1. A Virtual Montecarlo (VMC) Application for AMS-01
Alberto Oliva & Nicola Tomassetti, September 2009

2. Motivation
AMS-01 more precise trigger efficiency calculation  comparison between Geant3, Geant4 and Fluka An effordable estimation of the fragmentation systematics  contamination from high charges to low charges Estimation of the VMC framework capabilities in the AMS02 perspective

3. The Virtual Montecarlo (VMC)
Virtual Interface
User Code
(VMCApp)
VMC
TGeo
GEANT3 VMC
TGeant3
GEANT3
Particles
Hits
GEANT4 VMC
TGeant4
GEANT4
FLUKA VMC
TFluka
FLUKA
Output
ROOT framework
Abstract Layer
Physics
simulation
FORTRAN
Tested setup
root /+/ geant4.9.2 /+/ fluka dpmjet3 + rqmd_v2.4 /+/
geant321+_vmc.1.11 (GEANT3 + TGeant3) /+/ geant4_vmc.2.7 (TGeant4) /+/
fluka_vmc.0.4 (TFluka) /+/ rqmd_v interface (extracted from GBATCH)

4. Building an application
You have to derive the application from the virtual classes:
TVirtualMCApplication
TVirtualMCStack
In these classes there are the some virtual methods needed for the
interface with the generic MC:
TGeant3 : public TVirtualMC
TGeant4 : public TVirtualMC
TFluka : public TVirtualMC
In the TVirtualMC interface all the MC generic features should be
virtualized
TVirtualMCApplication
InitGeometry()
AddParticles()
AddIons()
BeginEvent()
BeginPrimary()
PreTrack() // Track (...geant/fluka)
Stepping() // Access to the step
PostTrack()
FinishPrimary()
FinishEvent()
TVirtualMCStack
PushTrack() // Real method
PopNextTrack() // Real method
PopPrimaryForTracking() // Real method
GetNtrack()
GetNprimary()
GetCurrentTrack()
GetCurrentParticle()
GetCurrentTrackNumber()

5. VMC Summary
Installation of the AMS01 Geant4 gbatch  Yue Zhou version, used for some production Installation of VMC framework Conversion of the AMS01 G4Geometry to TGeometry  Done automatically (overlaps …)  TEve event display Importing the magnetic field on TVirtualMagField class Output classes  Particles, MCHits Additional ions definitions Primary generator Physics list and cuts (…) From interpreted to compiled  This version is fully independent from GBATCH  No reconstruction, only MC information available

6. TGeometry AMS01 TEveManager TGeometry is quite similar to G4Geometry:
G4Solid (G4Box, G4Tubs, …)  TGeoShape (TGeoBBox, TGeoTubeSeg)
G4UnionSolid, G4SubtractionSolid  TGeoCompositeShape
G4DisplacedSolid  TGeoShape
G4Material  TGeoMedium
G4LogicalVolume  TGeoVolume
G4PhysicVolume  TGeoVolume + TGeoCombiTrans
A simple C++ routine transforms the Yue Zhou G4-AMS01-Geometry in a TGeometry
AMS01
TEveManager
Magnet

7. TGeometry: Overlaps Pay attention to region overlaps
 FLUKA and GEANT use a different geometric approach (Comb./Hyer.)
Overlap between
TOF PMTs &
TOF support
(solved ignoring
TOF supports)
Other founded overlaps:
 Structure overlaps (1 cm, solved)
 Markers (solved ignoring makers, better to use TGeoComposite shapes)
 TOF Counters and Silicon planes (solved moving PMTs by 4 mm)
 TOF Counters and TOF Counters (solved scaling by 2 mm the PMTs)
 Ladders overlaps (0.2 mm, solvable but not solved)

8. Ion Fragmentation Physics Lists
GEANT3 GEANT3 Physics + an hadronic pakage (GHEISHA, GFLUKA, GCALOR or MICAP):  works only up to Alpha particles  for heavy ions there are only EM processes In GBATCH there is a RQMD(v1.07) interface able to produce fragmentations:  The code has been adapted to work with TGeant3 in the VMC framework (IGFPAR, several function name, few lines here and there, …)
GEANT4 Ion inelastic fragmentations final states are simulated by two models:  Via the G4BinaryLightIonReaction class in the Binary Cascade interaction model (limitations: from 80 MeV to 10 GeV/n, one of the two nuclei must have A < 13)  Wilson Ablation Model (old fashioned geometric approach) For now I’m using the QGSP_BIC_HP Physic List:  wher BIC stands for Binary Ion Cascade  I’ve extended by hand the BIC validity limit!!!
FLUKA Uses RQMD(v2.4) between 0.1 and 5 GeV/n Uses DPMJET-3 above 5 GeV/n

9. Toward a VMC Validation: AMS-01 GSI Beam Test (1998)

10. AMS-01 GSI Beam Test (1998)
Helium runs: 27 M events (186, 446, 896, 2000 GeV/c/n)
Carbon runs: 11 M events (330, 446, 896, 2000 GeV/c/n)
Cosmic runs: 1.5 M events
Sample used » 2 GeV/c/n  » 50 k events
Mean GeV/n

11. GSI: Charge Selection Charge ID from STS-91 data
Nicola’s Toy MC
(constructed with STS-91 data)
Application of Nicola’s Charge ID
(constructed with STS-91 data)
Zreconstructed
Zgenerated
Deconvolution
Charge ID from STS-91 data
Toy MC for contamination estimation
GSI charge distribution
Exact unfolding (decontamination)
 GSI fragmentation estimation

12. GSI: VMC Simulation
2 GeV/n/c
2 GeV/c

13. GSI: Inelastic interaction probability
G3-VMC
G4-VMC
Low e shield
Upper TOF
Lower TOF
Tk1
Tk2
Tk3
Tk4
Tk5
Tk6
FL-VMC
The trigger is simulated according to a very simple digitization scheme.

14. GSI: MC/Data Comparison
G3-VMC
3.1%
1.1%
0.7%
G4-VMC
FL-VMC
4.9%
2.8%
1.5%
0.8%
0.8%
0.4%