This study shows chi2/DOF for best TPC tracks using current default BFC. (updated, June 2009)

SECTION A ---------------------------------

1. track selection preferring primary tracks pool of global tracks:
   • use only global tracks matched to BTOW,ETOW, or central membrane
   • require nFit/nPoss>51%, and Rxy @ DCA to X=Y=0 below 3 cm
   • require PPV finds valid vertex along Z direction, require track DCA delZ <3cm
   • require global pT in range [0.8, 8.0] GeV/c
2. Data sets:
   • real events from pp500 run in 2009
   • M-C Pythia events, QCD w partonic PT>15 GeV/c
3. Conclusions about assumed TPC hit errors for pp 500 data production:
   • real data : errors should be enlarged by  24% (by sqrt(1.53)) , unless TPC alignment is the cause 
   • M-C : errors are fine 

Fig 1, real pp 500 events from L2W stream

Fig 2, M-C Pythia events pp 500, ckin3=15. Expected sig(chi2/dof)=sqrt(2/Ndof)=sqrt(2/30)=0.26. Actual RMS=0.51. But RMS!=sig. 

 BFC chain details:

• pp500 DATA: root4star -b -q 'bfc.C(10,"pp2009a ITTF BEmcChkStat QAalltrigs btofDat Corr3 OSpaceZ2 OGridLeak3D beamLine -VFMinuit  VFPPVnoCTB -dstout -evout","/star/data03/daq/2009/Weve/083/10083013/st_W_10083013_raw_6180001.daq")' 
• M-C: root4star -b -q bfc.C\(10,'"trs -ssd y2009  Idst IAna l0 tpcI fcf -ftpc Tree logger ITTF Sti StiRnd  -IstIT NoSvtIt SvtCL,svtDb -SsdIt MakeEvent McEvent geant evout geantout IdTruth  bbcSim emcY2 EEfs bigbig -dstout fzin -MiniMcMk McEvOut clearmem -ctbMatchVtx VFPPVnoCTB eemcDb beamLine"','"/star/institutions/mit/balewski/2009-PPV-3D-simu/setD1_QCD500.fzd"'\)

SECTION B ---------------------------------

Evaluation of stability of chi2/DOF for best quality  global tracks (selection as sect A.1 )

I have used non-liner SC correction (lets call it ver 1).

Select 89 runs from 6 fills (see Attachment 1) , approximate run QA. First 3 fills have lower luminosity (zdc rate 60-120 kHz), last 3 fills have zdc rate 100-170kHz. For every run only 120 events is processed. Only L2W-triggered events are used.

Fig 3. Top: # of tracks per run (depends on the length of run). X=run index
Bottom : X=run index, Y=chi2/DOF. Every column was reweighted to  match integral of the 1st column.

Fig 4. Improved SC correction ver2. Now Joe used 64 runs equally spread over full pp 500 data taking period.

Fig 5. Yield of of reco global tracks per run & chi2 for 550 runs spanning whole pp 500 data taking. List of runs is in attachment 2.

 Observations:

• Chi2/dof decreases slightly with luminosity within fill, see e.g. F10412, 
• few fills have tracking problem

New iteration of SC ver 3

Fig 6.  SC ver 3 from Joe:

Fig 7. Resulting chi2/dof:

Top: # of qualified high PT primary tracks per input trigger event (L2W only).  This removes problem with too few events in some daq files. BFC runs on 150 or less events and Y axis get rescaled according to # of input events.

Middle: ZDC rate in kHz . I have no data for the 3rd fill and hardcoded 70kHz.
Note, ZDC rate changes a lot but chi2/dof (bottom) stays mostly very flat. The # of high PT tracks goes down at the beginning of the fill.
This smalls like triggering on

Bottom: chi2/dof per run, Z axis rescaled to preserve the integral for given run.
Mean chi2/dof is shown as black histo.