1. Status of tracking Thijs Cornelissen, Genova
Thanks to Claudia Gemme and Andrea Coccaro

2. Tracking algorithms
Default `NewTracking’ runs several different algorithms:
Inside-out tracking: start from spacepoint seeds in pixels & SCT
Backtracking and TRT tracking: start from TRT segments, use only hits that were not already used by inside-out
LowPt tracking: use remaining silicon hits to find tracks with 100 MeV < pT < 500 MeV
Standard set of tracking cuts designed to give best trade-off between efficiency and fake rates in busy events (e.g. ttbar)
Commissioning settings much more loose:
Allow more holes on track
Allow worse track fit quality (20 instead of 7), loose cuts on ‘outlier’ hits
Use conservative measurement errors (cluster width/√12)
Further increased by error scaling where needed

3. First collisions
First collisions on Nov Challenging for tracking:
Pixels off
SCT at 20V (only 50% efficient)
Momentum unknown (solenoid off), therefore cannot correct for material effects (scattering)
Limits quality of reconstructed tracks
Tracking ran with very loose cuts
Accept tracks with large impact parameters and/or large number of holes (i.e. sensors without a hit)

4. Pixel and SCT hits
Number of pixel and SCT hits on combined tracks, very good agreement between data and simulation
Can distinguish SCT endcap disks
Smooth variation in number of pixel hits due to disabled modules (up to 4%)

5. Tracking efficiencies
Reconstruct ‘tracklets’ in each subdetector, check efficiency relative to combined track
Pixel tracking most efficient, TRT least efficient
Due to problems with TRT tracking algorithm, see next slide

6. TRT tracking
CTB xK
Compare standard TRT tracking for collisions (‘xK’) with modified cosmics tracking (‘CTB’)
Higher efficiencies and lower fake rates with CTB
Modified version of cosmics TRT extension also under study

7. TRT tracking (cont’d) Rtrack (mm)
TRT simulation does not reproduce data well: less hits on track, worse resolutions (especially tails)
Default cuts in tracking are rather tight, gets rid of most of the tail
Commissioning settings much more loose, caused bad fit qualities in simulation (average c2/ndf around 2)

8. Silicon tracking efficiency
Check silicon tracking efficiency by using pixel tracklets to probe combined track collection (`SCT extension’)
Very good agreement between data and MC up to |h| ~ 2.2
Discrepancy at large |h| not yet understood

9. dE/dx per le tracce
La dE/dx si calcola come media troncata del rilascio di energia di tutti I cluster appartenenti alla traccia. Troncata perche’ si escude il cluster con rilascio di carica maggiore (in modo da tagliare le code della Landau).
dE/dx accesibile in InnerDetector/InDetRecTools/PixelToTPIDTool
Puo’ essere usato per studiare risonanze come il L, f e, piu` tardi, l’identificazione di esotici (low b)
Clusters cuts:
Barrel and EndCap
Residuals: abs(pullPhi)<5, abs(pullEta)<5
Spatial Incidence angle a : cos(a)>0.1 (i.e. a<1.47)
Cut on Incident Angle in rf plane: none
Tracks cuts:
Tracks: Good clusters >=3

10. PID
Runs ,
Per adesso solo una rozza identificazione per poter fare I primi plot. p k p p k
Positive: 9800 (p), 3000 (k) Negative: 2800 (p), 1800 (k) 10

11. Online tracking: pT spectrum
In 2009 run, HLT including SiTrack was run in pass-through mode
pT spectrum of reconstructed online tracks (red) and offline tracks (black)
Good agreement (online tracking cuts at |pT| > 1 GeV)

12. Online tracking: z0 spectrum
Vertex shifted along z during the run, due to LHC adjustments
Shift detected both by SiTrack and IDScan

13. Online tracking: efficiencies
Efficiencies of SiTrack (red) and IDScan (blue) relative to offline tracking (black)
Inefficiencies in SiTrack thought to be due to initial misalignments, to be checked with upcoming trigger reprocessing
Trigger reprocessing uses improved alignment (esp. endcaps)

14. Conclusions Tracking worked quite well from beginning
Items to be understood/improved:
TRT tracking efficiencies and fake rates
Silicon tracking efficiency near |h| = 2.5
Need to work towards removing commissioning settings
Would benefit from improved resolution (esp. pixels)
Better rejection against fake tracks
Online tracking works
Need to study performance with updated alignment
Will soon be needed to select events
dE/dx tool now available
Need to study efficiencies and fake rates of PID using data and MC

15. Backup

16. Tracking efficiency in MB analysis