12th Dec 2024 - Minutes

Intro by Alberto 

No questions

Gabriele Cosmo: Geant4 Kernel news in Geant4 11.3.0

• Pere: on sub-event parallelism, is the chunking of the subevent done automatically? Yes
• What is the use case of phase 2? how can you distinguish the different physics for different chunks
• Alberto: there will be the possibility to group together particles inside chunks, that can be assigned to different physics lists.
• Pere: the grouping will be user drive?
• Yes.
• Marilena: slide 6: sub-event parallelism - how to activate it, is it transparent to the user?
• There is an option in the run manager factory. It is one of the modes available, but no additional development is needed. 
• Is it working in the new Task based model? 
• Yes

Vladimir on EM physics

• Mikhail: Positronium physics is interesting not only for med physics but also to do material studies
• Marilena: Slide 5: Positron 3- gamma annihilation on fly, is it included in default physics lists? or is there a UI command? How to define the threshold?
• For now it is included only EMZ physics list, but not in the others (default) because we didn't study the CPU effect yet.
• lowestTripletEnergy : should not go below e- mass, it is connected to the structure of the calorimeter or structure of cells of fine grained calorimeters (it affects transferred and longitudinal energy distribution and total deposited energy at the per mille level)

Alberto on Hadronic physics

• Marilena: slide 10: how does QGSP_BERT_HP compare to FTFP_BERT_HP? Do you expect to be better for radiation modelling studies?
• We don't know yet, because this new HP model is still under development.
• Marilena on slide 11: improvement in the angular emission of ≥ 4 bodies in BERT, reducing the differences between BERT and FTFP -> do you suggest to use QSGP_BERT? 
• No, FTFP_BERT is still the suggested one for HEP, but QSGP_BERT can be interesting to test  given the recent big improvements. It can be used for comparison for example for systematic studies in Jet studies to see the effect of the string model. Big improvements in the QSGP

Witek on R&D

• Marilena on slide 9:  in Adept the geometry can be read only from gdml? What are the missing pieces for having a direct conversion from Geant4?
• A translator was implemented first in VecGeom  and then improved and maintained on the Celeritas side. Some dependencies have to be removed in order to make it standalone to link it in a clean way to Adept now. Work is in progress

----- USERS PART

Mikhail Kirsanov on hadronic physics models, NA64 experiment

• Alberto: EM calorimeter is made of ?
• Lead. The study was made in the Hadronic calorimeter made of Iron, the active material is scintillator in both
• Lorenzo: I observed this weird bump when shooting hadrons into calorimeters made of different materials, close to the energy of the hadron. What you present is interesting because I was attributing this bumps to some inaccuracy in the charge exchange process. What you suggest here is interesting also because when you turn-off this effect  charge exchange doesn't happen anymore.
• Alberto:  on slide 7 since G4 10.5 and up to recent versions the diffraction target and projectile diffraction are switched off for heavy nuclei (like lead). In principle it should not do anything
• Julia: This effect is not applied to all hadrons. Also there is no need to comment out the code, because there's an option to switch the projectile diffraction on/off
• Mikhail: idea to continue the study and compare the full shape of this curve. In the past we had problems like drifts in the calorimeters and the energy resolutiion. Now we have done work to improve the accuracy (energy resolution, control of drifts). So now we can resume this study and provide some data to improve the model in G4. 

Sunanda on CMS update

• Gabriele:  on slide 5 for the second case of stuck tracks you could try to use the latest version of VecGeom that contains some fixes

Xiaorong Zhou on antiproton simulation

• Alberto: The change that you see btw G4 9 and 10, is due to the chips model that was dropped from 10.0 on - today for annihilation of antiproton, we use FTF model. If you take 11.2 or 11.,3 FTFP_INCLXX, INCL will be used for antiproton annihilation in fly and at low energy, that should be the best description for antiproton physics. If FTFP_INCLXX doesn't work well in general, and FTF_BERT is the best physics list for you it could be possible to use INCL only for antiprotons.
• Jean-Christophe David: INCL and FERMI motion. To be honest, it's an assumption in INCL, we don't take into account the FERMI motion. It would be good to see the result with INCL, to see if FERMI motion is really important. In some cases like slide 13, it might be important.

Mustafa on ATLAS update

• Vladimir: on the requirement about the physics stability: this is the time to develop new things
• Marilena: the requirement is about having a frozen physics list/models during a Run
• Gabriele: this is very difficult to obtain because even changes to the kernel can change the physics output
• Marilena: we don't expect bitwise compatibility but statistical one. It would be useful to exclude that changes are due to changes in the underlying physics
• LHCb guy (sorry I don't know your name): about the hackhaton, how much time was spent on getting the build to work?
• Ben: in LHCb Gaussino you have AdePT working, so if it works with AdePT is going to work with Celeritas also

Patrick on python bindings for G4

• Maxim Potekhin: are you running with MT G4?
• No, we had problems with that. 
• How do you map symbols in c++ sense to your library, where these class resides in the .so is not entirely transparent. Your framework has to find the .so for the symbols to be loaded.
• For G4 we load all the Geant4.so files. But the headers are not loaded. Only the global namespaces that cppyy has only the one that you ask at runtime
• The par that crashes is the load of some headers
• Did you have problems with namespaces because H2 can mean Hydrogen but also 2D histos?
• We haven't seen these problems. It might be due to the lazy loads of includes/libraries. 
• Ben Morgan: the work in this ares is very interesting. We got interest also from the Gate collaboration, they have partial bindings as well. We could coordinate these efforts together at the beginning of next year.  https://indico.cern.ch/event/1338689/contributions/6015924/  might be of interest for MT in Python