496
J. Apostolakis
theory [50], which is used in GEANT 3, but is limited to small angles. The approach
of GEANT4 borrows from Lewis’ description [51].
Key effects of multiple scattering are angular deflections and straggling. The
latter’s most important effect is the shortening of the distance travelled in the
direction of the initial momentum. This must be modeled in order to obtain correct
energy loss for the passage through material, as is done in GEANT 3. The second
effect of straggling, the displacement in the lateral directions is correlated with
angular deflection. Similarly to EGS4, FLUKA and GEANT4 also sample this
displacement, each using different algorithms. This enables longer steps while
maintaining accuracy. The best algorithms allow longer steps, or more accurate
modeling of the correlations between the affected changes in the state of the particle.
The algorithm for multiple scattering has a significant effect on the results
obtained in many detectors and setups. Examples include the partition of energy
in sampling calorimeters, the correlation between the deflection of muons and their
positions after substantial material. In particular, many quantities are very sensitive
to the details of its formulation and implementation. These include the fraction of
low-energy electrons (T << 1 MeV) scattering backwards at the interface between
low and high Z materials and the correlation between the direction of a particle
exiting a detector (e.g. muon) and its position.
In addition to the sampling of the final state, high accuracy for electron transport
necessitates careful treatment of multiple scattering of low-energy electrons at
boundary crossing [52]. New algorithms have been developed for exact electron
transport without special treatment for boundary crossing [53, 54]. These algorithms
have been implemented in electron–photon Monte Carlo codes: PENELOPE [55],
EGSnrc [56] and EGS5 [57]. A comparison [58] in 2007 benchmarked the
algorithms in several of these codes and in GEANT4, using data from custom setups
with thin slabs. PENELOPE and EGSnrc demonstrated the best performance, while
GEANT4 obtained good results only with specific settings.
The new GEANT4 GS model [59], available since GEANT4 release 10.3, implements Kawrakow’s approach to provide angular deflections for any size step without
free parameters and offers the option for accurate boundary crossing. It achieves the
best agreement, amongst GEANT4 models, with a wide range of benchmark data
including backscattering data.
Models for specialized processes, such as transition radiation, exist in some tools
including GEANT4 [60].
11.3.6 Hadronic Interactions and Their Modeling
In contrast to the simulation of EM physics processes, the simulation of hadronic
physics processes from first principles is not possible, except partially at the high
energy limit. At all energies, the cross-sections and the models used are based,
directly or indirectly, on measured data of hadron–nucleon and hadron–nucleus
interactions, and on phenomenology.
J. Apostolakis
theory [50], which is used in GEANT 3, but is limited to small angles. The approach
of GEANT4 borrows from Lewis’ description [51].
Key effects of multiple scattering are angular deflections and straggling. The
latter’s most important effect is the shortening of the distance travelled in the
direction of the initial momentum. This must be modeled in order to obtain correct
energy loss for the passage through material, as is done in GEANT 3. The second
effect of straggling, the displacement in the lateral directions is correlated with
angular deflection. Similarly to EGS4, FLUKA and GEANT4 also sample this
displacement, each using different algorithms. This enables longer steps while
maintaining accuracy. The best algorithms allow longer steps, or more accurate
modeling of the correlations between the affected changes in the state of the particle.
The algorithm for multiple scattering has a significant effect on the results
obtained in many detectors and setups. Examples include the partition of energy
in sampling calorimeters, the correlation between the deflection of muons and their
positions after substantial material. In particular, many quantities are very sensitive
to the details of its formulation and implementation. These include the fraction of
low-energy electrons (T << 1 MeV) scattering backwards at the interface between
low and high Z materials and the correlation between the direction of a particle
exiting a detector (e.g. muon) and its position.
In addition to the sampling of the final state, high accuracy for electron transport
necessitates careful treatment of multiple scattering of low-energy electrons at
boundary crossing [52]. New algorithms have been developed for exact electron
transport without special treatment for boundary crossing [53, 54]. These algorithms
have been implemented in electron–photon Monte Carlo codes: PENELOPE [55],
EGSnrc [56] and EGS5 [57]. A comparison [58] in 2007 benchmarked the
algorithms in several of these codes and in GEANT4, using data from custom setups
with thin slabs. PENELOPE and EGSnrc demonstrated the best performance, while
GEANT4 obtained good results only with specific settings.
The new GEANT4 GS model [59], available since GEANT4 release 10.3, implements Kawrakow’s approach to provide angular deflections for any size step without
free parameters and offers the option for accurate boundary crossing. It achieves the
best agreement, amongst GEANT4 models, with a wide range of benchmark data
including backscattering data.
Models for specialized processes, such as transition radiation, exist in some tools
including GEANT4 [60].
11.3.6 Hadronic Interactions and Their Modeling
In contrast to the simulation of EM physics processes, the simulation of hadronic
physics processes from first principles is not possible, except partially at the high
energy limit. At all energies, the cross-sections and the models used are based,
directly or indirectly, on measured data of hadron–nucleon and hadron–nucleus
interactions, and on phenomenology.