144
5 Direct Searches for New Physics
falling background, causing the function to be less constrained in the high mass
region. This leads to a spurious-signal uncertainty, which describes a bias observed
when testing the signal+background model against pseudo data generated under the
background-only hypothesis. Intrinsic limitations, associated with the fit range and
number of free parameters, create a systematic difference between the estimated
and real backgrounds, producing a spurious signal. The corresponding uncertainty is
approximately 30–40% up to 4 TeV, but increases rapidly to 200% at 5 TeV, where the
lack of data can not constrain the background model sufficiently well. Nevertheless,
the advanced top tagging algorithm in conjunction with the parametric background
model provide an improvement of 65% in the expected cross-section upper limit at
4 TeV compared to the previous ATLAS analysis [809], when using the same data
set.
Searches for resonant tt production in the +jets channel have been performed
using 7 and 8 TeV data by ATLAS [527, 812, 813] and CMS [805, 806, 814]. While
ATLAS uses t tagging already in the first analysis in this channel [813], CMS has
adopted a cascading selection, where small-R jets are assigned to the leptonic or
hadronic top quark decay [805, 814]. In a given event, all permutations of jets are
tested. The final assignment is chosen by the permutation minimising a χ
2 -metric,
consisting of squares of differences between the reconstructed and expected top quark
masses. No assumption on the number of small-R jets assigned to the reconstructed
top quark hypotheses is made, such that a smooth transition between resolved and
boosted final states is obtained. In order to probe resonance masses of 500 GeV and
below, both ATLAS and CMS have dedicated analyses in resolved final states [527,
805, 812, 814]. In addition, dedicated analyses have been performed in search for
scalar resonances by ATLAS [815] and CMS [816], taking into account interference
with SM tt production. ATLAS [527] and CMS [806] have explored the use of
a cascading selection for boosted and resolved final states. The boosted topology
is defined by a t-tagged jet opposite of the selected lepton. At high resonances
masses, the lepton is in close proximity to the b jet from the t decay, and is identified
with mini-isolation in ATLAS and a two-dimensional isolation requirement in CMS
(see Sect. 4.2.3). If an event fails the boosted selection, it is reconstructed using
a resolved selection based on the χ
2 -metric from above. This approach has been
shown to improve the sensitivity of the searches by about 30% with respect to the
χ
2 assignment of small-R jets without t tagging.
Analyses at 13 TeV in the +jets channel have been performed by CMS using
data corresponding to 2.6 fb
−1 [807] and 35.9 fb
−1 [808], and by ATLAS with
36.1 fb
−1 [817]. In these analyses, the same t tagging algorithms have been employed
as in the corresponding all-hadronic channels, to allow for a combination of the
results. An important difference is the omission of subjet b tagging. In the +jets
channel, the largest irreducible background is W +jets production, which can be
sufficiently suppressed by substructure and kinematic requirements, as well as b tagging on the small-R jet from the leptonic top quark decay chain. This leaves tt as the
dominant background, which can not be suppressed by a requirement on a b-tagged
subjet. Since such a selection would reduce the signal and background efficiency
by the same amount, it would lead to a lower overall sensitivity. Because of the
5 Direct Searches for New Physics
falling background, causing the function to be less constrained in the high mass
region. This leads to a spurious-signal uncertainty, which describes a bias observed
when testing the signal+background model against pseudo data generated under the
background-only hypothesis. Intrinsic limitations, associated with the fit range and
number of free parameters, create a systematic difference between the estimated
and real backgrounds, producing a spurious signal. The corresponding uncertainty is
approximately 30–40% up to 4 TeV, but increases rapidly to 200% at 5 TeV, where the
lack of data can not constrain the background model sufficiently well. Nevertheless,
the advanced top tagging algorithm in conjunction with the parametric background
model provide an improvement of 65% in the expected cross-section upper limit at
4 TeV compared to the previous ATLAS analysis [809], when using the same data
set.
Searches for resonant tt production in the +jets channel have been performed
using 7 and 8 TeV data by ATLAS [527, 812, 813] and CMS [805, 806, 814]. While
ATLAS uses t tagging already in the first analysis in this channel [813], CMS has
adopted a cascading selection, where small-R jets are assigned to the leptonic or
hadronic top quark decay [805, 814]. In a given event, all permutations of jets are
tested. The final assignment is chosen by the permutation minimising a χ
2 -metric,
consisting of squares of differences between the reconstructed and expected top quark
masses. No assumption on the number of small-R jets assigned to the reconstructed
top quark hypotheses is made, such that a smooth transition between resolved and
boosted final states is obtained. In order to probe resonance masses of 500 GeV and
below, both ATLAS and CMS have dedicated analyses in resolved final states [527,
805, 812, 814]. In addition, dedicated analyses have been performed in search for
scalar resonances by ATLAS [815] and CMS [816], taking into account interference
with SM tt production. ATLAS [527] and CMS [806] have explored the use of
a cascading selection for boosted and resolved final states. The boosted topology
is defined by a t-tagged jet opposite of the selected lepton. At high resonances
masses, the lepton is in close proximity to the b jet from the t decay, and is identified
with mini-isolation in ATLAS and a two-dimensional isolation requirement in CMS
(see Sect. 4.2.3). If an event fails the boosted selection, it is reconstructed using
a resolved selection based on the χ
2 -metric from above. This approach has been
shown to improve the sensitivity of the searches by about 30% with respect to the
χ
2 assignment of small-R jets without t tagging.
Analyses at 13 TeV in the +jets channel have been performed by CMS using
data corresponding to 2.6 fb
−1 [807] and 35.9 fb
−1 [808], and by ATLAS with
36.1 fb
−1 [817]. In these analyses, the same t tagging algorithms have been employed
as in the corresponding all-hadronic channels, to allow for a combination of the
results. An important difference is the omission of subjet b tagging. In the +jets
channel, the largest irreducible background is W +jets production, which can be
sufficiently suppressed by substructure and kinematic requirements, as well as b tagging on the small-R jet from the leptonic top quark decay chain. This leaves tt as the
dominant background, which can not be suppressed by a requirement on a b-tagged
subjet. Since such a selection would reduce the signal and background efficiency
by the same amount, it would lead to a lower overall sensitivity. Because of the