The CDF Collaboration
March 28, 2003
We present dedicated silicon tracking algorithm for reconstruction of particles created at a significant distance from the primary vertex (e.g. outside of the beampipe). This algorithm is designed to deal with daughters of long lived particles such as or lambdas. Adding the correct silicon information to COT tracks is, obviously, very helpful for improving track parameter resolution. For rejecting background in long lived particle reconstruction it is especially beneficial because the improved track's resolution (mostly z) allows to reject random two-track combinations more easily at the vertex fitting stage.
Long lived particles like or require special tracking (on average they decay just before the last layer of the SVX detector but within the ISL). Adding just a single ISL small-angle stereo cluster to one of the daughters of a or improves the track's z-resolution from 1cm to 2 mm. This improves the determination of the and vertices and consequently the reconstruction of the parent of those particles (i.e., going upstream , or ).
Two special silicon pattern recognitions are used for the daughters of the or . In both of them, in order to compute its inv. mass, hits are added to tracks and energy loss correction is applied taking into account the position of the secondary production vertex for those tracks.
The default pattern recognition has:
The dedicated pattern recognition has:
The default approach might be too agressive for long lived particles.
The dedicated algorithm tends to :
We look ( by ) at the mass, vertex and # hits ( and z) difference between default and dedicated tracks-pairs.
Show z-distance of the two daughter tracks at the default-vertex position in the xy plane () and inv. mass comparison as well.
Compute the dedicated efficiency wrt default.
Look at Z-vertexing, in parituclar, at Z = z0 - z being z0 = z - (wo. vertex constrain).
This section summarizes the numbers which have been blessed for this analysis.
=0.9 MeV (2 MeV width)
= 0.1 mm (2.5 mm width)
= 8.6 MeV (default) / 8.2 MeV (dedicated)
= 0.82 cm (default) / 0.33 cm (dedicated)
Efficiency of the dedicated alg.:
4 % def-track pairs rejected by the dedicated algorithm ( none of those are )
100% efficient in this case + S/N improvement.
= 6 mm (default) / 4 mm (dedicated)
=0.024 MeV (2 MeV width)
= 5 (50 width)
= 4.6 MeV (default) / 4.4 MeV (dedicated)
Dedicated efficiency :
8 % def-track pairs rejected by the dedicated alg. ( drawback: 1/2 of those are )
misaligned z layers are responsible: using z, 90 stereo, rejects ; not using 90 stereo does not; improved alignment in progress.
= 1.82 mm (default) / 1.48 mm (dedicated)
The following figures have been blessed for the dedicated reconstruction studies. More text can be found in CDF note 6337.
Figure 1:
invariant mass subtraction (top left),
vertex position in the transverse plane (top right),
number of hits (bottom left) and number of z hits (small angle stereo and
90 degree stereo hits) (bottom right) subtraction ( candidates
obtained with the dedicated algorithm - candidates
obtained with default tracks).
The sample used corresponds to default track pairs without silicon hits.
Figure 2:
Distance on z of the two daughter tracks at ,
before refitting, obtained with the dedicated tracks (light green histogram),
so tracks that passed through the dedicated algorithm, and
with default tracks (black).
The sample used corresponds to default track pairs without silicon hits.
(default) = 0.86 cm.
(dedicated) = 0.36 cm.
Figure 3:
The invariant mass.
The dedicated (default) algorithm has been used for the light-green (black) histogram.
The starting sample corresponds to default track pairs without silicon hits.
No selection cuts have been applied. = 8.6 MeV. = 8.2 MeV.
Figure 4:
The invariant mass of those default track pairs rejected by
the dedicated algorithm. The starting sample corresponds to default track pairs without silicon hits.
The increasement of the background on the edges is a feature of the algorithm (it
has a shrinking mass-window selection on every different step taken)
Figure 5:
Z = z0 - z distribution;
z0 = z - (without vertex constrain).
The starting sample corresponds to default track pairs without silicon hits.
Dedicated s point better to PV (this sample contains mostly prompts).
Figure 6:
invariant mass subtraction (top left),
vertex position in the transverse plane (top right),
number of hits (bottom left) and number of z hits (small angle stereo and
90 degree stereo hits) (bottom right) subtraction ( candidates
obtained with the dedicated algorithm - candidates
obtained with default tracks).
The sample used corresponds to default track pairs with silicon hits
(in addition an ISL hit should be present in every track).
Figure 7:
Distance on z of the two daughter tracks at ,
before refitting, obtained with the dedicated tracks (light green histogram)
(so tracks that passed through the dedicated algorithm) and
with default tracks (black).
The sample used corresponds to default track pairs with silicon hits
(in addition an ISL hit should be present in every track).
Figure 8:
The invariant mass.
The dedicated (default) algorithm has been used for the light-green (black) histogram.
The starting sample corresponds to default track pairs with silicon hits.
No selection cuts have been applied. See CDF note 6337 for more details.
= 4.6 MeV. = 4.4 MeV.
Figure 9:
The invariant mass of those default track pairs rejected by
the dedicated algorithm. The starting sample corresponds to default track pairs with silicon hits.
The 90 degree z-hits have been used on the black histogram and not
on the red histogram by the dedicated algorithm.
Figure 10:
Z = z0 - z distribution;
z0 = z - (without vertex constrain).
The starting sample corresponds to default track pairs with silicon hits.
Dedicated s point better to PV (this sample contains mostly prompts).
Dedicated reconstruction studies
Text for the blessed web page - CDF note 6337
This document was generated using the LaTeX2HTML translator Version 96.1-h (September 30, 1996) Copyright © 1993, 1994, 1995, 1996, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html blessed-dedks-reco.tex.
The translation was initiated by Juan Pablo Fernandez on Fri Mar 28 16:47:32 CST 2003