11 Detector Simulation
509
observables compared include the linearity of the response, the energy resolution
and the shower longitudinal and lateral shapes.
The comparisons identified areas where improvement was necessary: there were
problems in the large dependence of the electron energy response on the production
thresholds. These were also reported in studies related to the use of GEANT4 in
medical applications [110] and were corrected in GEANT4 release 8.0.
11.5.1 Calibration of EM Calorimeter Using Monte Carlo
The calibration of the ATLAS liquid argon (LAr) Electromagnetic Calorimeter for
electrons was developed utilizing a detailed Monte Carlo simulation of the detector
and its test beam [111]. The simulation included a very detailed description of the
complex geometry, collection of the energy deposition into hits and conversion to
digitized signal.
The simulation was shown to describe well all the relevant measurements,
including the mean reconstructed energy, the distributions of energy deposition in
particular compartments and the energy profiles in the longitudinal compartments
and lateral sections. For example, the mean reconstructed energy was described
to within 2% in the Pre-Sampler and the first two (of three) compartments of the
accordion. Also the distribution of the total reconstructed energy is described very
well—within the uncertainty due to the upstream material ahead of the setup and in
front of the PreSampler.
Based on this agreement, the simulation was used to correct for several effects,
which could not be measured or could only be estimated indirectly. One involved the
average energy deposited in the dead material between the pre-sampler and the first
compartment; this provided an estimate of this energy deposit. Another effect was
Bremsstrahlung in upstream material, due to which a fraction of events arrives at the
detector with reduced energy. A quantitative description was made and compared
with the measured total energy deposit.
Dedicated Monte Carlo simulations were undertaken to study the systematic
uncertainties induced by each effect. As a result, the reconstructed energy response
in the energy range from 15 to 180 GeV was found to be linear within 0.1% (an
exception is at 10 GeV, where it was found to be 0.7% lower). The systematic
uncertainties due to incomplete knowledge of the detector, the test beam, or the
reconstruction were found to be about 0.1% at low energies and negligible at
high energies. The effect of the non-linearity at about 40 GeV and above on the
measurement of the W + /W − mass was found to match the aimed precision of 15
MeV—provided it can be extended from the section tested to the full calorimeter.
509
observables compared include the linearity of the response, the energy resolution
and the shower longitudinal and lateral shapes.
The comparisons identified areas where improvement was necessary: there were
problems in the large dependence of the electron energy response on the production
thresholds. These were also reported in studies related to the use of GEANT4 in
medical applications [110] and were corrected in GEANT4 release 8.0.
11.5.1 Calibration of EM Calorimeter Using Monte Carlo
The calibration of the ATLAS liquid argon (LAr) Electromagnetic Calorimeter for
electrons was developed utilizing a detailed Monte Carlo simulation of the detector
and its test beam [111]. The simulation included a very detailed description of the
complex geometry, collection of the energy deposition into hits and conversion to
digitized signal.
The simulation was shown to describe well all the relevant measurements,
including the mean reconstructed energy, the distributions of energy deposition in
particular compartments and the energy profiles in the longitudinal compartments
and lateral sections. For example, the mean reconstructed energy was described
to within 2% in the Pre-Sampler and the first two (of three) compartments of the
accordion. Also the distribution of the total reconstructed energy is described very
well—within the uncertainty due to the upstream material ahead of the setup and in
front of the PreSampler.
Based on this agreement, the simulation was used to correct for several effects,
which could not be measured or could only be estimated indirectly. One involved the
average energy deposited in the dead material between the pre-sampler and the first
compartment; this provided an estimate of this energy deposit. Another effect was
Bremsstrahlung in upstream material, due to which a fraction of events arrives at the
detector with reduced energy. A quantitative description was made and compared
with the measured total energy deposit.
Dedicated Monte Carlo simulations were undertaken to study the systematic
uncertainties induced by each effect. As a result, the reconstructed energy response
in the energy range from 15 to 180 GeV was found to be linear within 0.1% (an
exception is at 10 GeV, where it was found to be 0.7% lower). The systematic
uncertainties due to incomplete knowledge of the detector, the test beam, or the
reconstruction were found to be about 0.1% at low energies and negligible at
high energies. The effect of the non-linearity at about 40 GeV and above on the
measurement of the W + /W − mass was found to match the aimed precision of 15
MeV—provided it can be extended from the section tested to the full calorimeter.