254 The ATLAS Collaboration / Physics Letters B 763 (2016) 251–268
Fig. 4. The E
miss
T
distribution of the events in the signal region after the profile-
likelihood
fit to the data under the background-only hypothesis. The inset shows
the ratio of the data to the total background. Also shown is the E
miss
T
distribution
for the simplified model with a vector-boson mediator, scaled by a factor of 10
4
for
m
χ
=10 GeV and m
med
=10 TeV. The total background before the fit is shown as a
dashed line. The hatched bands represent the total uncertainty in the background.
(For interpretation of the references to color in this figure legend, the reader is re-
ferred
to the web version of this article.)
Samples of simulated W χ
¯
χ and Z χ
¯
χ events are generated us-
ing MadGraph5_aMC@NLO [55],
and the underlying event and par-
ton
showering are simulated with Pythia8.186 [53]. Two theoret-
ical
models are used as benchmarks: a seven-dimensional VVχχ
EFT [19] model (V meaning W or Z ) and a vector-mediated sim-
plified
model [56]. The strength of the EFT interaction is controlled
by a mass scale, M
, and the strength of the simplified model
interaction is controlled by the product of the couplings of the
mediator to the SM and the dark matter (DM) particles, g
SM
g
DM
.
The EFT model samples were generated with M
= 3000 GeV,
and the simplified model samples were generated with couplings
g
SM
= 0.25 and g
DM
= 1. The samples were generated as a func-
tion
of dark-matter particle mass m
χ
for the EFT model and in a
grid of mediator mass m
med
and m
χ
for the simplified model.
Major
sources of systematic uncertainty are uncertainties in the
modelling of large-R jet observables, which have a 5–13% impact
on the expected background and signal yields, and the energy
scale of the narrow jets, which contribute a 1–5% uncertainty to
the expected yields. Other sources of uncertainty include theoret-
ical
uncertainties in the simulated event samples used to model
the background processes (1–10%), parton distribution functions
(10–15%), and lepton reconstruction and identification efficiencies
(up to 2%).
A
profile-likelihood fit [57] to the E
miss
T
(E
miss
T
,noμ
) distribution in
the signal region (control regions) is used to constrain the W bo-
son,
Z boson, and t
¯
t backgrounds and extract the signal strength,
μ, for each model as an overall normalization factor for the signal
prediction. Besides the signal strength, three overall normalization
factors for the W boson, Z boson, and t
¯
t backgrounds are pa-
rameters
in the fit. The diboson and single-top backgrounds are
estimated from simulation, and the multijet background is negli-
gible.
The likelihood function is defined as the product of Poisson
distributions over all bins in E
miss
T
and E
miss
T
,noμ
, and the likelihood
is simultaneously maximized over the signal and control regions.
Variations
of the expected signal and background to allow for
their systematic uncertainties are described with nuisance parame-
ters
constrained by Gaussian probability distribution functions, and
correlations across signal and background processes and regions
are taken into account.
Table 1
Predicted
and observed number of events in the signal re-
gion.
The yields and uncertainties of the backgrounds are
shown after the profile-likelihood fit to the data under the
background-only hypothesis. For comparison, the expected
yield in the VVχχ EFT model with M
= 600 GeV and
m
χ
=500 GeV is 10.1 ±0.4events.
Process Events
Z + jets 544 ±33
W
+ jets 275 ±24
t
¯
t and single-top 211 ±19
Diboson 89
±12
Total background 1120
±47
Data 1121
Table 2
Background
normalization factors relative to the initial
theoretical prediction, extracted from the profile-likelihood
fit under the background-only hypothesis.
Process Normalization factor
Z + jets 1.01 ±0.16
W
+ jets 0.90 ±0.16
t
¯
t 0.91 ±0.18
A background-only (μ = 0) fit, shows no deviation from SM
predictions, and Figs. 3 and 4 show kinematic distributions af-
ter
the profile-likelihood fit. The floating background-normalization
parameters are consistent with unity within one standard devia-
tion.
Tables 1 and 2 show the expected event yields after applying
the signal selection and the background normalization scale fac-
tors,
respectively. The values in these tables are estimated for the
background-only hypothesis.
Upper
limits at 95% confidence level (C.L.) on μ are calculated
using the CL
s
method [58]. For the VVχχ EFT model, these lim-
its
are translated into constraints on the mass scale, M
. Fig. 5(a)
shows the limit on the mass scale, M
, in the EFT model, as a func-
tion
of m
χ
. Fig. 5(b) shows the limits on the signal strength, μ,
for a vector-mediated simplified model generated with couplings
g
SM
=0.25 and g
DM
=1in the plane of m
χ
and m
med
.
In
conclusion, this Letter reports ATLAS limits on dark-matter
production in events with a hadronically decaying W or Z bo-
son
and large missing transverse momentum. These limits from
3.2 fb
−1
of 13 TeV pp collisions at the LHC improve on earlier AT-
LAS
results. No statistically significant excess is observed over the
Standard Model prediction.
Acknowledgements
We thank CERN for the very successful operation of the LHC,
as well as the support staff from our institutions without whom
ATLAS could not be operated efficiently.
We
acknowledge the support of ANPCyT, Argentina; YerPhI, Ar-
menia;
ARC, Australia; BMWFW and FWF, Austria; ANAS, Azerbai-
jan;
SSTC, Belarus; CNPq and FAPESP, Brazil; NSERC, NRC and CFI,
Canada; CERN; CONICYT, Chile; CAS, MOST and NSFC, China; COL-
CIENCIAS,
Colombia; MSMT CR, MPO CR and VSC CR, Czech Re-
public;
DNRF and DNSRC, Denmark; IN2P3-CNRS, CEA-DSM/IRFU,
France; GNSF, Georgia; BMBF, HGF, and MPG, Germany; GSRT,
Greece; RGC, Hong Kong SAR, China; ISF, I-CORE and Benoziyo
Center, Israel; INFN, Italy; MEXT and JSPS, Japan; CNRST, Mo-
rocco;
FOM and NWO, Netherlands; RCN, Norway; MNiSW and
NCN, Poland; FCT, Portugal; MNE/IFA, Romania; MES of Russia and
NRC KI, Russian Federation; JINR; MESTD, Serbia; MSSR, Slovakia;
ARRS and MIZŠ, Slovenia; DST/NRF, South Africa; MINECO, Spain;
SRC and Wallenberg Foundation, Sweden; SERI, SNSF and Cantons