=============================================
Minutes of Tracking Meeting, November 1, 2007
=============================================
taken by U. Husemann


Antonio Boveia: Backward track refit and COT hit search
=======================================================
- 40% resolution gain by single Kalman track refit, now
  slightly better than IO tracks
- COT hit search does not currently work: lose 60% of events 
  (under investigation)
- High failure rate of linking stereo segments -> under investigation
  (done in BackwardTracking before search for Silicon hits)
- Investigation of IO forward tracking
  o All IO tracks: large background
  o Cut on zero bin of hadronic energy: cleaned up
- Tested other cleanup variables (just from tracking): 
  o bcd0 is flat for events with zero hadronic energy
  o Missed hits, i.e. difference between observed and expected Si layers
    (suggestions: exclude L00, only 70% efficient)
- Suggestion: check if forward OI tracks would profit from quality cuts?
- Question: should we store bad quality tracks in events 
  (120 instead of 240 tracks)? -> Rick Snider will check


Matt Herndon: Forward tracking
==============================
- Tested several changes to new forward tracking
- Overall CPU consumption only 20% more than Gen6
- Some regions: up to 8% efficiency loss, 
  but 80% of tracks removed (probably junk)
- Changes to PadTrackMaker and IOTrackingModule: 
  o Require only one missed r-phi hit (not including L00)
- Changes to Kal tracking:
  o Seed only from highest pt vertex or vertex with sumpt > 10 GeV
  o Require >=20 ADC counts
  o Pattern reco seed hits must be good
  o Several other details of hit assignment
  o At least 51% of possible hits found
  o Seed pt cuts: 0.2 GeV (SiSA), 1.0 GeV (FWD)
- Plans:
  o Learn how to use SiExpected information in Kal tracking
  o Add tcl switches for the algorithm changes
  o Check in changes
  o Test within Antonio's framework and Ulrich's timing test


All: Next Gen7 Pre-Release
==========================
- Aart: Si looks ok, sign-off next week
  o Compared Period11 processed with Gen7 and Gen6 from 2006
  o Findings: similar residuals, ladder-by-ladder variations correlated
- ISL alignment: expect gain for new forward tracking algorithms 
  from from good forward ISL alignment
- Tracking changes are major part of next release
- Pasha updated muon finder (drift time ambiguity solved with track)
- Doug's plan: reprocess electron and muon samples with all fixes in
- Aseet's dE/dx container is independent of tracking changes, 
  but nobody seems interested in testing?