500
J. Apostolakis
general principles for unmeasured reactions and energies. The number of nuclear
shells varies from one for the light, three for medium, and a maximum six for the
heaviest elements.
As its main use is the simulation of LHC experiment detectors, a number of
modeling and implementation choices were made to optimize computation speed.
These include linear interpolation for the sampling of partial cross sections and
large 10-degree bins in angular distributions—justified by the smoothing effect of
additional interactions.
The newest cascade model in GEANT4, the INCL Liege cascade, is the one under
the most active development. This event generator, under development since the
1990s [75], was developed to reproduce spallation data for reactions at 100 MeV–
1.5 GeV using a parameter-free model. It interfaced with the ABLA code for
deexcitation. Its original Fortran version up to version 4.6 [76] has interfaced to
MCNPX and MCNP6. Recent development focused on the re-engineered INCL++
[77], which reproduced the performance of 4.6 and was extended to handle light
ion projectiles (up to carbon16) and to higher energies, up to 15 GeV producing
multiple pions.
Large suites of benchmark data are used to tune and verify the modeling of each
cascade, covering neutron, proton and pion production at energies from 60 MeV to
3 GeV. Spallation data from inverse kinematic reactions on hydrogen targets at GSI
with a range of projectiles from 56 Fe [78] to 238 U [79] at 1 GeV/nucleon provide
different challenges for modeling and complement these data.
Comparisons of hadronic and in particular cascade models have been undertaken
periodically under the auspices of the IAEA [80]. These benchmarks use a large set
of thin target data to probe the accuracy and predictive capabilities of each model.
INCL was found to be one of the competitive models. Similar test suites are utilized
as a part of internal benchmarks and for tuning of models.
The details of the intranuclear cascade and the pre-equilibrium model determine
all the emission of higher energy particles, but the details of the coupling and
the quality of the de-excitation module are also critical to the performance of
many applications from activation to calorimeter simulation. Good modelling of
the resulting nuclei is required to ensure that the energy lost to nuclear breakup, a
key component of non-compensation in hadronic calorimeters, is accurate. Recent
results from RD52 [81] were interpreted as deficiencies in the modelling of nuclear
breakup in GEANT4 version 9.4, showing that there is room for further improvement
in this energy range.
11.3.9 High-Energy ‘String’ Models
Models for interactions at high energies (above 5–10 GeV) simulate quark-level
interactions and rely on phenomenological descriptions of soft QCD interactions to
generate low-energy hadrons from the remnants of the high-energy collisions. They
are applicable to all hadron projectiles. Three variants are available: the Dual Parton
J. Apostolakis
general principles for unmeasured reactions and energies. The number of nuclear
shells varies from one for the light, three for medium, and a maximum six for the
heaviest elements.
As its main use is the simulation of LHC experiment detectors, a number of
modeling and implementation choices were made to optimize computation speed.
These include linear interpolation for the sampling of partial cross sections and
large 10-degree bins in angular distributions—justified by the smoothing effect of
additional interactions.
The newest cascade model in GEANT4, the INCL Liege cascade, is the one under
the most active development. This event generator, under development since the
1990s [75], was developed to reproduce spallation data for reactions at 100 MeV–
1.5 GeV using a parameter-free model. It interfaced with the ABLA code for
deexcitation. Its original Fortran version up to version 4.6 [76] has interfaced to
MCNPX and MCNP6. Recent development focused on the re-engineered INCL++
[77], which reproduced the performance of 4.6 and was extended to handle light
ion projectiles (up to carbon16) and to higher energies, up to 15 GeV producing
multiple pions.
Large suites of benchmark data are used to tune and verify the modeling of each
cascade, covering neutron, proton and pion production at energies from 60 MeV to
3 GeV. Spallation data from inverse kinematic reactions on hydrogen targets at GSI
with a range of projectiles from 56 Fe [78] to 238 U [79] at 1 GeV/nucleon provide
different challenges for modeling and complement these data.
Comparisons of hadronic and in particular cascade models have been undertaken
periodically under the auspices of the IAEA [80]. These benchmarks use a large set
of thin target data to probe the accuracy and predictive capabilities of each model.
INCL was found to be one of the competitive models. Similar test suites are utilized
as a part of internal benchmarks and for tuning of models.
The details of the intranuclear cascade and the pre-equilibrium model determine
all the emission of higher energy particles, but the details of the coupling and
the quality of the de-excitation module are also critical to the performance of
many applications from activation to calorimeter simulation. Good modelling of
the resulting nuclei is required to ensure that the energy lost to nuclear breakup, a
key component of non-compensation in hadronic calorimeters, is accurate. Recent
results from RD52 [81] were interpreted as deficiencies in the modelling of nuclear
breakup in GEANT4 version 9.4, showing that there is room for further improvement
in this energy range.
11.3.9 High-Energy ‘String’ Models
Models for interactions at high energies (above 5–10 GeV) simulate quark-level
interactions and rely on phenomenological descriptions of soft QCD interactions to
generate low-energy hadrons from the remnants of the high-energy collisions. They
are applicable to all hadron projectiles. Three variants are available: the Dual Parton