A GEANT4 Simulation of the Derek Glazier University of Edinburgh

Transcription

1 A GEANT4 Simulation of the Derek Glazier University of Edinburgh

2 GEANT4 Overview Geant4 is the successor of GEANT3, the world-standard toolkit for HEP detector simulation Geant4 is an object-oriented C++ toolkit the goal is to provide all that is needed to build a wide variety of physics simulation applications code is open, modular available for all to download In particular a variety of geometries and physics models can be plugged in Additionally a number of independent visualisation tools can be used Extensive documentation and tutorials provided Principal references: NIM A506, 250 (2003) and IEEE Trans. Nucl. Sci. 53, 270 (2006)

3 User Packages Define material and geometry G4VUserDetectorConstruction Controls detector and target geometries Select appropriate particles and processes and define production threshold(s) G4VUserPhysicsList Define the way of primary particle generation G4VUserPrimaryGeneratorAction Import mkin, AcquMC files Define the way to extract useful information from Geant4 G4UserSteppingAction, G4UserTrackingAction, etc. G4VUserDetectorConstruction,G4UserEventAction G4SensitiveDetector, G4VHit, G4VHitsCollection Defines information passed to AcquRoot via A2CBOutput

4 Electromagnetic Physics standard package (1 kev and up) We use multiple scattering, ionization, bremsstrahlung Compton, pair production, photo-electric, annihilation synchrotron, Cerenkov, transition radiation, high energy muon Also _EMV version =G4.7.1 EM physics ~20% faster low energy package uses database information to extend interactions below 1 kev many of the same processes as offered in standard Possible to track optical photons (from Cerenkov, Scintillation) reflection/refraction, absorption, Rayleigh, wavelength shifting Requires production cuts (minimum distance a particle can travel to be tracked) Default 1mm

5 Hadronic Physics Low Energy and High Energy Parametrized (LEP, HEP) models for all hadrons LEP and HEP models are the re-engineered versions of the GHEISHA models (parametrized from data) fast ~Standard G3 energy is conserved on average, not event-by-event Bertini-style cascade for low energies (< 10 GeV) classical cascade model, uses free-space cross sections Pros and cons: designed for use in HEP trackers, collider detectors good for neutrino beams, kaon interactions Gamma-nuclear model added for E < 3.5 GeV Binary cascade for low energies (< 3 GeV) detailed theory-driven model upper limit due to dependence on resonances Recomended G4 LEP, HEP for hyperons, anti-baryons, LE kaons Gamma-nuclear model added for E < 3.5 GeV: l

6 CrystalBall Geometry Import from GEANT3 complicated by GEANT4 handling of overlapping volumes Could not use major/minor triangle construction Each crystal placed individually Additional materials (skirting, equator) taken from cbsim with someupdates from UCLA Cut crystals in tunnel region realised through Boolean solids i.e CCUT cylinder from G3 is rotated and subtracted from each cut crystal Geometry with RayTracer

7 TAPS Geometry Originally Implemented as in cbsim Additional interactivity added to go between MAMI-B and MAMI-C Boxes, vetos copied from cbsim Dummy crystals only for MAMI-B Vetos read out independent of BaF2 PbWO4 crystals have been Introduced Can be read out into combined BaF2-PbWO4 AcquRoot class BaF2_PWO_09.dat

8 MWPC Geometry MAMI-B MWPC implemented as for cbsim (Jamie Robinson) Just materials no individual strips Can use Sven's ReadDecoded (smears initial 4 vectors for tracks) Updated for MAMI-C (David Howdle) Supports etc from Paulo Individual wires included (not effect CPU much) Position readout for each strip cylinder New AcquRoot ReadDecoded (A2/acquroot/TA2CylMWPC.*) Smears strip position : Phi by wire spacing, Z by Then calculates position in each chamber and track as for real data Should be included in new MWPC classes

9 PID and Targets PID1 and PID2 implemeted Different mountings required for each Accurate lightguide geometrys Right angular wedge shaped scintillators (cbsim uses trapezoid) Standard solid and cryo targets are as for cbsim New solid target for 2008 C beamtimes (J. Robinson) Variable cell size allowed for cryo target(m. Firminger) Users should check their own solid target geometry Do not trust standard implementation!!! Sim. will check to be sure given vertex is inside cell and choose a new one if not!

10 Polarised Target Longitudinal target created by Monica Firminger, Sackville Full geometry including coils and butanol cell Approximate magnetic field consisting 1Tesla in target volume, zero out Requires field map Gives noticable deflection of 100 MeV pions coils cell

11 Primary Generator Action Number of different options for input events AcquMC, (PLUTO?) and mkin ntuple.root files as for cbsim Particle phasespace and overlap new for A2PrimaryGeneratorAction Interactive G4 particle gun useful for testing AcquMC mkin h2root particle phasespace Particle overlapping G4 Interactive ParticleGun Ntuple or TMCPartice /gun/particle pi+ /gun/energy 1 MeV /gun/direction... Run macro A2PrimaryGeneratorAction Original 4 vectors A2CBOutput (Use MCNtuple.h for G3 to PDG particle ID Includes nuclei) G4 Tracking

12 Running A2 Simulation : DetectorSetup.mac ####Use the crystal ball? /A2/det/useCB 1 #####Use TAPS? /A2/det/useTAPS 0 ####Configure TAPS /A2/det/setTAPSFile taps07.dat /A2/det/setTAPSZ 145 cm /A2/det/setTAPSN 384 /A2/det/setTAPSPbWO4Rings 2 ####Use the PID /A2/det/usePID 2 /A2/det/setPIDZ 0. cm /A2/det/useMWPC 2 /A2/det/useTOF 0 /A2/det/setTOFFile TOF.par ##Set the target #/A2/det/useTarget Cryo ###Cryo targets : G4_lH2, A2_lD2### #/A2/det/targetMaterial G4_lH2 /A2/det/setTargetLength 4.8 cm #/A2/det/useTarget Solid /A2/det/useTarget Polarized /A2/det/targetMaterial A2_HeButanol W A R N I N G Please check default DetectorSetup.mac before running. Make sure it is the configuration you want

13 Running A2 Simulation : Run Configuration Macro File doppi0.mac #####Pre-Initialisation #Choose a physics list, for a full listing type #/A2/physics/ListPhysics /A2/physics/Physics QGSP_BIC ####Initialise /run/initialize ##the initial random number seed /A2/generator/Seed #Set the number of particles to be tracked from the input ntuple /A2/generator/NToBeTracked 3 #give the indexes of the particles to be tracked #(have a look at the branch names in the input file) /A2/generator/Track 2 /A2/generator/Track 3 /A2/generator/Track 4 #Open the file and set up the ntuple for reading /A2/generator/InputFile /scratch/dglazier/kin_pi0p_ root #####Output #Open the output file for writing /A2/event/setOutputFile /scratch/dglazier/testqgsp_bic.root Also possible to give input filename as command line Argument, output is then the same appended with tr_

14 Equator material Coherent pi0 analysis, noticed detection efficiency Differed by ~5% between G3/G4 G4 G3 Large deviation in ϕ~0 region Realised G3 had much larger Fe thickness at equator Note, G3 version used Marc/Sven's mod. With same thickness G3/G4 agree to about 1% UCLA prefer thin version Suggestion, make cross sections with 10o gap in ϕ acceptance at equator If the right thickness is used the cross sections should not change

15 Recoil polarimetry (Sikora) Scattered proton angular distribution G4 Binary Cascade model does excellent job of reproducing proton nucleus interactions, Cross sections and angular distributions Φ (π-proton) versus Φπ -structure consistent with Top hemisphere offset ~3mm Is it possible to use the wirechambers to measure the CB positions?

16 Hadronic interactions K+ cross sections Nice agreement with previous measurements of K+Σ cross section Sensitive to K+ nucleus interaction, as cannot tag K+ if inelastic interaction July 07 April 08 JLAB

17 Bug fixes/updates For next release.. Inner can material set to iron (M. Sikora) (sent round already) Vertex position restriction (makes sure event comes from target material) only worked first event! (D. Werthmueller) Allow adjustment of hemisphere seperation S. Prakhov's updated geometry Some significant changes to iron geometries and cut crystals Other suggestions...

18 Open Issues Contributions welcome/needed Detectors : Targets : Cerenkov (Class exists, requires details/volunteer) Polarised Needs transverse geometry (E.Downie) Field map (A.Mushkarenkov) 3He cryo and polarised?? AcquRoot.Offline,.dat, ReadDecoded normalised with standard user classes (see A2/acquroot directory) What additions do experiments need???

19 The Past See presentation R. Brune confid= The first version of GEANT appeared in It was a very simple framework for simulation between NA3, NA4 and Omega experiments. (about 5000 LOC) GEANT2 came in 1977 with more functions to control the initialisation, stepping phases (10000 LOC). GEANT3 came in 1981 in OPAL, then many experiments. It was a huge step. A powerful geometry system, electromagnetic physics based on EGS3/4 and interfaces with hadronic shower packages like Tatina, Gheisha and Fluka. ( LOC)

20 The Present GEANT4 came in 1995 following the directions taken by GEANT3 but written in C++. The geometry system was along the same lines as in GEANT3 and the electromagnetic physics was a continuation (with the same authors) of what was in GEANT3. GEANT4 had a long list of developments and improvements in the physics sector, in particular hadronic physics and this work is going on. Recent reports from LHC have demonstrated the high quality of the simulations with GEANT4 physics.

21 The Future GEANT5 = GEANT4 + ROOT +... = Next year Motivation Primarily from LHC These tools should evolve in a more compatible framework Should incorporate FastMCs (G4 too slow for many analysis) From ROOT want to use I/O, Interpreters, Graphics (event display), Math, Infrastructure, parallelisation, Geometry TGeo i.e. Will look like ROOT will transport like GEANT4 (by default, but other transporters (fast/statistical) will possible)

22 Benefits The Future for the CB? Highly flexible tool (lots of configuration options) Easier to install Parallel processing (including multicore or GPU) Will be further physics developments Predictive modelling Interface to HEP event generators... Keep pace with computational developments Realistic GEANT4 already quite flexible, probably not used (in this collaboration) Flexibility ~ more complicated to run, users just want a start button! Physics developments may have limited impact at MAMI energies GEANT4 physics will continue to be developed seperately Migration would be ~ 2 years away