Setup

SetupProject Brunel HEAD --nightly lhcb-head --build-env
SetupProject Brunel HEAD --nightly lhcb-head 
for package in Pr/PrAlgorithms  Pr/PrFitParams  Pr/PrKernel  Pr/PrMCTools Pr/PrUtils  Pr/PrVeloUT; do
getpack ${package} head
done
gaudirun.py Brunel_newFTGeoDownstream.py Upgrade-RichPmt.py | tee some_file.out

• Change the order of Match and Downstream in RecoUpgradeTracking.py

Suggested studies:

• Mail from Steve Blusk:

Dear all,
  I finally got a chance to look at the note that Sasha sent around... Some basic questions, and
suggestions:
(1) Do we know that the better Y segmentation in UT is used properly in PatDownstream?
(2) It would be very nice to see what happens if one removes hits used on long tracks and VeloUT tracks
that have "good" chisquare  (whatever those values are)... Once really wants to get rid of those
hit combinations that are "lined up"...
(3) It would be nice to confirm that the right set of hits in UT were among the possible choices, and
understand why the wrong set of hist was chosen. If there are 25 possible track stubs then, it's hard to imagine
we choose the correct one with any reasonable efficiency.

I'm assuming the real issue is that there are too many hits in the search region as you move to the forward rapidity.
Since most of the hits in UT should be either from long or upstream tracks, getting rid of them "should" make a big difference.

Cheers
Steve

• Try and re-tune the Z magnet parameters and re-measure effs.
• T-seed P resolution.
• Think about Hough transform of the forward tracking (idea from Sascha).
• How many T-seeds to we have?
• Plot the efficiency versus radial impact parameter
• What is the momentum resolution of the T-seeds?

My commits:

• Package: Tf/TrackSys, Revision number or local tag: r165492 (Move the matching algorithm to before the downstream in RecoUpgradeTracking.py).

My own questions

• Apparently the T-seeds have been through a full Kalman fit before they are given to the downstream tracking. How much time does it cost to do the fit? How would the downstream tracking performance be degraded by not doing this first?
• What documentation exists for the UT?

Why we care about downstream tracking.

• Compete with the B factories on sin2beta with B -> J/psi Ks.
• Ks -> mumu discovery
• B0 -> KsMuMu angular measurements
• B->KsHH, B->KsKs
• D->KsHH, D->KsH (note that we may be limited by the K0/K~0 interaction asymmetry uncertainty for DD Ks)

Discussion with Sascha and Michel

Notes:

What track states are available?

• /afs/cern.ch/lhcb/software/releases/LHCB/LHCB_v36r3/Event/TrackEvent/Event/State.h

Documentation

• Sascha's thesis
• O. Callot, Downstream Pattern Recognition, CERN-LHCb-2007-026
• Manuel's thesis (pat-seeding)
• Sascha, Stephie, Update on downstream tracking, LHCb-INT-2009-032
• Note on clone killing

Interesting talks

• JC Wang, Upgrade tracking, 7th November 2013.
• Adam Davis, UT software meeting, 4th October 2013.
• Adam Davis, UT software meeting, 20th September 2013.
• Adam Davis, UT software meeting, 14th June 2013.
• Adam Davis, UT software meeting, 10th May 2013.