Event Reconstruction

1
Event Reconstruction
Ties Behnke, SLAC and DESY

• Event Reconstruction in the BRAHMS simulation
  framework
• The BRAHMS framework
• Tracking Reconstruction (a brief reminder)
• Calorimeter Reconstruction

• The Goal
• Reconstruction of all 4-vectors in the event
  (charged and neutral)

• The Method
• Use information from all available subdetectors
  (tracker, calorimeter, etc)
• Currently implemented in BRAHMS
• Tracker
• ECAL, HCAL (tile option)
• Muon system still missing (under development)

2
(No Transcript)
3
Calorimeter Reconstruction
The Goal
Reconstruct the 4-momentum of all particles
(charged and neutral) in the event
Particle / Energy Flow in this context does not
deal with event properties but only with
particles Event properties are part of the
analysis
tt event at 350 GeV, no ISR
4
The tracking package

• A very brief reminder
• Patrec done separately in VTX, TPC, FCH
• Merging done for the complete event
  simultaneously

Performance measured in tracking efficiency in
dd events, full background simulation
5
The Calorimeter Reconstruction
Currently available in BRAHMS SNARK package
(author Vasiliy Morgunov)

• The philosophy behind SNARK
• Assume tracks have been found and are perfect
• Start with tracks, associate hits in calo with
  the tracks
• Look for hits in a tube
• Iterate the size of the tube
• Use the information from the track to determine
  the tube parameters
• remove the hits associated to tracks
• Do cluster finding (conventional)
• Identify neutral objects

• Advantages
• During clustering more information is
  availabel charged/ neutral/ ..
• Treatment of overlaps uses full information of
  the event
• Utilise the strong tracking system of the LC
  detector

6
The Algorithm

1. Collect hits in the calorimeter along the
   predicted track (track core) within a
   distance of /- one electronic cell.
2. Make a first particle hypothesis (e.g. MIP,
   ...)
3. Predict the transverse shower profile, collect
   more hits within the expected road
4. Iterate, until measurement and expectation agree
   best
5. Any hits which at the end of the procedure are
   not associated belong to a neutral particle.
   Run conventional clustering, determine
   properties of neutral particle

• The system depends on
• high granularity both in ECAL and HCAL
• excellent linking between Tracker ECAL HCAL
• extensive use of amplitude info (optimised for
  tile HCAL)

Note a similar program, but optimised for the
digital HCAL, is also under development (Ecole
Polytechnic)
7
Performance Single Particles
Photon
Electron
Muon
Kaon
Kaon (neutral)
Pion
PiZero
Particle identification as given by the SNARK
algorithm
8
Performance Single Particles
1 gamma 2 electron 3 muon 4 kaon 5 kaon 0 6
pion 7 kaon 0
Efficiencies
9
Performance Single particles
Photons
Electrons
Pions
10
Single Particle Performance

• Decent single particle identification
  probabilities
• Based on simple selections intrinsic to the
  program
• More sophisticated algorithms can be applied
  post mortem

• The difference in neutral and charged particle
  treatment is visible in the single particle
  reconstruction performance
• Larger number of fake objects in charged
  particles
• Larger tail at high energies for charged objects

• Overall performance quite ok, though (of course)
  further imporvements are possible

11
Final Reconstructed Particle Objects

• Output of BRAHMS with SNARK Reconstructed
  particle 4-vectors

3-momentum px, py, pz Energy E particle ID
hypotheses link to track(s) used link to
cluster(s) used

• The user works with these objects
• Build jets
• Find vertices
• Calculate event properties
• ....

• The system does work (see talk (V.
  Morgunov) in top session on top
  reconstruction

• Under development common data model for all
  simulation and reconstruction systems (US, EU,
  J(?), ...)

Fully hadronic top decay (6 jets), full
background
12
Conclusion

• BRAHMS offers a complete simulation and
  reconstruction framework for a LC detector
• Tracking implemented for a complicated geometry,
  easily adaptable to other geometries
• Tracking interface to MOKKA (Geant4) does exist
• One version of calorimeter reconstruction
  software is included
• Optimised for SI-W ECAL and tile type HCAL
• Port to other systems is (at the moment) not easy
  
• Full implementation of the energy flow algorithm
• First results based on this full reconstruction
  do look promising

• Further developments
• Tuning and improvements of the calorimeter
  reconstruction software
• Port of simulation part to GEANT4 (MOKKA)
• Implementation of the new LCIO standard for
  persistency and data model to easy portability
  of software between systems and regions