A Full Rank Pileup Deconvolution Scheme Suitable for Calorimeter Online Trigger Primitive Generation

TL;DR

This paper presents a novel full-rank deconvolution scheme for calorimeter data in high energy physics, enabling real-time trigger primitive generation without relying on mathematical guessing or sparse assumptions.

Contribution

It introduces a determined, full-rank deconvolution method suitable for online processing, overcoming limitations of traditional underdetermined approaches.

Findings

Deconvolution scheme is robust over long time windows.

The method achieves full-rank convolution matrices for online trigger tasks.

It eliminates the need for sparse representation assumptions.

Abstract

In this document, a pileup deconvolution scheme not relying on any mathematics guessing is presented. In high energy physics experiment, as the luminosity increases, pile-up issues on detectors such as calorimeters become non-negligible. Deconvolution approaches developed for data taken from DAQ systems are usually rank-deficient or underdetermined, having less equations than unknowns, even with the ADC values from multiple beam crossings are collected. These deconvolution approaches need mathematic pre-assumptions such as Sparse Representation. For online computation tasks such as for trigger primitive creation, signal availability is significantly different as in offline data analysis stage, and therefore, it is possible to use different (yet simpler) algorithms. In this situation, number of ADC values of the calorimeter outputs is the same as the number of beam crossings (or 4 times number of beam crossings, depending on the ADC sampling rate), and therefore, the number of equations can be arranged to be the same as number of unknowns. This way, a determined deconvolution scheme with a full-rank squared convolution (and deconvolution) matrix becomes possible. The robustness of deconvolution over long time windows is also studied in this paper.