Neural tracking in ALICE

1
Neural tracking in ALICE

• Alberto Pulvirenti University and I.N.F.N. of
  Catania
• ACAT 02 conference
• Moscow, June 26 2002

2
Outline

• The ALICE experiment
• Tracking in ALICE
• Why an ITS stand-alone tracking?
• Implementation
• Results
• Work in progress and outlook

3
The Large Hadron Collider
4
The 4 LHC experiments
5
ALICEs objective QGP study
PbPb _at_ LHC (5.5 A TeV)
The Little Bang
The Big Bang
6
ALICE track multiplicity
A sketch
7
ALICE track multiplicity
A sketch of 1/100 of a typical ALICE event
Simulation and reconstruction of a full
(central) PbPb collision at LHC (about 84000
primary tracks!) takes about 24 hours of a
top-PC and produces an output bigger than 2 GB.
8
The ALICE detector
9
Tracking in ALICE

• Time Projection Chamber.
• 180 points per track ? main contribution.
• Inner Tracking System.
• 6 points close to primary vertex ? improves
  resolution near to the production vertex.
• Standard procedure
• Points in the TPC outermost pad-rows are arranged
  into suitable track seeds.
• the seeds are propagated through the TPC towards
  its innermost pad-row, according to a Kalman
  filter algorithm for both recognition and
  reconstruction.
• each track found in the TPC is propagated in the
  ITS and its parameters are refined with the aid
  of the six best matched ITS points.

10
Why an ITS stand-alone tracking?

• because the TPC is a slow detector
• some events could be produced in a high-rate
  acquisition mode, by turning on only the fastest
  ALICE modules (ITS, Muon Spectrometer), to
  produce large amounts of data useful for all
  analyses needing high statistics.
• in this case, we need at least a satisfactory
  efficiency for high transverse momentum (pt gt1
  GeV/c).
• because some particles decay within the TPC
  barrel volume, and the standard TPC tracking
  doesnt manage to create seeds for them.
• in this case, the tracking is performed after
  completing the standard Kalman procedure, and
  working only on the points which the Kalman
  method didnt use.

11
Implementation 1 definitions
12
Implementation 1 definitions
Neuron oriented track segment ? 2 indexes
sij links two consecutive points in the
particles path according to
a well-defined direction
13
Implementation 1 definitions

• Weight geometrical relations between neurons ? 4
  idxs wijkl
• Geometrical constraint
  
  only neurons which share a point have a non zero
  weight

14
Implementation 1 definitions

• Weight geometrical relations between neurons ? 4
  idxs wijkl
• Geometrical constraint
  
  only neurons which share a point have a non zero
  weight
• Case 1 sequence
• guess for a track segment
• good alignment requested

15
Implementation 1 definitions

• Weight geometrical relations between neurons ? 4
  idxs wijkl
• Geometrical constraint
  
  only neurons which share a point have a non zero
  weight
• Case 1 sequence
• guess for a track segment,
• good alignment requested
• Case 2 crossing
• negative weight
• leads to a competition
  between units

16
Implementation 1 definitions

• Weight geometrical relations between neurons ? 4
  idxs wijkl
• Geometrical constraint
  
  only neurons which share a point have a non zero
  weight
• Case 1 sequence
• guess for a track segment,
• good alignment requested
• Case 2 crossing
• negative weight
• leads to a competition
  between units

17
Neural Network Simulation Specifics

• Associative memory topology
  (single layer of fully connected units).
• Real valued (sigmoidal) activation function,
  limited between 0 and 1.
• Random initialization.
• Asynchronous updating cycle (one unit at a time).
• Stabilization threshold on the average activation
  variation after a complete updating cycle.
• Resolution of competitions to the advantage of
  the unit with the greatest real activation.
• Binary mapping of on and off units with a
  threshold of 0.6 on the final real neural
  activation.

18
Implementation 2 cuts

• Needed to limit the number of point pairs
  used to create neurons
• Check only couples on adjacent layers
• Cut on the difference in polar angle (q)
• Cut on the curvature of the projected circle
  passing through the two points and the calculated
  vertex
• Helix matching cut

where a is the corresponding circle arc of the
projection in the xy plane
19
Implementation 3 procedure

• Step by step procedure
• (removing the points used at the end of each
  step)
• Many curvature cut steps, with increasing cut
  value
• Sectioning of the ITS barrel into N azymuthal
  sectors

RISK edge effects the tracks crossing a sector
boundary will not be recognizable by the ANN
tracker
20
Implementation 4 reconstruction

• Track reconstruction Kalman Filter.
  (ref. A. Badalà et al., NIM A(2002) in press and
  references therein).
• vertex constrained seed.
• A helix is estimated by using the two outermost
  points and the experimental vertex
  (the same which is used for neuron creation
  cut).
• two operational phases
• vertex ? layer 6.
• layer 6 ? vertex.

21
Test trial ingredients

• Test on a simulation produced with the HIJING
  event generator interface (developed within the
  AliRoot framework), and tracks transported
  through the detector by GEANT 3.21
• All detectors and all physical effects turned
  on.
• Fully detailed geometry, simulation and
  reconstruction in the ITS.
• ALICE default number of primary tracks
  (84210 in the
  pseudorapidity region h lt 8.0).

Track definition for efficiency evaluation Track definition for efficiency evaluation Track definition for efficiency evaluation
Criterion GOOD TRACK (fake otherwise) FINDABLE TRACK
SOFT at least 5 right points Has at least 5 points in ITS
HARD all 6 point must be correct Has a point for each layer
Efficiency good tracks (fake tracks) / findable tracks Efficiency good tracks (fake tracks) / findable tracks Efficiency good tracks (fake tracks) / findable tracks
22
Signal-to-noise ratio
Layer 1 2 3 4 5 6 Average
Good / All 46 60 65 69 77 74 65
Unused good / All unused 21 37 45 51 68 63 47
23
Stand-alone tracking results (I)
Number of found tracks, efficiency and CPU time
as a function of the of sectors. Only one event
analyzed.
Test choice 18 sectors CPU time 10 of the
time requested the whole ITS at once PC used
PIII 1 GHz
24
Stand-alone tracking results (II)
SOFT
good
fake
25
Stand-alone tracking results (III)
26
Stand-alone tracking results (III)
27
Stand-alone tracking results (III)
Parameters resolution Parameters resolution Parameters resolution
Neural Kalman (without vertex. constr.)
pt () 13.4 ? 0.3 1.57 ? 0.02
? (mrad) 4.71 ? 0.01 1.40 ? 0.08
l (mrad) 3.69 ? 0.01 1.60 ? 0.08
Dt (?m) 79.7 ? 0.1 50
Dz (?m) 265.6 ? 0.4 150
Efficiency for tracks with pt ? 1 GeV / c Efficiency for tracks with pt ? 1 GeV / c Efficiency for tracks with pt ? 1 GeV / c
Efficiency () Fake ()
Neural soft 78.2 ? 3.0 9.9 ? 0.9
Kalman soft 72.8 ? 2.9 4.9 ? 0.6
28
Combined tracking results (III)
Results for Pt ? 1 GeV / c Results for Pt ? 1 GeV / c Results for Pt ? 1 GeV / c
Kalman
Efficiency 72.8 ? 2.9
Fake prob. 4.9 ? 0.6
Efficiency per particle pt ? 1 GeV/c Efficiency per particle pt ? 1 GeV/c Efficiency per particle pt ? 1 GeV/c
? K
Kalman 74.7 ? 2.7 64.5 ? 0.8

29
Combined tracking results (III)
Results for Pt ? 1 GeV / c Results for Pt ? 1 GeV / c Results for Pt ? 1 GeV / c
Kalman Combined
Efficiency 72.8 ? 2.9 83.0 ? 3.0
Fake prob. 4.9 ? 0.6 7.0 ? 0.7
Efficiency per particle pt ? 1 GeV/c Efficiency per particle pt ? 1 GeV/c Efficiency per particle pt ? 1 GeV/c
? K
Kalman 74.7 ? 2.7 64.5 ? 0.8
Combined 84.7 ? 2.9 76.2 ? 0.8
The findable tracks are counted among all ITS
findable tracks (even the ones which are NOT
findable in the TPC)
10 increase!
30
Conclusions work in progress

• The Neural Network tracking algorithm has been
  successfully adapted to the unprecedented ALICE
  multiplicity
• Implementation has been done in the official
  AliRoot off-line framework based on ROOT.
• Recognition efficiency is comparable with the
  Kalman Filter one, in the range of pt gt 1 GeV/c.
• Under study
• Improving the neural algorithm performances for
  LOW transverse momentum tracks pt lt 0.2 GeV/c
  (not a trivial task!).
• Alternative possible techniques for the same
  purpose (adapting some existing algorithms like
  elastic tracking, elastic arms algorithm, or
  developing a genetic algorithm).
• Future developments (for combined tracking).
• Improving track parameter resolution by including
  also the TPC/TRD points unused by Kalman
  tracking.

31
(No Transcript)