Geant4 8.3 Release Notes

May 4th, 2007

The code and binary libraries for the supported systems are available through our Source Code Web page.

Please refer to the Geant4 User Documentation for further information about using Geant4.

Contents

1. Supported and Tested Platforms
2. CLHEP and AIDA
3. Items for migration of the user code
4. Expected effects on physics and performance
5. Known Run-Time Problems
6. Compilation Warnings
7. Known Run-Time Warnings
8. Geant4 Software License
9. Detailed list of changes and fixes

1. Supported and Tested Platforms

• SUN Solaris 5.8, C++ CC-5.5.
• Linux, gcc 3.2.3 or gcc 3.4.5.
  This configuration was tested on 32 bits architectures (Intel) and 64 bits architectures (AMD) with the Scientific Linux CERN 3 (SLC3) distribution (based on RedHat Linux Enterprise 3) and Scientific Linux CERN 4 (SLC4) distribution (based on RedHat Linux Enterprise 4). Versions of Geant4 have also been compiled successfully on other Linux distributions, like Debian, Suse or other RedHat systems.
• Windows/XP and CygWin Tools with: Visual C++ 7.1 .NET

• SUN Solaris 5.8, C++ CC-5.4 Patch 111715-02
• Linux, gcc-4.1.2
• Linux, Intel-icc 9.1
• MacOS 10.4, gcc-4.0.1
• Windows/XP and CygWin Tools with: Visual C++ 8.0 (Visual Studio 2005)

• AIX 4.3.2, xlC 6.0
• DEC V4.0, cxx C++ V6.1-027
• HP 10.20, aCC C++ B3910B A.01.23
• SGI V6.5.5, CC 7.2.1

2. CLHEP and AIDA

3. Items for migration of the user code

4. Expected effects on physics and performance

• No visible effects in CPU performance.
• Calorimeter response: stable within 1% for visible energy and resolution.
• Energy deposition in thin layers: stable for step size down to 1 um.

• Revised hadronic modeling, resulting in longer shower shapes:
  the FTF model now has improved treatment of diffraction. As a result the longitudinal shower component now starts later and ends later, which goes in the direction of the data. The addition of quasi-elastic scattering (G4QuasiElasticChannel) to the QGS model is also expected to improve shower shapes by reducing the deep inelastic component.
• New cross section data sets added:
  Barashenkov parameterised cross sections are now available for elastic and inelastic nucleon-nucleus and elastic and inelastic pion-nucleus scattering.
• Cross sections for user-defined elements:
  cross sections are now calculated based on user-defined isotope abundances, if they are provided. Natural abundances are used otherwise.
• Improved muon capture model:
  G4QCaptureAtRest model provides more accurate final state electron spectrum for muon capture and more accurate secondary hadron sprectra for hadron capture.
• Error reporting:
  when a segmentation fault occurs in a hadronic process, the information about the process and model name where the fault has occurred will be printed out.

• FTFP and FTFC use a significantly revised FTF model. Comparisons to thin target experiment at ~10 GeV show much better agreement with data; the model is still under validation at higher energies. Prelimirary tests display slightly longer showers than the previous version of FTFP/FTFC.
• QGS type physics lists (QGSP, QGSC, QGSP_BERT, etc) now include a simple quasi-elastic model for a small fraction of high energy inelastic interactions. The cross-section parameterisations for inelastic interactions include the quasi-elastic cross-section, while QGS only models deep inelastic interactions.
  • For comparison or backward compatibility, the new QGSP_NQE, QGSP_EMV_NEQ, and QGSP_BERT_NQE ignore this new quasi-elastic, and are therefore the equivalent of QGSP, QGSP_EMV, and QGSP_BERT available in release 8.2 and before.
  • Also, the obsolete QGSP_LEAD/QGSP_LEAD_HP do not include quasi-elastic.
• QGSC physics list and its variants use CHIPS G4QCaptureAtRest to model capture of muons at rest instead of G4MuonMinusCaptureAtRest. This model better reproduces the spectrum of secondaries close to the kinematic limit.
• The new physics list QGSP_BERT_TRV is a variant of QGSP_BERT where Bertini cascade model is used only up to energies of 5.4 GeV for protons, neutrons, pions and kaons. For comparisons QGSP_BERT uses Bertini cascade model up to 9.9 GeV.
• HElastic model for projectile energy above 1 GeV is used for all targets with Z>1.
• In all lists for electro-magnetic physics, the "skin" for multiple-scattering on boundaries is set to zero, meaning that no computation of the linear distance to the geometrical boundaries is made by the multiple-scattering model.
• In the QBBC physics lists, the Binary Cascade is used up to 4 GeV, the FTFP model between to 3.5 to 12 GeV, and QGSP above 11.5 GeV.

5. Known Run-Time Problems and Limitations

6. Compilation Warnings

7. Known Run-Time Warnings

    G4PropagateInField: Warning: Particle is looping 
    -  tracking in field will be stopped. 
       It has performed 1000 steps in Field while a maximum of 1000
       are allowed. 

8. Geant4 Software License

9. Detailed list of changes and fixes

• Protected generation of dependencies using internal make variable MAKECMDGOALS in binmake.gmk and common.gmk, following suggestion made in problem report #930. This protection resolves the problem of generation of dependencies when issueing the clean target.

• Fixed way of counting outward steps in flux scorers.

• Muons
  • G4MuPairProductionModel: added protection against zero or negative partial cross section inside SelectRandomAtom() method. Fixes problem reported when runnning tutorial example A01.
  • Set linLossLimit to 1.E-6 in G4EnergyLossForExtrapolator.
• Polarisation
  • G4PolarizedGammaConversionModel: fixed non-initialised variable.
• Standard
  • G4UrbanMscModel:
    • Fix in the single scattering code for heavy particles, now the code can be used with skin=1 e+/e- and for heavy particles as well.
    • Added protections against NaN values in two places: sampling of theta when tau=0 in lateral displacement, and in method ComputeTrueStepLength() for the case of zero true length.
    • Stepping algorithm changed for skin=0, now there are 3 stepping modes:
      1. Stepping similar to that performed in release 7.1;
      2. skin=0 - designed for high energy simulations with or without magnetic field;
      3. skin=1 - designed for 'precision' simulations without magnetic field.
    • If skin>1, perform small steps with single scattering before and after boundary the boundary. The step is reduced before boundary for geomlimit less than geombig only.
    • Use tPathLength inside ComputeStep() instead of geombig.
  • G4MultipleScattering, G4hMultipleScattering:
    • Use skin parameter from the base class (default skin=0).
  • G4eBremsstrahlungModel:
    • Added reinitialisation of LPM flag between runs.
    • Corrected LPMconstant by factor 2 (according to SLAC recent publication).
  • G4BetheHeitlerModel: for any initial energy select atom of the material on which conversion occur (needed for polarisation module).
  • G4UniversalFluctuation: GLANDZ part of the fluctuation model has been completely rewritten, there is no longer distinct loss regime.
  • G4hIonisation: removed dependence on K+ and K- particles.
• Utils
  • G4VEnergyLossProcess:
    • Use G4SafetyHelper instead of G4Navigator for determining distance to boundary, and added protection agains zero step length.
  • G4VMultipleScattering:
    • Added protection against zero step or energy.
    • Added SetSkin() and Skin() methods.
  • G4EmModelManager: added maxCutInRange variable for sub-cutoff option.
  • G4EmCalculator: added method ComputeEnergyCutFromRangeCut().
  • G4EnergyLossMessenger: send message /run/physicsModified after all commands requiring reinitialisation.
  • G4LossTableManager, G4EmProcessOptions, G4EnergyLossMessenger: added commands for MscLateralDisplacement, MscSkin and linearLossLimit.
  • G4LossTableBuilder: use more precise algorithm to build inverse range table using G4LPhysicsFreeVector.
  • G4EmCorrections: taking mass of ion from G4IonTable to avoid division by zero.

• Prevent from unnecessary creation of G4String objects in G4Event, which were default for random number status. Use new flag in G4EventManager.

• Cuts
  • Added method ConvertRangeToEnergy() to G4ProductionCutsTable.
• Scoring
  • Fix to set step status for ghost pre-step point.

• Magnetic field
  • Corrected integration constants in G4MagHelicalStepper.
  • Added implementation of G4RKG3_Stepper, stepper derived from Geant3.
  • G4MagIntegratorDriver: added protections against zero initial or internal step h=0; revised (re)calculation of h, and protect more against very small steps.
• Management
  • Added accessors to stores for retrivial of volumes/solid by name.
  • Modified signature of G4VPhysicalVolume::CheckOverlaps() to introduce ability of handling tolerances for overlaps.
  • Do not reset RegionRootFlag in RemoveRootLogicalVolume() if volume is the world, since the volume itself may be already removed from the store if G4LogicalVolumeStore::Clean() is called.
• Navigation
  • In G4Navigator, forcing call to CheckOverlaps() on the current volume before event abortion for cases of stuck tracks.
  • G4PropagatorInField: implemented ClearPropagatorState() to clear track-dependent state (safety, zero-steps, loop flag. etc...). Call to this method becomes mandatory in G4Transportation for each new track to overcome side effects of safety inconsistency in presence of geometry overlaps.
  • Fixed case of out of range access in G4GeomTestSegment responsible for crashes observed when recursive-grid test for overlaps was run. Addressing problem report #784.
• Solids (Specific)
  • Introduced new shape G4ExtrudedSolid, a solid representing the extrusion of an arbitrary polygon with fixed outline and multiple Z sections.
  • G4TessellatedSolid:
    • Made relevant solid's methods virtual, to be used as base class.
    • Added implementation for GetSurfaceArea() and GetPointOnSurface() methods.
    • Added protection in SurfaceNormal() method in case of facet not found.
    • Check and return kOutside in Inside() if point out of solid's extent.
    • Bug fix in G4TriangularFacet::Intersect() to allow for correct update of Distance in condition. Cleared unecessary declaration of IsInside() methods in both G4TriangularFacet and G4QuadrangularFacet.
    • Fix in G4PolyconeSide::DistanceAway() for correct estimation of distance for the case of Phi sections. Addresses problem report #598.
    • Fix in GetPointOnSurface() for G4Polycone to consider also points laying on rings when consecutive Z-sections are coincident. Added method GetPointOnRing().
• Volumes
  • Extensions to the CheckOverlaps() algorithm:
    • Added possibility to define a tolerance for reporting overlaps; added new argument to signature, by default tolerance is set to zero.
    • Optimised implementation for detecting daughter volumes which are totally overlapping and checked in sequence.
    • Added check on current volume to optimise and correctly handle the case of explicit call of the method when geometry is constructed.
    • Made warning message more clear.

• Added G4TwoVector type, typef of CLHEP's Hep2Vector class.
• Changed date for release 8.3.

• G4VisAttributes:
  • Bug fix in SetForceLineSegmentsPerCircle(); now correctly implementing the reset.
  • Minor refinement: introduced IsForceLineSegmentsPerCircle() method.
• G4Colour: replaced static data members by static functions that create and copy out the relevant colours.

• Added model name for hadronic whiteboard error reporting.
• Cross sections
  • Extend user-chosen isotope abundance capability to several classes: G4UPiNuclearCrossSection, G4UInelasticCrossSection, G4UElasticCrossSection, G4GlauberGribovCrossSection, G4NucleonNuclearCrossSection.
  • G4PiNuclearCrossSection: removed special treatment ('tuning') for Copper (Z=29),i.e. the Cu cross section is no longer lowered by 4%.
  • New classes added:
    • G4BGGNucleonElasticXS: provided proton and neutron elastic and cross section class (G4BGGNucleonElasticXS) which uses the Barashenkov parameterization below 100 GeV and Glauber-Gribov model above.
    • also provided proton and neutron inelastic cross sections (G4BGGNucleonInelasticXS), pion elastic cross sections (G4BGGPionElasticXS), and pion inelastic cross sections (G4BGGPionInelasticXS) which combine the Barashenkov parameterization below 100 GeV and Glauber-Gribov model above.
    • G4NucleonNuclearCrossSection which contains Barashenkov cross sections for nucleon-nucleus scattering
  • Added protection against division by zero in method G4CrossSectionDataStore::SelectRandomIsotope().
  • Added handling of user-defined isotope abundances for elements and proper cross section and abundance weighted choice of isotopes in reactions. New virtual methods for IsApplicable() and GetCrossSections() added to classes G4VCrossSectionDataSet and G4CrossSectionDataStore.
  • G4HadronCrossSections::GetParticleCode(): following report from CMS, replaced code which tested by particle name with more performant code that tests by PDG ID. Also, use switch-case instead of else-if.
• Management
  • G4HadronicProcess:
    • Removed unused AlwaysKillLeadingHadron() method.
    • Fixed bug which was killing primaries in elastic scattering.
    • Fix weighting error in leading particle killer.
    • Added handling of user-defined isotopes: re-implemented methods GetMeanFreePath() and ChooseAandZ() to use new methods in G4CrossSectionDataStore.
    • No longer use dispatch pointer to call data store methods.
    • Remove data members currentZ, currentN and methods GetCurrentZ(), GetCurrentN(). They are no longer needed.
• Binary Cascade
  • Added protection against integrating zero timeStep in G4RKPropagation. This used to cause in some cases division by zero in integration.
• Cascade
  • Added protection in G4ElementaryParticleCollider against negative value to sqrt().
• Chiral Invariant Phase Space (CHIPS)
  • Fix for NaN, protected coulomb barrier against divide by zero, now defaults to zero. Fix for NaN in G4QNucleus when using pow() and in G4QEnvironment.
  • Completed implementation of G4QuasiFreeRatios which enables quasi-elastic scattering.
  • To avoid early particle definition the CHIPS world is now created at run time in the processes G4QElastic, G4QCollision, and G4QCaptureAtRest.
  • G4QElastic:
    • Fix for mis-use of units.
    • Treat particles with E=0 as alive for tracking purposes.
  • CHIPS model for muon capture now available. It is handled by G4QCaptureAtRest.
  • Bug fix for muon-nuclear interactions.
  • Bug fix for binomial distribution with p=1,q=0 in G4QNucleus.
• Elastic scattering
  • Speed up coherent hadron-nucleus scattering (G4ElasticHadrNucleusHE) by using iterator search on bin instead of computing log().
  • G4HadronElastic:
    • Use the most probable isotope for given Z to call the high energy part of the model (CPU/memory more efficient).
    • HElastic model for projectile energy above 1 GeV is used for all targets with Z>1.
  • G4UHadronElasticProcess: fixed bug in number of neutrons in CHIPS interface.
  • G4HadronElastic, G4ChargeExchange: more accurately compute Z and A.
  • G4UHadronElasticProcess and G4ChargeExchangeProcess:
    • Fixed bug of undefined isotope vector
    • Migrated to interface for isotope-dependent cross sections; do not use QElastic total cross sections in G4ChargeExchangeProcess.
• High Energy
  • Added G4QuasiElasticChannel to simulate quasi-elastic scattering. Relies on CHIPS class G4QuasiFreeRatios.
  • Modified G4TheoFSGenerator to use G4QuasiElasticChannel and add quasi-elastic scattering for string models, and use these in competition to the string model.
• Low Energy
  • G4LEAntiProtonInelastic: for low energy (< 0.1 MeV) anti-protons, the model was simply returning the particle change which had not yet been initialized, thus leading to an infinite loop. Now model returns particle change initialized from the original kinematics. Addresses problem report #888.
  • All Low Energy models: added model name to constructor for use in traceback and debugging.
  • G4LElastic: added protection against sqrt() of negative value.
• Parton String
  • Several modifications made to G4VLongitudinalStringFragmentation for use in FTF model. This was taken from the original QGS code.
  • Revised FTF model now available. Single diffraction was added and hadronization was improved. Management and qgsm modules also affected.
  • G4ExcitedStringDecay::EnergyAndMomentumCorrector(), no longer causes exception when not converging. Added more info when failing and return correct status. Added protection in G4ExcitedStringDecay against division by 0.
• Pre-Compound
  • Added protection against division by zero in G4PreCompoundTransitions.
• Utils
  • G4HadronicWhiteboard: added methods to set process and model names.
  • G4ReactionDynamics: added pi0 to anti-particle test.

• Added protection against negative argument to sqrt() in G4DecayProducts::Boost().

• Added quasi-elastic in QGSP physics list; this required changes to:
  • QGSP{proton,Neutron,PiK} Builder, adding QuasiElasticChannel on conditional flag.
  • HadronPhysicsQGSP passes flag (default=false) to Builders.
  • QGSP uses flag to include quasi-elastic.
• New list QGSP_NQSE, a copy of QGSP without quasi-elastic.
• For QGSC, QGSC_EMV, and QGSC_EFLOW enable use of G4QCaptureAtRest() also for muons. Changes made in G4QStoppingPhysics as well.
• Modifications for Quasi-elastic. Added to all QGS physics lists execpt the two obsolete QGSC_LEAD/QGSC_LEAD_HP. This also required adapting of HadronsPhysicsXYZ and {QGSP;QGSC;QGSC_EFLOW}{Neutron,Proton,PiK} Builders.
• Removed unnecessary HadronPhysics...{EMV,EMX} classes.
• Adopting recent upgrades in QGS and FTF for the QBBC physics list.
• Set skin to zero in G4EmStandardPhysics7 (i.e. no use of ComputeStep() in multiple-scattering).
• G4HadronHElasticPhysics uses new cross section classes G4BGGNucleonElasticXS and G4BGGPionElasticXS.
• G4HadronElasticPhysics: set limits on low energy and QElastic to zero.
• Fixed special cuts for LBE physics list.
• Added set min/max Energy and Inelastic Energy to NeutronHP builder.
• G4EmStandardPhysics:
  • Use G4hMultipleScattering for muons and removed extra printouts.
  • Use default multiple-scattering setup in G4EmStandardPhysics72.
  • Use skin=0 (no use of ComputeStep() in multiple-scattering) in G4EmStandardPhysics and G4EmStandardPhysics71.
• Added new physics list QGSP_BERT_TRV, based on QGSP_BERT, but changing to Bertini Cascade occurs at 5.0-5.4 GeV for proton, neutron, pion and kaon.
• Increased version number to 5.2.

• Prevent from unnecessary creation of G4String objects in G4Event, which were default for random number status.
• Added new G4RunManager flag and corresponding UI command.

• Management
  • Added command /vis/abortReviewKeptEvents and improved guidance of command /vis/reviewKeptEvents.
  • Introduced command /vis/geometry/set/forceLineSegmentsPerCircle.
  • Fixed bug concerning call to /vis/open in /vis/reviewKeptEvents loop.
  • G4VSceneHandler:
    • Introduced utility LoadAtts() for G4AttValues and G4AttDefs.
    • Improved other utilities (changed signature of GetLineWidth()).
  • Reduced WARNING noise and other minor improvements.
• Modeling
  • Bug fix: corrected treatment of ForcedLineSegmentsPerCircle when clipping.
• OpenGL
  • Implemented bitmap markers (for MarkerSizeType == screen).
  • Bug fix: added missing break in X colormap lookup routine.
  • Adapted to change of signature of GetLineWidth().
  • Fixed Windows problem with DBL_MAX and compilation warnings.
  • G4OpenGLXViewer: Small bug fix in print(); context was switched but not switched back.
• OpenInventor
  • Moved LoadAtts() to vis management module.
• RayTracer
  • Bug fix: use of window size hints.
• ASCIITree
  • Bug fix: format of total mass.

• Updated reference outputs.
• extended
  • electromagnetic/TestEm2
    • Uncommented hadron ionisation in Livermore builder.
  • electromagnetic/TestEm3
    • Fixed compilation problem on SUN.
    • Added command /testem/run/limitEdep allowing to compute restricted mean values to supress big statistical fluctuations.
    • Updated command /testem/run/acceptance.
    • Parameter skin=1 is set as default PhysList.
    • Added macro emtutor.mac.
    • Added physics-lists names: penelope, livermore.
    • Restored random seed storage.
    • By default do not store random seed and do not activate VisManager in batch.
    • Added G4standard_exp PhysList. Added material Lucite.
    • Return Z-position computation.
  • electromagnetic/TestEm5
    • Use all default values in default PhysList.
    • Use hadronic builders from physics_lists library.
  • electromagnetic/TestEm7
    • Parameter skin=1 is set as default PhysList.
    • Added physics-lists names: penelope, livermore.
    • Fixed usage of elastic builders in PhysicsList.
  • electromagnetic/TestEm9
    • Compute truncated mean values to reduce statistical fluctuations and fixed output of statistics.
    • Added physics-lists names: penelope, livermore.
    • Uncommented hadron ionisation in Livermore builder.
  • electromagnetic/TestEm12
    • Uncommented hadron ionisation in Livermore builder.
    • HistoManager: by default, get csdaRange from G4EmCalculator.
  • electromagnetic/TestEm13, TestEm14
    • vis.mac: suppressed command drawTracks.
  • electromagnetic/TestEm15
    • SteppingAction: added protection if no lateral displacement.
  • electromagnetic/TestEm18
    • New example allowing for studying the energy lost by a charged particle in a single layer, due to ionization and bremsstrahlung. Results are compared to 'reference' values.
  • field/field03
    • Fixed bug in F03FieldSetup for the activation of local fields.
  • hadronic/Hadr01
    • PhysicsList update according to QBBC and quasi-elastic modifications.
    • Fix in QGSP_BIC_HP physics list. Addressing problem report #937.
  • medical/fanoCavity
    • New example for computing the dose deposited in an ionization chamber by a monoenergetic photon beam.
  • radioactivedecay/exrdm
    • Fixed usage of physics-lists.
  • runAndEvent
    • RE01: Removed large-N reference output.