Run II Muon Software An Overview

1
Run II Muon SoftwareAn Overview

• Pushpa Bhat
• 
  Muon Vertical Review
• 
  May 22, 2000

2
Muon Algorithms/ID Group
Paul Balm (Level3, Hitreco) Pushpa Bhat (Muon
Alg. Co-leader, Geometry,muo_explore) Kors
Bos (Level3, TIM, d0ve) Laurent
Chevalier(Tracking) Frederic Deliot (Tracking,
alignment) Oleg Eroshin (FAMUS tracking code)
Mike Fortner (Unpack/pack reco, muon_index)
Claude Guyot (Tracking) Dave Hedin (Muon ID
Co-leader, muo_examine,online) Sharon
Hagopian (Graphics) Rick Jesik (Geometry)
Sasha Kozelov (Segment reco, track reco) Chun
hui Luo (Muon ID, muo_analyze) Onne
Peters (Level3, hitreco,segmentreco,trf,d0ve)
Harrison Prosper (Graphics, Geometry,
muo_explore) Nemat Rakmatov (Geometry)
Christophe Royon (Muon-AlgCo-leader,D0sim,release,
debugtest) Marc Virchauex (Tracking) Daria
Zieminska (MuonID Co-leader, muonid,muo_analyze,
trf) Andrzej Zieminski(Geometry, Calibration,
Alignment) Ex-members Andrei Mayorov (mureco,
muon_index) Vladimir Sirotenko (D0gstar, McToUnp
Reco,
muon_util, muo_examine) Taka Yasuda (muo_digi,
mureco)
3
Outline

• Introduction
• Data Processing in a Nutshell
• Detailed Data Flow
• Existing and Planned Packages
• Software Review
• Status Plans

4
What does the software do?

• Access and unpack Raw Data
• Sort the data, process and create hit objects
• Reconstruct segments and tracks
• Local Tracking
• Global Tracking (TRF)
• Link reconstructed segments/tracks to tracks from
  SMT, CFT and produce muon objects
• Analyze hits, segments, tracks, muon particles
  and MC data make ntuples
• Detector and data visualization.
• (Muon Algorithms and ID groups run as an
  integrated entity has helped all parties )

5
Data Processing in a Nutshell
Raw Data
Pack
Unpack
Monte Carlo Data
Unpacked Data
Digitize Index
Sort, Calibrate Align,..
Smearing TrigSim,..
Processed Data
Find Segments, Tracks
Reconstructed Data
Physics Data
Muon ID
6
Data Objects
Unpacked Data PDTChannel MDTChannel MSCChannel
Monte Carlo Data SimTkHit
Processed Data PDTHit MDTHit MSCHit
Reconstructed Data MuoSegment MuoTrack
Physics Data MuonParticle
7
SimMUONHit Chunk
MuoMcToUnp
Reco
RawData
MuoUnp Chunk
Chunk

MuoHit
Reco
MuoHit Chunk
MuoSegment Reco
GlobalTrackReco
(trf)
MuoSegment Chunk
MuoTrack
MuoTrackReco
Reco
MuoTrack Chunk
GlobalTrack
MuonParticle Reco
Chunk
MuonParticle
Chunk
8
Packages
9
Muon Geometry

• muon_geometry
• Defines ideal geometry for the muon system
  detectors
• Provides access to shapes, sizes and positions
  of the detectors and their elements, number of
  wires, wire lengths etc.,
• Handles conversions of local to global
  coordinate systems and vice versa
• MuoBaseGeometry, MuoGeomDetector,
• MuoGeometer (user interface to geometry tree)

10
muon_geometry package
The muon geometry is represented by a
hierarchical tree structure.

• each element is represented by a simple geometric
  shape and position
• when an element is moved, the change of position
  is automatically propagated to the elements below
  it in the tree.

11
Muon Index

• muon_index
• An indexing/addressing scheme by which every
  section and every readout cell in the muon system
  can be uniquely identified.
• MuoSectionIndex, MuoCellIndex
• Each hit will have a muon index
• Translation to other addressing schemes used in
  the muon system are provided (MC, Geometry,
  Electronics)
• MuoIndexTrans
• Greatly simplifies searching, sorting, and
  tagging hits, hence improving efficiency and
  elegance of software.
• (performs as a key)

12
RECO Packages

• muo_hit
• muo_data muo_segment
• muo_track
• Specifies data formats for processed and
  reconstructed objects.
• muo_hitreco
• Take T0 corrected times, apply calib/alignment,
  make time to distance conversions and provide
  geometrical information of the hits to the user.
• muo_segmentreco
• Find A BC layer segments using MuoHits. Allows
  more than one segment finding algorithm.

13
Track RECO

• muo_trackreco
• Builds muon tracks (and track objects) starting
  with pairs of A BC segments.
• Allows more than one algorithm
• Currently two algorithms implemented
• Combinatorial Algorithm
• Simple matching of A BC segments in the
  Toroid, assuming a uniform magnetic field
• Fit Algorithm
• Pair of segments are fit into a muon track via
  track propagation through the non-uniform
  magnetic field, taking into account multiple
  scattering and energy loss in the toroid.

14
Muon ID Package

• MuonParticleReco is the steering framework
  package for muon ID
• Provides for linking the tracks found in the muon
  system to the central tracks (reconstructed in
  SMT and CFT)
• Uses the matched A BC segments from the muon
  reconstruction and charged particles from
  central tracker (Gtracks)
• The quality of match between the central track
  and A-layer segment is stored as MuonQualityInfo.
• Working definition of Muon ID
• Gtrack A segment loose m
• Gtrack (ABC)segments tight m

15
Analysis Packages

• muo_analyze
• MuoIDAnalyze
• Reads D0reco output and makes a HEPtuple and a
  ROOT ntuple with quantities relevent to muons,
  based on the contents of appropriate chunks.
• MC, Gtracks (ptgt1.5GeV/c), hits, segment, tracks,
  MuonID
• muo_explore
• An alternate analysis package which currently
  stores MC and muon local tracking info in an
  HBOOK ntuple

16
Detector and Event Display

• d0ve_muon
• D0 Visualization Environment
• Developed by the NIKHEF group
• Has been a great tool for debugging geometry and
  reconstruction code
• Now being used for on-line display in the
  commissioning runs.
• Mr Datae
• Work underway
• Beautiful 3D displays
• D0Scan
• Based on Open Inventor

17
Software Review Design Implementation

• Very thorough and very useful review of hit (and
  segment) reconstruction code(s) by Marc Paterno
  and Jim Kowalkowski
• Design Concerns
• - Template strategy for MuoDataProcessor
• - Problem Decomposition in hit processing
• - Hit storage in Hit Collection
• - EDM and Framework usage
• Coding Concerns
• - Poor memory management due to
• extensive copy of pointers
• - Error reporting

18
Software ReviewActions Taken

• Done away with MuoDataProcessor
• Added new utilities to access unpack data
• Chunk selector
• SelectModules
• SelectDataChannels
• Revised Hit and HitCollection classes
• Provided new hit access and insert methods
• Separated different hit types
• Written Channel and Hit Adaptors
• Revised HitReco and HitChunk
• Fixed other minor technical problems
• Error Logging

19
Status and Plans

• Revisions of hitreco almost done
• -sign-off this week
• Revisions of segment reco and track reco are
  underway
• -concrete schedule this week
• New algorithms for segment-finding and tracking
  underway 3 months
• trf Global tracking
• -Ramp up after 2 months
• Muon tracking in Calorimeter (MuoTrackInCal) ?
  Gallas, a few months
• Improve efficiency, purity, resolution
• ( Performance studies will cover these topics in
  detail)

20
Major Milestones

• First version of D0gstar Geometry and
• Protvino C code for FAMUS tracking Fall 1997
• Extend to WAMUS, convert to D0standard
• Summer 1998
• Invented MuoIndex June 1998
• D0ve event display available
• Muon Geometry written, mostly debugged early
  1999
• Redesign and revise code to decouple EDM and
  framework stuff from algorithms. Spring -Fall
  1999
• MCData ? Unpdata ? hit ? segment ? track
  complete. First real data (cosmic) in Run II
  format read, processed and displayed
  Aug .99
• A thorough Software Design and Code Review
  (Paterno and Kowalkowski)
  Dec. 99
• 4th major Iteration Redesign and Revision
  
  Jan.00-present
• Three Mini Workshops
• April 2000, May 1999, Dec. 1998

21
Summary

• We can process MC and Raw data (in Run II format)
  and find tracks in the muon system.
• The momentum resolution obtained is as expected.
  Work needs to be done on improving efficiencies
  and reducing fake segments.
• Design and coding revisions are underway for
  local tracking software. Global tracking (trf)
  work has to be ramped up.

Momentum Resolution ?PT
vs. PT