recotracks revisited
DESCRIPTION
RecoTracks Revisited. A quick word on association by hits Clear up a question on the low pt cut-off RS vs CTF Charge of recoTracks The resolutions plots revisited CMSSW 1_6_X - PowerPoint PPT PresentationTRANSCRIPT
UCSD meeting November 2, 2007 R. Kelley
RecoTracks Revisited
UCSD meeting November 2, 2007 R. Kelley2
RecoTracks Revisited
• A quick word on association by hits• Clear up a question on the low pt cut-off RS
vs CTF• Charge of recoTracks• The resolutions plots revisited• CMSSW 1_6_X
– Ran over ~5k events from TTbar release validation sample(/RelValTTbar/CMSSW_1_6_0-RelVal-1189470628/GEN-SIM-DIGI-RECO )
UCSD meeting November 2, 2007 R. Kelley3
Association by Hits
• Definition of association by hits: A matched recoTrack is a recoTrack that has 50% or more of its tracker hits overlap the simulated hits from a simTrack.
UCSD meeting November 2, 2007 R. Kelley4
Fraction of Shared Hits
Note: Most of the associated recoTracks share 100% of their hits with their corresponding simTrack.
UCSD meeting November 2, 2007 R. Kelley5
RS vs CTF
Last time there was some confusion on why it seemed that RS reconstructed many less tracks than CTF
UCSD meeting November 2, 2007 R. Kelley6
A Quick and Dirty on RS Tracks(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley7
RS vs CTF
• This shows clearly that the RS algorithm has a higher minimum seed pT value (≈1.5GeV).
• The ‘bleed off’ is due to misestimate of pT from only a few [seed] hits.
• So in order to be fair to RS, we should make a minimum pT cut.
UCSD meeting November 2, 2007 R. Kelley8
Efficiency (pT > 3.0 GeV)
UCSD meeting November 2, 2007 R. Kelley9
Charge Asymmetry
• There seems to be less negatively charged recoTracks than positive. Why?
+ charge
- charge
UCSD meeting November 2, 2007 R. Kelley10
Charge Asymmetry
• One possible explanation is that π+ reacts differently than π- at lower energies.
• some of the charged tracks are never reconstructed after they react with the material.
• Since σπ-p > σπ+p, more negatively charged tracks are lost in the material.
1 GeV
≈ 60 mb
≈ 20 mb
UCSD meeting November 2, 2007 R. Kelley11
Resolution Revisited• Resolution = RecoTrack – SimTrack (denoted by δ)• All these plots were made with CTF tracks and associated by Hits.• There seems to be a similar trend in variables defined in the r-z plane.
– An initial improvement in resolution followed by a steady decrease.
UCSD meeting November 2, 2007 R. Kelley12
Steady Rise• Touch light affect: as you move in |η|, the distance from the first
measurement point to the point of closest approach increases.• This causes this point to be reconstructed with less precision which
is responsible for the steady rise in the resolution plots for variables defined in the r-z plane.
• The r-z plane reco variables to be less precise as |η| increases.
UCSD meeting November 2, 2007 R. Kelley13
Small |η| improvement (Charge Sharing)
• The improvement for small |η| can be seen from considering charge sharing.
UCSD meeting November 2, 2007 R. Kelley14
More Resolution Plots
• Consider the plots in the r-φ plane.• They suffer from a similar rise in resolution for
increasing |η|.• They do not have the improvement at low |η|.
UCSD meeting November 2, 2007 R. Kelley15
Tracker Geometry
• This can be explained by looking at the geometry of the tracker.
UCSD meeting November 2, 2007 R. Kelley16
R-Z plane vs R-Phi Plane
z
2
720
0.3 4x TT
T
pp
p BL N
L2
L1
UCSD meeting November 2, 2007 R. Kelley17
What About pT?
• There is an additional affect for the pt measurement to multiple scattering.
UCSD meeting November 2, 2007 R. Kelley18
Accounting for the Effects
Total Affect
UCSD meeting November 2, 2007 R. Kelley19
Pull Distributions
• Pull = resolution / error
• Since the pull distribution depends also on error, it is a bit more complicated since the error determination may not be fully accurate.
• We’ll get back to you on these.
UCSD meeting November 2, 2007 R. Kelley20
Pull in r-z Plane
UCSD meeting November 2, 2007 R. Kelley21
Pull in r-φ Plane
UCSD meeting November 2, 2007 R. Kelley22
Back up
UCSD meeting November 2, 2007 R. Kelley23
CombinatorialTrackFinder (CTF)(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley24
CombinatorialTrackFinder (CTF)(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley25
CombinatorialTrackFinder (CTF)(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley26
CombinatorialTrackFinder (CTF)(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley27
RoadSearch Tracking(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley28
A Quick and Dirty on RS Tracks(Kevin Burkett)
UCSD meeting November 2, 2007 R. Kelley29
RoadSearch Tracking(Kevin Burkett)
• Above pT~5GeV, efficiencies about equal.• The overall inefficiency dominated below 2 GeV.
UCSD meeting November 2, 2007 R. Kelley30
Efficiency (pT > 3.0 GeV)
UCSD meeting November 2, 2007 R. Kelley31
design of the tracking system (Boris)
Some formulae stolen from books. The L.Rolandi’s one on your shelf contains them
(“Particle Detection with Drift Chambers” by Blum and Rolandi)
UCSD meeting November 2, 2007 R. Kelley32
design of the tracking system (Boris)