TrAnalyse

TrAnalyse is the code that monitors the efficiency and ghost rate of tracks found and fitted by other routines. The output is a table of the form:


      Type       Ghosts      Bad Chi2     with hits  3V+3Stations  3V+3T+>5 GeV   V+T+B-child   K0S<-B,3V3S
        MC                              6747           657           420            55             4
 VeloROnly   801  30.7%      0   0.0%   1811  26.8%    641  97.6%    415  98.8%     54  98.2%      3  75.0%
                                         762           339           221            31             1         clones.
      Velo   284  15.1%      0   0.0%   1597  23.7%    616  93.8%    401  95.5%     53  96.4%      2  50.0%
                                          78            35            23             2             0         clones.
   Forward    77  11.5%    265  39.5%    594   8.8%    548  83.4%    364  86.7%     48  87.3%      1  25.0%
                                          17            14             8             1             0         clones.
      Seed    74  12.6%      2   0.3%    515   7.6%     72  11.0%     36   8.6%      4   7.3%      2  50.0%
                                           2             1             1             0             0         clones.
     Match    36  53.7%     17  25.4%     31   0.5%     21   3.2%     11   2.6%      2   3.6%      1  25.0%
   KsTrack    79  42.0%      1   0.5%    109   1.6%     25   3.8%     14   3.3%      1   1.8%      1  25.0%
Unique Fwd    77  11.5%    263  39.2%    594   8.8%    548  83.4%    364  86.7%     48  87.3%      1  25.0%
                                           5             5             2             0             0         clones.
Unique Mat    36  53.7%     17  25.4%     31   0.5%     21   3.2%     11   2.6%      2   3.6%      1  25.0%
Unique Ks     60  40.0%      0   0.0%     90   1.3%     23   3.5%     14   3.3%      0   0.0%      0   0.0%
     Final   504  20.5%    285  11.6%   1951  28.9%    636  96.8%    410  97.6%     53  96.4%      3  75.0%
                                          80            43            25             4             1         clones.

TT1 :   MC Clusters on track :     any      >0      >1      >2      >3
Forward     number of tracks :     598     566     549     532     442 |   2103 clus 3.52/track
Forward  with >n MC clusters :     598     336     292      99      30 |    757 OK   1.27/track Eff.   36.0%
Forward  some wrong clusters :      73      41      40      40      28 |    106 Bad  0.18/track Frac   12.3%
Forward       efficiency (%) :   100.0    59.4    53.2    18.6     6.8

Match       number of tracks :      31      20      17      13      12 |     62 clus 2.00/track
Match    with >n MC clusters :      31       9       6       2       0 |     17 OK   0.55/track Eff.   27.4%
Match    some wrong clusters :      11       0       0       0       0 |     22 Bad  0.71/track Frac   56.4%
Match         efficiency (%) :   100.0    45.0    35.3    15.4     0.0

KsTrack    number of tracks :     184      97      97      91      81 |    369 clus 2.01/track
KsTrack with >n MC clusters :     184      90      89      75      61 |    315 OK   1.71/track Eff.   85.4%
KsTrack some wrong clusters :      99      12      12       8       6 |    207 Bad  1.12/track Frac   39.7%
KsTrack      efficiency (%) :   100.0    92.8    91.8    82.4    75.3

Track types

The code tracks the number of MC tracks that fill several categories of increasing usefulness.

Ghosts
Tracks in which less than 70% of the hits are from the same track
Bad Chi2
Tracks with poor fits, so rejected after the Kalman fit
with hits
Tracks found from MC particles leaving at least one hit a detector
3V+3Stations
Tracks found from MC particles leaving at least 3 VELO R and phi clusters and 3 T station clusters
3V+3T+>5 GeV
As above but with a true momentum of greater than 5 GeV
V+T+B-child
As 3V+3Stations and the MC truth says it was from a decay of a particle with a b quark
K0S<-B,3V3S
A Ks with at least 3 VELO R and 3 phi clusters and 3 T station clusters
clone
One of two or more tracks created from the same MC particle

Pattern Recognition Algorithms

The type column lists the algorithms that generated the tracks these are:

MC
Number of true MC particles fitting the category
VeloROnly
The number of 2D (r/z) VELO tracks found (not normally printed)
Velo
Number of 3D Velo tracks found, uses as seeds for later tracking stages
Forward
Long tracks found by extrapolating Velo tracks into T and TT stations and adding hits
Seed
Tracks found by pattern recognition in the T stations only
Match
Tracks found by pairing Velo and Seed tracks that extrapolate to the same point, TT hits are also added
KsTrack
tracks found by adding TT hits to seed tracks to find Ks decays after the VELO

After the different pattern recognition algorithms are run a CloneKiller is run that removes multiple copies of the same track. The tracks with the most hits and best chi^2 are kept and the others discarded. The Unique here refers to the fact that the clones of this track were removed.

Unique Fwd
Tracks from the Forward algorithm passing the clone killer
Unique Mat
Tracks from the Match algorithm passing the clone killer
Unique Ks
Tracks from the KsTrack algorithm passing the clone killer
Final
All tracks after the clone killer, sum of the above unique tracks

-- DavidHutchcroft - 09 Aug 2005

Edit | Attach | Watch | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r3 - 2005-08-12 - DavidHutchcroft
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb 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