Primary and secondary track Performances

1 - First study : coarse primary and secondary comparison.
2 - Second study : comparison for each particle type.

When looking at global results in terms of efficiency and purity the secondary tracks feature lower performances than primary tracks. One reason for this discrepancy can be the different momentum distributions of these two types of tracks (the mean pt of secondaries is lower than the one of primaries). Tracking performances have been studied as a function of the tranverse momentum in order to assess this argument. In order to avoid edge effects (sligh acceptance differences of the layers) only tracks leaving a hit in each SVT and SSD layer are considered. These first results have been obtained with three events from a set of standard density cocktail events whereas 5 events have been used for the second set of results

1 - First study.
The table below summarizes the global results cumulated over the 3 events. The figure below shows the efficiency and ghost track probability for primaries, secondaries and all the tracks. The tracking efficiency curve for secondary particles does not exhibit a trend different than the primary efficiency curve. The global result differences can thus be explained by the momentum distributions and is fortunately not due to a particular (bad) feature of EST such as the initial track refitting with the primary vertex.

Global Results
All Primaries Secondaries
Ideal tracks 2978 2403 575
Good tracks 2615 2152 463
Bad tracks 251 135 116
Efficiency 87.8 % 89.6 % 80.5 %
Ghost fract. 8.7 % 5.9 % 20.0 %

2 - Second study.
When looking at the especially high ghost contamination in the secondary tracks , we discovered that a large fraction of these wrong tracks are muons. These muons are coming from pions decaying often at large distance and with track parameters close to their parent track parameters (small kink). The reconstructed TPC tracks are tagged as muons but pick the SVT/SSD hits of their pion parent during the tracking phase (the tpc track itself in some cases contains a fraction of tpc hits belonging to the pion). In the first study these tracks were flagged as bad tracks and we decided in the second step to consider them as good tracks if they collect their own hits or the hits from their charged parents. However, at this moment the evaluation considers both the muon and its parents as ideal tracks.

The following results are based on this prescription. These results are a compilation of the numbers printed out by the evaluation method. The number of ideal tracks are not corrected for double counting of the segmented tracks (except for the "all tracks" results). The efficiency values reported should be considered as lower limits. One can expect the real efficiencies to be of the order of 10% higher in overall, higher by 5% for the primaries and 15% for the secondaries. Since the number of good tracks counted twice is very low (zero in this sample) the ghost fractions can be considered as correct. For all tracks and each specie, the transverse momentum distributions the effiency and ghost fraction curves are plotted on the figures below.
Detailed Results
All Tracks Electrons Muons Pions Kaons Protons
All Prim. Seco.All Prim. Seco. All Prim. Seco.All Prim. Seco. All Prim. Seco.All Prim. Seco.
Ideal tracks 544541321313 67067 3550355 40743364710 4894890 457279178
Good tracks 42743558716 21021 74074 33512876475 4344340 394248146
Bad tracks 267164103 24024 12012 18313350 16160 311516
Efficiency 78.5%86.1%54.5% 31.3%\31.3% 20.8 %\20.8% 82.2%85.5%66.9% 88.7%88.7%\ 86.2%88.9%82.0%
Ghost fract. 5.9 %4.4 %12.6% 53.3%\53.3% 14.0 %\14.0% 5.2%4.4%9.5% 3.5%3.5%\ 7.3%5.7%9.9%
Real Ideal 488239381097\
Real Effi. 87.5%90.1%65.0%\

All tracks :

Electrons :

Muons :

Pions :

Kaons :

Protons :


Last modified: Fri Nov 24 17:15:53 CET 2000