140
5 Direct Searches for New Physics
5.2.1 t t Resonances
Searches for resonances with masses larger than 1 TeV decaying to tt have been the
first applications of top tagging algorithms at the LHC. Interest in models predicting
resonant tt production has been fuelled by measurements of the forward-backward
asymmetry A FB in p p → tt at the Tevatron (see Sect. 4.2.3). As a consequence,
the first searches at the LHC have already been performed with 7 TeV data recorded
in 2011 [803, 804]. Even though no excess has been found in these early analyses,
searches for resonant tt production have been carried out on each dataset recorded
by ATLAS and CMS. These searches can give testimony of the improvements in
terms of sensitivity due to developments in the field of jet substructure.
The tt system can be classified into all-hadronic, +jets and di-leptonic final states.
In the searches described here, decays of the type W → τ ν are included through
leptonic τ decays, and are part of the +jets final state. Hadronic τ decays are not
considered. The three distinct final states have very different reducible backgrounds,
while the irreducible background from tt production is common to all. In addition to
tt production, also single top quark production in the t W channel in association with
an additional b quark contributes to the irreducible background. In the all-hadronic
channel, with a branching fraction of 45.4%, multijet production constitutes the
largest reducible background. This background originates mostly from light quark
and gluon jets produced via the strong force, but also includes hadronic decays of W
and Z bosons produced in association with jets. Compared to the multijet background,
other backgrounds from processes not involving top quarks are negligible. In the
+jets final state with a branching fraction of 34.1%, the largest reducible background
originates from W +jets production with the decay W → ν. The dilepton channel
has the smallest branching fraction of 6.4%. The dominant reducible background in
this channel is Z +jets production with Z → . Due to the small branching fraction,
this channel contributes mostly at masses below 1 TeV. Above this value, searches
in the all-hadronic and +jets channels achieve higher sensitivity.
Analyses in the all-hadronic channel rely heavily on top tagging, as this is the only
possibility to reduce the large multijet background. At 7 TeV, CMS has used the
CMSTT to identify boosted t jets [803]. In addition to topologies with two t jets, also
events with one t jet and one W jet have been considered, where the W jet is paired
with a small-R jet in angular proximity. The multijet background is estimated using
the mistag probability, which is obtained in a sample with two high- p T jets without ttagging requirements. The fraction of events with one t-tagged jet defines the mistag
probability, which depends on p T . A sideband region is defined by requiring a dijet
topology with one t-tagged jet. Events in this sideband region are weighted by the
mistag probability as a function of p T of the non-tagged jet. Since these jets have on
average a smaller jet mass than t-tagged jets in the signal region, their jet mass is
set to a random number drawn from the simulated jet mass distribution in the signal
region. The obtained distribution in m tt constitutes the background estimation for
the signal region with two t-tagged jets. A similar procedure is applied for semimerged events. In the signal region, the multijet background is about an order of
5 Direct Searches for New Physics
5.2.1 t t Resonances
Searches for resonances with masses larger than 1 TeV decaying to tt have been the
first applications of top tagging algorithms at the LHC. Interest in models predicting
resonant tt production has been fuelled by measurements of the forward-backward
asymmetry A FB in p p → tt at the Tevatron (see Sect. 4.2.3). As a consequence,
the first searches at the LHC have already been performed with 7 TeV data recorded
in 2011 [803, 804]. Even though no excess has been found in these early analyses,
searches for resonant tt production have been carried out on each dataset recorded
by ATLAS and CMS. These searches can give testimony of the improvements in
terms of sensitivity due to developments in the field of jet substructure.
The tt system can be classified into all-hadronic, +jets and di-leptonic final states.
In the searches described here, decays of the type W → τ ν are included through
leptonic τ decays, and are part of the +jets final state. Hadronic τ decays are not
considered. The three distinct final states have very different reducible backgrounds,
while the irreducible background from tt production is common to all. In addition to
tt production, also single top quark production in the t W channel in association with
an additional b quark contributes to the irreducible background. In the all-hadronic
channel, with a branching fraction of 45.4%, multijet production constitutes the
largest reducible background. This background originates mostly from light quark
and gluon jets produced via the strong force, but also includes hadronic decays of W
and Z bosons produced in association with jets. Compared to the multijet background,
other backgrounds from processes not involving top quarks are negligible. In the
+jets final state with a branching fraction of 34.1%, the largest reducible background
originates from W +jets production with the decay W → ν. The dilepton channel
has the smallest branching fraction of 6.4%. The dominant reducible background in
this channel is Z +jets production with Z → . Due to the small branching fraction,
this channel contributes mostly at masses below 1 TeV. Above this value, searches
in the all-hadronic and +jets channels achieve higher sensitivity.
Analyses in the all-hadronic channel rely heavily on top tagging, as this is the only
possibility to reduce the large multijet background. At 7 TeV, CMS has used the
CMSTT to identify boosted t jets [803]. In addition to topologies with two t jets, also
events with one t jet and one W jet have been considered, where the W jet is paired
with a small-R jet in angular proximity. The multijet background is estimated using
the mistag probability, which is obtained in a sample with two high- p T jets without ttagging requirements. The fraction of events with one t-tagged jet defines the mistag
probability, which depends on p T . A sideband region is defined by requiring a dijet
topology with one t-tagged jet. Events in this sideband region are weighted by the
mistag probability as a function of p T of the non-tagged jet. Since these jets have on
average a smaller jet mass than t-tagged jets in the signal region, their jet mass is
set to a random number drawn from the simulated jet mass distribution in the signal
region. The obtained distribution in m tt constitutes the background estimation for
the signal region with two t-tagged jets. A similar procedure is applied for semimerged events. In the signal region, the multijet background is about an order of