• Add Tracking POG link in main CMS twiki
  • Survey among TRK institutes
  • Skims or filters for signal events (loose and tight) for track reco validation

Physics Coordination

  • Update Workbook
  • Tracking POG public twiki

--++ Data Taking conditions

  • Understand DBX distributions
    • DBX vs BX

--++ Analyses to be followed/commented

Possible papers

Track Reconstruction

  • iterative tracking
  • efficiency
  • outlier rejection
    • start discussion
  • fakes
  • momentum resolution?
  • generic track distributions?

Tracking Studies


  • how do the pile-up events without reconstrcuted PV look like ?

Study Probe Tracks

  • properties of probes from tag and probe method applied to J/psi and Z

Beam background

  • mix collision events with beam background events and check track distributions

Beam Spot and Primary Vertex

Distributions of PV-BS positions
The tails are completely different between data and MC. Understand why it is so:
  • Beam Spot time dependence and LS merging algorithm ?
  • Beam Spot calibration uncertainties? Test by repeating BS calibration on MC with the same statistics
  • Beam shape?
  • PV resolution tails? Check with splitted PV analysis


EDAnalyzer to study reco and true primary vertices collections

Pixel tracking and vertexing

Pixel track IP resolution

  • data vs DESIGN and STARTUP
    • use mixed PV resolutions
    • discss with alignment group
  • compute delta(1/pt)

Pixel Vertex cut tuning in FinalTrackSelector

Pixel vertex cut tuning in FinalTrackSelector to optimize response to pile-up. A Z->mumu event has been shown (tracking 958) with one muon not highPurity because its pixel vertex was reconstructed but did not pass the selection cut (prob>0.01). Possible investigations:
  • study pixel vertices as Wolfram studied primary vertices in pileup events
  • correlate pixel vertices with primary vertices
  • look for events with many tracks with low quality and/or from late iterations
  • correlate events with too few "good" pixel vertices with anomalous rate of low quality or late iteration tracks
  • review new selection cuts once the pixel track error has been fixed
  • something else

Tracing efficiency and fake rate

  • Tracking efficiency in jet with data in data embedding
  • Fake rate from embedded jets

Efficiency from J/psi T&P

  • followup the forward inefficency
    • only in onia selection?
    • pixel hits?
    • outlier rejection?

MC tuning and Minimum Bias

  • Review event selection: BSC, HF, ZeroBias


  • Proper simulation of beam pipe (and pixel detector) geometry
  • Improve data/MC mixing
  • Fastsim vs Fullsim
  • Beamspot simulation parameters in DB
  • IOV-based MC conditions


  • Shared clusters: improve treatment
  • Review PV reconstruction configuration: z separation = 2 mm
  • PV reconstruction: annealing fitter
  • beam spot with LS-based IOVs: validation of algorithm
  • pixel vertex in HLT
  • Use information about inactive pixel detector parts
  • allow bowed and kinked sensors

Acceptance vs PV z position

  • check which iterations are affected
  • introduce monitoring of dependence on z

Use actual sigma Z in DB for reconstruction

It affects:
  • Tracking region acceptance if n*sigma option is used
  • FinalTrackSelector acceptance when no pixel vertex
In the past (details to be recovered) a change like this caused a drop in the V0 acceptance:
  • lower pixel vertex efficiency? (not optimized Divisive, minimum bias)
  • smaller sigma z (about 2.8cm) than now (about 6 cm)
Giuseppe's tests: To Do
  • add in the release the proper absolute z cuts for 0th and 2nd step
  • add in the release the proper absolute z cut and PSet block for pixelTracks reconstruction
  • prepare a GT with actual sigma z for RelVal
  • analyze Giuseppe's events
  • review all the values for tracking regions

Thresholds on cluster and seed mutliplicities

Monitor cluster multplicity and track distributions on logerror events Statistics (361p4)
Dataset events Reco conditions events with tracking errors TooManySeeds TooManyClusters Lumi
/MinimumBias/Run2010A-Skim_logerror-v2/RAW-RECO 5005 standard 3263 3214 372  
new config 495 488 7  
/MinimumBias/Run2010A-Skim_logerror-v6/RAW-RECO 483227 standard 318033 (or 318035??) 306098 14067  
new config 4106 3987 119  

To Do
  • Monitor number of seeds (when threshold is set high)
  • TooManySeeds errors are possible from CkfTrackCandidateBaseMaker: follow it up

Pixel detector read out inefficiencies

  • How to monitor them?

Reported problems

  • Investigate loss of tracks from b-jets in data/data mixed events
  • Tracks with ndof < 0 (fixed already?)

Oddly reconstructed tracks (muons) at eta 2.1-2.4

Hypernews threads where the issue has been discussed Talks where the issue has been addressed:

Questions of mine:

  • is forward/backward asymmetry observed in 2009 still present?
  • why only muons ?
    • decays in flight?
    • bias toward leptons for high pt tracks?
    • we did not look for hadrons like these
    • Kevin's answer: hadrons with bad delta(pt) are already rejected
  • Is a large pt error a sensible signature? If so, why?
    • is there any peculiarity in the hits associated to these tracks?
  • What is the impact of the TEC/TID-like track reconstruction: 1/pz is the actually measured parameter.
  • Are the missed hits a key feature? What's the cause?
    • APE orientation is not optimal
    • error linearization affects significance computation
Ideas after Tracking POG meeting:
  • Check particle gun results with MC without defect to check correlation with "hot spots" in the distribution of the track last points for short tracks
    • why hot spots are correlated with defects since they are "known" defects? 2nd order effect?
  • Check the impact of IDEAL vs STARTUP alignment
  • Investigate events with lost hits and understand the cause of the loss:
    • pattern recognition/track extrapolation
    • outlier rejection
    • residual significance computation
    • Is it possible to use STA muon part to try to recover missed hits?
  • Perform the study focusing on 1/pt and 1/pz
  • Impact of APE
  • Interplay with TEC alignment (see here)
  • How many times two tracks are reconstructed from a single particle? is it correlated with bad tracks?

DQM and data distributions

  • Study performance with various failure scenarios
  • Forward acceptance: eta distribution data vs MC in 7 TeV data
  • Review of tracking DQM validation (Patrice)
  • More automatization of tracking validation for PVT
  • monitor pixel cluster compatibility quantities (seeding and outlier rejections)
  • Detailed track distributions: eta, phi, pt, Nhit

Look for 2.5 ns displaced vertices as requested by RunCoord

  • As explained by Marcin, pixel track reconstruction with origin not inside the pixel first layer(s) is not expected to work.

Conversions and Material Budget

  • Results with altered material MC
  • List of possible improvements in geometry description

Photon conversion analysis

  • Conversion Gun
  • Review results with Nhit > 4: where does the excess come from
  • Understand number of selected conversions with six iterations and with six+two iterations
  • review efficiency computation and possible double counting with associator by position
  • investigate loss of efficiency in eta: seeding? Nhit>5 ? poor electron tracking with material ?
  • investigate delta(z) distribution/bias
Vertex fitter
  • Check impact on Ks mass with centered beam spot
  • Check status of conversion RECO collection
  • Study constrained fit for endcap tracks
  • Investigate better paramerization of the constraints when vertex is not on tracks.

Conversion rate discrepancy at eta > 2
  • Check description of material at the end of the TIB section of the pixel support tube

Momentum Scale

Phi resonance analysis

  • Can the difference between phi vertex and BS center introduce a systematic bias? What about phi from displaced vertices?
  • Use K mass hypothesis in the phi track reconstruction

Ks mass analyses

  • Recall Ks charge asymmetry discussion in 900 GeV data
  • Investigate Ks mass discrepancies among data/MC/truth in the endcaps
  • Ks analysis from ULB: TRK DPG/POG 2010/7/30
  • Nick's analysis
    • investigate tracks with large distance between vertex and first layer
      • why band at constant eta? Does it correspond to Ring3? Seeding effect?
    • what do the track with late first hit correspond to? displaced vertex or missed hit?
    • re-use particle gun for conversion studies
Ks mass azimuthal dependence
  • Interplay with material distribution
  • Position of the (displaced) beam spot used as reference point for the track parameters
    • Try to use centered beam spot MC
  • Try to replace track parameters computed at BS with track parameters computed at innermost point
  • Investigate dependence on decay length cut
  • magnetic field maps used to do propagation back and forth
  • study bias of fitted track parameters w.r.t. MC truth
Ks mass and hint for a B field misalignment
  • My results
  • try to simulate and check the pattern
  • endcap track puzzle
  • what if corrections are mapped in MUSCLE?
  • try a 3D B field map
  • coordinate with alignment group

Tracker Local Reco

Signal Collection

My model

  • Signal propagation along strips
  • Review old results and comparison with measurements
  • Effective LA model

LA analysis

  • Detector timing in 2010
  • step in LA during 2010 run: Yuri's results
    • Impact on alignment if not corrected for
    • correlate with timing change and compare with predictions
  • Compute effective thickness for Lorentz Angle cluster position correction
  • patches in TOB map with different LA: timing?

Ben's analysis

  • TEC back plane corrections
    • sign flip in Ring 7
  • TEC results
  • Use 2010 cosmics during collisions


TEC results with MP and 2010 data

Possible consequences
  • Momentum scale analysis
    • understand why Ks mass is not very much affected (see Jula's results )
  • Z mass difference with and without new geometry
  • single track momentum difference
  • very forward muon reconstruction
  • is there any effect in MC?

Kinked and bowed sensors

  • Patterns in the sensor sagitta maps
  • Consistency of the parameters for kinked modules
References to results

APV cycle noise

  • Fix APV phase producer: Done ?
  • Understand noise in event 2 correlated to position in the cycle of event 1

Signal Normalization

  • input tickmark correction
  • review impact of dead/masked strips in low dE/dx tracks
  • Reconcile DPG/QCD gain calibration: relativistic rise in very forward sensors

Timing run studies in 2009/2010

  • AOH setting and APV setting during timing runs
  • stability of timing run results
  • look for gainscan runs
  • correlate with particle calibrations

116626 116634 116608 116635 116636
117801 116634 116608 116635 117801
120696 120650 120651 120649 120696
121763 121762 121773 121768 122380
122710 122720 122707 122706 XXX (124260?)
122710 123242 122707 123263 YYY (124261?)

TRKPFG common job

  • better path creation

Analyzers to be written or improved

  • to compare two (reconstructed) vertex collections
  • to study vertex reco efficiency and purity with MC truth
  • to study lepton tracks
  • trackCount
    • 2D plots
    • iter distributions

-- AndreaVenturi - 05-Jan-2010

Edit | Attach | Watch | Print version | History: r50 < r49 < r48 < r47 < r46 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r50 - 2011-01-28 - AndreaVenturi
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback