An Overview of Pattern Recognition in the Central Arms of the PHENIX Detector

Paper: 355
Session: A (talk)
Speaker: Mitchell, Jeffery, Brookhaven National Laboratory, Upton
Keywords: algorithms, analysis, simulation, visualization


An Overview of Pattern Recognition in the
Central Arms of the PHENIX Detector

Jeffery T. Mitchell
Brookhaven National Laboratory
Physics Department, Bldg. 510C,
P.O. Box 5000, Upton, NY 11973-5000
for the PHENIX Collaboration

Abstract

The PHENIX detector at the Relativistic Heavy Ion Collider
at Brookhaven National Laboratory is being built to search
for a state of deconfined quarks and gluons, the Quark-Gluon
Plasma, in 200 GeV/A Au+Au collisions. These collisions will
produce roughly 500 charged particles traversing the two
central arms of the PHENIX detector in a single event, providing
a very hostile environment for pattern recognition. The
central arms consist of a wide variety of detectors designed
for measurements of electron pairs, hadrons, and photons.
A set of three pixel pad chambers and 40 drift chamber planes
are used for track reconstruction and momentum measurement
while ring-imaging Cerenkov detectors, 4 time expansion
chamber planes, time-of-flight, and calorimeters are used
for particle identification. This paper will describe the
algorithms used for track reconstruction in this high multiplicity
environment with an emphasis on the problems encountered due
to the high track density along with their solutions. With
the large number of tracks that must be reconstructed, many
software tools have been developed to understand the performance
of the track reconstruction algorithms. This includes a
suite of evaluation modules designed to provide calculations
of efficiency and numbers of ghost tracks, for example. The
evaluators also allow the developer to easily isolate flaws
in the reconstruction and provide comparisons of competing
algorithms on common ground. The concepts used in the
evaluation suite, which are applicable to any track
reconstruction package, will be described, and the
performance of the PHENIX package will be presented. Also,
a 3-dimensional interactive event visualizer written with
OpenGL that allows experimenters to easily manipulate these
large events and perform hands-on reconstruction evaluation
will be presented.