Calorimeter Clustering Algorithms : Description and Performance

TLDR

ATLAS uses two principal clustering algorithms. This note evaluates the performance of ATLAS calorimeter clustering algorithms, detailing their reconstruction steps and summarizing their effectiveness for particle identification. The reconstruction software employs a sliding‑window algorithm that clusters cells in fixed‑size rectangles for electron, photon, and tau identification, and a topological algorithm that groups neighboring cells with significant signals relative to noise. The topological algorithm’s clustering results improve jet and missing transverse energy reconstruction, as demonstrated using ATLAS ATHENA releases 12 and 13. Publication identifier: ATL‑LARG‑PUB‑2008‑002.

Abstract

This note describes the performance of the ATLAS calorimeter clustering algorithms, which provide inputs for particle identification. ATLAS uses two principal alg orithms. The first is the “sliding-window” algorithm, which clusters calorimeter cells within fixe d-size rectangles; results from this are used for electron, photon, and tau lepto n identification. The second is the “topological” algorithm, which clusters together neighboring ce lls, as long as the signal in the cells is significant compared to noise. The results of this seco nd algorithm are further used for jet and missing transverse energy reconstruction . This note first summarizes the steps of the calorimeter reconstruction softwar e. A detailed description of the two clustering algorithms is then given. A last section su mmarizes their performance. The results presented in this note are obtained with the ATLAS ATHENA software releases 12 and 13. ATL-LARG-PUB-2008-002