138
5 Direct Searches for New Physics
category is introduced for large-R jets failing all requirements. This last category
improves the sensitivity for masses larger than 4 TeV and extends the mass reach up
to 6.8 TeV. The analysis has comparable sensitivity for Z γ resonances to the CMS
search with the same amount of data [781]. Limits on spin-1 W γ resonances are
derived as well and are equivalent to the results for Z γ . The b-tagged category is
used to search for an H γ resonance for the first time. In this case, the jet mass
range is changed to 93 < m jet < 134 GeV and no selection on D 2 or n trk is applied
to achieve high signal efficiency. The signal efficiency for b-tagged jets in the H γ
search is 25% for masses of 1 TeV, decreasing to 7% at 3 TeV. The decrease in efficiency can be attributed to a b-tagging inefficiency for highly boosted H jets with
ghost-associated R = 0.2 track-jets (see Sect. 3.5.3). The obtained upper limits on
σ ( pp → Z
)B(Z
→ H γ ) vary between 10 and 4 fb for resonance masses between
1 and 3 TeV. In a dedicated search for H γ resonances by CMS [784], the double-b
tagging algorithm is used, which is designed for boosted H tagging. Together with
a selection on the PUPPI-corrected soft drop jet mass of 110 < m jet < 140 GeV,
a signal efficiency is achieved of 36% at a mass of 1 TeV and 20% at 3 TeV. The
sensitivity of the search for masses above 2 TeV is further improved by the addition
of an untagged category. The gain in signal efficiency, together with the improved
suppression of SM backgrounds by the double-b tagger for low masses, compared
to the ATLAS analysis [783] results in a significant gain in sensitivity by factors
between 2 and 10. The analysis reports upper limits of 25 fb for a mass of 720 GeV,
5.3 fb for 1 TeV and 0.4 fb for 3.25 TeV.
The latest result in search for an H γ resonance is reported by ATLAS using
139 fb
−1 [785]. The analysis has been optimised for this channel, making use of
the centre-of-mass subjet reconstruction for subjet b tagging for the first time. Two
categories are defined for this search, based on single and double b-tagged large-R
jets. The combined signal efficiency is about 19% for a mass of 1 TeV, increasing to
30% at 3 TeV. For masses below 2.7 TeV double-b tagged jets have higher efficiency,
for masses larger than that single-b tagged jets have higher efficiency. At 4 TeV the
efficiency of double-b tagged jets is still 10%. The new subjet b tagging approach
results in an excellent signal-to-background ratio even at the highest masses. This
constitutes a significant improvement relative to the first ATLAS analysis in this
channel. The reconstructed H γ distributions are shown in Fig. 5.6 for the single
and double b-tagged categories. The background in each category is modelled by a
function with three free shape parameters, validated in control regions without subjet
b and jet mass requirements. The data are well described by the background-only
fit, and the two categories are combined to set upper limits on σ ( pp → Z
)B(Z
→
H γ ). The analysis achieves the best limits to date, of 10 fb for a mass of 720 GeV, 3 fb
for 1 TeV and 0.24 fb for 3.25 TeV. It is an excellent example on how improvements
in jet substructure tagging can improve the sensitivity of searches at the LHC.
5 Direct Searches for New Physics
category is introduced for large-R jets failing all requirements. This last category
improves the sensitivity for masses larger than 4 TeV and extends the mass reach up
to 6.8 TeV. The analysis has comparable sensitivity for Z γ resonances to the CMS
search with the same amount of data [781]. Limits on spin-1 W γ resonances are
derived as well and are equivalent to the results for Z γ . The b-tagged category is
used to search for an H γ resonance for the first time. In this case, the jet mass
range is changed to 93 < m jet < 134 GeV and no selection on D 2 or n trk is applied
to achieve high signal efficiency. The signal efficiency for b-tagged jets in the H γ
search is 25% for masses of 1 TeV, decreasing to 7% at 3 TeV. The decrease in efficiency can be attributed to a b-tagging inefficiency for highly boosted H jets with
ghost-associated R = 0.2 track-jets (see Sect. 3.5.3). The obtained upper limits on
σ ( pp → Z
)B(Z
→ H γ ) vary between 10 and 4 fb for resonance masses between
1 and 3 TeV. In a dedicated search for H γ resonances by CMS [784], the double-b
tagging algorithm is used, which is designed for boosted H tagging. Together with
a selection on the PUPPI-corrected soft drop jet mass of 110 < m jet < 140 GeV,
a signal efficiency is achieved of 36% at a mass of 1 TeV and 20% at 3 TeV. The
sensitivity of the search for masses above 2 TeV is further improved by the addition
of an untagged category. The gain in signal efficiency, together with the improved
suppression of SM backgrounds by the double-b tagger for low masses, compared
to the ATLAS analysis [783] results in a significant gain in sensitivity by factors
between 2 and 10. The analysis reports upper limits of 25 fb for a mass of 720 GeV,
5.3 fb for 1 TeV and 0.4 fb for 3.25 TeV.
The latest result in search for an H γ resonance is reported by ATLAS using
139 fb
−1 [785]. The analysis has been optimised for this channel, making use of
the centre-of-mass subjet reconstruction for subjet b tagging for the first time. Two
categories are defined for this search, based on single and double b-tagged large-R
jets. The combined signal efficiency is about 19% for a mass of 1 TeV, increasing to
30% at 3 TeV. For masses below 2.7 TeV double-b tagged jets have higher efficiency,
for masses larger than that single-b tagged jets have higher efficiency. At 4 TeV the
efficiency of double-b tagged jets is still 10%. The new subjet b tagging approach
results in an excellent signal-to-background ratio even at the highest masses. This
constitutes a significant improvement relative to the first ATLAS analysis in this
channel. The reconstructed H γ distributions are shown in Fig. 5.6 for the single
and double b-tagged categories. The background in each category is modelled by a
function with three free shape parameters, validated in control regions without subjet
b and jet mass requirements. The data are well described by the background-only
fit, and the two categories are combined to set upper limits on σ ( pp → Z
)B(Z
→
H γ ). The analysis achieves the best limits to date, of 10 fb for a mass of 720 GeV, 3 fb
for 1 TeV and 0.24 fb for 3.25 TeV. It is an excellent example on how improvements
in jet substructure tagging can improve the sensitivity of searches at the LHC.