126
5 Direct Searches for New Physics
In searches in the ν J final state, the presence of an isolated lepton results in the
QCD multijet background being largely suppressed, such that the most important
backgrounds are tt and W +jets production. This leads to less stringent requirements
on V tagging selections relative to all-hadronic analyses. It also offers a number of
possibilities to estimate the SM backgrounds. In an early analysis of 13 TeV data in
the ν J final state, CMS uses the pruned jet mass to define a signal-depleted sideband
region with 40 < m jet < 65 GeV [492]. The main background from W +jets production is obtained from this sideband region, by scaling the reconstructed diboson mass
distribution m V V in data with a transfer function α(m V V ). The function α(m V V ) is
constructed from the ratio of probabilities to observe an event at a given value of m V V
in the signal region to the one in the control region. This ratio is obtained from the
simulation of W +jets production and is largely insensitive to modelling uncertainties.
However, when scaling the data in the sideband region, the expected background from
tt production has to be subtracted, which introduces additional uncertainties. Also,
the shapes of the probability density functions in the sideband and signal regions are
affected by modelling uncertainties. In an analysis of 36.1 fb
−1 of data, ATLAS considers weak vector-boson fusion (VBF) in addition to the usual gluon-gluon fusion
(ggF) and qq production modes [495]. The VBF process pp → V V j j is characterised by the presence of two jets with large separation in rapidity, originating from
the initial state quarks from which a vector boson is radiated. This leads to a total of
twelve signal regions, categorised by VBF or ggF/qq production, HP or LP V tags,
W W or W Z signals, and merged or resolved final states. The background estimation
uses the simulated shapes of the twelve m V V distributions, which are dominated by
W +jets and tt production. Given an accurate modelling of these backgrounds, this
has the advantage of constraining modelling uncertainties in a simultaneous fit of all
backgrounds and a possible signal contribution to the data. In order to improve the
precision of this procedure, W +jets and tt control regions are defined by large-R jets
failing the V tagging requirements and the presence of additional b-tagged small-R
jets. Similar to the all-hadronic final state, it is possible to make use of the full information contained in the m jet and m V V distributions to improve the precision of the
background estimation. A CMS analysis in the ν J final state, based on 35.9 fb
−1 of
data, has used conditional probabilities of m V V as a function of m jet to model the SM
backgrounds [710]. The signal is modelled by a conditional probability of the signal
mass as a function of m V V and m jet . The two-dimensional fit to data in the m V V -m jet
plane results in an extended reach in signal mass and better sensitivity compared to
the α(m V V ) method.
A diboson search in the νν J final state has been carried out by CMS, using
35.9 fb
−1 of data [711]. The requirement of p
miss
T
> 200 GeV results in a stable trigger efficiency of about 96% and a negligible amount of background events from QCD
multijet production. Similar to the +jets analysis, HP and LP V -tagging categories
are used to achieve an optimal signal-to-background ratio for low and high resonance masses. The main difference to analyses in leptonic final states is the unknown
momentum in z direction of the Z → νν decay. Hence, the transverse mass m
Z V
T is
used instead of the diboson mass. The main background in this search are events from
Z +jet production with Z → νν, and W +jet production with W → ν, where the
5 Direct Searches for New Physics
In searches in the ν J final state, the presence of an isolated lepton results in the
QCD multijet background being largely suppressed, such that the most important
backgrounds are tt and W +jets production. This leads to less stringent requirements
on V tagging selections relative to all-hadronic analyses. It also offers a number of
possibilities to estimate the SM backgrounds. In an early analysis of 13 TeV data in
the ν J final state, CMS uses the pruned jet mass to define a signal-depleted sideband
region with 40 < m jet < 65 GeV [492]. The main background from W +jets production is obtained from this sideband region, by scaling the reconstructed diboson mass
distribution m V V in data with a transfer function α(m V V ). The function α(m V V ) is
constructed from the ratio of probabilities to observe an event at a given value of m V V
in the signal region to the one in the control region. This ratio is obtained from the
simulation of W +jets production and is largely insensitive to modelling uncertainties.
However, when scaling the data in the sideband region, the expected background from
tt production has to be subtracted, which introduces additional uncertainties. Also,
the shapes of the probability density functions in the sideband and signal regions are
affected by modelling uncertainties. In an analysis of 36.1 fb
−1 of data, ATLAS considers weak vector-boson fusion (VBF) in addition to the usual gluon-gluon fusion
(ggF) and qq production modes [495]. The VBF process pp → V V j j is characterised by the presence of two jets with large separation in rapidity, originating from
the initial state quarks from which a vector boson is radiated. This leads to a total of
twelve signal regions, categorised by VBF or ggF/qq production, HP or LP V tags,
W W or W Z signals, and merged or resolved final states. The background estimation
uses the simulated shapes of the twelve m V V distributions, which are dominated by
W +jets and tt production. Given an accurate modelling of these backgrounds, this
has the advantage of constraining modelling uncertainties in a simultaneous fit of all
backgrounds and a possible signal contribution to the data. In order to improve the
precision of this procedure, W +jets and tt control regions are defined by large-R jets
failing the V tagging requirements and the presence of additional b-tagged small-R
jets. Similar to the all-hadronic final state, it is possible to make use of the full information contained in the m jet and m V V distributions to improve the precision of the
background estimation. A CMS analysis in the ν J final state, based on 35.9 fb
−1 of
data, has used conditional probabilities of m V V as a function of m jet to model the SM
backgrounds [710]. The signal is modelled by a conditional probability of the signal
mass as a function of m V V and m jet . The two-dimensional fit to data in the m V V -m jet
plane results in an extended reach in signal mass and better sensitivity compared to
the α(m V V ) method.
A diboson search in the νν J final state has been carried out by CMS, using
35.9 fb
−1 of data [711]. The requirement of p
miss
T
> 200 GeV results in a stable trigger efficiency of about 96% and a negligible amount of background events from QCD
multijet production. Similar to the +jets analysis, HP and LP V -tagging categories
are used to achieve an optimal signal-to-background ratio for low and high resonance masses. The main difference to analyses in leptonic final states is the unknown
momentum in z direction of the Z → νν decay. Hence, the transverse mass m
Z V
T is
used instead of the diboson mass. The main background in this search are events from
Z +jet production with Z → νν, and W +jet production with W → ν, where the