, it is essential to have no statistical trends, as there will be the irrecoverable disposal of events. In high energy physics experiments, online particle identification is crucial for any new discovery or data analysis. In particular, the calorimeters are essential for this task in different experiments, as they measure the energies of the incident particles and also reveal their nature from the energy profile they leave in the detector. Thus, in the ATLAS experiment, at the LHC (Large Hadron Collider), the calorimetry system plays a fundamental role in online selection of by products resulting from collisions, which occur every 25 nanoseconds. Based on calorimetry information, an ensemble of 30 neural networks (NeuralRinger) is currently used for the selection of electrons, which are messengers of important Physics processes of interest in the LHC. In this work, we evaluated the use of the fusion of information in the detection of boosted electrons, combining the original information of NeuralRinger (concentric energy deposition rings, built around an estimate of baricenter from the particle shower produced by the incident particle) with other calorimetry variables, recognized for their power of electron discrimination. Finally, we evaluate the relevance of these variables for the construction of the new online filter.