Reconstrução Online para Calorímetros Operando em Condições de Altas Luminosidades

Abstract

This paper presents an online energy reconstruction algorithm for calorimeters operating in high-luminosity environments, along with its FPGA implementation. Based on the Gradient Descent algorithm, the proposed method has its properties defined through the COF (Constrained Optimal Filter) approach, which aims to design a luminosity-independent signal amplitude estimator. In addition, COF enables the recovery of both the central signal amplitude and the amplitudes of each superposed component. However, the method involves matrix inversion, which is computationally expensive. To address this, the Gradient Descent algorithm is applied to the COF method, reducing the computations to sums and products, which are more suitable for FPGA-based online applications. To support this approach, a dedicated RISC (Reduced Instruction Set Computer) processor with a pipelined architecture was developed to optimize FPGA resource usage. Additionally, a C compiler was created to facilitate the development of Gradient Descent algorithm subroutines.