A framework and implementation for data-driven trigger efficiency estimation at LHCb

Johannes Albrecht addr1, James Andrew Gooding e1,addr1, Maxim Lysenko addr2, Abhijit Mathad addr3, Alessandro Scarabotto addr1, Tomasz Skwarnicki addr4 · 2025

Read A framework and implementation for data-driven trigger efficiency estimation at LHCb by Johannes Albrecht addr1, James Andrew Gooding e1,addr1, Maxim Lysenko addr2, Abhijit Mathad addr3, Alessandro Scarabotto addr1, Tomasz Skwarnicki addr4 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-15951.txt.

Introduction

Six authors led by Johannes Albrecht build a shared tool to gauge how well a particle detector picks the events it wants. Such pick rates, called trigger efficiencies, sit at the base of most results at big colliders. Their method reads the traits of rebuilt events from real data, not from sims, to set the rate. The work serves the LHCb detector at CERN, which hunts rare particle decays.

The tool, called TriggerCalib, is the first joint home for these sums in that group. It drops into normal study code with no fuss for the user. Error bars, both from count noise and from method bias, get a full treatment in the paper. Design picks that keep the tool fast on huge data sets are laid out. Labs gain a shared yardstick in place of home-grown scripts.

Catalog record

CREATORJohannes Albrecht addr1, James Andrew Gooding e1,addr1, Maxim Lysenko addr2, Abhijit Mathad addr3, Alessandro Scarabotto addr1, Tomasz Skwarnicki addr4
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-15951

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A framework and implementation for data-driven trigger efficiency estimation at LHCb Johannes Albrecht addr1, James Andrew Gooding e1,addr1, Maxim Lysenko addr2, Abhijit Mathad addr3, Alessandro Scarabotto addr1, Tomasz Skwarnicki addr4 Abstract Estimations of trigger efficiencies are essential to modern particle physics analyses. A data-driven method provides a framework in which to estimate these efficiencies from the properties of reconstructed candidates, described in this paper. This paper also presents the design, implementation and performance of a software package, TriggerCalib, which provides a first centralised implementation of these calculations and can be seamlessly employed in physics analyses. Additionally, the estimation of statistical and systematic uncertainties is discussed. ∎ 1 1 1 E-mail: [email protected] Journal: Eur. Phys. J. C 1 Introduction Modern particle physics experiments employ multi-stage trigger systems, with the final stage usually performing full event reconstruction Albrecht and others (2025) . These reconstruct candidates and apply selection criteria to retain only those containing physics of interest to the experiment. The accurate estimation of the efficiency to select these reconstructed candidates is a crucial part any modern particle physics analysis. However; these efficiencies cannot simply be evaluated in simulated samples alone, as the complex nature of trigger selections is difficult to model accurately. To evaluate such efficiencies as the naive fraction of candidates accepted in recorded data would require prohibitively large samples of unfiltered data. Instead, a data-driven approach, dubbed the \mathrm{TISTOS} method, is employed to estimate efficiencies of the LHCb experiment trigger selection using the reconstructed candidates recorded for physics analysis. This paper presents both the framework of the \mathrm{TISTOS} method and its first centralised implementation as a Python-based software package. The LHCb experiment Alves Jr. and others (2008) ; Aaij and others (2015) is a forward single-arm spectrometer at the LHC, optimised for the study of heavy-flavour hadrons and primarily instrumented in the pseudorapidity range 2<\eta<5 . It collected proton-proton ( {\mathrm{p}}{\mathrm{p}} ) collision data between 2011 and 2018 during the LHC Run 1-2, reaching an integrated luminosity of 9 \text{\,fb}^{-1} . Subsequently, it underwent a major upgrade Aaij and others (2024) of the detectors and data acquisition system, with the aim of collecting data with a five times greater instantaneous luminosity than the original experiment, reaching {\mathcal{L}=2\times 10^{33}{\text{\,cm}}^{-2}{\text{\,s}}^{-1}} during the LHC Run 3 from 2022 to 2026. Prior to the upgrade, LHCb employed a hardware-based Level 0 (L0) trigger, and a two-stage High Level Trigger (HLT), operated on central processing units (CPU) LHCb collaboration (2003) . The hardware trigger was removed in the upgrade, such that a fully software-based trigger is operated in Run 3 LHCb collaboration (2014b) . The first trigger stage (HLT1) operates on graphical processing units, performing a partial event reconstruction at 30\text{\,MHz} , whilst the second stage (HLT2) operates on CPUs to perform a full offline-quality event reconstruction at 1\text{\,MHz} . Finally, data are further processed through the offline analysis framework Mathad and others (2024) ; Skidmore et al. (2022) . The software used in both trigger stages includes a set of algorithms processing information from the LHCb subdetectors to reconstruct and select processes of interest to the LHCb physics programme. The electronic responses of the subdetectors are used for four main reconstruction tasks: • tracking, reconstructing trajectories of particles from hits in the LHCb subdetectors; • vertexing, searching for the origin locations of {\mathrm{p}}{\mathrm{p}} collisions or of decaying particles; • particle identification, distinguishing charged final state particles’ nature; • neutral particle reconstruction, reconstructing neutral particles, such as photons and {\uppi}^{0} mesons, from calorimeter information. The selection algorithms applied to the resulting objects are referred to as trigger “lines”. Simulation is required to model the effects of the detector efficiency and the imposed selection requirements. In the simulation, {\mathrm{p}}{\mathrm{p}} collisions are generated using Pythia Sjöstrand et al. (2008) ; Sjöstrand et al. (2006) with a specific LHCb configuration Belyaev and others (2011) . Decays of unstable particles are described by EvtGen Lange (2001) , in which final-state radiation is generated using Photos Davidson et al. (2016) . The interaction of the generated particles with the detector, and its response, are implemented using the Geant4 toolkit Allison et al. (2006) ; Agostinelli and others (2003) as described in Clemencic and others (2011) . This paper provides first an overview of the \mathrm{TISTOS} method in Sec. 2 , which has already been used in many LHCb physics analyses. The implementation of this in the TriggerCalib software package is then introduced and demonstrated in Sec. 3 . Finally, Sec. 4 is dedicated to the estimation of statistical and systematic uncertainties. 2 The TISTOS method for trigger efficiencies Estimating efficiencies of trigger algorithms in data is non-trivial. whilst the recorded dataset contains only triggered events, there is sufficient redundancy between different trigger selections and signals to allow the trigger efficiencies to be estimated from these data. 2 2 2 A small dataset in which events need not pass a particular trigger is recorded for detector studies, but provides insufficient statistics for the evaluation of trigger efficiencies. Since this redundancy is insufficient to construct an efficiency as the simple fraction of events passing a given trigger selection, the LHCb collaboration developed a fully data-driven approach to evaluate such efficiencies. This method, referred to as the \mathrm{TISTOS} method, was first introduced in Aaij and others (2013) , is described in Tolk et al. (2014) ; Aaij and others (2019) and has been used extensively in the LHCb physics program. The method is defined in the context of b -decays, though in some cases can be extended to other processes. Whilst Refs. Aaij and others (2013) ; Aaij and others (2019) discuss the performance of the LHCb trigger and Tolk et al. (2014) offers a documentation of the \mathrm{TISTOS} method, this paper serves as a comprehensive description of the \mathrm{TISTOS} framework. 2.1 Trigger categories The \mathrm{TISTOS} method is built upon a tag -and- probe approach. The efficiency of a trigger of interest to select signal candidates is estimated by identifying a subsample of tag candidates selected by a certain trigger, assumed to be representative of all candidates in the dataset, and evaluating the fraction of these which are also probe candidates, selected by the trigger of interest. In the \mathrm{TISTOS} method, tag and probe categories, named Trigger Independent of Signal (TIS) and Trigger On Signal (TOS), respectively, are defined to construct data-driven trigger efficiency estimators. These categories characterise individual/aggregate trigger decisions with respect to a user-selected physics-signal candidate, e.g. , a reconstructed candidate corresponding to a {{{{B}^{+}}}\!\to{{J\mskip-3.0mu/\mskip-2.0mu\uppsi}}\left({\mu^{+}\mu^{-}}\right){{K}^{+}}} decay. Since trigger decisions are based on finding one or more reconstructed objects satisfying certain trigger criteria, the categorization task reduces to classification of individual trigger-selected objects with respect to each signal candidate selected during the data analysis phase. At HLT1 level, examples of the selected objects would be those of the HLT1TrackMVA and HLT1TwoTrackMVA trigger lines. The former selects significantly displaced high-momentum charged particle tracks; the latter selects displaced high-momentum two-particle vertices. Similarly, at HLT2 level an entire {{{{B}^{+}}}\!\to{{J\mskip-3.0mu/\mskip-2.0mu\uppsi}}\left({\mu^{+}\mu^{-}}\right){{K}^{+}}} candidate may be reconstructed. Signal candidates and trigger objects both originate from specific electronic signals (“detector hits”) in LHCb subdetectors, thus the basic \mathrm{TISTOS} algorithm classifies a set of trigger-related detector hits with respect to a set of detector hits associated with a physics signal

Read the complete text