5.1 Diboson Resonances
131
total cross section comes from events where the H pair is produced at the kinematic
threshold, such that boosted final states do not play a major role in the search of
SM H H production. However, in several BSM scenarios the existence of massive
resonances decaying to H H may enhance the cross section at high H p T significantly and result in a peak in the di-Higgs mass distribution, m H H . Early searches
for heavy resonances decaying to H H have been performed by ATLAS [735–738]
and CMS [739–741] at 8 TeV and on a small subset of 13 TeV data.
The highest sensitivity for high resonance masses is achieved by all-hadronic
analyses targeting the H H → bbbb final state [742], with a branching fraction of
33.8%. In these analyses, QCD multijet production constitutes the largest background
with 80–95%, depending on the number of b-tagged jets or subjets, with the remainder
being tt events. The shape of the multijet background in the reconstructed m H H signal
region is obtained from signal-depleted sideband regions with less b-tagged jets.
Care has to be taken to correct for differences in jet kinematics, which are expected
because requiring b tags generally affects the jet kinematics. ATLAS has carried out
an analysis using 36.1 fb
−1 of 13 TeV data [743], considering resolved final states
with four b-tagged small-R jets, and merged final states with two H -tagged large-R
jets. The signal and sideband regions are defined by circles in the reconstructed H
masses. In the resolved category, the four small-R jets with the highest b-tagging
score are paired to construct two Higgs boson candidates. Requirements derived
from the kinematics of the H decay are used to select the best pairing. This leads to a
sculpting of the reconstructed H mass distributions, as shown in Fig. 5.3 (left). Even
though the H H signals feature a more pronounced peak (not shown), the similarity
between the signal and background distributions leads to a reduction in sensitivity.
In contrast, in the merged category the background shapes are not sculpted, as shown
in Fig. 5.3 (right), where trimming shifts the jet mass to lower values for light quark,
gluon and b jets. The structure at 170 GeV in leading m jet originates from merged
t jets in tt production, which are absent in the resolved category. In the merged
category, in addition to the requirements on the H masses, the signal regions are
defined by the number of b-tagged track jets, ghost-associated to the large-R jets.
The background shapes in m H H are obtained from data, selected with one less btagged jet than in the corresponding signal region. Kinematic corrections due to the b
tagging requirement are derived from distributions with loose m jet requirements. The
resulting m H H distributions are fitted with a smoothly falling power-law function to
model the multijet and tt backgrounds. The background normalisation is given by
the b tagging pass-fail ratio, which is calculated in the sideband region (the outer ring
in Fig. 5.3). The procedure is validated in the control region (the intermediate ring
in Fig. 5.3), and after finding good agreement between the data and the background
estimation, applied in the signal region.
In a similar analysis by CMS, only the fully merged final state with two H -tagged
large-R jets is considered [744]; the resolved final state is covered by another analysis [745]. An H jet is defined through the soft drop jet mass, 105 < m jet < 135 GeV,
τ 21 < 0.55, and a loose or tight selection on the double-b tagging discriminator. The
reduced dijet mass is used as final discriminant, where the soft-drop masses of the H
tagged jets are replaced by m H . This removes fluctuations in the jet masses, leading
131
total cross section comes from events where the H pair is produced at the kinematic
threshold, such that boosted final states do not play a major role in the search of
SM H H production. However, in several BSM scenarios the existence of massive
resonances decaying to H H may enhance the cross section at high H p T significantly and result in a peak in the di-Higgs mass distribution, m H H . Early searches
for heavy resonances decaying to H H have been performed by ATLAS [735–738]
and CMS [739–741] at 8 TeV and on a small subset of 13 TeV data.
The highest sensitivity for high resonance masses is achieved by all-hadronic
analyses targeting the H H → bbbb final state [742], with a branching fraction of
33.8%. In these analyses, QCD multijet production constitutes the largest background
with 80–95%, depending on the number of b-tagged jets or subjets, with the remainder
being tt events. The shape of the multijet background in the reconstructed m H H signal
region is obtained from signal-depleted sideband regions with less b-tagged jets.
Care has to be taken to correct for differences in jet kinematics, which are expected
because requiring b tags generally affects the jet kinematics. ATLAS has carried out
an analysis using 36.1 fb
−1 of 13 TeV data [743], considering resolved final states
with four b-tagged small-R jets, and merged final states with two H -tagged large-R
jets. The signal and sideband regions are defined by circles in the reconstructed H
masses. In the resolved category, the four small-R jets with the highest b-tagging
score are paired to construct two Higgs boson candidates. Requirements derived
from the kinematics of the H decay are used to select the best pairing. This leads to a
sculpting of the reconstructed H mass distributions, as shown in Fig. 5.3 (left). Even
though the H H signals feature a more pronounced peak (not shown), the similarity
between the signal and background distributions leads to a reduction in sensitivity.
In contrast, in the merged category the background shapes are not sculpted, as shown
in Fig. 5.3 (right), where trimming shifts the jet mass to lower values for light quark,
gluon and b jets. The structure at 170 GeV in leading m jet originates from merged
t jets in tt production, which are absent in the resolved category. In the merged
category, in addition to the requirements on the H masses, the signal regions are
defined by the number of b-tagged track jets, ghost-associated to the large-R jets.
The background shapes in m H H are obtained from data, selected with one less btagged jet than in the corresponding signal region. Kinematic corrections due to the b
tagging requirement are derived from distributions with loose m jet requirements. The
resulting m H H distributions are fitted with a smoothly falling power-law function to
model the multijet and tt backgrounds. The background normalisation is given by
the b tagging pass-fail ratio, which is calculated in the sideband region (the outer ring
in Fig. 5.3). The procedure is validated in the control region (the intermediate ring
in Fig. 5.3), and after finding good agreement between the data and the background
estimation, applied in the signal region.
In a similar analysis by CMS, only the fully merged final state with two H -tagged
large-R jets is considered [744]; the resolved final state is covered by another analysis [745]. An H jet is defined through the soft drop jet mass, 105 < m jet < 135 GeV,
τ 21 < 0.55, and a loose or tight selection on the double-b tagging discriminator. The
reduced dijet mass is used as final discriminant, where the soft-drop masses of the H
tagged jets are replaced by m H . This removes fluctuations in the jet masses, leading