TrkLoopFix

1
TrkLoopFix

• A module in TrkFixup
• Gerry Lynch
• October 1, 2005

2
What is a Looper?

• Simple definition
• A track that goes at least a full turn in the
  detector.
• Alternate Definition
• A track that is fragmented into two or more
  tracks, which share many layers.
• Operational Definition
• Pairs of low momentum tracks that are 180 degrees
  apart.

3
Looper Azimuth difference

• Before TrkLoopFix a selection is made that keeps
  low momentum track pairs that are within 0.2
  radians of 180 degrees in both azimuth and dip.
  These go in the work list..

There is a small (7) background under the
peak.
4
Looper Sample

• We see that most of the tracks that are in the
  looper sample are tracks that hit the outer wall
  of the DCH.
• The preselection for the work list used by
  TrkLoopFix has a cut at 100 cm,

5

• Looper Multipicity

Most looper candidates have two tracks
The number of tracks for each looper size is
shown here.
6
Apogee, Parigee

• In the looper at the right,Track 1 is first.
• Tracks 1 and 2 form an apogee pair.
• Tracks 2 and 3 form a parigee pair.
• Our goal is to keep 1 and reject 2 and 3.
• The procedure for classifying the looper
  candidates starts with labeling each pair as
  apogee. perigee, neither, or either

In any looper pair one tracks is going out and
the other going in. But we fit the tracks as if
they were going out. So one is wrong.
7
Method 1

• When we plot Sum against Sum(d0), we See a clear
  band of apogee pairs.
• Very few are rejected.
• The vertical black band is a mixture of apogee
  and perigee pairs.

8
Method 2

• In this plot of the calculated number of
  turns from z02-z01, omega, and tanDip,we see
  clear apogee and (impure) perigee bands at dip
  angles above about 0.15.
• The number rejected is small, but larger than for
  method 1

9
Beam-line Chi-squared

• Pairs with both tracks primary are not loopers.
• Pairs with one track primary are probably apogee
  pairs.
• Note that many of the good apogee pairs have no
  primary track.

10
Position Chi-squared

• We get chi-squareds for the comparison of the
  orbits at parigee and apogee.
• The full chi-squared is not very useful, but the
  position component is valuable.
• This shows how these chi-squared agree with
  method 2 decisions.
• This position chi-squared is used to classify the
  pairs that were not decided before.
• 
  

11
Procedures

• The result of the classification of single pair
  loopers is
• 67 apogee 17 neither ( many of
  these are not loopers)
• 15 perigee 1 either
• For a single pair looper we keep the best track
  and discard the other
• for apogee and perigee pairs and keep both
  for the others.
• For multi-pair loopers the procedure is to choose
  the best pair
• by a formula that prefers the apogee pair
  with the highest
• momentum. If that pair is apogee or
  parigee, the preferred track in that
• pair is kept and all others rejected.
• For multi-pair loopers that are classified
  either or neither, both
• tracks in the highest momentum pair are kept.

12
MC Checks

• If our apogee pairs are right, the sign of
  curvature will agree with the Monte Carlo truth.
• The present program gets agreement with the MC
  98 of the time, showing that the correct track
  is is being selected.
• For perigee pairs the fraction in the opposite
  curvature sign is 61. I think that this
  should be larger.
• Another check is to compare fitted track
  parameters, To do this one needs to pay
  attention to the track direction and what sheet
  of the helix the parameters represent.

13
MC Parameter Test

• The parameters of the tracks are compared with
  the Monte Carlo truth.
• There is excellent agreement. Multi-pair loopers
  look almost as good.

14
Problems

• As indicated on the MC checks page, the
  performance for parigee pairs is suspect.
• One is tempted to conclude from the MC parameters
  test that things are ok because when the
  chi-squared is small the track was accepted.
• But all the small chi-squareds are from tracks in
  apogee pairs. The parigee pair tracks do not
  have small chi-squareds, probably because they
  are fitted in the wrong direction,.
• One approach to this would be to fit all of the
  parigee candidate tracks and work with two
  possible solutions when searching for best
  solution.
• We will have to refit the retained perigee track
  in any case because it is in the wrong direction.