SOA IT Top 5 Critical Factors for Successful PeopleSoft 9.2 upgrade.
EM Standard Physics Upgrade for Geant4 9.2
description
Transcript of EM Standard Physics Upgrade for Geant4 9.2
1
EM Standard Physics Upgrade for Geant4 9.2
V.N.Ivanchenko, 3 December 2008
2
J.Apostolakis, A.Bagulya, A.Bogdanov, S.Elles, V.Grichine, P.Gumplinger, O.Kadri, R.Kokoulin, M.Maire, A.Schaeliske, T.Toshito, L.Urban
33
Outline List of new developments Highlights on upgrade of standard EM
physics Bremsstrahlung Multiple Scattering Spline option ApplyCuts option
Some validation results Conclusions
4
New physics will available with g4 9.2 Relativistic bremsstrahlung model for E > 1 GeV
Updated density effect Hadron induced bremsstrahlung and e+e- pair production
processes/models Updated positron annihilation to hadrons G4eMultipleScattering process with G4UrbanMscModel2 G4MuMultipleScacttering process with G4WentzelVIModel G4Cerenkov process updated G4Scintillation process includes Birks low ICRU’73 data are inserted Ion ionisation updated
5
New options and infrastructure components will be available with g4 9.2 Spline option for physics tables Option3 Standard Physics builder
Optimized for medical and space applications New utility classes:
G4EmSaturation – Birks effect (used in G4Scintillation) NIEL and gamma interactions
G4ElectronIonPair – helper class to sample of ionisation clusters at charge particle step using ICRU’31 approach
G4EmProcessSubType – enumerator of EM processes G4EmConfigurator – helper class to configure EM models per
energy range and G4Region
6
Highlights on new developments which is available with 9.2
7
Bremsstrahlung
Added new relativistic model bremsstrahlung model for e± Updated density and LPM
effects Applied above 1 GeV
Added new model for hadron bremsstrahlung and e+e- pair production In predefined PhysLists for
π± and protons
287 GeV e- at Ta target (4%X0)
88
New relativistic bremsstrahlung modelA.Schaeliscke, IEEE NSS, 2008 Bethe-Heitler formula
with corrections Complete screening
with Coulomb correction Valid for E > 1 GeV
Density & LPM-Effect consistent combination
a’la Ter-Mikaelian
LPM old
LPM new
LPM off
Data from the CERN experiment: H.D.Hansen et al, PR D 69, 032001 (2004)
287 GeV e- at Ta target (4%X0)
9
Updated multiple scattering
New specialized process for e± G4eMultipleScattering Uses G4UrbanMscModel2 tuned for electron data FR = 0.04
Old model G4UrbanMscModel is frozen Validation versus data and alternative models:
G4WentzelVIModel G4CoulombScattering
1010
Test of e- transport versus Sandia data(details in O.Kadri et al, NIM B258 (2007) 358) Sensitive to
multiple scattering Directly connected
with LHC calorimeters results
Tuned Urban’s msc model#2 is best in describing
data Label e-Msc, skin=3
Ta Al
1111
Upgrade of multiple scattering model
Old default version will be kept for backward compatibility
1212
Spline option for Physics Tables Spline interpolation
of dEdx, range and other tables
Zoomed Bragg peak Default EM physics
without spline (Opt0) strong cut
dependence Non-statistical
variations due to numerical instability
Spline is activated (Opt3) Stable result
13
“ApplyCuts” option
“ApplyCuts” was inside the package for a long time By default is “false”
Production thresholds (cuts) are applied only for ionisation and bremsstrahlung
If “true” production thresholds are applied for all EM standard processes
Provides more fast simulation but may bias results if cuts are high (above 1 mm)
1414
ATLAS barrel type calorimeter EMV and EMX in this
plot use “ApplyCuts” Default Physics List
results unchanged Cut on gamma
processes (EMX) reduces accuracy of energy response but has no effect on EMV
15
ATLAS HEC type calorimeter
Visible energy increased by 0.5%
1616
CMS ECAL type calorimeter
5x5 matrix of PW04
crystals Reduced (~0.5%)
energy deposition in central crystal both for the default and EMV Physics Lists Result of upgrade of
Urban multiple scattering model
1717
CPU benchmarking Static build on dedicated SLC4 PC
no libraries from afs
EM1 EM2 EM3 EM1_EMV EM2_EMV EM3_EMV
8.3 1.33 2.30 1.84 1.0 1.0 1.0
9.0 1.21 2.05 1.65 0.92 0.93 0.94
9.1p02 1.16 2.05 1.64 0.92 0.94 0.93
9.2 (ApplyCuts for EMV) 1.11 1.84 1.49 0.72 0.81 0.87
CPU gain with 9.0 mainly due to c++ software cleanup CPU gain with 9.2 due to optimization of physics
step limitation by multiple scattering energy threshold for gamma processes
18
Summary
There are number of modifications in EM standard in g4 9.2, it is expected: better CPU performance and better agreement with test-
beam No change in total response and resolution for ECal Minor increase of response in sampling calorimeters with
low sampling fraction Apply Cuts option may significantly improve CPU
performance BUT Should be applied with care Cuts should be reviewed