Search for electroweak-scale dijet resonances using trigger-level analysis with the ATLAS detector in $132$ fb$^{-1}$ of $pp$ collisions at $\sqrt{s} = 13$ TeV

TL;DR

The study searches for sub-TeV to TeV dijet resonances in $pp$ collisions at $\sqrt{s}=13$ TeV using ATLAS trigger-level jets to access masses from $375$ to $1800$ GeV with $132\ \mathrm{fb}^{-1}$. It employs a data-driven, smooth background model and a BumpHunter-based scan to identify localized excesses, finding none and setting 95% CL limits on a benchmark leptophobic axial-vector $Z^{\prime}$ coupling $g_q$ and on cross-sections for Gaussian-shaped resonances. The results extend prior TLA coverage, provide a robust calibration and background treatment, and illustrate the power of trigger-level analyses to enhance sensitivity to electroweak-scale dijet resonances. Overall, the work constrains new dijet states in the electroweak regime and demonstrates the practical utility of trigger-level approaches for high-statistics searches at the LHC.

Abstract

This article reports on a search for dijet resonances using $132$ fb$^{-1}$ of $pp$ collision data recorded at $\sqrt{s} = 13$ TeV by the ATLAS detector at the Large Hadron Collider. The search is performed solely on jets reconstructed within the ATLAS trigger to overcome bandwidth limitations imposed on conventional single-jet triggers, which would otherwise reject data from decays of sub-TeV dijet resonances. Collision events with two jets satisfying transverse momentum thresholds of $p_{\textrm{T}} \ge 85$ GeV and jet rapidity separation of $|y^{*}|<0.6$ are analysed for dijet resonances with invariant masses from $375$ to $1800$ GeV. A data-driven background estimate is used to model the dijet mass distribution from multijet processes. No significant excess above the expected background is observed. Upper limits are set at $95\%$ confidence level on coupling values for a benchmark leptophobic axial-vector $Z^{\prime}$ model and on the production cross-section for a new resonance contributing a Gaussian-distributed line-shape to the dijet mass distribution.

Figures (9)

• Figure 1: Trigger rate as a function of time for the total L1 trigger and the L1 triggers feeding the TLA stream. The single LHC fill shown occurred in September 2018. While the J100 L1 item (dark blue histogram) collected events for TLA throughout the fill, the single-jet J50 and dijet J50_DETA20-J50J triggers (light blue and red histograms) activated as the LHC instantaneous luminosity (dashed line) and total L1 trigger rate (solid line) declined during the fill.
• Figure 2: Distributions of dijet invariant mass $m_{jj}$ for events collected via the TLA stream and reconstructed in the $\textrm{HLT}$, combining events selected at $\textrm{L1}$ by the $\pT > 100\text{Ge V}\xspace$ trigger (filled circles) and the prescaled single-jet and dijet $\pT > 50\text{Ge V}\xspace$ triggers (open squares), showing the increase in events recorded at lower values of dijet mass for this analysis compared with events collected via the main stream and reconstructed offline (histogram). For this illustrative comparison, each sample must satisfy the analysis selection described in \ref{['sec:selection']}, except for rejection of the tile gap region, and includes events with at least two trigger or offline jets, each with $\pT>85\text{Ge V}\xspace$ and $|\eta|<2.4$, with half a rapidity difference smaller than 0.6. The final sample used for the search is discussed and shown separately in \ref{['sec:selection']}.
• Figure 3: Stages and substeps of the trigger jet calibration and how they differ from the offline jet calibration. Following the jet area correction performed by the $\textrm{HLT}$ during data collection (yellow hatched), this analysis uses dedicated calibration steps (blue hatched) correcting for pile-up and differing energy response of hadronic signals throughout the detector. Steps of the offline calibration employing tracking are omitted due to the lack of tracking in the TLA stream (orange hatched). Trigger jets in data are further corrected with a dedicated $\eta$-intercalibration while the same absolute in situ calibration as derived for offline jets is applied (purple hatched). This is followed by a new trigger/offline step (green hatched) to correct for remaining differences between offline and trigger jets.
• Figure 4: Dependence of the reconstructed trigger jet on \ref{['fig:pileup_mu']} the average number of interactions per bunch crossings $\mu$ and \ref{['fig:pileup_npv']} the number of primary vertices $N_{\textrm{PV}}$ before (open squares) and after the two steps of pile-up calibration described in \ref{['sec:calibration']}, the area based correction (open triangles), and residual correction (filled circles) in simulated dijet events, as a function of the pseudorapidity $|\eta^\text{truth}|$ of the corresponding truth jet.
• Figure 5: \ref{['fig:etajes_response']} Trigger jet energy response, $E^{\textrm{trigger\xspace}}/E^{\textrm{truth}}\xspace$, in bins of detector pseudorapidity $\eta^{\textrm{det}}$ of the truth jet , $\pT^{\textrm{truth}}$, before applying the EtaJES calibration step (open markers in ranges of truth jet energy from 801500$\text{Ge V}$) and closure after applying the correction (filled circles). \ref{['fig:etajes_resolution']} Resolution $\sigma(\pT^{\textrm{trigger\xspace}}/\pT^{\textrm{truth}})$ of trigger jet transverse momentum as a function of truth jet after the EtaJES calibration step (open squares) and after each of the calorimeter-based GSC steps, Tile0 (open triangles) and LAr3 (filled circles) in the central detector $\eta$ bin $|\eta^{\textrm{det}}\xspace|<0.1$.