son. To discriminate against backgrounds, the jets are required to have small values of τ
21
,
and events are divided into categories of high-purity (HP) if τ
21
< 0.4, and low-purity (LP)
if 0.4 < τ
21
< 0.75. A jet is V tagged if it fulfills the soft-drop jet mass and τ
21
requirements.
The normalization scale factors 0.99 ±0.06 for the HP category and 0.96 ±0.11 for the LP
category [56] are applied to simulated events with genuine hadronic boson decays. Higgs
boson jet candidates are classified according to the number of subjets (1 or 2) that pass the
b tagging selection. Subjet b tagging is not used for jets compatible with W or Z candidates
and no N-subjettiness requirement is applied to the Higgs boson candidate jet. If neither
the N-subjettiness nor the b tagging requirements are satisfied, the event is discarded.
Events are divided into categories depending on the number of identified τ
h
candidates
(1 or 2), and on the classification of the large jet cone, i.e. either HP or LP in τ
21
, or either
1 or 2 b-tagged subjets.
Since the undetected neutrinos carry a significant fraction of the momentum in the
ττ system, signal events are expected to have a large ~p
miss
T
, thereby justifying the use of
triggers that require large p
miss
T
or H
miss
T
. A stringent offline requirement of greater than
200 GeV is applied to the reconstructed p
miss
T
, to ensure a stable trigger efficiency and to
suppress the background contribution from multijet events. Events with top quark pairs or
single top quarks are suppressed by removing events in which any AK4 jet not overlapping
with the AK8 jet is b-tagged.
Several selection requirements are applied to remove SM backgrounds, such as meson
and baryon resonances, Z+jets, W+jets and tt and single top quark production. The
angular distance ∆R
τ τ
should be smaller than 1.5, in order to reject W+jets events in
which a jet misidentified as a τ lepton is typically spatially well-separated from the genuine
lepton. To further increase the signal purity, the di-tau mass, as estimated from the SVfit
procedure [64–66], should be between 50 and 150 GeV. The SVfit algorithm, based on a
likelihood approach, estimates the di-τ system mass using the measured momenta of the
visible decay products of both τ leptons, the reconstructed ~p
miss
T
, and the ~p
miss
T
resolution.
Finally, the resonance candidate mass m
X
, defined as the invariant mass of the H → ττ
candidate and the hadronically decaying boson jet, is required to be larger than 750 GeV
in order to ensure full trigger and reconstruction efficiencies.
After these selection requirements, the selection efficiency of a radion signal in the 1
and 2 b tag categories is 1–6% in the τ
h
τ
h
channel, and 3–10% in the `τ
h
channel, for
low and high resonance masses, respectively. The efficiencies for a W
0
signal passing the
V-tagging selection are 2–9% and 8–19% in the τ
h
τ
h
and `τ
h
channels, respectively. Events
in which H → VV are found contribute up to an additional 20–30% of the total signal yield
in the `τ
h
channels, and less than 10% in the τ
h
τ
h
channels.
6 Background estimation
The main sources of background originate from top quark pair production and from the
production of a vector boson in association with jets (Z+jets and W+jets), while minor con-
tributions arise from single top quark, diboson, and multijet production. These background
– 7 –