=============================================
Minutes of the Tracking Meeting, June 6, 2007
=============================================
taken by U. Husemann


Ben Brau & Antonio Boveia: Studies of Missing ISL/L00

- Spokes asked for study on impact of missing ISL/L00 on B tagging to show to P5
- Checked tag rates and impact parameter significance
- Denominator for efficiency measurements: traditional "taggable" may normalize away efficiency drop, "all" gives more honest accounting
- Examine heavy flavor rich sample of electron jets in Jet100 data set vs. kinematics (Et etc.), also checked away jet
- Impact on impact parameter significance: no L00 causes resolution loss ("sloshing" of decay length measurement close to cut), no ISL causes overall reduced efficiency
- Relative change in b tagging rate without ISL and L00: 85% (approx +/-2%)

- Ulrich reports that also Higgs group has done studies on ISL/L00 impact on H->WW:
  o 40-50% reduction for forward muons (CMIOPES category is using IO tracks)
  o CMIOPES is only 8% of total yield, so total acceptance hit is 3.7%

- Antonio studied impact on Phoenix tracking
  o Efficiency w/o ISL and L00 drops from 79% to 74% for Phoenix in |eta|<1.8
  o Efficiency loss because of both fewer axial small-angle stereo hits w/o ISL

- Closer look at SecVtx algorithm (Conveners of B tagging group present):
  o Pass 1 track selection for SecVtx only uses only 70% of good tracks (according to Matt's definition for OI tracks in the Silicon)
  o Missing ISL/L00 reduces this number to 60%
  o Track chi^2 for Pass 1 tracks shows a lot of low chi^2 tracks in the set of good tracks

- Discussion with B tagging conveners: 
  o Can SecVtx track selection be improved by cutting on track chi^2?
  o How can such a change be tested, included, validated, ...
  o Ben et al. will try to include the change and look at rates in jet data


All: Discussion on Int9

- Ben's study saw many jobs crashing: not related to int9, but b tagging
- Antonio: backward tracking not yet validated
- Koji has sent around plots on COT validation: no surprises
- Oliver reports that Aseet promised final COT segment merging changes by the end of the week
- Matt promised to run his validation code