212
6 Summary
and photon identification algorithms also profit from these developments, as their
performance is influenced by contributions from pileup. Virtually every analysis
performed by the ATLAS and CMS Collaborations uses some aspect of jet substructure methods, such that a rudimentary understanding of these is beneficial for
everyone working in this field.
Jet substructure studies in heavy ion collisions constitute an aspect that has not
been discussed in this book. In these dense environments, grooming methods allow
to access the partonic structure of jets, while mitigating the effects of the underlying
event and hadronisation. The partonic energy loss and p T -broadening of jets induced
by gluon radiation from traversing a large nucleus can be enhanced significantly in hot
matter, as produced in heavy-ion collisions [1286]. These can be accessed through
measurements of the groomed momentum sharing of the two-prong jet substructure in pp and heavy ion collisions [1287]. Insights into the interactions of quarks
and gluons with the hot medium can also be obtained from groomed [1288] and
ungroomed [1289] jet mass measurements. Future measurements can benefit from
jet substructure algorithms to separate the partonic energy loss and gluon-induced
radiation from non-perturbative effects in the scattering with the medium [1290].
Jet grooming may also help to develop a factorisation theorem in SCET, allowing
to resum the large logarithms arising from final state measurements when summing
over multiple interactions of the jet with the medium [1291].
In the future, jet substructure methods will continue to gain in importance. This is
in part due to the upgrade to the HL-LHC and the resulting larger data sample, together
with an increase in centre-of-mass energy to 14 TeV, allowing for more precise studies of high p T processes. Even the high-energy tails of rare electroweak processes
will become accessible, and jet substructure taggers will play an important role in
their measurements. The drawback of the high instantaneous luminosities aimed at
with the HL-LHC are higher levels of pileup. These will require advanced mitigation
strategies, where jet substructure methods will be indispensable. The upgrades of the
ATLAS and CMS detectors will enable the use of particle flow and pileup mitigation algorithms already at the first trigger level, allowing for online jet substructure
selections at a rate of 40 MHz. Upgrades of the inner tracking detectors will extend
the tracking coverage to pseudorapidities of about 4, and higher granularities of the
calorimeters will ensure an excellent angular resolution in the reconstruction of particles. Overall, the detector capabilities for jet substructure analyses will be enhanced
with respect to the present detectors. Another reason for the future importance of
jet substructure algorithms are studies using advanced machine learning techniques.
These can be used to design powerful jet substructure taggers, but they can also help
to identify variables carrying additional information, which have not been considered so far. For example, machine learning can help to construct a single observable,
which is the product of a number of jet substructure observables, carrying most of the
discrimination power of a complex artificial neural network [1292, 1293]. Similar
techniques can also help to identify jet substructure observables for the identification
of electrons inside jets [1294] or additional variables for jet tagging [1295]. Classi-
6 Summary
and photon identification algorithms also profit from these developments, as their
performance is influenced by contributions from pileup. Virtually every analysis
performed by the ATLAS and CMS Collaborations uses some aspect of jet substructure methods, such that a rudimentary understanding of these is beneficial for
everyone working in this field.
Jet substructure studies in heavy ion collisions constitute an aspect that has not
been discussed in this book. In these dense environments, grooming methods allow
to access the partonic structure of jets, while mitigating the effects of the underlying
event and hadronisation. The partonic energy loss and p T -broadening of jets induced
by gluon radiation from traversing a large nucleus can be enhanced significantly in hot
matter, as produced in heavy-ion collisions [1286]. These can be accessed through
measurements of the groomed momentum sharing of the two-prong jet substructure in pp and heavy ion collisions [1287]. Insights into the interactions of quarks
and gluons with the hot medium can also be obtained from groomed [1288] and
ungroomed [1289] jet mass measurements. Future measurements can benefit from
jet substructure algorithms to separate the partonic energy loss and gluon-induced
radiation from non-perturbative effects in the scattering with the medium [1290].
Jet grooming may also help to develop a factorisation theorem in SCET, allowing
to resum the large logarithms arising from final state measurements when summing
over multiple interactions of the jet with the medium [1291].
In the future, jet substructure methods will continue to gain in importance. This is
in part due to the upgrade to the HL-LHC and the resulting larger data sample, together
with an increase in centre-of-mass energy to 14 TeV, allowing for more precise studies of high p T processes. Even the high-energy tails of rare electroweak processes
will become accessible, and jet substructure taggers will play an important role in
their measurements. The drawback of the high instantaneous luminosities aimed at
with the HL-LHC are higher levels of pileup. These will require advanced mitigation
strategies, where jet substructure methods will be indispensable. The upgrades of the
ATLAS and CMS detectors will enable the use of particle flow and pileup mitigation algorithms already at the first trigger level, allowing for online jet substructure
selections at a rate of 40 MHz. Upgrades of the inner tracking detectors will extend
the tracking coverage to pseudorapidities of about 4, and higher granularities of the
calorimeters will ensure an excellent angular resolution in the reconstruction of particles. Overall, the detector capabilities for jet substructure analyses will be enhanced
with respect to the present detectors. Another reason for the future importance of
jet substructure algorithms are studies using advanced machine learning techniques.
These can be used to design powerful jet substructure taggers, but they can also help
to identify variables carrying additional information, which have not been considered so far. For example, machine learning can help to construct a single observable,
which is the product of a number of jet substructure observables, carrying most of the
discrimination power of a complex artificial neural network [1292, 1293]. Similar
techniques can also help to identify jet substructure observables for the identification
of electrons inside jets [1294] or additional variables for jet tagging [1295]. Classi-