ATLAS CONF Note

ATLAS-CONF-2018-052
26th November 2018

Search for boosted resonances decaying to two

b-quarks
√ and produced in association with a jet at
s = 13 TeV with the ATLAS detector

The ATLAS Collaboration

A search for new resonances decaying into a pair of bottom quarks in the merged regime is
√
reported using an integrated luminosity of 80.5 fb−1 of proton-proton collisions at s = 13 TeV
recorded by the ATLAS detector at the Large Hadron Collider. For the resonance to be
merged, an extra jet with high transverse momentum is required. The mass range from 70 to
ATLAS-CONF-2018-052

230 GeV is inspected. For the Standard Model Higgs boson, the observed signal strength is
µ H = 5.8 ± 3.1 (stat.) ± 1.9 (syst.) ± 1.7 (th.), consistent with the background-only hypothesis
27 November 2018

at 1.6 standard deviations. No evidence of a significant excess of events beyond the expected
background is found and limits on leptophobic Z 0 bosons with democratic axial couplings to
all quark generations are set in the probed mass range.

© 2018 CERN for the benefit of the ATLAS Collaboration.

Reproduction of this article or parts of it is allowed as specified in the CC-BY-4.0 license.
1 Introduction

Analyses searching for dijet resonances have been thoroughly explored by the LHC experiments [1–3]. The
main focus of these analyses are resonances at high masses, with the lowest mass boundary being dictated
by the trigger and data-acquisition systems of the experiments. Searching for resonances below the TeV
mass range is well motivated and alternative approaches employing more sophisticated trigger or analysis
strategies has flourished within the ATLAS and CMS collaborations when analysing the pp collision data
provided by the LHC [4–11].
An analysis searching for boosted resonances decaying to a pair of bottom quarks using an integrated
√
luminosity of 80.5 fb−1 of proton-proton collisions at s = 13 TeV is presented. For the resonance to
be boosted, an extra jet with high transverse momentum (pT ) is required to be present in the event. The
decay products of a hadronically-decaying resonance can be reconstructed in a single jet with a large
radius parameter (large-R jet), vastly reducing the amount of background from Standard Model (SM)
processes. The final state of interest therefore consists of a signal large-R jet plus an additional jet. The
additional jet is also taken with a large-radius cone to simplify the analysis and the propagation of the
related uncertainties.
Many physics models predict low-mass dijet resonances decaying to bottom quarks. Yukawa couplings
being proportional to the mass of the decay particle favour decays to b-quarks if the resonance mass is
lower than two-times the top-quark mass, as is true for the Higgs boson. Although the Higgs boson has
been discovered in Refs. [12, 13] and its coupling with bottom quarks recently observed in Refs. [14,
15], an inclusive analysis also targeting the gluon-fusion production mode and the decay to b-quarks was
conducted only recently in Ref. [16]. At high pT , the Higgs boson production starts being sensitive to
the top-quark loop in the gluon-fusion production mode [17], including possible contributions from new
resonances, and to the presence of anomalous couplings that give rise to an effective gluon-gluon-Higgs
interaction. In the presence of new physics contributions, increases in the high-pT Higgs boson production
are predicted to be as high as 50% [18]. A leptophobic Z 0 decaying to two quarks with a mass ranging
between 100 and 200 GeV is also searched for.

2 The ATLAS detector

The ATLAS experiment [19] at the LHC is a multi-purpose particle detector with a forward-backward
symmetric cylindrical geometry and a near 4π coverage in solid angle1. It consists of an inner tracking
detector surrounded by a thin superconducting solenoid providing a 2 T axial magnetic field, electromagnetic
and hadronic calorimeters, and a muon spectrometer. The inner tracking detector covers the pseudorapidity
range |η| < 2.5. It consists of, in ascending order of radius from the beam-line, silicon pixel, silicon
microstrip (SCT), and transition radiation tracking detectors. The pixel detectors are crucial for b-jet
√
identification. For the second LHC data-taking period at s = 13 TeV a new inner pixel layer, the Insertable
B-Layer (IBL) [20, 21], was added at a mean sensor radius of 3.2 cm from the beam-line. The calorimeter
system covers the pseudorapidity range |η| < 4.9. Within the region |η| < 3.2, electromagnetic calorimetry
is provided by barrel and endcap high-granularity lead/liquid-argon (LAr) electromagnetic calorimeters. An
1 ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point (IP) in the centre of the detector
and the z axis along the beam pipe. The x-axis points to the centre of the LHC ring, and the y axis points upwards. Cylindrical
coordinates (r, φ) are used in the transverse plane, φ being the azimuthal angle around the z-axis. The pseudorapidity
p η is
defined in terms of the polar angle θ as η = − ln tan(θ/2). Angular distance is measured in units of ∆R = ∆η2 + ∆φ2 .

2
additional thin LAr presampler, covering |η| < 1.8, is used to correct for energy loss in material upstream
of the calorimeters. Hadronic calorimetry is provided by a steel/scintillator-tile calorimeter, segmented into
three barrel structures within |η| < 1.7, and two copper/LAr hadronic endcap calorimeters. The solid-angle
coverage is completed with forward copper/LAr and tungsten/LAr calorimeter modules optimised for
electromagnetic and hadronic measurements, respectively. The muon spectrometer is composed of separate
trigger and high-precision tracking chambers, measuring the deflection of muons in a magnetic field
generated by superconducting air-core toroids. The precision chamber system covers the region |η| < 2.7
with three layers of monitored drift tubes, complemented by cathode strip chambers in the forward region,
where the particle flux is highest. The muon trigger system covers the range |η| < 2.4 with resistive plate
chambers in the barrel, and thin-gap chambers in the endcap regions. A two-level trigger system, using
custom hardware followed by a software-based level, is used to reduce the event rate to a maximum of
around 1 kHz for offline storage [22].

3 Data and simulated event samples

The data were collected by the ATLAS detector in pp collisions at the LHC with a centre-of-mass energy
of 13 TeV during the years 2015, 2016 and 2017. The data used correspond to an integrated luminosity
of 80.5 fb−1 . Events used for this search were recorded using a trigger requiring the presence of at least
one reconstructed anti-k t jet with radius parameter R = 1.0 [23]. The trigger threshold ranged between
pT > 360 GeV and 460 GeV across the years and was adapted following the different instantaneous
luminosity profiles and the inclusion of pile-up suppressing techniques at the trigger level [24].
Monte Carlo (MC) samples of the expected backgrounds are used for developing the non-resonant
background estimation procedure and for modelling the resonant backgrounds themselves, as detailed in
Section 6.
Simulated QCD dijet events were generated using the Pythia 8.186 [25] generator with the NNPDF 2.3
at next-to-leading order (NLO) parton distribution function [26] and the A14 set of tuned parameters
(tune) [27]. To maintain a constant statistical power over a large energy range, the weighted events were
generated with a flat jet pT spectrum and split into several samples. The hadronically decaying W and Z
events, with up to 4 additional partons, were generated using Sherpa 2.1.1 [28] with the CT10 [29] parton
distribution function. Matrix elements were calculated at leading order (LO) and the cross sections were
corrected to NLO using K-factors derived from corresponding samples with leptonic vector-boson decays
generated at NLO. The procedure, described in Ref. [5], results in a K-factor of 1.28 for W + jets and 1.38
for Z + jets with relative uncertainties of 10% based on the analysis described in Ref. [30]. To evaluate
the uncertainty due to the choice of MC generator, an alternative set of hadronically decaying W and Z
samples were generated using Herwig++ 2.7 [31] with the CTEQ6L1 [32] parton distribution function.
The simulated t t¯ events were generated at tree-level using Powheg-Box 2 [33–35] and the NNPDF 2.3
NLO parton distribution function. The hadronization was performed using Pythia 8.230 [36] with the A14
tune and the NNPDF 2.3 LO parton distribution function. To evaluate the uncertainty due to the choice of
MC generator, a second set of t t¯ events was generated by Sherpa 2.2.1 using the NNPDF 3.0 NNLO parton
distribution functions [37]. A sample containing single-top events in the Wt channel was generated at
tree-level using Powheg-Box 2 with the NNPDF 3.0 parton distribution function. The process is showered
using Pythia 8.230 configured with the A14 tune and the NNPDF 2.3 LO parton distribution function.

3
The signal samples considered are the Standard Model Higgs boson decaying to two b-quarks and a
leptophobic Z 0 with democratic axial coupling (gq ) to all quark generations. Several Higgs boson samples
were simulated corresponding to the three main production mechanisms; gluon-gluon fusion (ggF), vector
boson fusion (V BF) and Higgs-strahlung (W/Z associated production). All contribute in a non-negligible
manner to the signal region. The ggF and V BF Higgs boson events were generated at NLO with
top-mass-effect corrections following the MiNLO procedure [38, 39] using Powheg-Box 2 [40, 41] with
the NNPDF 3.0 NNLO parton distribution function. They were showered using Pythia 8.212 with the
AZNLO [42] tune and the CTEQ6L1 parton distribution function. The Higgs boson events produced in
association with a W or Z boson and initiated by quarks were generated using Pythia 8.212 with the
AZNLO tune and the CTEQ6L1 parton distribution function. The cross-sections are then scaled to the
LHCHXSWG recommendation [43], including extra factors to account for Z H production initiated by
gluons, which were not generated in the utilized samples. All decay modes of the vector boson were
enabled. The MiNLO procedure, with finite top-mass effects, is known to be a good approximation of the
ggF Higgs production at NNLO with parton shower matching and at high transverse momentum, with an
inclusive cross section of 29+24%
−21% fb for matrix-element level Higgs bosons with pT > 400 GeV, which is
consistent with fixed-order calculations at NNLO with top-mass effects corrections [39].
The events containing a leptophobic Z 0 decaying to two quarks and produced in association with a jet were
generated using a Simplified Dark Matter (DM) model [44] implemented in MadGraph5_aMC@NLO [45]
with the NNPDF 3.0 LO parton distribution function. The showering was performed using Pythia 8.212
with the A14 tune and the NNPDF 2.3 LO parton distribution function. Signal samples were generated at
several Z 0 masses spanning the search region. The values of mZ 0 are between 100 and 200 GeV in steps of
25 GeV. The simplified model defines an absolute axial coupling, gq , to an quark-antiquark pair that is
taken to be the same for all generations. The signal events were simulated with gq = 0.25. The specific
value does not have a significant effect on the kinematics of the reconstructed events because the natural
width of the Z 0 is smaller than the large-R jet mass resolution. The cross-section at other values can be
obtained by scaling as gq2 . The calculation was implemented using the DM width calculator in Ref. [46].
The DM particle mass is set to 10 TeV to not alter the Z 0 branching ratios.
All simulated events were created using the full detector simulation in Geant 4 [47, 48] and the same event
recontruction algorithms were used as for the data. The decay of b- and c-hadrons, with the exception of
the events generated by Sherpa, is consistently performed using EvtGen [49]. To account for the presence
of multiple interactions in the same and neighbouring bunch crossings (pile-up), all samples were overlayed
with minimum bias events simulated using Pythia 8.1 with the A3 tune [50] and the NNPDF 2.3 LO
parton distribution function. All simulated events are weighted to match the pile-up distribution observed
in the data.

4 Event reconstruction

The analysis makes use of the reconstruction of large-R jets with b-tagging applied to the associated small-R
jets. The associated small-R jets are track-jets with the radius dependent on their pT (variable-radius
track-jets). Muons are used to define a t t¯ control region as detailed in Section 6.
Collision vertices are reconstructed from at least two reconstructed tracks with transverse momentum
pT > 0.5 GeV. The primary vertex is selected as the one with the highest p2T , calculated considering all
Í
associated tracks with pT > 0.5 GeV.

4
Large-radius jets are reconstructed from topological clusters of energy deposited in the calorimeters using
the anti-k t algorithm with radius parameter of R = 1.0 and are trimmed [51] to improve the mass resolution
and reduce dependence on pile-up. The input clusters are calibrated using the local calibration method [52].
The trimming is achieved by reclustering the constituents of the initial jet, using the k t algorithm [53, 54],
into smaller R = 0.2 subjets and then by removing any subjet that has a pT less than 5% of the pT of the
parent jet [55]. The jet energy and pseudorapidity is corrected using pT - and η-dependent calibrations
derived from simulation [56]. The jet mass is calibrated by deriving pT - and η-dependent calibration
factors and by combining calorimeter and tracking information as described in Ref. [57]. A smaller mass
resolution and a better estimate of the median mass value is obtained than using only calorimeter energy
clusters. Only trimmed large-R jets with pT > 250 GeV and |η| < 2 are considered in the final selection.
Track-jets are clustered using the variable-radius jet algorithm [58, 59] with ρ = 30 GeV, Rmin = 0.02 and
Rmax = 0.4 on a subset of the reconstructed tracks. The Rmin and Rmax parameters, respectively, control the
minimum and maximum cone size, preventing the jets from becoming too large at low pT and going beyond
the detector resolution at high pT . The ρ parameter instead determines how fast the jet size decreases with
the jet pT . The considered tracks are required to have pT > 0.5 GeV and |η| < 2.5, at least 7 hits in the SCT
and pixel detector layers, no more than one hit shared by multiple tracks in the pixel detector, no more than
one missing hit in the pixel detector and no more than two missing hits in the SCT detector. Fake tracks
and tracks originating from pile-up vertices are suppressed by requiring the longitudinal impact parameter
(z0 ) to be within |z0 · sin θ| < 3 mm of the primary vertex. Only track-jets with pT > 10 GeV, |η| < 2.5, at
least two track constituents and satisfying ∆R/min R > 1, with min R the smallest radius of the track-jets
and ∆R the distance among any pair of them, are considered. The last requirement discards collinear
track-jets to prevent anomalous b-tagging response. The signal efficiency is not affected by imposing this
requirement. The reconstructed track-jets are ghost-associated [60] to the large-R jets before trimming.
Track-jets containing b-hadrons are identified using the multivariate b-tagging algorithm MV2c10 [61],
which makes use of observables with discriminating information provided by an impact parameter algorithm,
an inclusive secondary vertex finding algorithm and a multi-vertex finding algorithm. The MV2c10
algorithm is applied to a set of charged-particle tracks that satisfy quality and impact-parameter criteria
and are matched to reconstructed track-jets. Two b-tagging requirements are employed in the analysis for
defining the signal and control regions that, respectively, correspond to efficiencies of 77% and 85% for
b-jets with pT > 20 GeV when evaluated in a sample of simulated tt events. These two requirements are
referred to as ‘tight’ and ‘loose’ working points in the following. Scale factors to account for efficiency
differences observed in data and simulated events are applied.
Muons are reconstructed by combining tracks in the inner detector with tracks in the muon spectrometer
and are required to have pT > 10 GeV, |η| < 2.5 and pass the ‘medium’ identification and the ‘loose’
isolation working points [62] to supress fake reconstucted muons.

5 Event selection

Events containing jets induced by calorimeter noise bursts, beam-induced background or cosmic rays
are removed using the same criteria as in Ref. [63]. At least one reconstructed vertex and at least two
large-R jets are then required. Events are further preselected by requiring the leading large-R jet to satisfy
pT > 480 GeV and the sub-leading one pT > 250 GeV. The requirement on the leading large-R jet ensures
full trigger efficiency across years.

5
 Fraction

Arbitrary Units
ATLAS Preliminary BB CC CL ATLAS Preliminary
1 CRQCD
1.2 Simulation, s = 13 TeV
Simulation, s = 13 TeV BL LL BC
Pythia 8 QCD SR
Pythia 8 QCD - SR
1
10−1
0.8

0.6
10−2
0.4
Search Region

CRQCD
0.2 1.20 50 100 150 200 250

SR
1.1
1 Signal candidate large-R jet mass [GeV]
0 0.9
0 50 100 150 200 250 0.8
0 50 100 150 200 250
Signal candidate large-R jet mass [GeV] Signal candidate large-R jet mass [GeV]
(a) (b)

Figure 1: (a) Predicted flavour composition of the dijet background in the SR based on the truth-matched hadron
content of the two leading-pT track-jets associated to the signal candidate large-R jet, with the B/C labels indicating
the presence of a b/c-quark and L indicating the presence of a light quark or a gluon. (b) The expected shape of the
dijet background in the SR and CRQCD normalised to the same event count between 70 GeV < mJ < 230 GeV.

Next the large-R jet assumed to contain the decay products of the Z 0 or Higgs boson, referred to as the
signal jet candidate in the following, is chosen as the highest-pT large-R jet among those passing the
following requirements: large-R jet pT > 480 GeV, at least two associated track-jets and 2mJ /pT < 1
with mJ being the invariant mass of the large-R jet. The last requirement ensures that boosted decays are
selected.
The events are further classified based on the number of b-tagged jets in the two leading-pT track-jets
associated to the signal jet candidate. Events with exactly 0 loose b-tagged track-jets form the control
region for the non-resonant background estimation studies (CRQCD ). Events with exactly 2 tight b-tagged
track-jets form the signal region (SR). The 77% b-tagging working point used in the SR was optimised for
best signal significance. The mass range considered in the analysis is from 70 GeV to 230 GeV. The lower
bound is selected to avoid turn-on effects while the upper bound excludes high masses not resulting in
merged decays.
Figure 1 (left) shows how the dijet background in the SR is dominated by double b-quarks plus combinations
of (b- or c-) quarks. Events with no heavy (b- or c-) quarks make up only 5%. Figure 1 (right) shows
the simulated distribution of the signal candidate large-R jet mass of the dijet background in the SR and
CRQCD and how the shapes are close to each other in the mass range of interest. The signal region has a
higher turn-on curve ending roughly at 70 GeV which is not included in the analysis.
The efficiencies and yields in the CRQCD and SR for the considered resonant backgrounds and the Higgs
boson and Z 0 signals are shown in Table 1. The composition of the vector boson, t t¯ and H → bb̄ resonant
components in the SR and CRQCD is shown in Table 2. The vector boson component is dominated by the
W boson in the CRQCD . However, as expected by imposing the b-tagging requirement, Z → bb̄ is the
main contribution in the SR. The t t¯ contribution is divided equally in the two regions with ∼ 60% coming
from hadronic decays, ∼ 40% from semi-leptonic events and only a small percentage from di-leptonic
decays of the top quarks. The H → bb̄ component is mainly produced via ggF in all regions, contributing
53% of the Higgs boson signal in the SR. The alternative production mechanisms also contribute in a
non-negligible manner; VBF production contributes 25% and Higgs-strahlung the remaining 22%. This

6
 Process CRQCD Eff. (%) CRQCD Yield CRQCD Yield SR Eff. (%) SR Yield
80.5 fb−1 1.4 fb−1 80.5 fb−1
Z → q q̄ + jets 46.2 84400 1470 3.4 6200
W → q q̄ + jets 51.3 219000 3810 0.4 1500
t t¯ 25.9 110900 1929 2.5 10550
H → bb̄ (ggF) 23.6 140 2 19.4 115
H → bb̄ (V BF) 15.8 41 1 20.7 53
H → bb̄ (W H) 32.4 71 1 12.0 26
H → bb̄ (Z H) 30.5 40 1 15.8 21
H → bb̄ (Total) 24.3 292 5 17.9 216
Z 0 (m = 100 GeV) 43.9 87700 1530 4.1 8200
Z 0 (m = 125 GeV) 43.6 82800 1440 3.8 7300
Z 0 (m = 150 GeV) 43.5 78300 1360 3.7 6700
Z 0 (m = 175 GeV) 42.8 71100 1240 3.3 5550
Z 0 (m = 200 GeV) 41.8 64800 1127 3.2 4910
Data 38.7 29883000 519710 0.6 484600

Table 1: The efficiencies and yields in the CRQCD and SR for the non-QCD background, the Higgs boson and Z 0
boson signals and data. The yields in the CRQCD are given for the full luminosity 80.5 fb−1 and for the luminosity
employed, as explained in Section 6, for the background estimate of the non-resonant dijet process. The efficiencies
are relative to leading large-R jet pT > 480 GeV requirement.

CRQCD SR
Z + jets 0.28 0.80
V + jets
W + jets 0.72 0.20
All hadronic 0.58 0.63
Semi-leptonic 0.38 0.34
t t¯
Dileptonic 0.04 0.03
ggF 0.49 0.53
V BF 0.17 0.25
H → bb̄
WH 0.21 0.12
ZH 0.12 0.10

Table 2: The fractional composition of the different resonant contributions in the defined regions. The fraction is
evaluated using the given contribution type as the total.

indicates that a dedicated Higgs boson analysis with selection optimized for each of the production modes
could improve the sensitivity of the search.

6 Background modelling

The dominant background contribution in the SR is the non-resonant dijet process. Its estimate through
MC is not reliable due to the statistical precision and the underlying accuracy of the event generation. A
data-driven estimate is therefore employed by fitting the mass distribution mJ in the SR with a parametric
function and by validating the procedure using the data in the CRQCD . The shapes of the Z + jets and
W + jets contributions are taken from MC while their normalisation is fitted to data. The t t¯ contribution is

7
also estimated from MC with extra scale factors taken from a specific control region, denoted as CRtt in
the following.
The background estimation procedure is tested on the CRQCD . It is found that the statistical precision in
∼ 1 fb−1 of CRQCD data is comparable to the SR when using the luminosity of 80.5 fb−1 . The CRQCD is
therefore sliced into independent datasets used as probes for fitting the dijet background in the SR. The
CRQCD slices are constructed by grouping adjacent data runs. A total of about 60 slices is formed, with an
average of 1.4 fb−1 of data per slice.
To find a parametric function that fits the dijet background well, the following family of polynomial
exponential functions is considered:
n
!
   Õ
fn x  θ = θ 0 exp θi x ,
® i
(1)
i=1

where x = (mJ − 150 GeV)/80 GeV and θ i are the parameters of the fit. The parameterization of
the polynomial exponential function maps the independent variable to x ∈ [−1, 1] for the fit range of
[70, 230] GeV, which is empirically seen to provide improved numerical stability.
A slice is chosen randomly for performing studies to select the appropriate fit function. The Z +jets, W +jets
and t t¯ contributions, scaled by their cross sections times the luminosity, are subtracted. The t t¯ contribution
is corrected using the scale factor derived from the CRtt . A likelihood ratio test, in combination with Wilks’
theorem [64], and the F-test [65] are both used to determine the minimum number of model parameters
to describe the shape of the distribution. With the former preferring a five-parameter model and latter
favouring a four-parameter model, the five-parameter model is chosen to be conservative.
The fit is performed by adding the templates for the Z + jets, W + jets, t t¯ and single top contributions. The
fit function is validated using the data slices previously described. Within the statistical precision given by
the different data slices, the χ2 /ndf from the individual fits follows the expected distribution of a good
fit.
Boosted t t¯ events represent a significant background contribution in the SR. The t t¯ yield in the SR is
corrected with a data-driven normalization scale factor obtained by fitting the t t¯ MC template in the CRtt .
The CRtt is achieved by requiring a muon from the semi-leptonic decay from the top quark in the opposite
hemisphere of the top selected as the signal jet candidate.
The t t¯ events show the expected back-to-back topology of the muon and signal jet candidate, hence a
∆φ > 2π/3 requirement between the muon and the signal jet candidate greatly reduces the contribution
from muons coming from dijet and V + jets processes. Additional requirements on muon pT > 40 GeV and
∆R < 1.5 between the muon and the closest jet, plus a b-tagged jet at the loose working point are required
to eliminate soft muons from dijet events. The resulting CRtt is 90% pure in t t¯ events. The t t¯ MC template
is fitted to the CRtt data in the large-R mass region between 100 and 200 GeV. A normalisation factor of
0.84 is extracted and used to constrain the t t¯ contribution in the final fit to the SR.

7 Systematic uncertainties

Sources of systematic uncertainty that could potentially result in biases in the search are related to
experimental calibrations and to MC modeling of signal and background processes. These uncertainties

8
contribute both to the uncertainties in the overall yield (“normalisation”) and differential shape of the
large-R jet mass observable (“shape”) that is used in the statistical procedure for the final search.
Systematic uncertainties associated with the QCD background are estimated with pseudoexperiments.
Poissonian toys are generated from the QCD component of the fit in the SR containing all components
(QCD, V + jets, t t¯ and exotic signal or Higgs boson components) and no nuisance parameters. The toys are
refitted with the nominal QCD function, and an alternative parametric function (formal Laurent series [66]).
These two sets of fits to toys, nominal and alternative, provide a measure of the statistical uncertainty on
the QCD parametrization (the spread of fit parameters) and of the systematic uncertainty from the choice
of fitting function (the difference between the two fitted shapes).
Uncertainties affecting all MC templates are related to the large-R jet energy and mass calibrations [67] and
the calibration of the MV2c10 algorithm [61], which impacts different jet flavors with different magnitudes.
Large-R jet energy and mass calibration uncertainties affect both the shape and normalisation of the utilized
MC templates, therefore, their impact on the analysis is assessed by varying the jet energy and mass
within their uncertainties and propagating those variations through the analysis chain. The effect of the jet
energy resolution uncertainty is also tested and found to be negligible. The impact of uncertainties on the
calibration of MV2c10 have been found to be independent of the large-R jet mass for both the signal and
background samples, and hence only affecting the signal normalisation.
Extra shape uncertainties are applied to the V + jets and t t¯ MC templates related to modeling uncertainties.
To assess these uncertainties, the large-R jet mass shapes from two different MC generators are compared
in both cases. For the V + jets component, the nominal shape, generated using Sherpa 2.1.1, is compared
to an alternative shape, generated using Herwig++ 2.7. Similarly, for the t t¯ component, the comparison is
performed between the nominal Powheg-Box 2 shape, and the Sherpa 2.2.1 alternative shape.
√
Impact on Signals ( ∆σ 2 /µ)
Source Type V+jets Higgs Z’ (100 GeV) Z’ (175 GeV)
Jet energy and mass scale Norm. & Shape 15% 14% 23% 18%
Jet mass resolution Norm. & Shape 20% 17% 30% 20%
V + jets modeling Shape 9% 4% 4% < 1%
t t¯ modeling Shape < 1% 1% < 1% 11%
b-tagging (b) Normalisation 11% 12% 11% 15%
b-tagging (c) Normalisation 3% 1% 3% 5%
b-tagging (l) Normalisation 4% 1% 4% 7%
t t¯ scale factor Normalisation 2% 3% 2% 58%
Luminosity Normalisation 2% 2% 2% 3%
Alternative QCD function Norm. & Shape 4% 4% 3% 17%
W/Z and QCD (Theory) Normalisation 14% – – –
Higgs (Theory) Normalisation – 30% – –

Table 3: Summary of the impact of the main systematic uncertainties on the uncertainty σ on the measurement of the
signal strength µ for the V + jets, Higgs boson and Z 0 signals.

As previously mentioned, the t t¯ normalisation in the signal region is constrained with the scale factor
obtained in the t t¯ enriched control region. The t t¯ scale factor has been measured with an uncertainty
of 13%, which is treated as a systematic uncertainty on the t t¯ normalisation.

9
The uncertainty in the integrated luminosity is 2.1%. It is derived following a methodology similar to that
detailed in Ref. [68] and it is applied to all processes modelled using MC simulations.
In order to compare the observed number of events from the V + jets and Higgs components of the signal
region fit, we apply theory uncertainties to their normalisations. Theory uncertainties on the V + jets
processes come from the impact of higher order electroweak and QCD corrections to their differential cross
sections [69]. Among the Higgs boson production theory uncertainties, the gluon-gluon fusion uncertainty
is dominant. It is taken as 30%, consistent with the cross section uncertainty as calculated with the MiNLO
procedure and including top-mass effects for Higgs bosons with pT > 400 GeV. The same uncertainty is
also applied to the other production mechanisms at high pT . This results in a total uncertainty on the Higgs
cross-section of 30%.
The effects of the main systematic uncertainties on the measurement of the signal strength µ, defined as the
scale factor multiplying the cross section times branching fraction predicted by the signal hypothesis, for
the V + jets, Higgs boson and Z 0 signals are displayed in Table 3. The impact of a systematic uncertainty is
defined as the difference in quadrature between the uncertainty σ in µ computed when all other uncertainties
are considered and when are fixed to their pre-fit values. The total systematic uncertainty is then defined as
the√difference in quadrature between the total uncertainty in µ and the total statistical uncertainty, denoted
as ∆σ 2 in the table.

8 Results

To search for and quantify the significance of V + jets production and to set limits on the Higgs boson
and Z 0 boson models, a Bayesian method is used to calculate a posterior likelihood as a function of
number of signal events for the signal hypotheses under consideration [70]. Systematic uncertainties enter
the likelihood as nuisance parameters. In this method, the Bayes’ theorem is utilised to build the final
conditional probability of the parameters of interest given the observed data by integrating over the nuisance
parameters (“marginalisation”) with a Markov Chain Monte Carlo procedure. The posterior probability
distribution is used to gauge the fitted signal statistical significance or to set 95% credibility-level upper
limits on the cross-section times acceptance times efficiency.
The first result obtained is the combined SM fit of the V + jets, H + jets and t t¯ plus QCD components.
With this fit, the signal strengths of the V + jets and H + jets processes, which are not constrained to
positive values, are obtained simultaneously. The normalisation of the t t¯ component is constrained with
the scale factor obtained in the dedicated CRtt . The comparison to data after the marginalisation of
the nuisance parameters is shown in Figure 2. The observed signal strength for the V + jets process is
µV = 1.5 ± 0.22 (stat.) +0.29
−0.25
(syst.) ± 0.18 (th.), corresponding to a significance of 5 standard deviations.
For the H + jets process, the observed signal strength is µ H = 5.8 ± 3.1 (stat.) ± 1.9 (syst.) ± 1.7 (th.),
consistent with the background-only hypothesis at 1.6 σ. The combined likelihood of µV and µ H is shown
in Figure 3.
The BumpHunter search procedure [71, 72], assuming the V + jets and t t¯ templates, is then performed for
looking for possible extra exotic signals in the large-R jet mass distribution. The effect of the Higgs boson
with a SM strength is smaller than the expected uncertainty on the Z’ limit and is therefore neglected. The
first step to this procedure is to perform a fit with the full set of systematic uncertainties and obtain the
best fit values for the nuisance parameters that are utilized in the analysis. Given these postfit shapes, the
BumpHunter algorithm looks for significant deviations from the background model. The largest deviation

10
 ×103
Data-QCD-tt Events / 5 GeV Events / 5 GeV

Data SM Higgs (µ = 5.8)

ATLAS Preliminary H
25 -1 V+Jets (µ = 1.5)
s = 13 TeV, 80.5 fb QCD Fit V

Signal Region QCD Fit ± 1σ Top

20

15 ×103
Data SM Higgs (µ = 5.8)
ATLAS Internal H
25
10 -1 V+Jets (µ = 1.5)
s = 13 TeV, 80.5 fb QCD Fit V

Signal Region QCD Fit ± 1σ Top

20
5
×103
15
2
QCD and Top ± 1σ

10
0
5
803
×10 100 120 140 160 180 200 220
1
Data-QCD-tt-V

Signal candidate large-R jet mass [GeV]

QCD, Top and V+Jets ± 1σ

−1
80 100 120 140 160 180 200 220
Signal candidate large-R jet mass [GeV]

Figure 2: Postfit plot of the SM Higgs boson, V + jets, t t¯ and QCD fit comparison to data. The middle panel shows
the postfit and data distributions with the QCD and t t¯ components subtracted. The lower panel shows the same
distributions when also the V + jets component is subtracted.

11
 V
µ
ATLAS Preliminary Best Fit Value
3.5 s = 13 TeV, 80.5 fb-1 SM Prediction
Signal Region 68% CL
3 95% CL

2.5

1.5

0.5
−5 0 5 10 15 µ
H

Figure 3: Combined probability distribution of µ H and µV from the signal region fit.

from the background found by the BumpHunter algorithm, as depicted in Figure 4, is found for a large-R
mass value around 125 GeV, although with a small statistical significance. The corresponding global
p-value is found to be 0.54.
This analysis also sets 95% confidence level limits on signals from dark matter mediators that decay
democratically to quarks, with masses between 100 and 200 GeV. These limits are shown in Figure 5,
in terms of cross section times branching ratio, acceptance and efficiency (left), and in terms of the gq
parameter that controls the coupling of the DM mediator to quarks and thus determines the cross-section
(right).

9 Conclusion

A search for boosted resonances decaying to two b-quarks and produced in association with an extra jet
with high transverse momentum is performed. The analysis uses an integrated luminosity of 80.5 fb−1 of
√
proton-proton collisions with a center-of-mass energy of s = 13 TeV recorded by the ATLAS detector at
the Large Hadron Collider. The search probes the mass range between 70 and 230 GeV. The observed
signal strenght for the V + jets process is µV = 1.5 ± 0.22 (stat.) +0.29
−0.25
(syst.) ± 0.18 (th.), corresponding to
a significance of 5 standard deviations. For the SM Higgs boson process, the observed signal strength is
µ H = 5.8 ± 3.1 (stat.) ± 1.9 (syst.) ± 1.7 (th.), consistent with the background-only hypothesis at 1.6 σ.
No evidence of a significant excess of events above the expected SM processes is found and limits on
leptophobic Z 0 with democratic axial couplings to all quark generations (gq ) are set in the probed mass
range.

12
 30000

Events / 5 GeV
ATLAS Preliminary Data
s =13 TeV, 80.5 fb -1 Background fit
BumpHunter interval
20000

p -value = 0.54
Fit Range: 70 - 230 GeV

10000
Significance

3
2
1
0
−1
−2
−3
80 100 120 140 160 180 200 220
mJ [GeV]

Figure 4: The reconstructed mass distribution mJ with the event reconstruction and selection as described in the text.
The solid red line depicts the background prediction, consisting of the non-resonant dijet, V + jets and t t¯ processes.
The vertical blue lines indicate the most discrepant interval identified by the BumpHunter algorithm. Without
including systematic uncertainties, the probability that fluctuations of the background model would produce an excess
at least as significant as the one observed in the data anywhere in the distribution, the BumpHunter probability, is
0.54. The low panel shows the bin-by-bin significances of the differences between the data and the fit, considering
only statistical fluctuations.

0.4
gq
σ × ∈ × A × BR [pb]

ATLAS Preliminary ATLAS Preliminary

s=13 TeV, 80.5 fb-1 0.35 s=13 TeV, 80.5 fb-1

10−1 0.3

0.25

0.2

10−2 0.15
Z’ (gq = 0.25)
0.1
Observed 95% CL upper limit Observed 95% CL upper limit
Expected 95% CL upper limit 0.05 Expected 95% CL upper limit
68% and 95% bands 68% and 95% bands
3
10−100 120 140 160 180 200 100 120 140 160 180 200
mZ’ [GeV] mZ' [GeV]

(a) (b)

Figure 5: The 95% credibility-level upper limits obtained from the invariant mass distribution on (a) the cross-section
times acceptance times branching ratio times efficiency for the Z 0 model described in the text and on (b) the gq
parameter that controls the decay width of the DM mediator into SM particles.

13
A Data/MC comparisons in the signal region

Comparisons between some key distributions in the data and simulated background samples in the SR,
for illustration only as the non-resonant dijet MC is not used for background estimation in the analysis,
are shown in Figure 6. The considered backgrounds are the non-resonant dijet, Z + jets, W + jets and t t¯
processes. Good modelling of the dijet background is not is expected. The normalization is corrected
by fitting, in the range 70 GeV < mJ < 230 GeV, the signal candidate large-R jet mass distribution with
an overall normalisation scale factor which is found to be 0.903 for SR. This factor is applied in the
distributions in Figure 6. Uncertainties in the ratio plots are only statistical.

Events [ / 10 GeV]
Events [ / 1 ]

ATLAS Preliminary W Z ATLAS Preliminary W Z

107 106
s = 13 TeV, 80.5 fb-1 tt QCD s = 13 TeV, 80.5 fb-1 tt QCD
106 SR MC Stat Data 105 SR MC Stat Data
105 104
104 103
103
102
102
10
10
1
1
10−1 10−1
Data/MC

Data/MC

1.5
1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 1.2 500 600 700 800 900 1000

1 Nfatjet 1 Signal candidate large-R jet pT [GeV]

0.5 0.8
1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 500 600 700 800 900 1000
Nfatjet Signal candidate large-R jet p [GeV]
T

(a) (b)

106 107
Events [ / 5 GeV]

Events [ / 1 ]

ATLAS Preliminary W Z ATLAS Preliminary W Z

10 5 s = 13 TeV, 80.5 fb-1 tt QCD s = 13 TeV, 80.5 fb-1 tt QCD
106
SR MC Stat Data SR MC Stat Data
104 105
3
10
104
102
103
10
102
1
10
10−1
Data/MC

Data/MC

1.2 80 100 120 140 160 180 200 220 1.5

1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5

1 Signal candidate large-R jet mass [GeV] 1 Ntrkjet

0.8 0.5
80 100 120 140 160 180 200 220 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5
Signal candidate large-R jet mass [GeV] Ntrkjet

(c) (d)

Figure 6: The data vs MC comparison in the SR (a) of the number of large-R jets with pT > 250 GeV in the event, (b)
of the leading signal candidate jet pT and (c) large-R jet mass, and (d) the number of track-jets associated to the
signal candidate are shown. All the distributions are extracted by looking at the SR. The simulated dijet contribution
is scaled by a constant factor of 0.90 in all the distributions, as determined when fitting the candidate large-R jet mass
distribution. Uncertainties in the ratio plots are only statistical. The gray band is the uncertainty from Monte Carlo
statistics.

14
References

[1] ATLAS Collaboration, Search for new phenomena in dijet events using 37 fb−1 of pp collision data
√
collected at s = 13 TeV with the ATLAS detector, Phys. Rev. D 96 (2017) 052004,
arXiv: 1703.09127 [hep-ex].
√
[2] CMS Collaboration, Search for dijet resonances in proton–proton collisions at s = 13 TeV and
constraints on dark matter and other models, Phys. Lett. B 769 (2017) 520,
arXiv: 1611.03568 [hep-ex].
[3] CMS Collaboration, Search for new physics in dijet angular distributions using proton–proton
√
collisions at s = 13 TeV and constraints on dark matter and other models, (2018),
arXiv: 1803.08030 [hep-ex].
[4] ATLAS Collaboration,
Search for new light resonances decaying to jet pairs and produced in association with a photon or
√
a jet in proton–proton collisions at s = 13 TeV with the ATLAS detector,
ATLAS-CONF-2016-070, 2016, url: https://cds.cern.ch/record/2206221.
[5] ATLAS Collaboration,
Search for light resonances decaying to boosted quark pairs and produced in association with a
√
photon or a jet in proton–proton collisions at s = 13 TeV with the ATLAS detector, (2018),
arXiv: 1801.08769 [hep-ex].
[6] ATLAS Collaboration, Search for resonances in the mass distribution of jet pairs with one or two
√
jets identified as b-jets in proton–proton collisions at s = 13 TeV with the ATLAS detector,
Phys. Rev. D 98 (2018) 032016, arXiv: 1805.09299 [hep-ex].
[7] CMS Collaboration, Search for low mass vector resonances decaying to quark–antiquark pairs in
√
proton–proton collisions at s = 13 TeV, Phys. Rev. Lett. 119 (2017) 111802,
arXiv: 1705.10532 [hep-ex].
[8] CMS Collaboration, Search for low mass vector resonances decaying into quark–antiquark pairs in
√
proton–proton collisions at s = 13 TeV, JHEP 01 (2018) 097, arXiv: 1710.00159 [hep-ex].
[9] CMS Collaboration, Search for low-mass resonances decaying into bottom quark-antiquark pairs in
√
proton-proton collisions at s = 13 TeV, Submitted to: Phys. Rev. (2018),
arXiv: 1810.11822 [hep-ex].
[10] ATLAS Collaboration, Search for low-mass dijet resonances using trigger-level jets with the ATLAS
√
detector in pp collisions at s = 13 TeV, Phys. Rev. Lett. 121 (2018) 081801,
arXiv: 1804.03496 [hep-ex].
[11] CMS Collaboration, Search for narrow and broad dijet resonances in proton–proton collisions at
√
s = 13 TeV and constraints on dark matter mediators and other new particles,
JHEP 08 (2018) 130, arXiv: 1806.00843 [hep-ex].
[12] ATLAS Collaboration, Observation of a new particle in the search for the Standard Model Higgs
boson with the ATLAS detector at the LHC, Phys. Lett. B 716 (2012) 1,
arXiv: 1207.7214 [hep-ex].
[13] CMS Collaboration,
Observation of a new boson at a mass of 125 GeV with the CMS experiment at the LHC,
Phys. Lett. B 716 (2012) 30, arXiv: 1207.7235 [hep-ex].

15
[14] ATLAS Collaboration,
Observation of H → bb̄ decays and V H production with the ATLAS detector,
Phys. Lett. B786 (2018) 59, arXiv: 1808.08238 [hep-ex].
[15] CMS Collaboration, Observation of Higgs boson decay to bottom quarks,
Phys. Rev. Lett. 121 (2018) 121801, arXiv: 1808.08242 [hep-ex].
[16] CMS Collaboration,
Inclusive search for a highly boosted Higgs Boson decaying to a bottom quark–antiquark pair,
Phys. Rev. Lett. 120 (2018) 071802, arXiv: 1709.05543 [hep-ex].
[17] M. Schlaffer, M. Spannowsky, M. Takeuchi, A. Weiler and C. Wymant, Boosted Higgs shapes,
Eur. Phys. J. C 74 (2014) 3120, arXiv: 1405.4295 [hep-ph].
[18] C. Grojean, E. Salvioni, M. Schlaffer and A. Weiler, Very boosted Higgs in gluon fusion,
JHEP 05 (2014) 022, arXiv: 1312.3317 [hep-ph].
[19] ATLAS Collaboration, The ATLAS Experiment at the CERN Large Hadron Collider,
JINST 3 (2008) S08003.
[20] ATLAS Collaboration, ATLAS Insertable B-Layer Technical Design Report,
(2010), CERN-LHCC-2010-013, ATLAS-TDR-019,
url: http://cds.cern.ch/record/1291633.
[21] ATLAS Collaboration, ATLAS Insertable B-Layer Technical Design Report Addendum,
(2012), CERN-LHCC-2012-009. ATLAS-TDR-19-ADD-1,
url: http://cds.cern.ch/record/1451888.
[22] ATLAS Collaboration, Performance of the ATLAS trigger system in 2015,
Eur. Phys. J. C 77 (2017) 317, arXiv: 1611.09661 [hep-ex].
[23] M. Cacciari, G. P. Salam and G. Soyez, The anti-k t jet clustering algorithm, JHEP 04 (2008) 063,
arXiv: 0802.1189 [hep-ph].
[24] ATLAS Collaboration, Performance of pile-up mitigation techniques for jets in pp collisions at
√
s = 8 TeV using the ATLAS detector, Eur. Phys. J. C 76 (2016) 581,
arXiv: 1510.03823 [hep-ex].
[25] T. Sjöstrand, S. Mrenna and P. Z. Skands, A brief introduction to PYTHIA 8.1,
Comput. Phys. Commun. 178 (2008) 852, arXiv: 0710.3820 [hep-ph].
[26] R. D. Ball et al., Parton distributions with LHC data, Nucl. Phys. B 867 (2013) 244,
arXiv: 1207.1303 [hep-ph].
[27] ATLAS Collaboration, ATLAS Pythia 8 tunes to 7 TeV data, ATL-PHYS-PUB-2014-021, 2014,
url: https://cds.cern.ch/record/1966419.
[28] T. Gleisberg, S. Höche, F. Krauss, M. Schönherr, S. Schumann et al.,
Event generation with SHERPA 1.1, JHEP 02 (2009) 007, arXiv: 0811.4622 [hep-ph].
[29] H.-L. Lai et al., New parton distributions for collider physics, (2010), eprint: arXiv:1007.2241.
[30] ATLAS Collaboration, Measurements of the production cross section of a Z boson in association
√
with jets in pp collisions at s = 13 TeV with the ATLAS detector, Eur. Phys. J. C 77 (2017) 361,
arXiv: 1702.05725 [hep-ex].
[31] M. Bahr et al., Herwig++ physics and manual, Eur. Phys. J. C 58 (2008) 639,
arXiv: 0803.0883 [hep-ph].

16
[32] J. Pumplin et al.,
New generation of parton distributions with uncertainties from global QCD analysis,
JHEP 07 (2002) 012, arXiv: hep-ph/0201195.
[33] P. Nason, A New method for combining NLO QCD with shower Monte Carlo algorithms,
JHEP 11 (2004) 040, arXiv: hep-ph/0409146.
[34] S. Frixione, P. Nason and C. Oleari,
Matching NLO QCD computations with Parton Shower simulations: the POWHEG method,
JHEP 11 (2007) 070, arXiv: 0709.2092 [hep-ph].
[35] S. Alioli, P. Nason, C. Oleari and E. Re, A general framework for implementing NLO calculations in
shower Monte Carlo programs: the POWHEG BOX, JHEP 06 (2010) 043,
arXiv: 1002.2581 [hep-ph].
[36] T. Sjöstrand et al., An introduction to PYTHIA 8.2, Comput. Phys. Commun. 191 (2015) 159,
arXiv: 1410.3012 [hep-ph].
[37] R. D. Ball et al., Parton distributions for the LHC Run II, JHEP 04 (2015) 040,
arXiv: 1410.8849 [hep-ph].
[38] K. Hamilton, P. Nason and G. Zanderighi, MINLO: Multi-Scale Improved NLO,
JHEP 10 (2012) 155, arXiv: 1206.3572 [hep-ph].
[39] K. Hamilton, P. Nason and G. Zanderighi,
Finite quark-mass effects in the NNLOPS POWHEG+MiNLO Higgs generator,
JHEP 05 (2015) 140, arXiv: 1501.04637 [hep-ph].
[40] P. Nason and C. Oleari,
NLO Higgs boson production via vector-boson fusion matched with shower in POWHEG,
JHEP 02 (2010) 037, arXiv: 0911.5299 [hep-ph].
[41] J. M. Campbell et al., NLO Higgs boson production plus one and two jets using the POWHEG BOX,
MadGraph4 and MCFM, JHEP 07 (2012) 092, arXiv: 1202.5475 [hep-ph].
[42] ATLAS Collaboration, Measurement of the Z/γ ∗ boson transverse momentum distribution in pp
√
collisions at s = 7 TeV with the ATLAS detector, JHEP 09 (2014) 145,
arXiv: 1406.3660 [hep-ex].
[43] B. Mellado Garcia, P. Musella, M. Grazzini and R. Harlander,
CERN Report 4: Part I Standard Model Predictions, (2016),
url: https://cds.cern.ch/record/2150771.
[44] D. Abercrombie et al., Dark matter benchmark models for early LHC Run-2 searches: report of the
ATLAS/CMS dark matter forum,
(2015), ed. by A. Boveia, C. Doglioni, S. Lowette, S. Malik and S. Mrenna,
arXiv: 1507.00966 [hep-ex].
[45] J. Alwall, R. Frederix, S. Frixione, V. Hirschi, F. Maltoni et al.,
The automated computation of tree-level and next-to-leading order differential cross sections, and
their matching to parton shower simulations, JHEP 07 (2014) 079, arXiv: 1405.0301 [hep-ph].
[46] Andrew Nelson, url: https://github.com/LHC-DMWG/DarkMatterWidthCalculation.
[47] S. Agostinelli et al., GEANT4: A Simulation toolkit, Nucl. Instrum. Meth. A 506 (2003) 250.
[48] ATLAS Collaboration, The ATLAS Simulation Infrastructure, Eur. Phys. J. C 70 (2010) 823,
arXiv: 1005.4568 [physics.ins-det].

17
[49] D. J. Lange, The EvtGen particle decay simulation package, Nuclear Instruments and Methods in
Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 462
(2001) 152, BEAUTY2000, Proceedings of the 7th Int. Conf. on B-Physics at Hadron Machines,
issn: 0168-9002,
url: http://www.sciencedirect.com/science/article/pii/S0168900201000894.
[50] ATLAS Collaboration, The Pythia 8 A3 tune description of ATLAS minimum bias and inelastic
measurements incorporating the Donnachie–Landshoff diffractive model,
ATL-PHYS-PUB-2016-017, 2016, url: https://cds.cern.ch/record/2206965.
[51] D. Krohn, J. Thaler and L.-T. Wang, Jet trimming, JHEP 02 (2010) 084,
arXiv: 0912.1342 [hep-ph].
[52] ATLAS Collaboration,
√
Jet energy measurement with the ATLAS detector in proton–proton collisions at s = 7 TeV,
Eur. Phys. J. C 73 (2013) 2304, arXiv: 1112.6426 [hep-ex].
[53] S. Catani, Y. L. Dokshitzer, M. H. Seymour and B. R. Webber,
Longitudinally invariant Kt clustering algorithms for hadron hadron collisions,
Nucl. Phys. B406 (1993) 187.
[54] S. D. Ellis and D. E. Soper, Successive combination jet algorithm for hadron collisions,
Phys. Rev. D48 (1993) 3160, arXiv: hep-ph/9305266 [hep-ph].
[55] ATLAS Collaboration, Identification of boosted, hadronically decaying W bosons and comparisons
√
with ATLAS data taken at s = 8 TeV, Eur. Phys. J. C 76 (2016) 154,
arXiv: 1510.05821 [hep-ex].
[56] ATLAS Collaboration, Performance of jet substructure techniques for large-R jets in proton–proton
√
collisions at s = 7 TeV using the ATLAS detector, JHEP 09 (2013) 076,
arXiv: 1306.4945 [hep-ex].
[57] ATLAS Collaboration, Jet mass reconstruction with the ATLAS Detector in early Run 2 data,
ATLAS-CONF-2016-035, 2016, url: https://cds.cern.ch/record/2200211.
[58] D. Krohn, J. Thaler and L.-T. Wang, Jets with Variable R, JHEP 06 (2009) 059,
arXiv: 0903.0392 [hep-ph].
[59] ATLAS Collaboration, Variable Radius, Exclusive-kT , and Center-of-Mass Subjet Reconstruction
for Higgs(→ bb̄) Tagging in ATLAS, ATL-PHYS-PUB-2017-010, 2017,
url: https://cds.cern.ch/record/2268678.
[60] M. Cacciari, G. P. Salam and G. Soyez, The Catchment Area of Jets, JHEP 04 (2008) 005,
arXiv: 0802.1188 [hep-ph].
[61] ATLAS Collaboration,
√
Measurements of b-jet tagging efficiency with the ATLAS detector using t t¯ events at s = 13 TeV,
JHEP 08 (2018) 089, arXiv: 1805.01845 [hep-ex].
[62] ATLAS Collaboration, Muon reconstruction performance of the ATLAS detector in proton–proton
√
collision data at s = 13 TeV, Eur. Phys. J. C 76 (2016) 292, arXiv: 1603.05598 [hep-ex].
[63] ATLAS Collaboration,
Selection of jets produced in 13 TeV proton–proton collisions with the ATLAS detector,
ATLAS-CONF-2015-029, 2015, url: https://cds.cern.ch/record/2037702.

18
[64] S. S. Wilks,
The Large-Sample Distribution of the Likelihood Ratio for Testing Composite Hypotheses,
Annals Math. Statist. 9 (1938) 60.
[65] G. Snecdecor and W. Cochran, Statistical Methods, Statistical Methods v. 276, Wiley, 1991,
isbn: 9780813815619.
[66] I. M. Gessel, A factorization for formal laurent series and lattice path enumeration,
Journal of Combinatorial Theory, Series A 28 (1980) 321, issn: 0097-3165,
url: http://www.sciencedirect.com/science/article/pii/0097316580900746.
[67] ATLAS Collaboration, In situ calibration of large-R jet energy and mass in 13 TeV proton-proton
collisions with the ATLAS detector, Submitted to: Eur. Phys. J. (2018),
arXiv: 1807.09477 [hep-ex].
√
[68] ATLAS Collaboration, Improved luminosity determination in pp collisions at s = 7 TeV using the
ATLAS detector at the LHC, Eur. Phys. J. C 73 (2013) 2518, arXiv: 1302.4393 [hep-ex].
[69] J. M. Lindert et al., Precise predictions for V+ jets dark matter backgrounds,
Eur. Phys. J. C77 (2017) 829, arXiv: 1705.04664 [hep-ph].
[70] ATLAS Collaboration, A search for new physics in dijet mass and angular distributions in pp
√
collisions at s = 7 TeV measured with the ATLAS detector, New J. Phys. 13 (2011) 053044,
arXiv: 1103.3864 [hep-ex].
[71] T. Aaltonen et al., Global Search for New Physics with 2.0 fb−1 at CDF,
Phys. Rev. D79 (2009) 011101, arXiv: 0809.3781 [hep-ex].
[72] G. Choudalakis, On hypothesis testing, trials factor, hypertests and the BumpHunter, (2011),
arXiv: 1101.0390 [physics.data-an].

19