GEANT4 for Future Linear Colliders

Document Sample
GEANT4 for Future Linear Colliders Powered By Docstoc
					GEANT4 for Future
 Linear Colliders

      Norman Graf
  Geant4 Workshop @ TRIUMF
      September 5, 2003
Linear Collider Environment
  Detectors   designed to exploit the
   physics discovery potential of e+e-
   collisions at s ~ 1TeV.
  Will perform precision measurements of
   complex final states.
  Require:
           Exceptional momentum resolution
           Excellent vertexing capabilities
           “Energy Flow” calorimetry
           Hermeticity
 4/8/2010                 Norman Graf          2
Mission Statement
    Provide full simulation capabilities for
     Linear Collider physics program:
       Physics simulations
       Detector designs

       Include machine backgrounds

    Need flexibility for:
           New detector geometries/technologies
    Limited resources demand efficient
     solutions, focused effort.
 4/8/2010                  Norman Graf             3
  Have a common simulation environment
   used in all LC studies which allows
   sharing of detectors, algorithms, and
  The system should be flexible,
   powerful, yet simple to install and

 4/8/2010         Norman Graf          4
LC Detector Full Simulation

  MC Event

             LCApplication           Raw Event



                     Reconstruction, Visualization, …

 4/8/2010       Norman Graf                     5
 Simulator for JLC Detectors based on Geant4
  and ROOT (JSF)
 JLC Unified Particle Interaction and Tracking
 A set of base abstract classes provide
  methods for installation and data-output.
 Specific parameters are run-time definable,
  but geometry structure hard-coded.

    4/8/2010         Norman Graf            6

                R-phi section of VTX
                (installed by Aso-lab)
                @Toyama National College
                of Maritime Technology

 Layer 10
 Cell 36~108/Layer
 Wire 5/Layer                              Event display of
                                           e+e- -> Z0H
 4/8/2010                    Norman Graf                      7
TESLA Full Simulation

 4/8/2010    Norman Graf   8
 Geant4 full simulation for the Tesla detector.
 Uses subdetector-specific geometry drivers.
     Relevant parameters stored in MySQL database.
     Tight coupling between Sensitive Detector and
      geometry volume definitions.
   LCIO persistence for generic hits & MC chain.

     4/8/2010          Norman Graf              9
      The Proto00 geometry driver





4/8/2010               Norman Graf   10
LCD Full Simulation
   Geometry defined in XML.
     Flexible, but simplified volumes.
     Projective readout of sensitive volumes.

 Dynamic topology, not just parameters.
 Have defined generic hit classes for
  sensitive tracker and calorimeter hits.
 Root and LCIO bindings for I/O.

 4/8/2010              Norman Graf               11
TPC Tracker, Si Disks, CCD VTX

 4/8/2010     Norman Graf        12
All Si Tracker, CCD VTX

 4/8/2010      Norman Graf   13
Generic Hits Problem Statement
  We wish to define a generic output hit
   format for full simulations of the
   response of detector elements to
   physics events.
  Want to preserve the “true” Monte
   Carlo track information for later
  Want to defer digitization as much as
   possible to allow various resolutions,
   etc. to be efficiently studied.
 4/8/2010          Norman Graf              14
Types of Hits
   “Tracker” Hits
     Position sensitive.
     Particle unperturbed by measurement.

     Save “ideal” hit information.

   “Calorimeter” Hits
     Energy sensitive.
     Enormous number of particles in shower
      precludes saving of each “ideal” hit.
     Quantization necessary at simulation level.

    4/8/2010             Norman Graf                15
Hits Summary
  Storing “ideal” hits gives detailed
   information about MC track trajectory.
  Deferring digitization allows studies of
   detector resolution to be efficiently
  Can approximate the same in
   calorimeter by defining small cells, then
   ganging later.

 4/8/2010           Norman Graf            16
 Persistency framework for LC simulations.
 Currently uses SIO: Simple Input Output
     on the fly data compression
     some OO capabilities, e.g. pointers

     C++ and Java implementation available

 Changes in IO engine designed for.
 Extensible event data model
     Generic Tracker and Calorimeter Hits.
     Monte Carlo particle heirarchy.

    4/8/2010            Norman Graf           17
 Persistency framework for LC simulations.
 Java, C++ and f77 user interface.
 LCIO is currently implemented in simulation
     hep.lcd
     Mokka/BRAHMS-reco

    -> other groups are invited to join

    4/8/2010          Norman Graf          18
Towards Internationalization
 Suggest that Tesla, NLC and JLC full
  simulation groups could run a single
  GEANT4 executable.
 Geometry determined at run-time (XML).
 Write out common “ideal” hits.
 Digitize as appropriate with plug-ins.
 Enormous savings in effort.
 Makes comparisons easy.

 4/8/2010         Norman Graf          19
Full Simulations
  LCD Full Sim     BRAHMS             JIM
   GISMO           GEANT3          GEANT3
   C++             FORTRAN         FORTRAN

             Common GEANT4
             Runtime geometry
             Generic Hit output
  4/8/2010         Norman Graf              20
 First version of mysql / xml interface exists
 SD detector fully modelled including
 Several TESLA detector versions modelled
 LCIO output implemented in beta version
 Interfaces to HEPEVT and STDHEP and
  background files implemented
 Interface to AIDA integrated

    4/8/2010          Norman Graf             21
SD in Mokka

 4/8/2010     Norman Graf   22
LC Detector Full Simulation
  MC Event
  (STDHEP)                         Raw Event
                                  (Generic Hits)
Geometry (XML)


  4/8/2010          Norman Graf              23
 Main Issues
   Need flexible method to describe geometry.
       Prefer G4 supported geometry input (GDML?)
 Beam Delivery System requires arbitrary
  magnetic fields, excellent tracking precision.
 Tracking System: (1/pT)5x10-5 GeV/c
       Multiple Scattering, tracking precision.
   Jet Reconstruction: E/E~30%/√E
       Excellent hadronic shower simulations.

        4/8/2010             Norman Graf             24
Thank you!
 To the small number of people doing an
  enormous amount of work on the Linear
  Collider simulations.
 To the Geant4 collaboration for providing
  me the opportunity to present this talk.
 To the workshop organizers for hosting this

    4/8/2010        Norman Graf           25
4/8/2010   Norman Graf   26
 Beam Delivery System
Beamlines are built up out of
modular accelerator components

 Full simulation
 of em showers
                                   All secondaries

   4/8/2010          Norman Graf             27
 G4 Stepper


Step-size from           Multipole
physics process          Stepper
                                            Each volume can have its
                                              own field or stepper

         x,xp,y,yp,z,E                    Multipoles up to Octupoles
                                                included so far
   4/8/2010                 Norman Graf                      28
Synchrotron Radiation
Generator of H. Burkhardt
Implemented for all components
Based on local curvature
Individual photons from
    individual parents

  4/8/2010        Norman Graf    29
    BDS Summary/Future Plans
• Accurate accelerator tracking within Geant4.
• Some modification of G4.4.0 is needed.
• Interaction with the G4 experts at CERN.
• Soon will be fully consistent with standard G4.
• Code at the status of an alternative tracker.
• Results on SR need to be checked.
• New processes incorporated - eg Planck
scattering, Laser wire…
• Serious collimation studies now possible ...
    4/8/2010          Norman Graf             30
 Why XML?
 Simplicity: Rigid set of rules, plain text
 Extensibility: Add custom features, data types
 Interoperability: between OS and languages
 Self-describing data
 Hierarchical structure  OOP
 Open W3 standard, lingua franca for B2B
 Many tools for validating, parsing, translating
 Automatic code-generation for data-binding

    4/8/2010          Norman Graf             31
 Why G4 XML?
 XML Schema very useful for “compile-time”
  type safety and bounds checking.
 Prefer a G4-supported XML-based solution.
     Had hoped for common LHC solution.
     Investigated GDML.
           Looks promising.
           Sensitive detector definitions needed.

           Support?

        4/8/2010                 Norman Graf         32
Track Definition

 Particles suffer various indignities while
  traversing the detector.
 Knowing track parameters at a single point
  (e.g. the point of generation) is insufficient
  for precision fits due to material effects
  (dE/dx, MCS, bremsstrahlung) and field
       No global functional form for the fit.
   Store track information at each volume.
    4/8/2010                Norman Graf          33


4/8/2010   Norman Graf        34
Tracker Hit
  MC Track Number
  Encoded detector ID (detector dependent )
  Global hit position at entrance to sensitive volume
  Global hit position at exit of sensitive volume
  Track momentum at entrance to sensitive volume
  Energy deposited by track in sensitive volume
  Time of track's crossing

  Hit number
  Local hit position at entrance to sensitive volume
  Local hit position at exit of sensitive volume
  Step size used by simulator in sensitive volume

 4/8/2010                 Norman Graf                   35

Volume ID


             x, p, t
 Track ID
  4/8/2010             Norman Graf   36
Calorimeter Hit
 Encoded detector ID (detector dependent )
 MC ID and energy deposited by each
  contributing particle

 Hit Number
 Cell position
   Radius, Phi, Z of cell
   X, Y, Z of cell
 Total energy deposited in cell

    4/8/2010        Norman Graf          37

Shared By: