3.1 ATLAS and CMS Detectors
63
a laterally segmented calorimeter, and additionally provide pointing information. The
difference between ATLAS and CMS for the HCAL resolution is particularly large
at higher energies: a 1 TeV jet has
σ (E)
E
∼ 2% in ATLAS, in contrast to
σ (E)
E
∼ 5%
in CMS. This is one reason why CMS fully adapted a particle flow technique since
the beginning of the LHC (see Sect. 3.2).
3.2 Jet Reconstruction and Calibration 4
The ATLAS and CMS experiments have dedicated algorithms to reconstruct particle
kinematics from calorimeter and tracker information designed to minimise the fake
rate, maximise the efficiency, and minimise the bias and resolution of the particle
candidate parameters. As there is no algorithm that can simultaneously optimise all
of these objectives, the various approaches trade off optimality under one metric
for improvements under another. ATLAS and CMS have also developed different
algorithms that cater to the experiment’s hardware as well as the collaboration’s goals
for the tradeoffs. By default, CMS combines tracker and calorimeter information
into unified particle flow (PF) objects as inputs to jet reconstruction [74, 75, 407].
ATLAS has traditionally used calorimeter-only information for jet reconstruction,
with tracking information used to augment/enhance the performance. While ATLAS
is current migrating to a variation of particle flow [408], most of this review will
focus on calorimeter-only jets as they are still the most widely used setup. ATLAS
benefits less than CMS from particle flow because of its weaker magnetic field, the
longitudinally segmented calorimeter and the better jet energy resolution obtained
with the calorimeter.
ATLAS and CMS combine calorimeter cells using topological clusters [407,
409]. These clusters are three dimensional in ATLAS as a result of the longitudinal
segmentation. Cluster seeds are started from highly significant energy (high cell
signal compared to average electronic and pileup noise) deposits which are combined
(or split) based on the distribution of the significance of energy in nearby cells.
Calorimeter-cell clusters in CMS are obtained using a Gaussian-mixture model,
which results in one or more calorimeter clusters within each topological cluster.
HCAL clusters can be split according to the number and energy distribution of
associated ECAL clusters. Cluster splitting is critical to achieve a better estimate of
the spatial energy distribution as input to jet substructure algorithms [410, 411].
The topological clusters are calibrated using simulations to account for the noncompensating calorimeter response to hadrons, signal losses due to energy deposited
in inactive detector material and signal losses on cluster boundaries caused by the
topological clustering algorithms. In ATLAS, the calibration scheme relies on a classification of clusters as hadronic or electromagnetic in origin based on the energy
and position of the cluster, the longitudinal depth (λ clus ) and normalised signal
4 The text in this section has been taken from [26] and has been written by the author together with
B. Nachman. It has been adjusted to fit this book.
63
a laterally segmented calorimeter, and additionally provide pointing information. The
difference between ATLAS and CMS for the HCAL resolution is particularly large
at higher energies: a 1 TeV jet has
σ (E)
E
∼ 2% in ATLAS, in contrast to
σ (E)
E
∼ 5%
in CMS. This is one reason why CMS fully adapted a particle flow technique since
the beginning of the LHC (see Sect. 3.2).
3.2 Jet Reconstruction and Calibration 4
The ATLAS and CMS experiments have dedicated algorithms to reconstruct particle
kinematics from calorimeter and tracker information designed to minimise the fake
rate, maximise the efficiency, and minimise the bias and resolution of the particle
candidate parameters. As there is no algorithm that can simultaneously optimise all
of these objectives, the various approaches trade off optimality under one metric
for improvements under another. ATLAS and CMS have also developed different
algorithms that cater to the experiment’s hardware as well as the collaboration’s goals
for the tradeoffs. By default, CMS combines tracker and calorimeter information
into unified particle flow (PF) objects as inputs to jet reconstruction [74, 75, 407].
ATLAS has traditionally used calorimeter-only information for jet reconstruction,
with tracking information used to augment/enhance the performance. While ATLAS
is current migrating to a variation of particle flow [408], most of this review will
focus on calorimeter-only jets as they are still the most widely used setup. ATLAS
benefits less than CMS from particle flow because of its weaker magnetic field, the
longitudinally segmented calorimeter and the better jet energy resolution obtained
with the calorimeter.
ATLAS and CMS combine calorimeter cells using topological clusters [407,
409]. These clusters are three dimensional in ATLAS as a result of the longitudinal
segmentation. Cluster seeds are started from highly significant energy (high cell
signal compared to average electronic and pileup noise) deposits which are combined
(or split) based on the distribution of the significance of energy in nearby cells.
Calorimeter-cell clusters in CMS are obtained using a Gaussian-mixture model,
which results in one or more calorimeter clusters within each topological cluster.
HCAL clusters can be split according to the number and energy distribution of
associated ECAL clusters. Cluster splitting is critical to achieve a better estimate of
the spatial energy distribution as input to jet substructure algorithms [410, 411].
The topological clusters are calibrated using simulations to account for the noncompensating calorimeter response to hadrons, signal losses due to energy deposited
in inactive detector material and signal losses on cluster boundaries caused by the
topological clustering algorithms. In ATLAS, the calibration scheme relies on a classification of clusters as hadronic or electromagnetic in origin based on the energy
and position of the cluster, the longitudinal depth (λ clus ) and normalised signal
4 The text in this section has been taken from [26] and has been written by the author together with
B. Nachman. It has been adjusted to fit this book.