For this study, I generated 1000 pions +/- per event with  :

• detp geom y2007
• user/vsig 0 20
• gkine 1000 9 0.1 5.0 -0.7 0.7 -3.1415 3.1415 -30 30
• trig 50

I then run the bfc in DEV with the following chain :

• "trs,srs,ssd,fss,y2007,Idst,IAna,l0,tpcI,fcf,Tree,logger,ITTF,Sti,genvtx,SvtIt,SsdIt,geant,-evout,IdTruth,tags,bbcSim,tofsim,emcY2,EEfs,-GeantOut,big,-dstout,fzin,StarMagField,MiniMcMk,McAss,McAna
• with the Lorentz effect
• without the lorentz effect

Fot these plots,  the cuts used are :

• pTMC>0.1,pT(MatechedPairs)>0.1
• |eta(MC)| <= 0.5
• quality >90
• mHitsFit >25
• mSsdHitMc ==1

The black points are the mean of the FitSlices()

1. for the pions minus

• with the lorentz effect 

Fig 1 : pT(MatchedPairs) - pTMC vs ptMC for primaries

Fig 2 : pT(MatchedPairs) - pTMC vs ptMC for globals :

• without the lorentz effect :

Fig 3 : pT(MatchedPairs) - pTMC vs ptMC for primaries

Fig 4 : pT(MatchedPairs) - pTMC vs ptMC for globals

• with the fast Simulator (where no Lorentz effect is taken into account)

Fig 5 :pT(MatchedPairs) - pTMC vs ptMC for primaries

Fig 6 :pT(MatchedPairs) - pTMC vs ptMC for globals

            2. for the pions plus

Fig 7 :pT(MatchedPairs) - pTMC vs ptMC for primaries (pions +)

Fig 8 :pT(MatchedPairs) - pTMC vs ptMC for globals (pions +) 

delta_pt for negative tracks

 

delta pt for positive tracks

 

PART II

I rerun with more statistic :

• 500 events with 500 pions plus per event
• 0.1 < pT < 5.0 [GeV/c]
• -1.0 < eta < 1.0
• -pi < phi < pi
• -10 < z_vertex < 10 [cm]

For both plots :

1. red : slow simulator with Lorentz effect
2. green : slow simulator without Lorentz effect
3. blue : fast simulator

delta_pt : pT(RECO) -pT(MC)

 

sigma_pt

 

 From delta_pt :

1. running without and with Lorentz effect in simulation gives the same result
2. There seems still to have a difference between fast and the slow simulator. It's a small effect but delta_pt for fast simulator deviates less for high pt tracks. 

 

 PART III:

 

• 300 events with 500 pions minus per event
• 0.1 < pT < 10.0 [GeV/c]
• -1.0 < eta < 1.0
• -pi < phi < pi
• -10 < z_vertex < 10 [cm]

For both plots :

1. red : slow simulator with Lorentz effect
2. green : slow simulator without Lorentz effect
3. blue : fast simulator

 

I increased the pT range to see this effect for larger pT

Fig III.a : delta_pt : pT(RECO) -pT(MC) (for 0< pT(MC)< 4, the same as the plot for the positive)

the same as above but for all pT range 

Fig III.b : delta_pt : pT(RECO) -pT(MC) 

 

Fig III.c : sigma_pt

the same as above but for all pT range

Fig III.d : sigma 

 

For negative tracks :

• the splitting is less seeable because of the errors bars(Fig III.a)
• for the all pT range [0,10], we see clearly the splitting (Fig III.b)

 

part IV :

for positive and negative tracks on the same plot :

part IV 

in this part, I remove one aspect of the new St_sls_maker.cxx, that is to project simulated tracks, associated to GEANT SSD hits, to the geometry that is used.

Running on simulation only, it should have no effect since the geometry used by GEANT is the same in reco.

Then the projection is correct.

It has been implemented for the embedding mode, where geometry from real data is used.

Fig IV.a : delta_pt : pT(RECO) -pT(MC)

Fig IVb : sigma_pt : pT(RECO) -pT(MC)

part V : comparison with old code.

I took the code from SL07b librairy, so it is prior the recent changes in StSsdSimulationMaker

From the last CVS message, you see that this version is at least 1 year old :

 

 $Log: StSsdWafer.cc,v $

// Revision 1.4  2007/03/27 23:11:48  bouchet

// Add a method to use the gain calibration for the Charge Matching between pulse of p and n sides

I run with the following chain :

 

root4star -q -b bfc.C\(1,500,\"trs,srs,ssd,fss,y2007,Idst,IAna,l0,tpcI,fcf,ftpc,Tree,logger,ITTF,Sti,genvtx,SvtIt,SsdIt,MakeEvent,geant,evout,IdTruth,tags,bbcSim,tofsim,emcY2,EEfs,GeantOut,big,-dstout,fzin,MiniMcMk,clearmem\"

For the negative tracks

Fig V.a delta_pt : pT(Reco) -pT(MC)

The 2 lines are a pol1 fit : within the errors bars, there is a slope for the recent code ; this slope seems not appear with old version.

The point that is interesting is that this code version (SL07b) has the Lorentz effect treatment for simulation  (ie we do and undo the Lorentz effect) 

Fig V.b : sigma_pt