Tracking in the CBM experiment

1
Tracking in the CBM experiment

• I. Kisel
• Kirchhoff Institute of Physics,
• University of Heidelberg
• (for the CBM Collaboration)

TIME05, Zurich, Switzerland October 03-07, 2005
2
Facility for Antiproton and Ion Research (GSI,
Darmstadt)

• Future accelerator complex FAIR at GSI,
  Darmstadt
• Research program includes
• Radioactive Ion beams Structure of nuclei
  far from stability
• Anti-proton beams hadron spectroscopy, anti
  hydrogen
• Ion and laser induced plasmas High energy
  density in matter
• High-energy nuclear collisions Strongly
  interacting matter at high baryon densities

SIS 100 Tm SIS 300 Tm U 35 AGeV p 90 GeV
Compressed Baryonic Matter (CBM) Experiment
3
Facility for Antiproton and Ion Research
Photomontage of the existing and the planned
research facility at GSI/FAIR.
4
Tracking Nuclear Collisions
Open charm measurement one of the prime
interests of CBM, and one of the most difficult
tasks

• Tracking challenge
• 107 AuAu reactions/sec
• 1000 charged particles/event
• momentum measurement with resolution lt 1
• secondary vertex reconstruction (? 30 ?m)
• high speed data acquisition and trigger system

5
The Compressed Baryonic Matter (CBM) Experiment

• Tracking, momentum measurement, vertex
  reconstruction Radiation hard silicon
  pixel/strip detectors in a magnetic dipole field
• Electron ID RICH TRD ( ECAL)
• Hadron ID TOF ( RICH)
• Photons, p0, m ECAL
• High interaction rates

ECAL (12 m)
RICH
magnet
beam
target
STS (5, 10, 20, 40, 60, 80, 100 cm)
TOF (10 m)
TRDs (4,6, 8 m)
6
Modular Structure of DAQ
MAPS, STS
RICH
ECAL
TRD
Detector
50 kB/ev
107 ev/s
SFn
Dt
MAPS
STS
RICH
TRD
ECAL
SFn
Dt
SFn
Dt
SFn
Dt
SFn
Dt
100 ev/slice
Event Builder Network
N x M
Scheduler
SFn
Dt
MAPS
STS
RICH
TRD
ECAL
SFn
available
Farm Control System
5 MB/slice
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
105 sl/s
PC Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
Sub-Farm
7
Cellular Automaton method for track finding

• Implementations
• ARES (NIM A329, 1993)
• NEMO (NIM A387, 1997)
• HERA-B (NIM A489, 2002 NIM A490, 2002)
• LHCb (LHCb note 2003-064, 2003)
• CBM (CHEP04, 2004)
• (see http//www-linux.gsi.de/ikisel/reco/ )

• Generate a set of tracklets (similar to seeding).
  Tracklets are created everywhere all chambers
  are seeding chambers. The same set of cuts can be
  applied as in the Kalman Filter track finder
  the cuts reflect geometrical acceptance of a
  detector that is common for all methods. As hits
  are sorted, tracklets are generated in groups
  with the same leftmost hit (due to inserted loops
  over chambers). Therefore, every hit stores two
  pointers to the first and last tracklets of his
  group. Every tracklet has a counter meaning
  possible position on a track (initially 0).
• Extrapolate tracklets back to the previous layer.
  Usually tracklets are created (as in KF) starting
  from the downstream chambers and moving to the
  target. Therefore, during generation of the next
  portion of tracklets (one or two chambers closer
  to the target) the algorithm applies the track
  model to the tracklets with a common point
  (simple selection of the tracklets using the
  stored pointers, see 1).
• Find neighbors and increase the counter. If
  neighbors (possible track continuations) are
  found, a counter of a current tracklet is
  incremented with respect to a neighbor with the
  largest counter.
• Continue to 1 for all chambers.
• Collect track candidates. Start with tracklets
  having the largest counter (max_counter), for
  each of them take a neighbor (at the right) which
  has a countermax_counter-1, continue similar to
  the Simple Kalman Filter but follow counters (!),
  make branches, but no empty layers, keep the best
  (chi2) track for each initial tracklet with the
  largest counter.
• Apply competition between track-candidates. After
  step 5 a set of track-candidates of the same
  length is created, therefore chi2 is well
  suitable criterion to sort them. After sorting
  start with the best (chi2) track and flag all
  hits of the track as used. Continue with the next
  track (with lower chi2), check if number of used
  hits is less than X (parameter, depends on track
  density) and flag his hits as used or delete the
  track. Proceed with the next track-candidate etc.
• Continue to step 5, but collect tracks starting
  with the counter max_counter-1. Proceed 5-7
  decrementing max_counter until the shortest
  tracks (usually of lengthtracklet_length1) are
  collected.
• Merge clones if necessary. In case of significant
  detector inefficiency merge short tracks into
  long tracks.
• Kill ghost. Apply additional cuts to kill ghost
  tracks, most of them are short tracks.

Drawing analogy to the Kalman method one can
consider steps 1-4 as Filter, 5-7 as Smoother,
and 8-9 as Cleaner.
8
Performance of track finding
S. Gorbunov, I. Kisel and I. Vassiliev, Analysis
of D0 meson detection in AuAu collisions at 25
AGeV, CBM-PHYS-note-2005-001
9
Tracking in non-homogeneous magnetic field

• The precision of extrapolation does not depend
  on a shape of the magnetic field.
• One can cut off the higher-order terms in the
  series.

S. Gorbunov and I. Kisel, An analytic formula for
track extrapolation in an inhomogeneous magnetic
field, CBM-SOFT-note-2005-001
10
Elastic net for the traveling salesman problem
R. Durbin and D. Willshaw, An analogue approach
to the travelling salesman problem, Nature, 326
(1987) 689
11
Standalone elastic net ring finder in RICH
All set N hits 5 Ref set N hits 15 Extra
set 5 N hits lt 15 Reconstructed 70 hits
from the same MC Clone MC reconstructed few
times Ghost lt 70 hits from the same MC
S. Gorbunov and I. Kisel, Elastic net for
standalone RICH ring finding, CBM-SOFT-note-2005-0
02
12
Summary

• High track density at high rate
• Most crucial blocks of the (off-line)
  reconstruction code ready
• Work on detector optimization
• CBM notes and other publications on
  reconstruction at
• http//www-linux.gsi.de/ikisel/reco/
• Participants from the CBM experiment
• Walter Müller, Johann Heuser, Iouri Vassiliev
  and Ivan Kisel