Docstoc

ILC Detector Simulation with Geant4

Document Sample
ILC Detector Simulation with Geant4 Powered By Docstoc
					                                                                                                   1
     ILC Detector Simulation Projects: Directions and Ideas
                                    k0L               pbar               p                 e-
SiD 00




                      R: ± 650 cm
MCParticle Endpoint
X  Z: 280 cm
Y  R: ± 650 cm
2000 ttbars
                                    Z: 280 cm
        pi-                         k+                 n                mu+                e+




        pi+                           k-              nbar              mu-              gamma




                                                                              ± 300 cm
                                                                                          140 cm

                                         Jeremy McCormick
                                           SLAC LCD Simulations Group
      Detectors: Detector Design                     2




• ILC is the first major HEP project to use Geant4
Monte Carlo detector simulations as the primary
benchmark for designing full detectors.
• Reconstruction algorithms are used to compare
detector performance on benchmark processes.
• The parameter space is fairly large.
    • Full detector designs
    • Subdetector geometry and readout parameters
    • Reconstruction algorithms
• Frameworks based on Monte Carlo simulations can
also be used on the raw data.  Get started now!
        Detectors: Performance                                                      3




Standard ILC Detector Performance Requirements

 a) Two-jet mass resolution comparable to the natural widths of W and Z for an
                 unambiguous identification of the final states.
 b) Excellent flavor-tagging efficiency and purity (for both b- and c-quarks, and
                           hopefully also for s-quarks).
 c) Momentum resolution capable of reconstructing the recoil-mass to di-muons
      in Higgs-strahlung with resolution better than beam-energy spread .
    d) Hermeticity (both crack-less and coverage to very forward angles) to
                 precisely determine the missing momentum.
      e) Timing resolution capable of tagging bunch-crossings to suppress
                    backgrounds in calorimeter and tracker.
    f) Very forward calorimetry that resolves each bunch in the train for veto
                                   capability.
                                                                        4
              Detectors: Overview

Subdetector       GLD               LDC               SiD
Vertex detector   Si pixel          Si pixel          Si pixel
                  r1= 2.0 cm        r1= 1.5 cm        r1 = 1.4 cm
Tracker           TPC               TPC               Si strips
EM CAL            Scintillator-W    Si-W              Si-W
HAD CAL           Scintillator-Pb   Scintillator-Fe   RPC-SS
                                    RPC/GEM-Fe
Muon system       Scintillator      RPC               RPC
Solenoid          3 Tesla           4 Tesla           5 Tesla
                  R = 3.5 – 4.5 m   R = 3 – 4.45 m    R = 2.5 – 3.3 m
                  L = 8.9 m         L = 9.2 m         L = 5.4 m

              Takashi Maruyama @ Snowmass 2005
       Detectors: Silicon Detector (SiD)   5




• Primarily a U.S. project
• Toolset covered in this talk
• Lots of different versions over the
years
   • sdjan03, sdmar05, sdaug05,
   sid00
• Tracking
   • Silicon Tracker
• Software
   • SLIC and org.lcsim
• Website
   • http://www-sid.slac.stanford.edu
                                                       6
   Detectors: Large Detector Concept (LDC)

• Primarily a European project
• Tracking
  • Time Project Chamber (TPC)
• Based on TESLA TDR (2000)



                          • Current version is D14
                          • Mokka + MARLIN
                          • Website
                             • http://www.ilcldc.org
 Detectors: Global Large Detector (GLD)   7




• Primarily Asian project
• Software
   • JUPITER and satellites
   • ROOT
• Largest detector
• Website
   • http://ilcphys.kek.jp/gld/
 Particle Flow Algorithms: Overview                               8




• aka PFA, Particle Flow / Pflow, Individual Particle Reconstruction
• Reconstruct each particle individually using information from all
subdetectors
• Requires precise tracking and calorimetry
• Capsule summary
    • VTX and trackers  charged
    • ECAL  gamma, electron
    • ECAL + HCAL  H0
• Documents
    • “Skeleton of Particle–Flow Reconstruction Program” by
    Morgunov
    • See also talks by Steve Magill, Norman Graf, Alexei
    Raspereza, et al
          Project Ideas: Full Detectors   9




• Performance comparison of
detectors on benchmark physics
• Case FOR or AGAINST 2 detectors.
• Optimization and “tweaking” of one
detector design’s subdetectors
• Effects from hardware inefficiencies
   • Crosstalk, noise, dead channels,
   quantum efficiency of readouts
   (use digi pkg such as DigiSim)
• Realistic modelling of a prototype
tracking or calorimeter device
   • Testbeam, layer stack
                 R&D: Calorimetry       10




• Hcal
   • Readout
      • Scintillator, RPC, GEM
      • Digital or analogue
   • Absorber
      • Stainless Steel, Tungsten
• ECal
   • Readout
      • Silicon, scintillator, hybrid
   • Absorber
      • Lead, Tungsten
   R&D: Tuning the Calorimeter Design                                       11




Many design parameters to adjust

  Overall        Inner radius of calorimeter
                 Outer radius of calorimeter
                 Transition from barrel to endcaps
                 Transition from endcaps to very forward calorimeters

  ECAL           Absorber thickness (uniform, varying with depth)
                 Number of layers
                 Segmentation of readout

  HCAL           Absorber choice    → Tungsten (2 X0) versus steel (1 X0)
                 Number of layers
                 Active medium (RPC, GEM, Scintillator)
                 Segmentation of readout
                 Resolution of readout (number of bits)

  Tail catcher   Needed?
                 Same technology as HCAL


Need reasonably well performing PFA to evaluate different designs
                 Jose Respond @ Snowmass 2005
       R&D: Calorimeter Test Beams    12




• Institutions
    • CALICE
    • NICADD
    • IN2P3
    •…
• Default framework is Mokka.
    • TB04, TB05, etc.
• U.S. Framework can also simulate
testbeam geometries.
• Compare real data with simulation
• Physics validation studies (LHC)
• “shower libraries” from testbeam
results for realistic shower topo
• Technology comparisons
 R&D: Trackers and Vertex Detectors             13




• TPC
   • Gas tech
   • Influenced by LEP experiences
   • O(200) hits per track
• Silicon Tracker
   • Silicon strip readout
   • Influenced by SLD experiences with VTX
   • O(4) hits per track
   • V0's, kinks and low pT particles challenging
• Vertex Detector
   • Seeds tracking (especially) in Si Tracker
       R&D: Interface with the Machine   14




• Beam Delivery System (BDS)
• Machine Detector Interface
• Forward calorimetry
   • Spectrometer
   • BeamCal
   • Luminosity monitor
• BDSIM
   • Geant4-based Linac sim
• Backgrounds
   • Beam fragments, halo, muons
                                             15
   Project Ideas: Subdetectors and B-field


• Calorimeter parameters
    • Inner radius + outer Z
    • Sensor technology
    • Sensor resolution (granularity)
    • Absorber material
    • Absorber thickness
    • Number of layers
• Tracker parameters
    • Number of layers
    • TPC vs. Silicon Tracker
    • other?
• B-field strength
    • 2-5 Tesla  How it affects perf?
                     Sim: Tools                 16




• C++ Language
• Geant4
    • Monte Carlo Toolkit
• Linear Collider IO (LCIO)
    • Sim and reco IO format
• StdHep
    • Physics event IO format
• Linear Collider Detector Description (LCDD)
    • Detector geometry format based on GDML
• Simulator for the Linear Collider (SLIC)
    • Geant4 simulator package
• LCDetectors
    • Repository of detector data
           Sim: Fast Simulation               17




• MCFast
   • org.lcsim
   • Smearing for trackers and calorimeters
• Lelaps
   • Good performance
   • Writes LCIO files
   • GODL geometry description language
      • Written by GeomConverter
• Website
   • http://www.lcsim.org/software#sim.fast
              Sim: Full Simulation                    18




• Full Sim is a difficult area to “jump into”.
• But it provides Monte Carlo data for most other
software in the toolchain, so it is very important.
• Lots of opportunities to learn new skills
   • Get started with Geant4
• MCParticle codes
   • Room for improvement + options
   • Performance optimization
• Website
   • http://www.lcsim.org/software#sim.full
                  Sim: Geant4                            19




• Tuning
   • Range cuts
   • Physics limits
   • Field stepper resolution
• Physics
   • Physics lists
   • H0 production  energies, multiplicities
   • Single particle enery resolutions in calorimeters
   • Validation with testbeam
• Website
   • http://geant4.cern.ch
Detector Description: GDML and LCDD         20




• Geometry format for detector simulation
• Include in reconstruction and analysis?
    • Interchange format
    • Low-level detector description
• Possibilities for additional features
    • Readouts
    • BREPs
    • Tessellation, divisions, replicas
    • Multiple files
    •…
• Website
    • http://www.lcsim.org/software/lcdd
       Detector Description: Java                      21




• A fully-featured geometry and detector description
subsystem for org.lcsim is on the wishlist.
    • But manpower is somewhat lacking.
• Inspirational C++ frameworks
    • Gaudi
    • Geant4
    • ROOT
• Java frameworks
    • GraXML (ATLAS)
• Good area for students
to make contrib
         Event Generation: Software   22




• Zoology
   • Pythia
   • Pandora-Pythia
   • HERWIG
   • WHIZARD
   • Guinea Pig
   • ISAJET
   • Java Diagnostic Generator
   • Geant4 GPS
   •…
• Wrap with Python?
   • SWIG, f2py
                     Reco: Tools   23




• Java Language
• Netbeans
    • IDE
• Maven
    • Build tool
• org.lcsim
    • Reconstruction Pkg
• LCIO
• JAS3
    • Analysis Environment
• AIDA
    • Plotting API
• WIRED4
    • Visualization
                   Reco: org.lcsim                 24




• Repository of ILC reco + analysis code
• ~20-30 contributors
• Contributions always welcome
   • Tracking, clustering, vertexing, PFA, geom,
   digitization, etc.
• Uses JAS3 and WIRED4
• Websites
   • http://confluence.slac.stanfordu.edu
   • http://www.lcsim.org/software
               Reco: Clustering                     25




• Algorithms
   • Cheater, Nearest Neighbor, Minimum Spanning
   Tree, Fixed Cone, Directed Tree, MIP, …
• First, understand what is there.
   • org.lcsim
   • MarlinReco
• Porting algorithms from C++ and FORTRAN to Java
would be a useful and instructive excercise.
• Detailed performance comparisons between clustering
algorithms will be needed for the PFA subparts.
• Do we need new clustering algorithms? (probably not)
        Reco: Particle Flow Algorithms                     26




• Need a canonical or baseline PFA for comparisons of
the detector designs
    • How will this work w.r.t. C++ and Java?
    • Use perfect PFA?  What does it tell us?
• Seem to have lots of different algorithms available.
    • org.lcsim  ANL
    • MarlinReco
    • BRAHMS (the Godfather)
• Integration, consolidation, and optimization should be
the main goals, now.
    • Find institutions and researchers to work with. 
    Do not need more “lone wolves” in the PFA area!
   Project Ideas: Particle Flow Algorithms   27




• Neighbor finding using hit positions
   • Most independent of detector
   topology
   • MarlinReco for inspiration
• Optimize the pieces
   • Which clusterer?  NN, FC, etc.
   • Parameter tuning
      • ex. - min cells f/ NN
   • Improve on weak spots
• MC truth vs. analysis
   • How close can you get?
   • 30% / sqrt(E)  get some booze
                                28
Physics Benchmarks

Reduced Benchmark List




  Tim Barklow @ Snowmass 2005
               Project Ideas: Physics                       29




• Reconstruction and analysis using …
     • Norman Graf’s StdHep files (FTP)
     • Single particles
          • Energy angular dependence
     • New physics / SUSY
          • Parameter space scan
     • Benchmarks (see documents)
• Crossing angle
     • 2, 5, 10, 20 mrad
• Machine backgrounds
     • PFA tolerance / robustness against
• Fixed target experiments
     • See Snowmass talk by Sekazi K. Mtingwa
• documents
     • “Questions for Further Study”
     • “Scorecard”
     • “Benchmark Analyses for Linear Collider Detectors”
     • Presentations by Tim Barklow @ Snowmass 2005
             Visualization: WIRED          30




• Reads HepRep XML Format
• Based on FreeHep 2D libs
• Well-supported by org.lcsim
   • Easy to add “hooks”
   • Generate HepReps on-the-fly
      • Detector geometry
      • LCIO objects
          • Hits, particles, clusters, …
      • B-field (not yet)
• Website
   • http://wired.freehep.org
   Visualization: A Few Software Ideas         31




• OpenInventor
   • Coin3D
   • OpenScientist
• CAD
   • OpenCascade
• VTK
   • 3D toolkit for scientific visualization
   • Not used much in HEP … but why not?
   • High performance
   • Lots of software packages based on it
                                             32
      Visualization: VTK Field Map Example


• This example
created with
VTK’s Python
bindings.
• ~300 line script
to read map and
plot in 3D +
interactivity
• Picking,
zooming,
panning, rotation
Computing: Batch and Grid Services                     33




• ILC batch and Grid services available
    • SLAC
       • LSF
    • NICADD
       • Condor
    • FNAL
       • PBS
    • DESY
       • ILC Grid
• Can these be integrated using the Grid so that ILC
researchers can have a “one stop shop” ?
• Should be fairly easy to run jobs with Java. Just drop
in the JAR and input files and GO.
         Computing: Integration        34




• End-to-end jobs
   • Event generation
   • Detector simulation
   • Reconstruction and analysis
   • Visualization
   • Plotting
• Services
   • Data server
   • WWW  plots, results, workbooks
   • Grid integration
         General: Documentation                  35




• We love it when people write docs.
   • Tutorials, HOWTOs
   • Cookbooks, Workbooks, Guides, Manuals
   • Topical Papers, Theses
• The ILC Wiki is available for writing your docs.
   • http://confluence.slac.stanford.edu
   • Contact tony_johnson@slac.stanford.edu if
   you need an account.
                                                        36
General: Feature Requests and Bug Reports

• We also love it when people report bugs and feature
requests.
    • Otherwise, we don’t know how to help.
• What bugs are you finding, especially “show
stoppers” ?
• What features are lacking?
• In general, what problems are you having?
• bug trackers
    • http://jira.slac.stanford.edu
    • http://bugs.freehep.org
• forum
    • http://forum.linearcollider.org
              Acknowledgements                            37




• SLAC
    • Norman Graf, Tony Johnson, Ron Cassell, Mark
    Donzelman, Makoto Asai, Dennis Wright, Jan Strube
• NICADD
    • Guilherme Lima, Rob McIntosh, Jerry Blazey, Serguei
    Uzunyan, Vishnu Zutshi, Dhiman Chakraborty
• Others
    • Frank Gaede, Guy Barrand, Ties Behnke, Roman Poeschl,
    Steve Magill
• People from Snowmass 2005 whose slides I borrowed
• The ILC community

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:22
posted:10/7/2011
language:English
pages:37