A%20tutorial%20on%20geant4%20hadronic%20physics%20(GHAD)

1
A tutorial on geant4 hadronic physics (GHAD)

• J.P. Wellisch,
• CERN/EP.

2
With contributions from

• T. Ersmark,
• G. Folger,
• V.Ivanchenko,
• A. Kiryunin,
• R. Nartello,
• B. Trieu,
• M. Verderi,
• J.P.Wellisch,
• D. Wright.

3
Program

• How to use GHAD?
• Implications for detector construction.
• A look inside (hadronic) processes, or how to
  tailor.
• What models and options are available?
• How good is it really?
• Where to find more information.

4
Part 1

• How to use GHAD?

5
Physics lists defining the physics

• GHAD physics, as all other physics, is used
  through geant4s physics lists.
• A physics lists is (user) code specifying the
  complete physics modeling used in your
  application.
• Particle types
• Decays
• Electromagnetic physics
• ...
• Hadronic physics
• It associates processes with particles.

6
The approach to physics lists

7
Geant4 physics lists versus geant3 packages

• In geant4, the physics lists serve the same
  purpose as the packages (GHEISHA, FLUKA,
  GCALOR) in geant3.
• Conceptually, the two are identical.
• They provide the physics and its modeling to an
  application.
• Each package is built of a complete and
  consistent set of models
• Early in geant4, the idea was that each user
  would (have to) build his package.

8
In the case if hadronic physics, the problem was
complexity.

• It takes 5 levels of implementation framework in
  geant4 to implement hadronics.
• These, and the models implementing them, are used
  to assemble the hadronic physics for the
  simulation engine.
• The number of options is quite large.
• Each comes with trade-offs in descriptive power
  and performance.
• There are 25 particle species to be tracked, that
  need complete and consistent physics.

9
Assume we want to study activation.
10
Hence the educated guess physics lists.

• It simply became clear that writing a good
  physics list is no trivial, in particular when
  hadronic physics is involved.
• It is nice to be able to exploit the full power
  in the flexibility and variety of hadronic
  physics modeling in geant4, but being forced to
  do so is not what we want.
• It is also nice to have the physics transparently
  in front of you and be able to exploit it in the
  best possible way, but being forced to understand
  it all is (very understandably) not what people
  want, either.

11
Because of this

• We have systematically accumulated experience
  with various combinations of cross-section and
  models over the last years.
• Today we provide a set of physics lists
  institutionalizing this knowledge.
• Publishing them to the general audience was one
  of the main milestones of the hadronic working
  group for 2002.

12
Use case packages of Physics Lists

• LCG simulation project.
• HEP calorimetry.
• HEP trackers.
• 'Average' collider detector
• Low energy dosimetric applicationswith neutrons
• low energy nucleon penetration shielding
• linear collider neutron fluxes
• high energy penetration shielding
• medical and other life-saving neutron applications

• low energy dosimetric applications
• high energy production targets
• e.g. 400GeV protons on C or Be
• medium energy production targets
• e.g. 15-50 GeV p on light targets
• LHC neutron fluxes
• Air shower applications (still working on this)
• low background experiments

• http//cmsdoc.cern.ch/hpw/GHAD/HomePage

13
To make tailoring easier, and the code more
readable, we introduced Builders.

14

15

Now, what does this mean ?

• You can now
• Just pick a physics list from the menu.
• Aggregate your own cocktail from limited
  complexity of the builders
• Use all 5 framework levels with their full power.
• A structured reduction in the level of complexity
  exposed to you.

16
The WWW pages a small demo.

• There is a physics lists topic on the geant4
  HyperNews.
• We go to
• http//cmsdoc.cern.ch/hpw/GHAD/HomePage

17
The recommended procedure

• Start by trying the provided physics lists.
• It makes it such that results by different groups
  can be compared.
• You will profit from validation and verification
  done by others.
• Of course you are still encouraged to tailor the
  physics lists that we provide, and/or build your
  own where you need.
• Please also let us know about your findings.
• Plots you may whish to provide can enter WWW for
  everyones benefit.
• ?Do not use examples/novice/N04 as example for a
  hadronic physics list.

18
The support process static view

19
The support process dynamic view

20
Part 2

• Some implications for detector construction

21
On Material construction

• There are three ways to construct materials in
  geant4
• From its isotopic composition
• From its elements
• As an effective material (Aeff, Zeff)

22
Effective materials

• Hadronics cross-section are not a function of
  material properties, but a function of nuclear
  properties.
• If you use effective numbers, the element
  composition cannot be automatically recovered.
• The cross-section will be highly approximativ
  at best.
• The final states will have wrong properties.
• Never use effective A, Z with hadronic physics
• (There are situations, here you will not be able
  to avoid it, so we cannot protect against it.)

23
Proton induced reactions
24
From elementary composition

• This is good enough for most high energy
  applications.

25
Proton induced reactions
26
Isotope wise composition

• When detailled simulation of low energy neutrons
  is important, element info is not sufficient
  (Elt20MeV) to get the cross-section and final
  states right.
• For different isotopes, the neutron nuclear
  resonances will be at entirely different
  positions
• For different isotopes, the final state channels
  open can differ drastically.
• ? You may be tempted to construct your materials
  from Isotopes in this case

27
Isotope wise composition

• In case the neutron_hp models are used (detailed
  neutron transport below 20MeV), geant4 recovers
  the natural isotopic composition, in case
  materials and mixtures are specified in terms of
  their constituting elements.
• If you have enriched isotopes (like Uranium-238),
  please use the isotopes directly, to specify your
  material.
• Normally you do not need to use the G4Isotope in
  your detector construction

28
Example Neutrons in Lithium

• Neutron inelastic cross-section at 150eV
• Li-7 0.00 millibarns
• Li-6 12.2 barns !
• Open inelastic channels
• Li-7 none
• Li-6 nLi?ta
• (which makes Li-6 a well known shielding isotope)

29
Part 3

• A look inside (hadronic) processes, or how to
  tailor.

30
What is tracked in GEANT4 ?

• Propagated by the tracking,
• Snapshot of the particle state and location.

• Momentum, pre-assigned decay

• The  particle type 
• G4Electron,
• G4PionPlus

•  Hangs  the
• physics sensitivity

• The classes involved in the building the
   physics list  are
• The G4ParticleDefinition concrete classes
• The G4ProcessManager
• The processes

• The physics
• processes

31
The process tracking interface.

• There are three situations, where lttrackinggt may
  want to ask information from ltprocessgt

• AtRest
• Decay, e annihilation

• AlongStep
• To describe continuous interactions,
• occuring along the path of the particle,
• like ionisation

AlongStep
PostStep

• PostStep actions
• Most hadronic interactions,

32
The process tracking interface.

• A process will implement any combination of the
  three AtRest, AlongStep and PostStep actions
• Eg decay AtRest PostStep
• Each action defines two methods

• GetPhysicalInteractionLength()
• Used to limit the step size
• because the process  triggers  an interaction,
  a decay, geometry boundary, a users limit

• DoIt()
• Implements the actual action to be applied on the
  track
• Typically final state generation.

33
G4VProcess G4ProcessManager

• In praxi the G4ProcessManager has three vectors
  of actions
• One for the AtRest methods of the particle
• One for the AlongStep ones
• And one for the PostStep actions.
• It is those vectors the user sets up in the
  physics list and which are used by the tracking.

34
A word of caution on processes ordering

• Ordering of following processes is critical
• Assuming n processes, the ordering of the
  AlongGetPhysicalInteractionLength of the last
  processes should be
• n-2
• n-1 multiple scattering
• n transportation
• Why ?
• Processes return a  true path length 
• The multiple scattering  virtually folds up 
  this
• true path length into a shorter  geometrical 
• path length
• Based on this new length, the transportation can
  geometrically limits the step.
• For other processes ordering does not matter.

?
35
A few examples of processes

• G4Transportation
• G4Decay
• G4eIonization
• G4ionIonization
• G4MuBremsStrahlung
• G4SynchrotronRadiation
• G4OpAbsorption
• G4HadronElasticProcess
• G4NeutronInelasticProcess
• Etc..
• ?These are registered with the process managers
  by the physics lists and the builders.

36
Hadronic vs. electromagnetic processes

• In EM physics (mostly)
• 1 process 1 model and 1 cross-section.
• In hadronic physics (mostly)
• 1 process an assembly and selection of many
  cross-sections data-sets, models, production
  codes, model components, sub-assemblies, options.
  
• Default cross-section are provided for each
  process.
• You decide in the physics list, what exactly you
  use.
• Mix, match, assemble.

37
A sample inelastic process.
38
Cross section logic AddDataSet() fills a
FILO stack
Reg. sequence
GetCrossSection() uses the first applicable
dataset
Data set 3
Data set 2
Data set 1
Cross section baseline
Energy, particle, material, isotope, anything else
39
A sketch of model management
40
Part 4

• What models and options are available?

41
Cross section implementations

• Default covers all possible situations for hadron
  interactions.
• Carried forward from geant3.21.
• Different kinds of cross-section data sets
• Some are parametrizations,
• Some are theory,
• Some read and use large databases.

42
Cross section implementations

• Low energy neutrons
• Data both for cross-sections and final state
  generation.
• G4NDL, based on a number of Evaluated Nuclear
  Data Libraries
• Data in a ENDF/B-VI derived data format.
• Uses the Unix file-system
• Data come from various revisions of ENDF/B,
  JENDL, FENDL, CENDL, Brond, Jef, MENDL, MENDL-P,
  EFF, etc..
• Recently we started to add geant4 native
  evaluations.

43
Cross section implementations

• Proton and neutron reaction cross-sections
  (Wellisch-Axen systematics)
• 0-20GeV for protons
• 14MeV-20GeV for neutrons
• Alternative pion cross-sections
• 0-1TeV
• Data for isotope production
• From MENDL-2 AND MENDL-2P, and
  Wellisch-Axen systematics (name was coined by Los
  Alamos).

44
Cross section implementations

• Ion reaction cross-sections
• Tripathis cross-section formula for E/Alt1GeV,
  and Agt2.
• Wellisch-Axen systematics for E/Alt20GeV and Agt4
  for scattering off Hydrogen.

45
Cross section implementations

• Photo and electro-nuclear cross-sections
• For all energies
• Gammas
• GDR, quasi-deuteron, Delta, Roper,
  reggeon-pomeron contributions parametrizised
• Based on 14 nuclei, tested on many more
• Electrons/Positrons
• Use method of equivalent photons, and fold the
  photo-nuclear cross-section
• Added hard scattering

46
Final state generators

• Three categories of modeling approaches
• Parametrization driven modeling
• Data driven modeling
• Theory driven modeling

47
Parameterisation driven
models

• Two domains
• high energy inelastic (Aachen, CERN)
• low energy inelastic, elastic, fission, capture
  (TRIUMF, UBC, CERN)
• Stopping particles
• base line (TRIUMF, CHAOS)
• mu- (TRIUMF, FIDUNA)
• pi- (INFN, CERN, TRIUMF)
• K- (Crystal Barrel, TRIUMF)
• anti-protons (JLAB, CERN)
• Electromagnetic transitions of the exotic atom
  prior to capture effects of atomic binding.
  (Novosibirsk, ESA)

48
Data driven models

• Low energy neutron transport (neutron_hp),
• Radioactive decay (DERA, ESA)
• photon evaporation (INFN)
• elastic scattering (TRIUMF, U.Alberta, CERN)
• internal conversion (ESA),
• etc..

49
Theory driven models

• Ultra-high energy models
• Parton transport model (U.Frankfurt, in
  discussion)
• High energy models
• Fritjof type string model (CERN)
• Quark gluon String (CERN)
• Pythia(7) interface (Lund, CERN)
• Intra-nuclear transport models (or replacements)
• Hadronic cascadepre-equilibrium (HIP, CERN)
• Binary and Bertini cascades (HIP, CERN,
  Novosibirsk, SLAC)
• QMD type models (CERN, Inst.Th.Phys. Frankfurt)
• Chiral invariant phase-space decay (JLAB, CERN,
  ITEP)
• Partial Mars rewrite (Kyoto, Uvic, in
  collaboration with FNAL)
• De-excitation
• Evaporation, fission, multi-fragmentation,
  fermi-break-up (CMS)

50
A not totally correct hadronic model summary
Absorption at rest m, p, K, p-bar, n-bar
(I fore sure left off something, like
G4LElastic)
CHIPS I
CHIPS (gamma)
LEP
Neutron_hp
HEP
Evap
FTF string
multifrag
Fermi
Binary cascade
Phot, ev.
QGS string
conversion
Bertini cascade
Rad. Dec.
Precompound
Fission
mars
LEpp, np
1 MeV 10 MeV 100 MeV 1 GeV 10 GeV 100
GeV 1 TeV 10 TeV 100 TeV
51
Recently released features

• Theoretical models
• Binary cascade
• Classical (Bertini) cascade
• A generic scattering term for cascade models.
• Internal conversion
• Chiral invariant phase-space decay (CHIPS) for
  electro-nuclear scattering with Q2gt0.
• Quark gluon string model for real and virtual
  gamma reactions
• Auger electrons were added.
• Complete re-implementation of pre-compound.
• HETC and GNASH transition and emission
  probabilities as options in evaporation.

52
More recently released features

• Data
• New photon evaporation data
• New radioactive decay data
• Neutron transport consistent with G4NDL 0.2
  through 3.7.
• Parametrized model
• Low energy parametrized models Better energy
  conservation for anti-particle and strange
  particle reaction
• Mars rewrite Neutron spectra extend below 1MeV.
• Elastic scattering re-coils added.

53
Even more recently released features

• Cross sections
• pion cross sections A new dataset class.
• protection for low nuclear masses
• protection for electro-nuclear (fixes CMS
  problem)
• management
• model classes get deleted, no matter what the
  physics list looks like.
• tracing of originator model now possible.
• Stopping
• neutrino-flavor fixed for stopping mu-
• Many other bug-fixes and small improvements
• ...

54
The recommended procedure

• Start by trying the provided physics lists.
• It makes it such that results by different groups
  can be compared.
• You will profit from validation and verification
  done by others.
• It makes your (and my) life much easier.

55
Part 5

• Trying to answer the question
• How good is it really?

56
Verification grouped into sections

• The verification effort of the geant4 hadronic
  working group is grouped into several sections
• Inclusive cross-sections
• Thin target comparisons
• Verification of model components
• Code comparisons (least effective)
• Complete application tests
• Robustness.
• I give few examples of each in the following
  slides.

57
A few total cross-section examples

58
Proton reaction cross-section
J.P.Wellisch
59
Pi reaction cross-sections dots data, open
symbols default and new.
60
Gamma-nuclear reaction cross-sections.
61
A few examples of thin target comparisons

62
Particle multiplicities, QGS model
63
Pion production examples, QGSRapidity
distributions and invariant cross-section
predictions in quark gluon string model

400GeV protons on Lithium
100 GeV pi on Gold
J.P.Wellisch
64
K scattering off Gold
QGS Model

Distributions of eta And transverse momentum.
65
Forward peaks in proton induced neutron
production
Lead
Beryllium
256 MeV data Neutrons at 7.5deg.

Iron
Aluminum
66
Binary cascade Neutrons from 597 MeV p on Pb
(PRC 22, p1184)
30 degrees
90 degrees
150 degrees

60 degrees
120 degrees
Neutron production At 30, 60, 90, 120 And 150
degrees
J.P.Wellisch
67
Low energy neutron capturegammas from 14 MeV
capture on Uranium
J.P.Wellisch
68
A few verification plots for model components

69
Nuclear densities Ex. 4He, 10B, 28Si, and 63Cu
4He
J.P.Wellisch
70
Predicting the Delta production cross-section in
pp scattering by binary casacde
J.P.Wellisch
71
prongs prediction in QGS model , single pomeron
exchange approximation.
72
A few code comparisons

73
Gammas and conversion electrons in 57Co geant4
vs. RADLIST
74
(No Transcript)
75
(No Transcript)
76
(No Transcript)
77
A few calorimeter simulation comparisons

78
HEC G4 5.0 (true geometry, my toy analysis)?

79
ATLAS HEC G4 5.0 (true geometry, my toy
analysis).

J.P.Wellisch
80
HEC shower shapes G4 5.0 (true geometry, my toy
analysis)

81
Part 6

• Where to find more information

82
Additional reading

• The GHAD WWW pages.
• http//cmsdoc.cern.ch/hpw/GHAD/HomePage/
• The LCG physics list pages.
• http//cmsdoc.cern.ch/hpw/GHAD/LCGPage
• The geant4 documentation.
• http//geant4.web.cern.ch/geant4/G4UsersDocuments/
  Overview/html/index.html