11 Detector Simulation
487
which reconstruction and calibration in many channels were achieved within the first
years of operation, a fraction of the time of previous hadron collider experiments [4].
A challenging aspect of the use of simulation in data analysis is the estimation
of the systematic errors of the simulation. The accuracy of a simulation depends on
the accuracy of the description of the detector’s geometry and material composition,
the knowledge of the beam or primary particles delivered and the intrinsic accuracy
of the simulation tool(s) used. The detector simulation tools, in turn, are limited by
several factors: the availability and known accuracy of measurements utilized to tune
or validate the physics models, in particular of the cross-sections; the limitations
of the physical models in reproducing the energy spectra and other properties of
interactions; the approximations utilized to obtain adequate computational speed, to
simulate the required number of events using the available computing resources.
There is an important tradeoff between the level of detail, both in the geometrical
description of a setup and the choice of physics modeling options, and the computational cost of large-scale simulation. In the past 5 years the LHC experiments
have been able to use detailed simulation to produce several billion events per year
[5] providing unprecedented support of analysis in hadron collider physics. The
increase in luminosity in the HL-LHC era will bring a need for much higher statistics
of simulated events, whereas projected growth in computing power is forecast
to be modest in comparison [6]. This is driving research to achieve substantial
performance gains in full simulation in GeantV [7], and the expectation that faster
approximate simulation will be relied upon once again for most analyses—leading
to efforts to create new kinds of fast simulation which more accurately capture
additional features of the full simulation, including fluctuations of key quantities.
11.2 Stages and Types of Simulation
• Event generators and detector simulation
• Scale from full detail to fast simulation
• Simulation of energy deposition or signal generation
• Assessment of radiation effects
• Key tools: Event generators, detector Monte Carlo, radiation transport
– Detector Monte Carlo: GEANT, FLUKA, GEANT4
– Radiation related MC: FLUKA, MARS, MCNP/MCNPX
– Signal generation: Garfield
The simulation of the passage of particles through a detector and the response
of the detector’s sensitive elements typically proceeds in different stages. In the
first stage an external event generator simulates the initial interaction and then
decays short-lived particles; the results are the “primary” tracks. The second stage
is detector simulation and involves the tracking of the primary particles through the
structures of the detector, sampling interactions with their components, and creating
487
which reconstruction and calibration in many channels were achieved within the first
years of operation, a fraction of the time of previous hadron collider experiments [4].
A challenging aspect of the use of simulation in data analysis is the estimation
of the systematic errors of the simulation. The accuracy of a simulation depends on
the accuracy of the description of the detector’s geometry and material composition,
the knowledge of the beam or primary particles delivered and the intrinsic accuracy
of the simulation tool(s) used. The detector simulation tools, in turn, are limited by
several factors: the availability and known accuracy of measurements utilized to tune
or validate the physics models, in particular of the cross-sections; the limitations
of the physical models in reproducing the energy spectra and other properties of
interactions; the approximations utilized to obtain adequate computational speed, to
simulate the required number of events using the available computing resources.
There is an important tradeoff between the level of detail, both in the geometrical
description of a setup and the choice of physics modeling options, and the computational cost of large-scale simulation. In the past 5 years the LHC experiments
have been able to use detailed simulation to produce several billion events per year
[5] providing unprecedented support of analysis in hadron collider physics. The
increase in luminosity in the HL-LHC era will bring a need for much higher statistics
of simulated events, whereas projected growth in computing power is forecast
to be modest in comparison [6]. This is driving research to achieve substantial
performance gains in full simulation in GeantV [7], and the expectation that faster
approximate simulation will be relied upon once again for most analyses—leading
to efforts to create new kinds of fast simulation which more accurately capture
additional features of the full simulation, including fluctuations of key quantities.
11.2 Stages and Types of Simulation
• Event generators and detector simulation
• Scale from full detail to fast simulation
• Simulation of energy deposition or signal generation
• Assessment of radiation effects
• Key tools: Event generators, detector Monte Carlo, radiation transport
– Detector Monte Carlo: GEANT, FLUKA, GEANT4
– Radiation related MC: FLUKA, MARS, MCNP/MCNPX
– Signal generation: Garfield
The simulation of the passage of particles through a detector and the response
of the detector’s sensitive elements typically proceeds in different stages. In the
first stage an external event generator simulates the initial interaction and then
decays short-lived particles; the results are the “primary” tracks. The second stage
is detector simulation and involves the tracking of the primary particles through the
structures of the detector, sampling interactions with their components, and creating