Changeset 1315 for trunk/source/processes/hadronic/models/cascade/History

Timestamp:

Jun 18, 2010, 11:42:07 AM (14 years ago)

Author:

garnier

Message:

update geant4-09-04-beta-cand-01 interfaces-V09-03-09 vis-V09-03-08

File:

• trunk/source/processes/hadronic/models/cascade/History (modified) (2 diffs)

trunk/source/processes/hadronic/models/cascade/History

@@ +1 @@
+$Id: History,v 1.90 2010/06/08 06:34:43 stesting Exp $
 -------------------------------------------------------------------
 
@@ -14 +15 @@
      * Please list in reverse chronological order (last date on top)
      ---------------------------------------------------------------
+
+08 June 2010 Gunter Folger  (hadr-casc-V09-03-42)
+-  trivial fix for compiler warning on gcc43 on empty body in for(...) in  
+    cascade/include/G4CascadeInterpolator.icc:67
+
+01 June 2010 Michael Kelsey (hadr-casc-V09-03-41)
+-------------------------------------------------
+- paraMaker.cc:  Fix vector initialization for Windows crash.
+
+26 May 2010 Dennis Wright (hadr-casc-V09-03-40)
+-----------------------------------------------
+- G4NucleiModel: remove conversion to fm, and conversion of cross 
+  sections to fm**2
+
+21 May 2010 Michael Kelsey (hadr-casc-V09-03-39)
+------------------------------------------------
+Final tag for inclusion in GEANT 4.9.4-beta.  Will be validated internal and
+included in global Hadronics tag.
+
+- G4CascadeSampler.icc:  Activate time-saving checks on single bins.
+- G4ElementaryParticleCollider.cc:  Remove special case for nucleon elastic
+  scattering.
+
+21 May 2010 Michael Kelsey (kelsey-20100521c)
+---------------------------------------------
+- G4CascadeInterface, G4PreCompoundCascadeInterface, G4InuclEvaporation:
+  Simplify code: fetch G4DynamicParticle out of outgoing G4InuclParticles,
+  rather than creating new ones.  Remove all of the *Collider data members
+  (instantiated within G4InuclCollider).  Move final state rotations to
+  G4CollisionOutput.
+
+- G4InuclParticle:  Provide accessor to return G4DynamicParticle.
+
+- G4CollisionOutput:  Add function to apply LorentzRotation to both lists;
+  used by G4CascadeInterface.
+
+- G4ElementaryParticleCollider:  Avoid memory churn with G4LorentzVector
+  temporaries.
+
+21 May 2010 Michael Kelsey (kelsey-20100521b)
+---------------------------------------------
+- G4VCascadeCollider:  New base class for all Colliders.  Carries
+  verboseLevel data member and set-function.  ::collide() is pure virtual.
+  Additional protected member functions will be defined to include all code
+  shared across multiple Colliders.
+
+- Migrate all of the "Collider" factories to use new base class, removing
+  common member functions, and replace passed-in sibling colliders with
+  local data members (pointers for now).  Since none of these classes have
+  persistent (state preserving) data members, the cost of more than one
+  instantiation is minimal.
+
+        G4BigBanger
+        G4ElementaryParticleCollider
+        G4EquilibriumEvaporator
+        G4EvaporationInuclCollider
+        G4Fissioner
+        G4IntraNucleiCascader
+        G4InuclCollider
+        G4NonEquilibriumEvaporator
+        G4PreCompoundInuclCollider
+
+- G4CascadeInterface.cc, G4InuclEvaporation.cc,
+  G4PreCompoundCascadeInterface.cc:  Remove siblings from constructors for
+  the Colliders.
+
+- G4InteractionCase.hh:  Eliminate unnecessary copying of std::pair<>'s, use
+  initializers in ctors, add clear() function to reset to 0-0-0.  Move
+  evaluation of bullet vs. target to new set() function, replacing separate
+  bulletTargetSetter() in the colliders.  Add boolean accessors to avoid
+  hard-coding inter_case values.  Change code numbers so nucleus collisions
+  are negative, and positive codes refer to the "is" or "rtype" numbers.
+
+- G4ElementaryParticleCollider.cc:  Use G4InteractionCase to compute "is",
+  instead of hardwiring the meaning.
+
+- G4Collider.hh:  Remove empty and never-used file.
+
+- G4EvaporationInuclCollider.cc:  Eliminate unnecessary creation of
+  duplicate G4InuclNuclei: pass input through to G4EquilibriumEvaporator.
+
+- G4LorentzConvertor:  Add interfaces to pass G4InuclParticles directly.
+
+21 May 2010 Michael Kelsey (kelsey-20100521a)
+---------------------------------------------
+Replace hand-coded interpolation of arrays with use of G4CascadeInterpolator.
+In all cases, local arrays are replaced with static const, and the
+interpolator object itself is declared static, to reduce memory churn.
+
+        G4NucleiModel::totalCrossSection()
+        G4EquilibriumEvaporator::getQF()
+        G4InuclSpecialFunctions::paraMaker[Truncated]() (in paraMaker.cc)
+
+- G4CascadSpecialFunctions:  Remove. totalCrossSection() redundant with 
+  G4NucleiModel.  Move absorptionCrossSection() to G4NucleiModel.
+
+- G4CascadeInterpolator.icc:  Bug fix to catch special case of exact upper
+  bin edge (occurs when !doExtrapolation).  If index == last + 0., just
+  return upper edge value, without calculation.
+
+- G4NucleiModel.cc:  Use G4Cascade{Pi*}Channel and G4Cascade{NN,NP,PP}Channel
+  classes to get total cross-sections for pi-N and N-N scattering.  Arrays
+  for kaon and hyperon scattering should be migrated as well, once binning
+  is resolved.  Move absorptionCrossSection() here from SpecialFunctions.
+
+20 May 2010 Michael Kelsey (kelsey-20100521)
+--------------------------------------------
+- G4CascadeInterpolator.icc:  Two bug fixes involving values at array edges.
+
+19 May 2010 Michael Kelsey, Dennis Wright
+-----------------------------------------
+- G4BertiniData.hh,.cc:  Remove unused and unnecessary class.
+- G4NucleiModel.cc: convert nuclear radii to fm and cross sections to 
+    fm**2.  Use six-zone model for A > 99, and make alpha a 1-zone
+    nucleus.
+- G4CascadSpecialFunctions: convert pion absorption cross sections from
+    mb to fm**2, correct spelling of method absorptionCrossSection
+
+17 May 2010 Michael Kelsey (hadr-casc-V09-03-38)
+------------------------------------------------
+- paraMaker.cc (G4InuclSpecialFunctions::paraMaker[Truncated]):  Upper edge
+  of interpolation array not properly handled.  Should be "if (Z >= 70.0)".
+  This affects G4NonEquilibriumEvaporator output.
+
+17 May 2010 Michael Kelsey (hadr-casc-V09-03-37)
+------------------------------------------------
+Bug fixes for compiler warnings (not seen on MacOSX, even with g++ -Wall)
+
+- G4CascadeData.hh: Dummy arrays needed double-braces for initialization.
+
+- G4CascadeInterpolator.icc: Extraneous decimal point in "for (i=0.;".
+
+- G4CascadeKzeroBarNChannel.cc:  Left three "old" arrays in place during
+  conversion.
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-36)
+------------------------------------------------
+Remove obsolete files following reorganization of two-body scattering
+lookup tables.
+
+        G4CascadeElasticInterface.hh,.cc
+        G4ElasticCascadeInterface.hh,.cc
+        G4FinalStateSampler.hh,.cc
+        G4NucleonSampler.hh,.cc
+        G4PionSampler.hh,.cc
+
+- G4CascadeChannel.hh,.cc:  All functionality removed, replaced with
+  namespace containing only getQnums and CheckQnums validation functions.
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-35)
+------------------------------------------------
+Replace independent parallel samplers with templated base class and single
+implementation.  This completes the reorganization of the channel tables.
+
+- G4CascadeSampler:  Redefine as templated base, taking binning array as
+  constructor argument.  Move implementation file from .cc to .icc.
+  NOTE:  G4CascadeSampler.cc is removed, since not consistent with new .hh.
+
+- G4PionNucSampler:  Subclass of G4CascadeSampler<30>, replaces
+  G4FinalStateSampler, using bins array from there.
+
+- G4KaonNucSampler:  Sublcass of G4CascadeSampler<31>, replaces previous
+  G4CascadeSampler, using bins array from there.
+
+- G4Cascade*Channel.hh:  Change all typedefs to pass appropriate of two new
+  samplers in template.
+
+- G4ElementaryParticleCollider.cc: Replace long "if (is==)"-cascades with
+  switch blocks.  Move all getOutgoing calls (for pions and nucleons) into
+  one function, and call it from outside multiplicity checks.  Use
+  G4CascadeInterpolator for angular distributions.  Set modelID=3 in all
+  G4InuclParticle constructions.
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-34)
+------------------------------------------------
+Extract lookup tables from G4PionSampler and G4NucleonSampler into
+individual "classes" defined and initialized with G4CascadeData.
+
+- G4CascadePi{Plus,Minus,Zero}{P,N}Channel.hh: Six new typedefs defined as
+  with the existing G4CascadeXXXChannel, to replace G4PionSampler.
+
+- G4CascadeT33piNChannel.cc: Lookup tables for both PiPlusP and PiMinusN,
+  implemented as with existing G4CascadeXXXChannel, from G4PionSampler.cc.
+
+- G4CascadeT31piNChannel.cc: Lookup tables for both PiMinusP and PiPlusN,
+  implemented as with existing G4CascadeXXXChannel, from G4PionSampler.cc
+
+- G4CascadeT11pizNChannel.cc: Lookup tables for both PiZeroP and PiZeroN,
+  implemented as with existing G4CascadeXXXChannel, from G4PionSampler.cc
+
+- G4Cascade{PP,NP,NN}Channel.hh: Three new typedefs defined as with the
+  existing G4CascadeXXXChannel, to replace G4NucleonSampler.
+
+- G4CascadeT1NNChannel.cc: Lookup tables for both p-p and n-n scattering,
+  implemented as with existing G4CascadeXXXChannel, from G4NucleonSampler.cc
+
+- G4CascadeT0npChannel.cc: Lookup tables for p-n scattering, implemented as
+  with existing G4CascadeXXXChannel, from G4NucleonSampler.cc
+
+- G4CascadeFunctions.hh, .icc: Pass std::vector as non-const argument to
+  getOutgoingParticleTypes() instead of returning internal buffer.
+
+- G4CascadeSampler.cc, G4FinalStateSampler.cc:  Add some checks in
+  findFinalStateIndex, fillSigmaBuffer, and sampleFlat(), such that
+  single-entry ranges don't go through the analysis, just return index 0.
+
+- G4ElementaryParticleCollider.cc: Follow change to getOutgoingParticleTypes()
+  in generateStrangeChannelPartTypes().  Replace all uses of G4PionSampler
+  and G4NucleonSampler with new channel "classes".
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-33)
+------------------------------------------------
+Continue consolidation of two-body collision lookup tables.  Unify
+interfaces between pion/nucleon and kaon/hyperon samplers.
+
+- G4CascadeData: Many changes to support use as data class for both
+  kaon/hyperon and pion/nucleon tables (migration will be next tag):
+
+  1) Add new first argument to template with number of bins in energy
+     interpolation.  This replaces the (temporary) use of
+     G4CascadeSampler::energyBins within the class definition.
+
+  2) Add two additional template arguments, N8 and N9, with default values
+     of zero.  These will be used by the pion/nucleon tables, but not by the
+     existing kaon/hyperon tables.
+
+  3) Define additional arrays for 8- and 9-body final states, using
+     conditional dimensioning (e.g., if N8==0, use [1]).  Define second
+     9-body constructor to accept and initialize the new arrays.
+
+  4) Eliminate static intialization of index[], moving it into
+     initialize().  Template specialization can't be used with data!
+
+  5) Add separate "sum" (summed N-body cross-sections) and "tot" arrays
+     (measured inclusive xsec).  Latter is a const-ref, initialized to "sum"
+     if not provided as ctor argument, with two new ctors as needed.
+
+  6) Add accessor to return maximum multiplicity (NM+1).
+
+- G4CascadeFunctions.hh:  Add second template argument to specify which
+  "final state sampler" is the base class.  Use both T::data.tot and
+  T::data.sum in GetMultiplicity(), to support sum/tot comparisons in
+  pion/nucleon channels.  Extend GetOutputParticleTypes() to support up to
+  multiplicity 9.
+
+- G4CascadeFunctions.icc:  Implementations of template class functions.
+
+- G4CascadeXXXChannel.hh: Add "31," as initial template argument for Data.
+  Add G4CascadeSampler as second template argument in G4CascadeXXXChannel
+  typedef.
+
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-32)
+------------------------------------------------
+Continue revisions to two-body collision lookup tables.  Make the kaon and
+hyperon channels compatible with the pion/nucleon samplers, before the next
+stage of unifying the structure.
+
+- G4CascadeSampler: NEW class, parallel to G4FinalStateSampler but with a
+  different energy binning.  Used as base class to G4CascadeFunctions<T>,
+  from which the cross-section and other arrays are passed.  Implement
+  findMultiplicity (cf. G4CascadeFunctions::getMultiplicity) to return the
+  true value, not the array index.
+
+  This class replaces functionality of G4CascadeChannel, which is redundant.
+
+- G4CascadeChannel: Remove all interpolation code, leaving only getQnums().
+
+- G4CascadeFunctions:  Inherit from G4CascadeSampler for interpolations,
+  except for final-state extraction (where FS type arrays are dealt with
+  locally, not passed).  Public static functions rely on private instantiated
+  version of class to do work, so that G4CascadeSampler gets instantiated.
+
+- G4CascadeData: Remove final template argument (total number of indices),
+  as redundant with simple sum of multplicity indices.  Use template args to
+  dimension data member arrays (const references), and to statically
+  initialize the start-stop index table.  Make large cross-sections table
+  G4double, not G4float.  Define ctor to pass in array refs for initialization.
+  Reduce index array to single dimension (start=index[i], stop=index[i+1]).
+
+- G4CascadeXXXChannel:  All of the initial-state channel "classes"
+  (typedef names) modified to follow G4CascadeData rewrite.  In .cc file,
+  the explicit lookup tables are static, and passed by const-ref to
+  G4CascadeData.  The "initializer" struct is dropped; with the static
+  "data" object initialized by ctor here.
+
+- G4ElementaryParticleCollider.cc:  In ::generateMultiplicity(), eliminate
+  extra "+2" and "-2" applied to local "mul" variables.  With change above,
+  all functions return return true multiplicity, not array index.
+
+- G4FinalStateSampler:  Add findFinalStateIndex() function, to make
+  interface match G4CascadeSampler exactly.  Reduce index array to single
+  dimension.
+
+- G4PionSampler.cc, G4NucleonSampler.cc:  Use new findFinalStateIndex()
+  function in GetFSPartTypesXXX() functions.  Reduce index array to single
+  dimension (start=index[i], stop=index[i+1]).
+
+14 May 2010 Michael Kelsey (hadr-casc-V09-03-31)
+------------------------------------------------
+Begin revisions to two-body collision lookup tables.  This will be a major
+reorganization, splitting the G4PionSampler and G4NucleonSampler into
+separate little "classes" (typedefs) for each initial state, in the same way
+that the kaon and hyperon states are done now.  The underlying templated and
+base classes will be unified across all of the channels.
+
+- G4CascadeInterpolator: NEW class to encapsulate interpolation procedure.
+  Uses C-style arrays passed by dimensioned reference, to allow compile-time
+  array bounds checking.  Template argument is array dimension, and array of
+  energy bins is saved by ctor.  User array to be interpolated is passed as
+  function argument.  Last value interpolated, along with result, is saved,
+  to allow fast multiple interpolations.
+
+  This class is substantially more lightweight than G4PhysicsVector.
+
+- G4FinalStateSampler: Replace manual interpolation with G4CascadeInterpolator.
+  Remove redundant sampleFlat(...) function with argument.  Rename member
+  functions in anticipation of merging with G4CascadeFunctions.
+
+- G4PionSampler.cc, G4NucleonSampler.cc: Follow use of G4CascadeInterpolator
+  above, and drop local caching of interpolation results.  Reduce offset
+  list for cross-sections to one dimension (start=index[i], stop=index[i+1]).
+
+11 May 2010 Michael Kelsey (hadr-casc-V09-03-30)
+------------------------------------------------
+Bug fixes in the two-body scattering code.
+
+- G4PionSampler.cc, G4NucleonSampler.cc:  Fix long-standing bug with offset
+  indices computed for cross-section tables.  XXXindex[m][0] (start) and
+  XXXindex[m][1] (stop) are computed in initChannels() as [start,stop]
+  inclusive range (i.e., "for (i=start; i<=stop; i++)"), but used with
+  G4FinalStateSampler::fillSigmaBuffer (and its predecessor code) as
+  [start,stop) exclusive range (i.e., normal C style "for (i=start; i<stop;
+  i++)").  This is fixed by properly setting XXXindex[m-1][1] =
+  XXXindex[m][0], and using the ranges consistently in initChannels().
+
+- G4ElementaryParticleCollider.cc:  In the six pion-nucleon scatters, the
+  two-body final state is incorrectly tested for charge-exchange.  All the
+  if statements check whether "particle_kinds[0]" is the initial nucleon,
+  but the outgoing baryon actually appears in particle_kinds[1].  Replace
+  the individual tests with the use of "finaltype", moved out of the
+  strangeness if-block.
+
+06 May 2010 Michael Kelsey (hadr-casc-V09-03-29, hadr-casc-V09-03-23-01)
+------------------------------------------------------------------------
+- G4Diproton.cc, G4Dineutron.cc, G4UnboundPN.cc:  After creating instance
+  of these "internal" particles, remove the pointer from G4ParticleTable.
+  This avoids an issue with the FTFP de-excitation code, which occasionally
+  picks up the diproton and puts it onto a final-state particle list.
+
+These three changes, and nothing else, are "backported" on top of -23 for
+continued validation tests.
+  
+30 Apr 2010 Dennis Wright (hadr-casc-V09-03-28)
+-----------------------------------------------
+- G4ElementaryParticleCollider: 
+    improved pi-nucleon two-body angular distributions, and made 
+    sampling of CM angle faster.  As a result, the methods 
+    getElasticCase and adjustIntervalForElastic were made redundant
+    and removed.  Also removed array containing old angular distribution
+    parameters.
+    
+    Now use nucleon-nucleon angular distributions for hyperon-nucleon
+    scattering, and pi-nucleon dsitributions for kaon-nucleon scattering
+    (until specific kaon distributions are added). 
+ 
+29 Apr 2010 Michael Kelsey (hadr-casc-V09-03-27)
+------------------------------------------------
+- G4InuclPartcleNames.hh:  Long name "gamma" conflicts with math function.
+  Change back to "photon".
+
+- G4InuclElementaryParticle.cc, G4InuclEvaporation.cc,
+  G4NonEquilibriumEvaporator.cc, G4NucleiModel.cc, G4NucleonSampler.cc,
+  G4PionSampler.cc, G4PreCompoundCascadeInterface.cc,
+  G4PreCompoundInuclCollider.cc:  Replace "gamma" with "photon".
+
+- G4InuclElementaryParticle.hh:  Change "photon()" to "isPhoton()" to avoid
+  conflicts with names enum.  Use names enum internally as well.
+
+- G4ElementaryParticleCollider.cc, G4InuclCollider.cc, 
+  G4PreCompoundInuclCollider.cc:  Change "photon()" to "isPhoton()".
+
+28 Apr 2010 Michael Kelsey (hadr-casc-V09-03-26)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4InuclParticleNames.hh:  Define enums for particle type codes, using both
+  "long" and "short" names copied from existing duplicated enums.  One
+  change is that "kp" and "km" are replaced by "kpl" and "kmi" respectively,
+  since "km" conflicts with GEANT4 "kilometers" unit.  Deploy in classes:
+
+        include/G4FinalStateSampler.hh
+        src/G4CascadeElasticInterface.cc
+        src/G4CascadeInterface.cc
+        src/G4ElasticCascadeInterface.cc
+        src/G4ElementaryParticleCollider.cc
+        src/G4InuclElementaryParticle.cc
+        src/G4InuclEvaporation.cc
+        src/G4NucleonSampler.cc
+        src/G4PionSampler.cc
+        src/G4PreCompoundCascadeInterface.cc
+
+- G4InuclElementaryParticle.cc:  Add Omega- and antinucleon partcles to lists.
+  No code uses these (yet).
+
+24 April 2010 Michael Kelsey (hadr-casc-V09-03-25)
+--------------------------------------------------
+Remove redundant classes, G4Ibertini and G4BertiniRegionModel.
+
+21 April 2010 Michael Kelsey (hadr-casc-V09-03-24)
+--------------------------------------------------
+- G4NucleiModel.hh,cc:  Eliminate last remaining hardwired baryon masses.
+
+18 April 2010 Michael Kelsey (hadr-casc-V09-03-23)
+--------------------------------------------------
+Final revisions to access std::vector<particle> via const-references:
+
+        G4CascadeInterface.cc
+        G4EquilibriumEvaporator.cc
+        G4IBertini.cc
+        G4InuclCollier.cc
+        G4NucleiModel.cc
+        G4PreCompoundCascadeInterface.cc
+
+- G4CollisionOutput:  Add new function to boost output lists to lab frame.
+
+14 April 2010 Michael Kelsey (hadr-casc-V09-03-22)
+--------------------------------------------------
+- G4CascadeInterface.cc, G4IBertini.cc, G4PreCompoundCascadeInterface.cc:
+  In ::ApplyYourself(), do the check on K0L/K0S _before_ passing G4PartDefn
+  to G4InuclElemPart::type().  This avoids unnecessary warning message.
+
+14 April 2010 Michael Kelsey (temporary: kelsey-memClean3)
+----------------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4FinalStateSampler: Replace "sigma" argument for sampling with data
+  member, add functions to fill sigma buffer by passing in array references.
+  Provide typedef for interpolator, and function to apply interpolation.
+
+- G4PionSampler, G4NucleonSampler:  Follow changes to base class,
+  eliminating much of the copy-and-paste code blocks via function calls.
+  Replace G4float arrays with G4double, to support passing into functions.
+
+13 April 2010 Michael Kelsey (hadr-casc-V09-03-21)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4InuclSpecialFunctions.hh, paraMaker.cc, paraMakerTruncated.cc:  Drop
+  return-by-value from paraMaker() and paraMakerTruncated().  Add second
+  argument with non-const reference to output buffer for return.
+
+- G4EquilibriumEvaporator.cc:  Follow changes above to paraMaker(), use
+  references to address parms.first and parms.second, instead of copying.
+
+- G4NonEquilibriumEvaporator.cc:  Follow changes above to
+  paraMakerTruncaed(), use references to address parms.first and
+  parms.second, instead of copying.
+
+- G4Dineutron, G4Diproton, G4UnboundPN:  Inherit from G4VShortLivedParticle,
+  per Kurashige.  Thought it would eliminate G4ElectronOccupancy churn, but
+  did not.
+
+12 April 2010 Michael Kelsey (hadr-casc-V09-03-20)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4NucleiModel: Move function implementations from .hh to .cc.  Replace
+  manual sorting (inefficient O(N^2)) with std::sort() and simple
+  "less-than" function for std::pair<> "partners".
+
+- G4CollisionOutput: Move function implementations from .hh to .cc.  Use
+  std::vevtor<>::insert() to add vectors together.
+
+- Replace return-by-value G4CollisionOutput with non-const reference buffer
+  passed into xxx::collide(), in the following classes:
+        G4BigBanger
+        G4CascadeElasticInterface
+        G4CascadeInterface
+        G4ElasticCascadeInterface
+        G4ElementaryParticleCollider
+        G4EquilibriumEvaporator
+        G4EvaporationInuclCollider
+        G4Fissioner
+        G4IBertini
+        G4IntraNucleiCascader
+        G4InuclCollider
+        G4InuclEvaporation
+        G4NonEquilibriumEvaporator
+        G4NucleiModel
+        G4PreCompoundCascadeInterface
+        G4PreCompoundInuclCollider
+
+  Note that some places where internal G4CollisionOutput buffers are being
+  used (and their contents appended to the global output) have not been
+  cleaned up yet.  These will require some rearrangement of the algorithms,
+  or/and elimination of debugging output.
+
+09 April 2010 Michael Kelsey (hadr-casc-V09-03-16-01)
+-----------------------------------------------------
+This modification is included in a pseudo-patch of hadr-casc-V09-03-16,
+without brining in any of the -17 through -19 modifications.
+
+- G4LorentzVector.cc:  Protect two instances of sqrt() against tiny negative
+  arguments (due to tiny-tiny subtractions).
+
+09 April 2010 Michael Kelsey (hadr-casc-V09-03-19)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4InuclParticle, G4InuclElementaryParticle, G4InuclNuclei:  Eliminate
+  unused G4String ctor argument, introduced in hadr-casc-V09-03-01.  Causes
+  a lot of pointless memory churn.
+
+- G4BigBanger.cc:  Fix type conversion warning I introduced by mistake.
+
+08 April 2010 Michael Kelsey (hadr-casc-V09-03-18)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4PionSampler, G4NucleonSampler: For all ::GetFSPartTypes*() functions,
+  eliminate return-by-value std::vector<>, and pass in non-const reference
+  for filling in situ.  Buffer is cleared.
+
+- G4ElementaryParticleCollider.cc:  Follow changes above for Sampler calls.
+
+07 April 2010 Michael Kelsey (hadr-casc-V09-03-17)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+Replace return-by-value std::vector<> with either const-ref returns, or
+with use of an internal buffer which can be reused, to reduce memory
+churn due to copying, allocation, and resizing.
+
+- G4BigBanger:  Create data buffer for lists of secondaries and momenta.
+  Fill buffers in ::generateBangInSCM() and ::generateMomentumModules(),
+  which are now non-const.  Reference buffers in ::collide().
+
+- G4CascadeFunctions::getOutgoingParticleTypes():  Make "kinds" a static
+  variable (buffer), and return it by const-ref.
+
+- G4ElementaryParticleCollider:  Eliminate return-by-value std::vector<>
+  on the functions listed below, by creating data buffers for particles,
+  momenta, and particle types.  The functions become non-const and void:
+  ::generateSCMfinalState(), ::generateMomModules()
+  ::generateStrangeChannelPartTypes(), ::generateSCMpionAbsorption()
+
+  Also, in ::generateStrangeChannelPartTypes(), use a switch statement
+  instead of the long cascade of if-else-else.
+
+- G4NucleiModel: Several member functions have different modifications:
+
+  ::initializeCascad(bullet,target) -- Replace return value with a passed-in
+        (non-const reference) buffer to be filled.  The calling code can
+        pass in a data member, local variable, or whatever.
+
+  ::generateParticleFate() -- Replace local "thePartners" variable with data
+        member, filled below.  Move "outgouing_cparticles" to data member
+        (fix the spelling!), and return by const-ref.
+
+  ::generateInteractionPartners() -- Eliminate return value; fill data
+        member instead, to be used by ::generateParticleFate().
+
+- G4IntraNucleiCascader.cc:  Follow changes to initializeCascad() above.
+
+- G4NucleonSampler, G4PionSampler: ** DON'T KNOW WHAT TO DO HERE **
+
+- G4WatcherGun:  Return const-ref to list of watchers, instead of copy.
+
+Replace all uses of std::vector<>::resize(0) with ::clear().  The former
+causes deallocation/reallocation/resizing cycles, while clear() just removes
+content, preserving the allocated buffer.
+
+        include/G4CollisionOutput.hh
+        src/G4ElementaryParticleCollider.cc
+        src/G4NucleiModel.cc
+        src/G4WatcherGun.cc
+
+07 April 2010 Dennis Wright (hadr-casc-V09-03-16)
+-------------------------------------------------
+- fix negative energy bug in G4BigBanger, and switch to using 
+  G4NucleiProperties::GetBindingEnergy
+
+07 April 2010 Michael Kelsey (hadr-casc-V09-03-15)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+1) Systematically replace pass-by-value function arguments with const-refs.
+This is the first step in reducing unnecessary memory churn, to be followed
+by replacing return-by-value std::vector<> with non-const argument buffers.
+
+        include/G4CascadeChannel.hh
+        include/G4FinalStateSampler.hh
+        include/G4InuclEvaporation.hh
+        include/G4NuclWatcher.hh
+        src/G4CascadeChannel.cc
+        src/G4FinalStateSampler.cc
+        src/G4InuclEvaporation.cc
+        src/G4NuclWatcher.cc
+
+2) Fix constness on some operator== and operator!= definitions.
+
+        include/G4CascadeElasticInterface.hh
+        include/G4CascadeInterface.hh
+        include/G4ElasticCascadeInterface.hh
+        include/G4IBertini.hh
+        include/G4PreCompoundCascadeInterface.hh
+
+07 Apr 2010 Michael Kelsey (hadr-casc-V09-03-14)
+------------------------------------------------
+- G4Dineutron.hh, G4Diproton.hh, G4UnboundPN.hh:  
+  Reset static "theInstance" pointer to zero in (now virtual) destructor.
+
+- G4Dineutron.cc, G4Diproton.cc, G4UnboundPN.cc: 
+  Implement constructor to do the G4Ions initialization, rather than in
+  ::Definition().  Drop use of G4ParticleTable; just monitor static pointer.
+  Set "ShortLived" flag in G4Ions initialization, to avoid interactions with
+  ProcessManager.
+
+- G4InuclNuclei.cc:  In ::makeNuclearFragment(), set "ShortLived" flag for
+  G4Ions, to avoid interactions with ProcessManager.
+
+26 Mar 2010 Dennis Wright (hadr-casc-V09-03-13)
+-----------------------------------------------
+- G4EquilibriumEvaporator.cc: remove line #include "G4types.hh"
+  (file doesn't exist)
+
+20 Mar 2010 Michael Kelsey (hadr-casc-V09-03-12)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4EquilibriumEvaporator.cc:  Move subtraction of Q1[i] outside new
+  GetBindingEnergy() call.
+
+- G4InuclNuclei.cc::makeNuclearFragment():  Add message warning users about
+  non-physical creation.  Use new GetBindingEnergy() function here as well
+  (V09-03-05 didn't have this function, so wasn't included in migration).
+
+19 Mar 2010 Michael Kelsey (hadr-casc-V09-03-11)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4InuclSpecialFunctions:  Two new functions:
+
+::generateWithRandomAngles() to encapsulate random-direction generation.
+  This function and existing generateWithFixedTheta() take optional mass
+  argument to produce proper four-vector.
+
+  Replace duplicated random angle code with new function above, and use new
+  optional mass argument where available:
+
+        G4BigBanger.cc
+        G4ElementaryParticleCollider.cc
+        G4EquilibriumEvaporator.cc
+        G4Fissioner.cc
+        G4NonEquilibriumEvaporator.cc
+        G4NucleiModel.cc
+
+::G4cbrt():  Define cube-root in terms of log and exp, rather than pow().
+  Note that "cbrt()" can't be used, as it conflicts with <math.h> on Linux.
+
+  Eliminate unnecessary uses of std::pow() with fixed arguments.  For
+  example, "pow(x,3)" -> "x*x*x", "pow(x,1/3)" -> G4cbrt(x):
+
+        G4BertiniRegionModel.cc
+        G4EquilibriumEvaporator.cc
+        G4EvaporationInuclCollider.cc
+        G4Fissioner.cc
+        G4IntraNucleiCascader.cc
+        G4InuclCollider.cc
+        G4NonEquilibriumEvaporator.cc
+        G4NucleiModel.cc
+        G4PreCompoundInuclCollider.cc
+        G4RegionModel.cc
+        bindingEnergyAsymptotic.cc
+        bindingEnergyKummel.cc
+
+  WARNING!  In G4IntraNucleiCascader.cc, the Coulomb barrier was calculated
+  using "std::pow(A, 0.333)", rather than the "exact" 1./3.  This leads to
+  per-mil differences in the value, and occasional differences in wheter an
+  outgoing particle tunnels through.  The new code is exact to double
+  precision, and therefore fixes a bug.
+
+- Move "using" directive from all .hh files to corresponding .cc, along with
+  some unnecessary #includes:
+
+        G4BigBanger.hh, .cc
+        G4ElementaryParticleCollider.hh, .cc
+        G4EquilibriumEvaporator.hh, .cc
+        G4EvaporationInuclCollider.hh, .cc
+        G4Fissioner.hh, .cc
+        G4IntraNucleiCascader.hh, .cc
+        G4InuclCollider.hh
+        G4InuclNuclei.hh
+        G4NonEquilibriumEvaporator.hh, .cc
+        G4NucleiModel.hh, .cc
+        G4PreCompoundInuclCollider.hh, .cc
+
+- G4NuclWatcher.hh and .cc (new):  Move implementation code from .hh to .cc.
+
+18 March 2010 Michael Kelsey (hadr-casc-V09-03-10)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+Missed one bug in repairs below, found when testing 4 GeV pi -> Pb.
+
+- G4Fissioner.cc:  Replace incorrect G4LorentzVector ctor arguments
+  (G4ThreeVector,m) with call to G4LV::setVectM().
+
+17 March 2010 Dennis Wright (hadr-casc-V09-03-09)
+-------------------------------------------------
+Fix warning in G4InuclNuclei.cc: G4Ions:G4Ions was being called with
+a double (a) in its 13th argument, when an integer is required.  Use
+G4lrint to convert.
+
+16 March 2010 Michael Kelsey (hadr-casc-V09-03-08)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+Fix bugs introduced during migration from G4CascadeMomentum to
+G4LorentzVector (see hadr-casc-V09-03-03).  With these bug fixes, 2000
+events generated with "ppi4GeV" test executable are equivalent to production
+output (GEANT 4.9.3), allowing for differences in hardwired nucleon masses.
+When compared with hadr-casc-V09-03-02 (which uses G4ParticleDefinition for
+all the particle masses) the 2000-event results are identical within
+floating point precision.
+
+- G4BigBanger.cc:  Line 72, fix unit conversion bug, GeV/MeV vs. MeV/GeV.
+  Replace incorrectly introduced cross-product with original vector
+  manipulation, written with G4LorentzVector accessors.
+
+- G4CollisionOutput::selectPairToTune(): Test on "i3>0" not appropriate with
+  G4LorentzVector indexing.  Replace with failure-test on "i3<0".  Also
+  simplify final if-cascade for readability.
+
+- G4EquilibriumEvaporator.cc:  Replace incorrect G4LorentzVector ctor
+  arguments (px,py,pz,m) with call to G4LV::setVectM().
+
+- G4IntraNuclearCascader.cc:  Replace leftover momentum_out[0] with .e().
+
+- G4InuclNuclei.cc,hh:  Add new function ::makeNuclearFragment(a,z,exc),
+  which "manually" construct a G4Ions object for cases where A/Z is not a
+  valid stable or unstable nucleus.  These unphysical objects are passed
+  back from G4IntraNuclearCascader.cc and immediately passed to G4BigBanger
+  for conversion to nucleons.  A future change should move this action into
+  G4INCascader and eliminate the need for unphysical nuclei.
+
+- G4LorentzConvertor::toTheTargetRestFrame():  Remove local declaration of
+  scm_momentum, restoring access to data member with that name.
+
+- G4NucleiModel::generateQuasiDeutron(): Undo consolidation of dmom with
+  local variables for each of the nucleon momenta.  Stack corruption led to
+  differences in unbound p-n state relative to pp and nn states.
+
+Minor updates:  Changed all the "verboseLevel()" initializers to (0), via a
+shell script.  This allows making the code uniformly verbose or quiet.
+
+        include/G4NucleiModel.hh
+        src/G4Analyser.cc
+        src/G4CollisionOutput.cc
+        src/G4ElementaryParticleCollider.cc
+        src/G4FissionStore.cc
+        src/G4Fissioner.cc
+        src/G4IntraNucleiCascader.cc
+        src/G4NonEquilibriumEvaporator.cc
+        src/G4NucleiModel.cc
+        src/G4WatcherGun.cc
+
+16 March 2010 Michael Kelsey (hadr-casc-V09-03-DO-NOT-USE)
+--------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+Restore version hadr-casc-V09-03-05 to HEAD of CVS in order to apply bug
+fixes and re-deploy migration from G4CascadeMomentum to G4LorentzVector.
+This reversion is tagged as a checkpoint only, and should not be used for
+testing or running (since it restores the known bugs in V09-03-03).
+
+The update to G4InteractionCase.hh by Gabriele Cosmo in hadr-casc-V09-03-07
+is not touched by this reversion!
+
+The following CVS commands were used to perform the reversion:
+
+        cvs update -j1.13 -j1.12 cascade/include/G4CascadParticle.hh
+        cvs update -j1.4 -j1.3 cascade/include/G4CascadeMomentum.hh
+        cvs update -j1.15 -j1.14 cascade/include/G4CollisionOutput.hh
+        cvs update -j1.23 -j1.22 cascade/include/G4ElementaryParticleCollider.hh
+        cvs update -j1.9 -j1.8 cascade/include/G4InuclCollider.hh
+        cvs update -j1.19 -j1.18 cascade/include/G4InuclElementaryParticle.hh
+        cvs update -j1.12 -j1.11 cascade/include/G4InuclNuclei.hh
+        cvs update -j1.16 -j1.15 cascade/include/G4InuclParticle.hh
+        cvs update -j1.13 -j1.12 cascade/include/G4InuclSpecialFunctions.hh
+        cvs update -j1.13 -j1.12 cascade/include/G4LorentzConvertor.hh
+        cvs update -j1.4 -j1.3 cascade/include/G4ParticleLargerBeta.hh
+        cvs update -j1.9 -j1.8 cascade/include/G4ParticleLargerEkin.hh
+        cvs update -j1.20 -j1.19 cascade/src/G4BigBanger.cc
+        cvs update -j1.13 -j1.12 cascade/src/G4CascadParticle.cc
+        cvs update -j1.6 -j1.5 cascade/src/G4CascadeElasticInterface.cc
+        cvs update -j1.68 -j1.67 cascade/src/G4CascadeInterface.cc
+        cvs update -j1.18 -j1.17 cascade/src/G4CollisionOutput.cc
+        cvs update -j1.7 -j1.6 cascade/src/G4ElasticCascadeInterface.cc
+        cvs update -j1.6 -j1.5 cascade/src/G4EvaporationInuclCollider.cc
+        cvs update -j1.5 -j1.4 cascade/src/G4IBertini.cc
+        cvs update -j1.3 -j1.2 cascade/src/G4InuclElementaryParticle.cc
+        cvs update -j1.10 -j1.9 cascade/src/G4InuclEvaporation.cc
+        cvs update -j1.3 -j1.2 cascade/src/G4InuclNuclei.cc
+        cvs update -j1.4 -j1.3 cascade/src/G4InuclParticle.cc
+        cvs update -j1.17 -j1.16 cascade/src/G4InuclSpecialFunctions.cc
+        cvs update -j1.19 -j1.18 cascade/src/G4LorentzConvertor.cc
+        cvs update -j1.5 -j1.4 cascade/src/G4PreCompoundCascadeInterface.cc
+        cvs update -j1.4 -j1.3 cascade/src/G4PreCompoundInuclCollider.cc
+
+Intermediate versions below had bindingEnergy changes from -05 re-applied
+
+        cvs update -j1.23 -j1.21 cascade/src/G4EquilibriumEvaporator.cc
+        cvs update -j1.22 -j1.20 cascade/src/G4Fissioner.cc
+        cvs update -j1.29 -j1.27 cascade/src/G4IntraNucleiCascader.cc
+        cvs update -j1.24 -j1.22 cascade/src/G4InuclCollider.cc
+        cvs update -j1.23 -j1.21 cascade/src/G4NonEquilibriumEvaporator.cc
+        cvs update -j1.38 -j1.36 cascade/src/G4NucleiModel.cc
+
+G4ElementaryParticleCollider::particleSCMmomentumFor2to2:  Restore new
+parametrization of elastic scattering added by D.Write in hadr-casc-V09-03-06
+
+        cvs update -j1.47 -j1.45 cascade/src/G4ElementaryParticleCollider.cc
+        + editing
+
+05 Mar 2010 Gabriele Cosmo (hadr-casc-V09-03-07)
+------------------------------------------------
+- Reinstated fix in G4InteractionCase.hh for initialisation in constructor to
+  allow for porting on C++0x Standard.
+
+07 Feb 2010 Dennis Wright (hadr-casc-V09-03-06)
+-----------------------------------------------
+- back out changes from tags V09-03-00 to V00-03-05 by restoring HEAD
+  to V09-02-11, and keeping the di-nucleon classes
+- replace Bertini method bindingEnergy() with 
+  G4NucleiProperties::GetBindingEnergy() in classes
+  G4IntraNuclearCascader, G4InuclCollider, G4NucleiModel, G4EquilibriumModel,
+  G4NonEquilibriumModel, G4Fissioner.  Note that G4Fissioner still uses the 
+  Bertini method bindingEnergyAsymptotic.
+
+- G4ElementaryParticleCollider::particleSCMmomentumFor2to2
+  replace original (incorrect) pp, pn, nn 2-body to 2-body scattering angular
+  distributions with a new parameterization of elastic scattering data using
+  the sum of two exponentials
+
+02 Feb 2010 Dennis Wright (hadr-casc-V09-03-05)
+-----------------------------------------------
+- replace all uses of bindingEnergy (local Bertini method) with the
+  Geant4 standard G4NucleiProperties::GetBindingEnergy(A,Z)
+  files modified:
+    G4IntraNucleiCascader.cc
+    G4InuclCollider.cc
+    G4NucleiModel.cc
+    G4EquilibriumEvaporator.cc
+    G4NonEquilibriumEvaporator.cc
+    G4Fissioner.cc
+- note that G4Fissioner still uses Bertini method bindingEnergyAsymptotic
+  
+28 Jan 2010 Dennis Wright (hadr-casc-V09-03-04)
+-----------------------------------------------
+- fix unused variable warnings in G4InuclParticle.hh, .cc and
+  G4NonEquilibriumEvaporator.cc
+
+26 Jan 2010 Michael Kelsey (hadr-casc-V09-03-03)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4ParticleLargerBeta.hh:  Fix to use of CVS "Name" tag; no code changes.
+
+- G4CascadParticle: Replace the position std::vector<> with G4ThreeVector.
+
+- G4NucleiModel: Replace all std::vector<> coordinates with G4ThreeVector,
+  and use vector math for computations.
+
+Remove all use of G4CascadeMomentum (a simple container with a C-style
+four-element array) with G4LorentzVector.  This is a fairly invasive change,
+since G4CascMom was used for all of the three- and four-vector operations
+throughout the Bertini code:
+
+        include/G4CascadParticle.hh
+        include/G4CollisionOutput.hh
+        include/G4ElementaryParticleCollider.hh
+        include/G4InuclCollider.hh
+        include/G4InuclElementaryParticle.hh
+        include/G4InuclNuclei.hh
+        include/G4InuclParticle.hh
+        include/G4InuclSpecialFunctions.hh
+        include/G4LorentzConvertor.hh
+
+        src/G4BigBanger.cc
+        src/G4CascadParticle.cc
+        src/G4CascadeElasticInterface.cc
+        src/G4CascadeInterface.cc
+        src/G4CollisionOutput.cc
+        src/G4ElasticCascadeInterface.cc
+        src/G4ElementaryParticleCollider.cc
+        src/G4EquilibriumEvaporator.cc
+        src/G4EvaporationInuclCollider.cc
+        src/G4Fissioner.cc
+        src/G4IBertini.cc
+        src/G4IntraNucleiCascader.cc
+        src/G4InuclCollider.cc
+        src/G4InuclElementaryParticle.cc
+        src/G4InuclEvaporation.cc
+        src/G4InuclNuclei.cc
+        src/G4InuclParticle.cc
+        src/G4InuclSpecialFunctions.cc
+        src/G4LorentzConvertor.cc
+        src/G4NonEquilibriumEvaporator.cc
+        src/G4NucleiModel.cc
+        src/G4PreCompoundCascadeInterface.cc
+        src/G4PreCompoundInuclCollider.cc
+
+The new code is not yet entirely "correct," since Bertini also does a lot of
+operations by way of three-momentum, imposing masses (and hence energy
+conservation) separately, after the fact.
+
+20 Jan 2010 Gabriele Cosmo
+--------------------------
+- G4InteractionCase.hh:  Minor fix in initialisation in constructor to
+  allow for porting on C++0x Standard.
+
+20 Jan 2010 Michael Kelsey (hadr-casc-V09-03-02)
+------------------------------------------------
+- G4LorentzConvertor.hh:  Fix scm_momentum data member to be G4ThreeVector
+  (as used in previous versions) rather than G4LorentzVector.
+
+NOTE:  I have *retagged* cascade/include/G4LorentzConverter.hh and this
+       History file, rather than creating a new tag, since these are private
+       development only, not public releases.
+
+12 Jan 2010 Michael Kelsey (hadr-casc-V09-03-02)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4CascadeMomentum: Replace use of static G4LV variables within functions
+  with mutable data member of object; this does introduce more memory
+  thrashing, but the plan is to get rid of the whole thing anyway.  Improve
+  implicit casting and add assignment from G4LV.
+
+- G4LorentzConverter:  Change function signatures to accept G4LV as input,
+  using implicit cast to convert existing G4CascadeMomentum argument
+  passing.  Use G4LV internally throughout, again using implicit cast to
+  return G4CascMom values.  This substantially increases (temporarily!)
+  memory thrashing.  Simplify ::rotate() functions to use G4LV and
+  G4ThreeVec operations rather than array index gymnastics.  Move some
+  function implementations out of .hh file into .cc.
+
+- G4ParticleLargerEkin.hh: Fix use of CVS "Name" tag.  No code changes.
+
+12 Jan 2010 Michael Kelsey (hadr-casc-V09-03-01)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4ParticleLargerBeta, G4ParticleLargerEkin: Add additional operator()
+  which takes pointers (for future mods to reduce ctor/dtor cycling), and
+  fix ordering bug in beta sorting.  Add debugging printout togged with
+  preprocessor flag.
+
+- G4InuclParticle, G4InuclNuclei, G4InuclElementaryParticle:  Rewrite to
+  carry G4DynamicParticle data member, and to instantiate particle
+  properties via G4ParticleDefinition subclasses.  This is first stage in
+  eventually eliminating the G4InuclParticle stuff entirely in favor of
+  direct use of G4DynamicParticle for internal propagation.
+
+These changes have been testing using debugging output on the production
+(4.9.3) code compared to the above changes.  |diff| shows mostly identical
+results, with occasional ~1e-5 floating point differences.
+
+08 Jan 2010 Michael Kelsey (hadr-casc-V09-03-00)
+------------------------------------------------
+All changes below within "cascade/" subdirectory:
+
+- G4BertiniNuclearModel:  REMOVE this class from the HEAD, and hopefully all
+  future tags.  It is redundant with the G4NuclearModel class, and is not
+  used anywhere in the G4 distribtion; the latter is referenced in both
+  regular G4 code and in examples.
+
+- G4Diproton, G4Dineutron, G4UnboundPN:  New G4ParticleDefinition
+  subclasses, which implement the Bertini-specific dibaryon states.  These
+  are not used anywhere as yet, but will be used in future modifications to
+  the G4InuclParticle subclasses.
+
+- G4CascadeMomentum:  Add interface to create from and cast to
+  G4LorentzVector.  Will be used (in future) by G4InuclParticle modifications,
+  and provides a transition to eliminating this class entirely.
 
 1 Dec 2009 Dennis Wright (hadr-casc-V09-02-11)