Calibration of the jet energy scale and resolution of small-radius jets using semileptonic $tar{t}$ events with the ATLAS detector

A measurement of correction factors for the hadronic jet energy scale and resolution in the ATLAS detector is presented. These correction factors account for differences between simulated and observed data. They are obtained by analysing a selection of top quark events collected in proton-proton collisions by ATLAS between the years 2015 and 2018 at a centre-of-mass energy $\sqrt{s} = 13$ TeV as well as in 2022 and 2023 at $\sqrt{s} = 13.6$ TeV. The forward-folding technique is used to quantify the impact of different jet energy scale or resolution corrections on the reconstructed mass of the hadronically decaying $W$ boson from top-quark decays in simulation. The correction factors are extracted from a fit to the parameterised reconstructed $W$-boson mass distribution to data. The energy scale and resolution corrections are measured as a function of the jet transverse momentum between 20 GeV and 200 GeV and absolute pseudorapidity less than 0.8. The uncertainties in the energy scale range from about 0.93% to about 1.7% for jets between 35 and 200 GeV, while for the energy resolution the uncertainties range from about 14% to 28%. The method presented will be used in conjunction with other techniques to further improve ATLAS jet energy scale and resolution precision.

💡 Research Summary

**
This paper presents a new in‑situ calibration of the jet energy scale (JES) and jet energy resolution (JER) for small‑radius (R = 0.4) jets in the ATLAS detector, using semileptonic top‑antitop (tt̄) events. The analysis exploits the fact that in semileptonic tt̄ decays one W boson decays hadronically into two jets, whose invariant mass distribution is highly sensitive to both JES and JER. By constructing templates of the reconstructed W‑boson mass with varied JES and JER assumptions (the forward‑folding technique) and fitting these templates to data, the authors extract correction factors that bring simulation into agreement with observation.

Two data sets are used: the full Run 2 sample (2015‑2018, √s = 13 TeV, 140 fb⁻¹) and the Run 3 sample (2022‑2023, √s = 13.6 TeV, 52 fb⁻¹). Events are selected with a single‑electron or single‑muon trigger, at least four jets (two of which are b‑tagged), and exactly one isolated lepton. The two jets most consistent with originating from the hadronic W decay are identified, and only central jets (|η| < 0.8) with transverse momenta between 20 GeV and 200 GeV are considered.

The forward‑folding procedure starts from the nominal Monte‑Carlo (MC) simulation of tt̄ events (generated with Powheg‑Box v2 at NLO, showered with Pythia 8, and passed through a full Geant4 detector simulation). An arbitrary JES shift α and JER scaling β are applied to each jet’s four‑momentum, producing a family of modified MC samples. For each (α, β) pair the invariant mass of the two selected jets is histogrammed, yielding a template of the W‑mass distribution. A binned maximum‑likelihood fit of these templates to the observed data determines the best‑fit values of α and β, together with their statistical uncertainties.

Systematic uncertainties are evaluated by varying several ingredients: alternative MC generators (Sherpa, MadGraph), parton‑shower tunes, PDF sets, pile‑up re‑weighting, background modelling (fake/non‑prompt leptons, Z+jets, diboson), trigger and reconstruction efficiencies, and the functional form of the forward‑folding parameterisation itself. Each source is varied independently, the fit is repeated, and the resulting shift in α or β is taken as the systematic contribution. The total JES uncertainty for central jets ranges from about 0.93 % at 35 GeV to 1.7 % at 200 GeV, while the JER uncertainty varies from roughly 14 % at low pT to 28 % at the highest pT considered.

These results are comparable to, and in some regions improve upon, the precision achieved with traditional in‑situ methods such as Z/γ+jet balance (for JES) and dijet‑balance or random‑cone techniques (for JER). Notably, the tt̄‑based method provides a robust handle on the low‑pT JER, where other methods suffer from limited statistics or larger modelling uncertainties. Moreover, the technique works consistently across the two centre‑of‑mass energies, demonstrating its applicability to future higher‑energy data taking, including the upcoming High‑Luminosity LHC runs.

The authors conclude that the semileptonic tt̄ W‑mass method offers an independent and complementary calibration tool. When combined with existing calibrations, it can reduce the overall JES/JER systematic budget, thereby improving the precision of ATLAS measurements that rely on jets—such as Standard Model cross‑section determinations, Higgs boson studies, and searches for new physics. The paper also outlines a path toward integrating this calibration into the ATLAS global jet‑energy‑scale framework, enabling continuous monitoring of jet response throughout Run 3 and beyond.