80
3 Jet Substructure at the LHC
of p T -dependent optimisation steps, but further experimental studies are needed to
commission this strategy for use in analyses.
For some analyses the requirement of p T 200 GeV is too restrictive, and hadronically decaying V bosons with lower p T need to be selected. This poses a particular
challenge due to the abundance of light flavour jets at the LHC and their indistinguishability from jets from resolved W /Z decays. An attempt was made by CMS to
discriminate ‘resolved’ (non-merged) hadronic W decays from multijet background
using the q/g likelihood, the sum of the jet charges of the dijet pair and the jet pull
angle. Combining these variables into a boosted decision tree, a misidentification
rate of about 25% is achieved for a signal efficiency of 50% [411]. While this is a
first success, the performance is about an order of magnitude worse than V tagging
for fully merged decays, showing the power of substructure techniques in this field.
In addition to developing tools for distinguishing boosted hadronically decaying
W and Z bosons from generic quark and gluon jets, ATLAS has also built a tagger to
further classify a boson jet as either originating from a W boson or a Z boson [499].
While theoretically clean due to the colour singlet nature of the W and Z bosons,
this task is particularly challenging because the jet mass resolution is comparable to
the difference m Z − m W . In order to improve the sensitivity of the tagger, jet charge
and b tagging information are combined with the jet mass. The jet mass distribution
depends on the type of W or Z decay due to semi-leptonic B and D decays, so a
full likelihood tagger is constructed by summing over the conditional likelihoods for
each flavour type. To maximise the discrimination power from b tagging, multiple
efficiency working points are used simultaneously in the tagger. A W
+ rejection
near 8 (corresponding to a misidentification rate of 12.5%) is achieved at a Z boson
efficiency of 50%. At this moderate Z boson efficiency, all of the inputs offer useful
discrimination information. At low efficiencies, below the bb branching ratio for Z
bosons, b tagging dominates over the jet mass and jet charge.
3.5.3 Higgs Boson Tagging
The Higgs boson decays with highest probability to bb and W W
∗
→ 4 quarks with
branching fractions of 58.1 and 9.8%, respectively (see Sect. 2.2.4). Both are hadronic
decays, which are difficult to identify. In fact, it has taken six years after the discovery of the Higgs boson to measure the H → bb decay with a significance of more
five standard deviations [35, 36]. The decay H → W W
∗
→ 4 quarks is still undiscovered, while the leptonic H → W W
∗ decay channels have been important for the
discovery and classification of the Higgs boson properties [1, 2, 500–503].
The identification of H → bb relies on the distinct signature of jets originating
from the fragmentation of b quarks (b tagging). The fragmentation leads to the
presence of B hadrons; whose decay results in a secondary vertex within a jet due
to the long lifetime of about 1.5 ps. The b tagging is crucial in many analyses at the
LHC and has been developed for small-R jets from isolated b production. ATLAS
and CMS use multivariate techniques with various input parameters related to the
3 Jet Substructure at the LHC
of p T -dependent optimisation steps, but further experimental studies are needed to
commission this strategy for use in analyses.
For some analyses the requirement of p T 200 GeV is too restrictive, and hadronically decaying V bosons with lower p T need to be selected. This poses a particular
challenge due to the abundance of light flavour jets at the LHC and their indistinguishability from jets from resolved W /Z decays. An attempt was made by CMS to
discriminate ‘resolved’ (non-merged) hadronic W decays from multijet background
using the q/g likelihood, the sum of the jet charges of the dijet pair and the jet pull
angle. Combining these variables into a boosted decision tree, a misidentification
rate of about 25% is achieved for a signal efficiency of 50% [411]. While this is a
first success, the performance is about an order of magnitude worse than V tagging
for fully merged decays, showing the power of substructure techniques in this field.
In addition to developing tools for distinguishing boosted hadronically decaying
W and Z bosons from generic quark and gluon jets, ATLAS has also built a tagger to
further classify a boson jet as either originating from a W boson or a Z boson [499].
While theoretically clean due to the colour singlet nature of the W and Z bosons,
this task is particularly challenging because the jet mass resolution is comparable to
the difference m Z − m W . In order to improve the sensitivity of the tagger, jet charge
and b tagging information are combined with the jet mass. The jet mass distribution
depends on the type of W or Z decay due to semi-leptonic B and D decays, so a
full likelihood tagger is constructed by summing over the conditional likelihoods for
each flavour type. To maximise the discrimination power from b tagging, multiple
efficiency working points are used simultaneously in the tagger. A W
+ rejection
near 8 (corresponding to a misidentification rate of 12.5%) is achieved at a Z boson
efficiency of 50%. At this moderate Z boson efficiency, all of the inputs offer useful
discrimination information. At low efficiencies, below the bb branching ratio for Z
bosons, b tagging dominates over the jet mass and jet charge.
3.5.3 Higgs Boson Tagging
The Higgs boson decays with highest probability to bb and W W
∗
→ 4 quarks with
branching fractions of 58.1 and 9.8%, respectively (see Sect. 2.2.4). Both are hadronic
decays, which are difficult to identify. In fact, it has taken six years after the discovery of the Higgs boson to measure the H → bb decay with a significance of more
five standard deviations [35, 36]. The decay H → W W
∗
→ 4 quarks is still undiscovered, while the leptonic H → W W
∗ decay channels have been important for the
discovery and classification of the Higgs boson properties [1, 2, 500–503].
The identification of H → bb relies on the distinct signature of jets originating
from the fragmentation of b quarks (b tagging). The fragmentation leads to the
presence of B hadrons; whose decay results in a secondary vertex within a jet due
to the long lifetime of about 1.5 ps. The b tagging is crucial in many analyses at the
LHC and has been developed for small-R jets from isolated b production. ATLAS
and CMS use multivariate techniques with various input parameters related to the