source: trunk/ReleaseNotes/ReleaseNotes4.3.1.txt

                 Geant4 3.1 Release Notes
                 ------------------------

                                                      6th April 2001

These are the main new features/fixes included in this release since
release 3.0 (for the detailed list of fixes/additions, please refer to the
related History files):

  o Configuration:
    -------------
    - Added/enabled code in configuration files to store path to dynamic
      library from example/test directory in the executable.

  o Digits+Hits:
    -----------
    - Migrated classes to use STL <vector> instead of RW wrappers.

  o Event:
    -----
    - Migrated classes to use STL <vector> instead of RW wrappers.
    - Added proper time to G4PrimaryParticle.
    - Fixed bug in G4GeneralParticleSource.

  o Electromagnetic Processes (Low-energy):
    --------------------------------------
    - Changes for STL migration of G4OrderedTable and G4DataVector.
    - Removed obsolete class G4Data.

  o Electromagnetic Processes (Standard):
    ------------------------------------
    - Small corrections in e/hEnergyLoss to ensure consistency between
      cut in range and step limitation due to energy loss (finalRange
      data member <-> cut in range)
    - Change in energy loss computation in eBrems process.
      Parametrisation of energy loss due to soft brems has been corrected
      for big cut values.
    - Bug-fix in e/hEnergyLoss processes for AlongStepDoIt().
      Do not use PostStep information in AlongStepDoIt() when generating
      sub-cutoff delta rays.
    - Small corrections in MuEnergyLoss process.
    - Bug fixed in MultipleScattering: there was an error in computation
      of the transport cross section, the cross section was too small for
      low energy (below few MeV for electrons).
    - Bug fixed in G4MuBremstrahlung::PostStepDoIt() in sampling energy
      of generated photons. Fixes the problem sometimes occuring of
      negative interaction length detected.
    - Fix at small cut value for error in loss fluctuations 
      (for example in Al 100 microns).
    - Suppressed protection in PhotoElectric process.
    - Adopt fStopButAlive in eplusAnnihilation process.
    - Introduced new module 'integral' for integral processes: moved
      existing integral processes (G4I*, G4VI*) therein.

  o General Processes, Parameterisation, Decay, Optical:
    ---------------------------------------------------
    - Modified G4OpRayleigh to allow for Rayleigh scattering length to
      be defined for arbitrary material through the use of 
      G4MaterialPropertyTable. This deals with bug report #199.
    - Fixes in the DoIt of G4OpRayleigh: to allow for positive and negative
      CosTheta and to force the new momentum direction of the photon after
      scattering to be in the same plane as the new and old polarization 
      vectors. Deals with bug report #207.
    - Added abstract class G4VExtDecayer for external decayer and modified
      G4Decay accordingly.
    - Modified Store/RetrievePhysicsTable() methods: added a bool argument
      to specify ascii/binary mode.

  o Geometry:
    --------
    - Boolean solids:
      o Minor fix to labels in G4Exception calls.
        Fixes problem report #200. 
      o G4IntersectionSolid:
        - Fix in CalculateExtent() for computation of 'pMin'
          and 'pMax', to avoid assignment in case CalculateExtent() of
          one component solid fails. 
        - Fix in Inside() to return 'kOutside' in case first component
          solid's Inside() returns 'kOutside'. 
      o G4UnionSolid:
        - Fix in DistanceToOut(p,v,...) for assignment of argument *n to
          happen only if calcNorm is false. 
        - Fix in CalculateExtent() for computation of 'pMin' and 'pMax'.

    - CSG solids
      o G4Tubs:
        - Corrections in phi-section algorithm of Inside(p). 
        - Fixed condition from (pPhi<0) to (pPhi<-0.5*kAngleTolerance) 
          in Inside(p). Fixes problem report #198.
        - Fixed bug in function Inside(p) in computation of pPhi 
          for points at phi = 0 if the (start phi)+(delta phi) is
          at phi = 0. 
        - Temporarly modified CalculateExtent(...), now simply based
          on G4Box::CalculateExtent(...).
      o G4Box:
        - Introduced checks against negative (or <2*kCarTolerance)
          dimensions in methods SetXHalfLength(), SetYHalfLength() and
          SetZHalfLength().
        - Fixes to algorithm for computation of rotated vertices in
          CalculateExtent(). 
      o G4Torus:
        - Extracted logic for G4PolynomialSolver now included
          in global/HEPNumerics. 

    - Specific solids:
      o Fix in G4Hype::DistanceToOut(p,v,...): added protection against
        zero v.z.
      o G4IntersectingCone: make small adjustment to help with roundoff
        error in LineHitsCone1() and LineHitsCone2(). 

    - Magnetic Field
      o Minor fix to G4MagneticField.hh: added explicit call to G4Field()
        in copy constructor. Fixes problem report #205.
      o Fixed G4MagHelicalStepper::AdvanceHelix for momentum, allowing
        the possibility of (correct) use of all Helical Steppers. 
      o Corrected field components in G4UniformElectricField. 

    - Management
      o Introduced G4AssemblyVolume: helper class to combine
        volumes together in an arbitrary way in the 3D space. 

    - Volumes
      o Fix in G4ReplicaNavigation::ComputeTransformation() to consider
        'pVol' for SetPhiTransformation() in the kPhi case.
      o Fix in G4ReplicaNavigation::ComputeStep() to initialise 'ourStep'
        to kCarTolerance.

  o Global:
    ------
    - Migrated G4PhysicsTable, G4PhysicsVector and all related classes
      to use STL <vector> instead of RW wrappers.
    - Fix to G4Allocator destructor: bug report #181.
    - Added generic class G4PolynomialSolver in HEPNumerics module.
    - Removed from G4PhysicsVector dependency over G4PhysicsTable. 
      Removed methods: LinkPhysicsTable(), IsLinkedTableExist() and 
      GetNextTable(). Removed data-member ptrNextTable. 
    - Added new unit 'parsec' to SystemOfUnits.h 
    - Fixed bug in constructor of G4PhysicsLnVector: initialised
      'dataVector' to zero. 
    - Added Store/RetrievePhysicsTable methods in G4PhysicsTable. 
    - Added Store/Retrieve and GetType methods in G4PhysicsVector and
      related. 
    - Introduced G4PhysicsVectorType class (enumeration). 
    - G4UnitsTable: resolved ambiguity of old/new style iostream for
      formatting in operator<<. 

  o Graphics Representations:
    ------------------------
    - Tidying for CodeWizard. Added virtual destructors to
      HepPolyhedronBox, etc. Also made all constructors non-inline. 
    - Removed G4Circle.icc and G4VVisPrim.icc.

  o Hadronic Processes:
    ------------------
    - Bug fix for very low energy ion cross-sections; should have no
      visible effect.
    - Moved inline implementation of G4Solver to G4Solver.icc. 

  o Materials:
    ---------
    - Bug fixed in G4IonisParamMat.cc : E100eV instead of E100keV
      This bug affected the values of the density correction parameters
      in the energy loss (dE/dx) computation: ~3-5 % error in dE/dx in
      the energy range 1 MeV - 50 MeV !
    - Added new method G4Material::GetNuclearInterLength().
    - Bug fix concerning units in G4SandiaTable::GetSandiaCofPerAtom().
      Reimplemented G4SandiaTable::ComputeMatSandiaMatrix().
    - Added protection in GetSandiaCofForMaterial().
    - fnulcof added as data member in G4SandiaTable. 
    - Revision of destructor, copy constructor, assignement operator in 
      G4IonisParam, G4Isotope, G4Element and G4Material, according to
      problem report #106.
    - Removed warning to cerr in GetIsotope(name), GetElement(name), 
      and GetMaterial(name). 

  o Particles:
    ---------
    - Added pre-assignedDecayTime to G4DynamicParticle.
    - Bug fix in calculation of electron energy in G4MuDecayChannel.
      Fixes problem report #224.
    - Fixed a bug in G4DynamicParticle::SetDefinition() for problem
      report #228.
    - Added RestoreCuts() methods to G4ParticleWithCuts class.

  o Persistency:
    -----------
    - G4PDCofThisEvent.ddl: fixed a bug in GetNumberOfCollections(). 

  o Readout:
    -------
    - Migrated classes to use STL <vector> instead of RW wrappers.

  o Run and Intercoms:
    -----------------
    - Migrated classes to use STL <vector> instead of RW wrappers.

  o Tracking & Track:
    ----------------
    - Changes for STL migration of G4OrderedTable and G4DataVector.
    - Migrated classes to use STL <vector> instead of RW wrappers.
    - Added comments to headers for the Software Reference Manual. 
    - Added copy constructor for G4Track.
    - Fix in G4Track::GetVelocity() for Gamma
    - Fixes to G4StepPoint::GetVelocity() for problem report #208
      and to return correct value in case of optical-photon.
    - Added call to SetVelocity() in G4Step::InitializeStep().
    - G4PArticleChange: fix in case momentum is zero, (even if
      energy>0.0 and very small) to overcome problems due to
      precision issues.

  o Transportation:
    --------------
    - Improved calculation of energy at the end of the field.
    - G4Transportation modified to use new FieldTrack, and get Kinetic
      Energy from it. This implements integration in Momentum (*c-light)
      instead of velocity. 

  o Visualization:
    -------------
    - Changed rotation of phi-replicas in G4PhysicalVolumeModel.cc to
      ensure correspondence with rule that phi=0 bisects the replica. 

      WARNING: Prior to this fix the graphics did NOT correspond to the
      replica navigation. If you have been relying on the graphics to
      position daughter volumes in a phi-replica, navigation will have
      been incorrect, possibly without causing a crash. The above fix
      will change the apparent position as shown by the graphics and you
      will have to correct the positioning of daughters. Make sure the 
      daughters are all properly enclosed within the replica. Graphics 
      and navigation now correspond. 
    - Added auto-refresh facility for OpenGL. 
    - Bug fixes for problems: 
      #120: improved up vector calculation to avoid view "flipping"
            at (90, 90). 
      #157: fixed crash when viewer was defined before scene. 
      #204: fixed failure to change number of sides per circle. 
    - Fixed problem when requested physical volume was at depth 2 or more. 
    - Migrated classes to use STL <vector> instead of RW wrappers.
    - Introduced many new commands (listed below). Old commands print a
      DEPRECATED warning. 
      o New compound command: 
        /vis/drawView [<theta-deg>] [<phi-deg>] 
                      [<pan-right>] [<pan-up>] [<pan-unit>] 
                      [<zoom-factor>] [<dolly>] [<dolly-unit>]
      o New primitive commands: 
        /vis/enable [true|false] 
        /vis/disable 
        /vis/verbose [] 
        /vis/viewer/set/lineSegmentsPerCircle [] 
        /vis/viewer/set/sectionPlane ... 
        /vis/viewer/set/autoRefresh 
        /vis/viewer/set/projection 
        /vis/viewer/dolly 
        /vis/viewer/dollyTo 
        /vis/viewer/lightsThetaPhi 
        /vis/viewer/lightsVector 
        /vis/viewer/pan 
        /vis/viewer/panTo 
        /vis/viewer/viewpointThetaPhi 
        /vis/viewer/viewpointVector 
        /vis/viewer/zoom 
        /vis/viewer/zoomTo 
        /vis/viewer/set/hiddenMarker 
        /vis/viewer/set/lightsMove

  o G4toG4 Tool:
    -----------
    - Fixed bug in G3toG4BuildTree(G3VolTableEntry*, G3VolTableEntry*) 
      when positioning volume copies. Index is now decremented. 
      Added -DG3G4DEBUG to CPPFLAGS in GNUmakefile.
      Fixes bug report #188.
    - Added support for replicated "PARA" shapes in
      G3Division::CreateSolid(...). Slices are placed with
      G4PVPlacement instead of G4PVReplica for this particular case.
    - G3toG4MakeSolid.cc: fixed bug in construction of G4Sphere where
      'theta' and 'phi' parameters were reversed according to G4Sphere
      constructor. Fixes problem report #230. 

  o Examples:
    --------
    - Introduced modular Physics List in exampleN04.
    - ExampleN02: fix to ExN02ChamberParameterisation and
      ExN02DetectorConstruction to cope with bug report #186.
    - Added G4MuonMinusCaptureAtRest in physics list of ExampleN04.


Notes: 

  o  Geant4 3.1 is based on STL and can be installed with strict
     ISO/ANSI C++ compliant setup or not. See notes below.

The code and some binary libraries are available through our "Source Code"
Web page:
- see our Geant4 Home Page (http://cern.ch/geant4).

Please refer to the Geant4 User Documentation
(http://cern.ch/geant4/G4UsersDocuments/Overview/html)
for further information about using Geant4.
Updates to the documentation for Geant4 3.1 are expected to be published as
usual within three weeks from the release date.


Contents
--------

  1. Supported and Tested Platforms
  2. ANAPHE and CLHEP
  3. The Standard Template Library (STL)
  4. Persistency
  5. Compiler Specific Problems
  6. Known Run-Time Problems
  7. Compilation Warnings
  8. Known Run-Time Warnings


1. Supported and Tested Platforms
---------------------------------

  o SUN Solaris 5.6/5.7, C++ 4.2 Patch 104631-04, C++ 5.2 Patch 109508-02.
  o HP 10.20, aCC C++ B3910B A.01.23
  o Linux 2.2.17-7, gcc C++ egcs-2.91.66 (egcs 1.1.2), gcc 2.95.2.
      This configuration was tested with the RedHat 6.1 distribution, but
      versions of Geant4 have also been compiled successfully on Debian
      and Suse distributions.
  o NT 4.0 with Service Pack 4 and Cygnus Tools b20 with:
      Visual C++ 6.0 Service Pack 2

Platforms also tested but giving rise to some problems - see below:

  o DEC V4.0, cxx C++ V6.1-027.

Platforms configured but not tested and not supported:
  o SGI V6.5.5, CC 7.2.1 with ObjectSpace
  o AIX 4.3.2, xlC compiler with ObjectSpace


2. ANAPHE and CLHEP
-------------------

Geant4 3.1 requires the installation of CLHEP. 

See CLHEP (http://cern.ch/clhep/INSTALLATION/clhep.html).
Tests have been performed with CLHEP 1.6.0.0.

By default, the CLHEP installation scripts will activate ISO/ANSI C++ features
if the compiler is capable. You need to explicitly disable CLHEP settings in
case you wish non-ISO/ANSI setup on specific compilers.

Geant4 3.1 coworks with ANAPHE 3.2.0 and components included therein:
  Objectivity 5.2.1 
  ObjectSpace 2.1
  HepODBMS 0.3.1.4
  CLHEP 1.6.0.0

For more information about ANAPHE, see:  http://cern.ch/anaphe


3. The Standard Template Library
--------------------------------

The following versions of STL have been tested:

   ObjectSpace STL on SUN (CC 4.2).

          See ObjectSpace (http://www.objectspace.com).

   "Native" STL on: Windows/NT, Linux, HP, DEC and SUN (CC 5.2).

This is selected at installation/compile time by the environment variable
G4USE_OSPACE in the case of the former - see documentation.

NOTES:
- Be aware that by default, the installation scripts will expect to use
  the native STL implementation. You must set the G4USE_OSPACE variable
  in the environment in order to select the ObjectSpace implementation.

- Platform specific setup:
  o DEC:   ObjectSpace is required in non ISO/ANSI setup.
           Native STL is required for ISO/ANSI setup.
  o HP:    either ObjectSpace or native STL can be used (non ISO/ANSI setup).
  o SUN:   ObjectSpace is required in non ISO/ANSI setup (CC 4.2).
           Native STL is required for ISO/ANSI setup (CC 5.2 only).
  o Linux: Native STL is required (non ISO/ANSI setup).
  o NT:    Native STL and ISO/ANSI setup required.

In the interest of stability, part of the code in Geant4 3.1 still makes use
of the "STL Interface" wrapper classes, used for allowing the migration from
Rogue Wave Tools.h++ container library.  For these the emphasis has remained
on correct behaviour, rather than performance.


4. Persistency
--------------

  See release notes for Geant4 3.0.   

5. Compiler Specific Problems
-----------------------------

  o DEC V4.0, cxx C++ V6.1-027.

    - We have found cases of mis-compilation of min and max templates
      on DEC.  We have traced those which cause serious malfunction
      but there is no guarantee that there are no other cases.
      Detailed inspection of assembler output suggests that this
      problem might not be confined to min and max.  This has been
      reported to DEC/Compaq.  We understand a new version 6.2 of the
      compiler is now available.  However we cannot guarantee correct
      execution of Geant4 on DEC with the current version of the
      compiler.
    - A bug in cin.getline() system function causes an anomalous
      behavior when running in interactive-mode on DEC-cxx 6.1 in
      ISO/ANSI mode. We tested the bug has been fixed in more recent
      versions of the compiler.


6. Known Run-Time Problems and Limitations
------------------------------------------

For a complete list of outstanding run-time problems and to submit any
problem you may find running this version of Geant4, please refer to the
Geant4 Problem Reporting System, available on the Web at this address:

    http://cern.ch/geant4/problemreport


7. Compilation Warnings
-----------------------

There are compilation warnings on some platforms.  We do not believe
that any will lead to incorrect run-time behaviour, but we are working
on reducing them.


8. Known Run-Time Warnings
--------------------------

The following messages can be written to error output while tracking.  We
believe none give rise to incorrect behaviour.

  o G4PropagateInField: Warning: Particle is looping 
    -  tracking in field will be stopped. 
       It has performed 10000 steps in Field  while a maximum of 10000
       are allowed.