4.2 Measurements Using Jet Substructure
109
can be largely suppressed, leaving W +jet and tt production as the main backgrounds.
This allows for the possibility to constrain these backgrounds sufficiently for a measurement of the small W W and W Z cross sections. The larger branching fraction in
relation to fully leptonic final states allows for a higher reach in p T . A W W +W Z
cross section measurement by ATLAS at 8 TeV uses trimmed anti-k T R = 1.0 jets
to reconstruct the ν J final state, where J stands for the single, large-R jet [605].
The jet mass distribution is used to fit the signal and background shapes to the data,
as shown in Fig. 4.8 (right). Thanks to the trimming algorithm, the light quark and
gluon jet background from W +jet production is observed to fall smoothly in m jet . No
selection on the substructure of the large-R jets is applied, such that the m jet distribution is not sculpted. The W +jet normalisation is constrained to an uncertainty of 4%
by the fit, but uncertainties in its modelling still result in the largest uncertainty in the
measured cross section. The second largest uncertainty originates from the modelling
of top quark production, which shows a resonant structure in m jet . It is controlled
by a dedicated sideband region. The cross section for W W +W Z production with
p T > 200 GeV is measured in the fiducial region to be 30 ± 11 (stat) ± 22 (syst) fb,
in agreement with predictions at NLO precision. A similar measurement by CMS at
8 TeV [606] uses pruned CA R = 0.8 jets. In contrast to ATLAS, a selection on τ 21
is applied, which suppresses the W +jet background by more than a factor of two.
This comes at the cost of sculpting the distribution in m jet , with large uncertainties
in modelling the W +jet background. Overall, the sensitivity is very similar to the
one by ATLAS, with an expected significance of about three standard deviations. A
recent CMS measurement of W W +W Z production at 13 TeV in the ν J final state
uses soft-drop groomed anti-k T R = 0.8 jets [607]. The jet substructure observables
are corrected with the PUPPI algorithm for pileup mitigation. A similar selection
on τ 21 as in the 8 TeV analysis is performed. No attempt is made to model the W +jet
background. Instead, it is derived from data in a control region using an extrapolation function (the α ratio method) [490, 492], with the benefit of reduced systematic
uncertainties. While the W W +W Z cross section is not explicitly measured in this
analysis, the most stringent limits to date on anomalous triple gauge couplings are
derived.
4.2.2 Higgs Boson Production
Before the start of the LHC, the associated H boson production channels VH with
the subsequent decay H → bb have been considered to be poor search channels due
to large backgrounds. The first study on high- p T H production, introducing the mass
drop tagger and filtering, has given hope to recover this channel [40]. However, the
experimental challenges have turned out to be far more difficult than anticipated,
and the expected significance of five standard deviations with data corresponding
to 30 fb
−1 has not been reached. Some of the experimental difficulties with respect
to the assumptions made in Ref. [40] are higher background levels, uncertainties
in the modelling of tt, W +HF and Z +HF, worse jet mass resolutions resulting in
109
can be largely suppressed, leaving W +jet and tt production as the main backgrounds.
This allows for the possibility to constrain these backgrounds sufficiently for a measurement of the small W W and W Z cross sections. The larger branching fraction in
relation to fully leptonic final states allows for a higher reach in p T . A W W +W Z
cross section measurement by ATLAS at 8 TeV uses trimmed anti-k T R = 1.0 jets
to reconstruct the ν J final state, where J stands for the single, large-R jet [605].
The jet mass distribution is used to fit the signal and background shapes to the data,
as shown in Fig. 4.8 (right). Thanks to the trimming algorithm, the light quark and
gluon jet background from W +jet production is observed to fall smoothly in m jet . No
selection on the substructure of the large-R jets is applied, such that the m jet distribution is not sculpted. The W +jet normalisation is constrained to an uncertainty of 4%
by the fit, but uncertainties in its modelling still result in the largest uncertainty in the
measured cross section. The second largest uncertainty originates from the modelling
of top quark production, which shows a resonant structure in m jet . It is controlled
by a dedicated sideband region. The cross section for W W +W Z production with
p T > 200 GeV is measured in the fiducial region to be 30 ± 11 (stat) ± 22 (syst) fb,
in agreement with predictions at NLO precision. A similar measurement by CMS at
8 TeV [606] uses pruned CA R = 0.8 jets. In contrast to ATLAS, a selection on τ 21
is applied, which suppresses the W +jet background by more than a factor of two.
This comes at the cost of sculpting the distribution in m jet , with large uncertainties
in modelling the W +jet background. Overall, the sensitivity is very similar to the
one by ATLAS, with an expected significance of about three standard deviations. A
recent CMS measurement of W W +W Z production at 13 TeV in the ν J final state
uses soft-drop groomed anti-k T R = 0.8 jets [607]. The jet substructure observables
are corrected with the PUPPI algorithm for pileup mitigation. A similar selection
on τ 21 as in the 8 TeV analysis is performed. No attempt is made to model the W +jet
background. Instead, it is derived from data in a control region using an extrapolation function (the α ratio method) [490, 492], with the benefit of reduced systematic
uncertainties. While the W W +W Z cross section is not explicitly measured in this
analysis, the most stringent limits to date on anomalous triple gauge couplings are
derived.
4.2.2 Higgs Boson Production
Before the start of the LHC, the associated H boson production channels VH with
the subsequent decay H → bb have been considered to be poor search channels due
to large backgrounds. The first study on high- p T H production, introducing the mass
drop tagger and filtering, has given hope to recover this channel [40]. However, the
experimental challenges have turned out to be far more difficult than anticipated,
and the expected significance of five standard deviations with data corresponding
to 30 fb
−1 has not been reached. Some of the experimental difficulties with respect
to the assumptions made in Ref. [40] are higher background levels, uncertainties
in the modelling of tt, W +HF and Z +HF, worse jet mass resolutions resulting in