1. Summary of dE/dx studies in silicon and MS in muon system
with different Gauss versions
 G4 7.1, 8.2 and 8.3
07 Feb. 2008

2. History Known for a long time that energy loss in Si in Gauss is wrong
ST solution: use own model
But problem remains for Velo + tracking
G4 7.1.patch01a

3. History DC 06 (G4 7.1.patch01a) G4 Fast summer (G4 8.2..patch01 EMV)
G4 Def summer (G4 8.2.patch01 Std)
Now (Gauss v30r4, G4 8.3.patch01 EMV)
Version of Geant 4 used in the summer dE/dx reduced
Difference between fast and standard settings
Question-mark whether Geant4 was using our tracking settings
Velo tracking efficiency went down but dp/p better by 10 %
New version dE/dx increased and also Velo tracking efficiency
i.e. our tracking cuts seemed to be ignored with 8.2
ADC threshold

4. Tuning Strategy Many parameters that can be tuned in Geant4
Big to stop
-rays production
Kinetic energy cuts
Range cuts
Important to tune the parameters consistantly
Standard energy cut settings -rays range > 5mm
If track down to 100 keV for electrons -ray range 100 micron
1 MeV -ray have range ~1 mm
Range of 100 keV
Electron is ~ 100 micron
Bichsel 1990
Brigida 2004

5. Particle Gun Studies Fast studies of energy loss with a Particle gun
Gauss v30r4 (G4 8.3.p01), Boole v30r6
Scan Geant settings: fast versus standard, -ray range
Generate samples of 50k muons at fixed energy
I am sure Geant is taking our tracking settings
Only simulate VELO to save speed
Previous studies show results are applicable to ST
Plot deposited energy in Si
Extract MPV value from parabola fit around maximum
FHWM by linear interpolation around ~ half height points

6. Particle Gun Studies
Want to count all the energy deposited by muon in Si
Primary + -rays
Most robust algorithm is to count all energy deposited in sensor
Only works because I have a single particle gun
Find some energy is ‘lost’ if cut settings not consistent
Choice of settings is important !
Standard energy thresholds
Range for -rays 1 micron
Lower energy thresholds

7. Particle Gun: Results Energy loss with new version of Gauss much
closer to expectations even
without -rays
Why is not clear…
No difference
between standard and fast settings
Different to studies in the Summer
Use only fast from now on
Gauss series  G4 7.1-patch01a
Gauss v30r4  G4 8.3-patch01
Summer  G4 8.2-patch01

8. Particle Gun: Results Turn on -rays shape of distribution versus 
become closer to expectation
from Landau theory
Results independent of -rays range cut for values between mm and 5 mm
Gauss v30r4  G4 8.3-patch01

9. Particle Gun: Results
The with of the distribution is still underestimated
Add smear for the atomic
binding things look more
reasonable. But…
Gauss series  G4 7.1-patch01a
Gauss v30r4  G4 8.3-patch01

10. Particle Gun: Results Smear increases the MPV
Counter this by scaling by ~ 6 %
Better: scale as function of 

11. Delta ray properties Properties of delta rays (120 GeV sample)
Spectrum looks too(?) hard (4 MeV MPV, long tail to 100 MeV)
Rate is suprising high 9480/ incident primaries ~0.014
Is this correct ?
Max energy transfer in single collision (relativistic kinematics)
Em = Mc2 2/[(M/2m) + 2m/M + )]
At high energy practically unbounded
For 500 MeV muons, limit 21 MeV
Geant4 gets this correct
Rate 5675/ ~ 0.9 %

12. Particle Type dependence
Differences for different particle types
Electrons, muon and hadrons behave differently
Energy loss 5 % higher for electrons than hadrons
This is not physical
Suggests differences due to different physics lists/threshold settings for different particle types
Again different -rays on/off
Different tuning/particle type
-rays off 
-rays on

13. Minimum Bias Data 2.8 GHz Opteron processor
4 samples of 1000 events generated:
Run
Sim event size
Time/event N
No -rays , 10 MeV cut for hadrons, 1 MeV for electrons
350 k (Gauss)
126 k (Boole)
21 s
270 ms LT
No -rays, 1 MeV cut for hadrons, 0.1 MeV for electrons
400 k (Gauss)
130 k (Boole)
29 s
300 ms D
5 mm range cut, 10 MeV cut for hadrons, 1 MeV for electrons
372 k (Gauss)
133 k (Boole)
24 s
290 ms
LTD
0.1 mm range cut, 1 MeV cut for hadrons, 0.1 MeV for electrons
572 k (Gauss)
160 k (Boole)
43 s
410 ms
From the previous discussion, results with low thresholds have large systematic errors

14. Multiple scattering in Muon system
Special production with very low thresholds (range and tracking) and {LHEP/QGSP}_BERT_HP to make parametrization of low energy and slow background to apply during digitization for massive productions

15. Multiple Scattering simulation
Muon trajectories are dominated by multiple scattering interaction in the Calorimeters and in the Muon Filters.
In GEANT8.0 Multiple Scattering simulation was not correctly simulated in the cases of dense materials and large step sizes (like Muon filters)
Correlation between displacement and angular deviation was not included.

16. Multiple Scattering Simulation
After calorimeters
After Muon Filters
Red = PDG, black = Gauss (GEANT 7.1.p01a)
muons p=[5-100]GeV/c (particle gun)

17. New Gauss v30r2 (GEANT8.2) Revised description of MS
Different option file tested so far (with muons) NO improvements with “standard options”. Ok if deltaRays production enabled in MuonFilters
Further investigation needed (check other particles)
EMPhysics
(new default)
EMfastPhysics
EMPhysics+
deltaRays in Mfilters

18. Multiple Scattering Simulation
GEANT 7.1.p01a
thin line = gauss;
thick line = PDG
M2: gauss > PDG;
M3,M4,M5 gauss < PDG
GEANT 8.2
EMPhysics+
deltaRays in Mfilters
Nice agreement