GEANT4 and EUDET/NA2

1
GEANT4 and EUDET/NA2
Geant4 Highlights VALSIM potential work items

• John Apostolakis, CERN
• for the G4/SFT team

V1.1
2
Content

• The Context Geant4, G4 team in PH/SFT
• Geant4 Highlights
• VALSIM potential work items

3
Part I The Context

• Geant4 toolkit
• Application Areas
• G4/SFT team (LCG)

4
Geant4 Overview

• Powerful structure and kernel
• tracking, stacks, geometry, hits,
• Extensive transparent physics models
• electromagnetic
• hadronic
• decay, optical,
• Interfaces
• visualization, GUI, persistency.
• Efficiency enhancing techniques
• Framework for fast simulation (shower
  parameterization)
• Variance reduction / event biasing

5
Application Areas

• Geant4 in HEP production
• BaBar
• ATLAS(Q1 2004)
• CMS (Q4 2003),
• LHCb (Q2 2004)
• Harp

• Medical applications
• Imaging (PET/SPECT)
• Dosimetry
• mBeam optics modeling
• Assessing treatment (hadrontherapy)
• Space
• Effects on electronics
• Planetary radiation environment
• Radiation in human flights

6
G4/SFT team work areas and people

• Geometry, Field and Transportation
• G Cosmo(coord.), V. Grichine, J. Apostolakis, O.
  Link (-Dec 2005)
• Software Management, System Testing
• G Folger, G Cosmo, I. McLaren, (also S. Sadilov)
• EM Physics
• V. Ivantchenko (resp./SFT), V. Grichine
• Hadronic Physics / Neutrons
• G. Folger (resp./SFT), M. Kossov, V Ivantchenko,
  V. Grichine
• A. Howard (from Oct 2005, also SI-Physics
  Validation)
• Regression testing / validation
• A. Ribon (also LCG-AA/SI/Physics Validation)
• Coordination / Release
• J. Apostolakis, G. Cosmo

7
Part II Physics

• Using the Physics via Physics lists
• Underlying physics modeling

8
Different types of hadronic shower models

• Parameterization-driven models
• Started from GHEISHA, revised/improved
• Theory driven models
• Pre-compound
• Cascades and CHIPS
• String models
• Data driven models
• Neutrons Elt20 MeV
• Photo-evaporation of nuclei

9
Models in hadronic framework
10
Theory driven models (1/2)

• Models derived from approximate/phenomenological
  models
• QCD, strings, chiral perturbation theory,
  statistical collective
• Only thin-target data used for verification
• Final states determined by sampling theoretical
  distributions
• Philosophy implies the usage physics lists,
  providing wanted collection of models, such as
• Parton string models at high energies,
• intra-nuclear transport models at intermediate
  energies, and
• statistical break-up models for de-excitation

11
Theory driven models (2/2)

• Parton string
• Projectiles with E gt 5 GeV
• Wounded nucleus is de-excited by attached model
  (pre-compound or Chips)
• Cascade energy-range models
• Bertini cascade (next slides)
• Binary cascade
• Chiral invariant phase space, CHIPS
• Quark-level event generator for the fragmentation
  of hadronic systems into hadrons
• All energies
• Interactions between hadrons are treated as
  purely kinematic effects of quark exchange
• Decay of excited hadronic systems is treated as
  the fusion of two quark-partons within the system
• Includes non-relativistic phase space of
  nucleons to explain evaporation
• Nuclear de-excitation and breakup

12
Bertini intra-nuclear cascade (1/2)

• Collection of theory driven models with
  parametrisation features
• Intermediate energies 100 keV 10MeV
• Models included
• Bertini INC model with exitons
• Pre-equilibrium model
• Nucleus explosion model
• Fission model
• Evaporation model

13
Bertini intra-nuclear cascade (2/2)

• For Agt4 a nuclei model is composed of three
  concentric spheres
• Impulse distribution in each region follows Fermi
  distribution with zero temperature
• Particle treated p,n, pions, photon evaporation
  and nuclear isotope remnats
• Latest addition include incident kaons up to an
  energy of 15 GeV
• Final states, will be included for K, K-, K0,
  K0bar, lambda, sigma, sigma0, sigma-, xi0 and xi-

Schematic presentation of the intra-nuclear
cascade. A hadron with 400 MeV energy is forming
an INC history. Crosses present the Pauli
exclusion principle in action.
14
Hadronic model inventory
15
pion production from 730 MeV proton on Carbon.
Geant4 LEP model (derived from Geant 3.21 and
improved)
16
Bertini cascade model pion production from 730
MeV proton on Carbon.
17
Tailored Physics lists

• Created and distribute educated guess physics
  lists
• correspond to major use cases of Geant4 involving
  hadronic physics,
• to use directly, and as a starting point for
  users to modify,
• facilitate the specialization of those parts of
  hadronic physics lists that vary.
• First released in September 2002
• Revised with experience of comparisons with data
• This provide tested options, with known
  performance
• Last major revision for physics models of Geant4
  6.2 (June 2004)
• Distribution
• Most up-to-date from the G4 hadronic physics web
  pages http//cern.ch/geant4/physics_lists
• Included in Geant4 releases
• Starting with Geant4 6.0 (Dec 2003), ported
  versions in major releases
• Current physics lists version is included in
  minor releases, patches.

18
Hadronic physics models, processes and lists
Components can be assembled in an optimized way
for each use case.

• Five level implementation framework
• Variety of models and cross-sections
• for each energy regime, particle type, material
• alternatives with different strengths and CPU
  requirements.

Pre-compound model
Cascade

• Illustrative example of assembling models into an
  inelastic process for set of particles
• Uses levels 1 2 of framework

Parameterized
QGSM
19
Use cases of Physics Lists

• HEP calorimetry.
• HEP trackers.
• 'Average' HEP collider detector
• Low energy dosimetric applicationswith neutrons
• low energy nucleon penetration shielding
• linear collider neutron fluxes
• high energy penetration shielding
• medical and 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
• low background experiments

20
Physics lists / engines for calorimetry

• LHEP is the fastest for CPU
• uses the LEP and HEP parameterized models for
  inelastic scattering.
• QGSP
• uses theory-driven modeling for reactions of ps,
  Ks, and nucleons
• for primaries with E starting at 12 GeV,
  dominant above 30 GeV
• It employs
• Quark Gluon String Model
• for the 'punch-through' interactions of the
  projectile
• A Pre-equilibrium decay model
• with an extensive evaporation phase that model
  the nucleus 'after the punch
• For other energies uses LHEP models

• QGSC, is similar to QGSP but uses CHIPS for
  fragmentation
• The CHiral Invariant Phase-Space decay (CHIPS)
• QGS now starts at 9 GeV
• FTFP starts with QGSP and replaces the string
• with a diffractive string excitation
• similar to that in FRITJOF, and the Lund
  fragmentation functions.
• Note g A (g nuclear) interactions recently
  added to all options.
• Previously available as _GN variants, eg QGSP_GN

21
Relevant comparisons for HEP A partial list

• ATLAS test beams
• FCAL m, e, p
• HEC m, e-, p
• EM Barrel
• TileCal
• TRT
• Muon chambers (extra hits)

• BaBar data
• Drift Chamber
• ALICE
• 100s MeV proton microscopic
• TIARA neutron benchm.
• CMS
• HCAL test beam
• BTeV
• ECAL test beam

Very hard to give just a few highlights
22
Hadronic Physics theoretical models

• Evaporation/pre-compound
• Bertini Cascade / INUCL
• extended to Kaons, up to 5-10 GeV, verified
  isotope production
• Binary cascade
• Extended to ion reactions, pion projectiles
• CHIPS Chiral Invariant Phase-space decay
• g-A, e-A, p absorption, p-bar annihilation at
  rest
• QGSM string model
• improved meson splitting

23
Latest
http//cern.ch/geant4

• Geant4 8.0 will include
• Refinements and new features in the kernel
• New models, improvements refinements in EM
• Revision of multiple scattering model
  (angle/lateral) process (step limits)
• Using extending model-based implementation of
  EM standard
• Improvements fixes models in hadronics
• Revisions to LElastic, and refined precision
  coherent process
• Neutrino nucleus using CHIPS
• Revised Physics-lists
• Revised to use EM builders, non-static particles
• Utilize revised physics processes (in most PLs).

24
Part III Validation / Testing

• Regression suite

25
Statistical testing 2004/5

• Establishment of statistical testing suite
• Automated comparison of physics quantities
• Simple setups for regression testing
• Simplified, typical LHC hadronic calorimeters
  (only E deposit, no digitisation)
• Additional testing suites
• Against standard data
• T. Koi (SLAC) hadron / ion comparison
• INFN EM interactions, per process X-sections
  vs. NIST
• Extensions of suite under consideration
• Further setups (EM calor.), quantities
• Reusing donated test-beam comparisons
• Full applications from ATLAS, CMS.
• For details see presentations of A. Ribon (Tech
  Forum, AA meeting)

26
Challenges / Ongoing

• Regression suite is identifying changes
• Identified problems (crashes) fixed
• Must establish more links to verification/sub-sy
  stem tests
• Limits to automated testing when revising models
• Not to forget user/experiment acceptance tests
• Performance improvement
• Large productions / always a goal
• Expanding use of best-practice
• Eg new methods to identify hard problems

27
Part IV Potential Work Issues
28
Verification of physical processes

• Review of physical processes for key detector
  materials (Verification)
• Materials Si, W, gases (Ar, CO2, CH4, CF2), ...
• Verification for key processes in relevant
  materials (using thin target data)
• Comparison/validation of detailed interaction
  products d2 s / dE dW in particular for hadronic
  interactions
• Potential additional 'details'
• catastrophic muon energy loss
• - e to hadrons (due to annihilation)
• synchrotron radiation in medium (LPM-like
  effects)

29
Benchmarking of neutron modeling

• Key aspects of use cases neutrons as background,
  activation, hazard/shielding.
• Identified/candidate benchmarks
• TIARA (shielding) ()
• Los Alamos 'thin target' (neutron generation) ()
• Particular CERF setups (shielding, activation,
  ...)
• TARC (spallation, elastic interactions, capture),
• new data (eg latest data with 14 MeV n)
• Note () comparisons exist, typically gt 2 years
  old
• Review / update data for data-driven neutron
  modelling (Elt20 MeV)
• Other aspects
• Radiation effects in silicon joint investigation
  with space community
• Radiation and effects on endcap detectors

30
3.) Hadronic shower development

• Neutron, EM, charged fractions
• identify and utilise available benchmarks for EM
  fraction, neutrons
• comparisons with test-beam results of
  existing/smaller-scale segmented-calorimeters
• hadronic shower shape comparisons with data and
  regression testing
• extend comparisons with data
• extend calorimeter regression suite with setups
  relevant to proposed detectors
• Effect of hadronic interactions in ECAL
• Identify aspects/elements affecting simulation's
  assessment of impact of ECal thickness, material
• Comparisons of modeling of jet showers and EM
  showers
• effect of choice of physics models
• Assessing strengths/weaknesses of available
  physics lists for shower simulation in segmented
  calorimeter

31
4.) Other/Technical aspects

• Coupled propagation/navigation in 'parallel'
  geometries in presence of fields
• Biasing
• revision of importance biasing (for parallel
  navigation)
• Scoring
• Re-factoring/improving existing and creating new
  'standard' tallies
• CPU performance
• Unique aspects for highly granular detectors (EM
  showers, field, neutrons?)
• Propagation in strong magnetic fields

32
THE END
33
Geant4 6.0 - general picture

• New capabilities in Geant4 6.0 for HEP
• Latest Physics lists distributed inside
• EM-std new model implementation by default
• Highlights of improvements
• to existing physics modeling models
• in physics process implementations
• in functionality
• The high level of user feedback is reflected in
  developments, fixes improvements

34
Hadronic Physics Lists

• The latest physics lists included since 6.0
• 8.0 ported from lists in 7.1 (June 2005)
• Porting (new particles), revised modeling (Mult.
  Scat., ..)
• Regression testing undertaken Nov/Dec on LCG/EGEE
  Grid
• New/revised versions of physics-lists to be
  released
• Revisions to be quickly included in Geant4
  patches, releases
• When required also via physics lists Web site
• Physics lists and builders are/can-be used
• As is, compiled in a deployment directory
• Altered (or additional/customized version) by
  user/experiment, in own installation

35
EM Physics Processes

• New model-based EM standard physics processes
  are now the default
• for maintaining and refining
• keeping user code unchanged
• Old (frozen) implementation is still available
• Issues encountered in transition
• Fixed in 6.0 patch 1 and 6.1
• Fix for repeatability issue
• Multiple scattering does not use tables (due to
  ions)

• Refinements
• Tail of multiple scat. angular distribution
• New in Low Energy EM
• New models (2BN, 2BS) for Bremstrahlung (Lisbon
  INFN)
• New processes for electrons positrons (a-la
  Penelope)

36
EM Physics Processes more

• Multiple scattering
• Tuning for tail of angular distribution
• Improvement for muons of Egt1 PeV
• Ionisation
• Updated energy intervals, fluctuation models
• Multiple scattering does not use table
• Needed to ensure repeatability
• Added PAI (Photon-Absorption-Ionisation) model
• EM low energy physics
• New models (2BN, 2BS) for Bremstrahlung
• New processes for electrons positrons (a-la
  Penelope)
• Optical processes
• New process for wavelength shifting
• Adoption of G4SurfaceProperty class for materials

37
Hadronic Models new

• Evaporation models
• Ablation new model for use with abrasion code
• GEM-like model implementation
• HETC emission probabilities for Weisskopf-Ewing
  evaporation model
• Ion Reactions
• Wilsons Abrasion for induced ion reactions.
• EM dissociation for ion-ion collisions
• High energy elastic scattering new
  Coherent_elastic model
• requires a new data set for elastic scattering
  data (provided)
• Diverse
• new m- nuclear absorption code
• Improved fast radioactive decay code
• GNASH2 transition probabilities now available
  from exciton precompound model

38
Hadronics Cross Sections Scattering

• Cross sections
• Newest pion scattering data Barashenkov, remove
  discontinuities
• Fix in high energy p-H cross-sections (G3 legacy
  bug)
• Ion-ion cross-sections
• Tripathi's systematics for ion-ion cross-sections
  for light ions
• Parameterizations from Shiver, Kox and Shen
• Scattering term
• extended for nucleon induced reactions to 8 GeV
• included s-wave absorption
• pion induced reactions (up to 1.5 GeV)

39
Models Cascade energy range

• Parameterized process (1997)
• Chiral Invariant Phase Space decay,CHIPS
• For g-Nucleus, p capture, string-backend
• First release Dec 2001 in Geant4 4.0
• Refinements and extension in 2002
• Bertini cascade (Dec 2002, Geant4 5.0)
• Re-engineered from HETC by HIP
• See the presentation of A Heikinen
• Binary cascade model (Frankfurt, CERN)
• First release for nucleon induced interactions
  (in G4 5.0)
• Extensive verification suite
• See CHEP 03 presentation by D. Wright, V.
  Ivantchenko, ..
• For further details,
• see the CHEP 03 presentation by J.P. Wellisch

M Kosov, P Degtyarenko, JP Wellisch
A Heikinen N Stepanov JPW
G Folger JPW
40
EM regression testing

• TestEm
• Added check for automating regression tests
• First observable on average energy deposit
• activated by UI command (in 3 of 9 tests)
• Improving regression/acceptance testing

41
User Interaction

• Geant4 Technical Forum (http//cern.ch/geant4/tech
  nical_forum)
• Quarterly meeting
• open to interested people
• Users developer dialog
• Identify prioritize issues
• LCG Physics Validation meeting
• Comparisons with test beam data
• Several new physics developments presented
• We continue to emphasize identifying problems
• To enable better use in large production
• To solve issues seen by diverse users
• Growing feedback
• Requests for refinements
• Problems reported (many identifying the
  underlying issue)

42
Kernel Propagation in EM Field

• Transportation can keep some loopers
• From 7.0, tracks with E gt Eimportant go on for n
  long steps
• Default Eimportant 250 MeV)
• Ability to specialize integration accuracy
• emin, emax now for each Field-Manager
• Choice of Field-Manager by track
• e.g. more precise for muon or for tracks Egt5
  GeV
• Ability to use variant Chord-Finder (5.2)
• Can use safety, radius of curvature, other info
• For performance improvement

43
Kernel summary

• Development
• Modular Run Manager
• Better HEPMC input
• Abstract Navigator
• New Biasing
• Biasingweight-window technique
• Refinements
• General Particle Source
• Design iteration
• Integration of motion in field
• Enabled tuning of accuracy parameters for
  particle type, Energy,

• Improvements
• Navigator
• New check mode
• better verbosity
• Fixes
• Corrected safety in solids
• Addressed propagation photon problems
• Reported by LHCb
• Fixes for case of missed intersections in field
• purging magnet example
• Fixes for a 'point outside' problem seen in
  solids
• problem in displacement in field

View of Atlas toroid Courtesy of Atlas
44
Debugging geometries

• It is easy to create overlapping volumes
• During tracking Geant4 does not check for
  malformed geometries
• The problem of detecting significant overlaps
  is now addressed by
• DAVID intersects graphics volumes
• Created by S. Tanaka, released ca 1997
• Commands to run verification tests
• Created by DC Williams released in 4.0
• New capabilities added in 5.2 (June 2003)
• New example with full tracking / navigation
• Created by M Liendl (CMS) released in 5.0

Thanks to S. Tanaka
45
Kernel Changes for 7.0

• Several changes in kernel are planned for the 7.0
  release.
• In order of the effects
• New scheme of storing/retrieving physics tables
• Enables user to read a portion generate the
  rest
• New particle unknown and new process unknown
  decay
• For particles whose physics is not simulated, we
  now create
• Enables full decay chains to be treated uniformly
  
• New dedicated class for user step limitation
• Separating step length limitation and track
  killing
• Possibility of altering detector sensitivity with
  the parameterized volume
• Categories affected Tracking, Track,
  Processes/transportation
• New design of particles, replacing
  static-singletons
• Under discussion may affect code that creates
  physics processes.

46
Platform changes

• OS / compiler movement
• Newly supported (June 2004)
• gcc 3.2.3 on Linux (RH 7.3 SLC 3)
• Visual C .net 7.1 on Windows XP
• Emerging platforms (verified)
• MacOs 10.3 with gcc 3.3
• icc 8.0 (IA-32 IA-64)
• Checking for porting
• Latest gcc 3.3.3 and now 3.4
• Dropped end-2003 egcs
• To drop end-2004 gcc 2.95.2 Vis C 6

• Enabled shared-library mechanism for Windows
• With 6.2, end-June 2004
• Request of LHCb
• Goal
• keep up with needs user communities
• Do integration testing on at least 3 platforms
• Not more than 5, if possible

47
Upcoming Releases

• Developments available
• In monthly development tags
• In open b releases each quarter
• Except if there is a scheduled or consolidating
  minor release.
• Upcoming releases
• Scheduled major release Geant4 7.0 in
  mid-December
• New developments
• Improvements and other refinements
• Any fixes, further performance improvements.
• 2004 work items planned release contents
• At URL http//cern.ch/geant4/source/planned_featur
  es.html
• Started from requirements and requests of
  users/experiments

48
Established new releases new features

• Established releases
• End of June (minor release)
• End of December (major release)
• Planning the new activities for 2004
• taking into consideration requirements of all
  users including those from LHC experiments / LCG
• Users Technical Forum at CERN
• February 5th, 1500-1730
• Requirements collection and first-level
  prioritization

49

• Geant4 is evolving
• Feedback from HEP experiments, and users in
  medical, space domains.
• Regular Users Technical Forum meetings to
  collect/sort requirements and prioritise

50
Part II The Kernel

• New developments improvements
• in Geometry, Tracking,
• Run Event handling

51
Kernel Geometry

• Describing model geometries
• Solids
• Navigation
• Field propagation
• Active areas
• Additional checking of navigation model
  geometry
• Abstraction of G4Navigator
• Solids
• Revision of surface normals (booleans)
• New shapes (twisted solids, ellipsoidal solids)
• New volume types
• Refinement/extensions to parameterised volumes
• Optimisation of field propagation

52
Variance reduction

• Importance biasing
• Splitting/Russian roulette (G4 4.1, June 2002).
• Importance values for a volume
• In the mass geometry or in a dedicated
  parallel geometry.
• Used for shielding (speedup demonstrated)
• Limited in case of fields to mass geometry
• To be addressed via coupled parallel navigation
• Other general methods (eg forced interaction)
• Some existing, others in development

M Dressel
N.Kanaya