Minutes of 21st January 2021 Meeting of Geant4 Technical Forum
Editor: Marilena Bandieramonte, 21st January 2021
- Starting at 16:00 (three hours format)
Introduction
============
+ see slides for detail
Major highlights:
- New Technical Forum chair:
- Marilena Bandieramonte (ATLAS, Pittsburg Univ.)
- Andreas Morsch (ALICE, CERN) is then stepping down. We express our deepest appreciation to him for his 10 years of service !
- The next TF meeting will take place during spring
- We plan a major release in 2021
- Today’s TF meeting dedicated to release 10.7
What’s new in 10.7 - non-physics part
============================================================
+ see slides for details
Major highlights:
- Updated VecGeom library, VecGeom v1.1.8
- Improved speed for overlaps detection
- New templated field classes for equation of motion and selected steppers
- New tasking system (Beta version) as option
What’s new in 10.7 - EM physics part
===========================================================
+ see slides for details
Major highlights:
EM developments for Geant4 10.7 was focused on:
- Technical improvements of code
- CPU performance for HEP applications
- New features for medical and low-energy applications
- Extension of capabilities for optical photon transport
- G4EmBuilder:
- A new utility class which will allow reduce code duplication for standard and low-energy EM physics constructors
- G4GammaGeneralProcess:
- Is available as an option in any standard Physics List
What’s new in 10.7 - Hadronic physics part
=================================================================
+ see slides for details
Major highlights:
- Update to Hadronic data sets
- For both FTF (Fritiof) & QGS (Quark Gluon String)
- Completed the extension to charm and bottom hadron projectiles
- Simplified treatment of the decays of secondary charm and bottom hadrons (i.e. created by inelastic nuclear interactions of primary hadrons)
- Stable hadronic showers in G4 10.7: we recommend to fit the Birks quenching coefficient from the h/e test-beam data !
- geant-val.cern.ch: only validation tool in Geant4, we rely heavily on it
Open requirements
=================
+ see slides for details
Major highlights: New requirements
- Extend “force collision” biasing to charged particles
- Include the Goudsmith-Sanderson multiple scattering model in the default EM option 0
- Improve inelastic cross-section of d-bar and antu_He3 at low energies
TO BE ADDED:
- Suggestion/Request from ATLAS to add a new Geant4 method (slide 12 of ATLAS report)
- G4PrimaryParticle::SetProperTimeFromDetectorFrameDecayLength (double GeneratorDecayLength)
ATLAS report
=========================================================================
+ see slides for details
Major highlights:
- Atlas is testing different Geant4 versions to select the RUN3 best candidate
- Effects of Birks constant tuning still to be understood
- Established in Sept 2020 a new Geant4 (G4) Optimisation Task Force (TF), responsible for optimising the performance of the ATLAS G4 simulation software
- Feedback on intrinsic Geant4.10.6 performance optimisations:
- Reduced n. of log calls in EM physics: ~5% speedup (fullsimlight simulation with ATLAS geometry and magnetic field)
- General Gamma Process: ~4.3% speedup measured in Athena with tt-bar events
- New way of setting the stepper/driver in Geant4.10.6 vs old way in Geant4.10.1.patch03.atlas07 : 11.82% speedup (comprehensive of ~10% speedup coming from optimisations across many Geant4 versions)
- Quasi-stable particle simulation was fixed (problem was on the ATLAS side)
- Suggestion/Request to add a new Geant4 method • G4PrimaryParticle::SetProperTimeFromDetectorFrameDecayLength(double GeneratorDecayLength)
- Many ongoing optimisation activities
CMS report
=========================================================================
+ see slides for details
Major highlights:
- Currently CMS uses the version Geant4.10.6.p02 by default
- Doing ultra-legacy MC production for Run2 data sets using the version Geant4.10.4.p03
- Considering to use Geant4.10.7 for Run3 MC production. This version is available since December 2020
- Planning to use the same physics list for ultra legacy as well as for the Run3 production • FTFP_BERT_EMM
- The validation is carried out using 2 sources of data:
- 2006 test beam with CMS calorimeter prototypes (hadron beams of different types and different energies)
- Collision data from the CMS experiment utilizing zero bias or minimum bias triggers from low luminosity runs
- All 3 versions (10.4.p03, 10.6.p02 and 10.7) provide good agreement with the data
On modelling multi fragment decays with Geant4
=========================================================================
+ see slides for details
Major highlights:
- Standalone tests of G4SMM and G4FermiBreakUp (v9.1 – v10.4): comparison with the FORTRAN versions of the respective models
- G4FermiBreakUp: As a result of debugging and tunes several changes were proposed by us to G4 developers and implemented in v9.2
- Comparisons of G4FermiBreakUp across v 9.2, 9.6, 10.3 and 10.4 show that from 9.6 on G4FermiBreakUp doesn’t work properly
- G4SMM works well in v10.4
- Problems with G4FermiBreakUp are discussed at Geant4 User Forum since July of 2020: https://geant4-forum.web.cern.ch/t/strange-results-of-g4fermibreakup-in -v10-4/3137
- Also reported within Geant4 Problem Tracking System: https://bugzilla-geant4.kek.jp/show_bug.cgi?id=2263