Reconstrução de energia para calorímetros finamente segmentados

Abstract

This thesis presents data processing techniques of signal detection and energy estimation for high energy calorimetry. Modern calorimeters have thousands of readout channels and operate at high event rate conditions. Typically, the energy reconstruction involves both detection and estimation tasks, and it is based on the amplitude estimation of the received digitized signal. The current methods employed by high energy experiments are based on variance minimization techniques, and the valid signals are selected based on the energy estimation. This work explores the use of a technique based on Matched Filter for signal detection, and it makes use of a calibration factor to estimate the energy. In the proposed approach, the stochastic parameters of the pulse (phase and deformation) and the statistics from the background are considered for the filter design in order to increase performance. In particular cases, where the signal pile-up is likely to occur, another promising technique, based on linear signal deconvolution is discussed. The techniques proposed in this thesis were implemented offline and applied on the ATLAS Tile Calorimeter (TileCal) at LHC. Both simulated signals and real data acquired during nominal LHC operation were used. The proposed estimators presented smaller error with respect to the methods currently used in modern calorimeter systems, and they have been extensively tested to be used in TileCal.