While running studies on secondary-particle production off a beryllium target for the North Area at CERN, I found a systematic drop in the π/K yield when going from Geant4 11.0 to 11.1 with FTFP_BERT.
I built a model with a 400 GeV/c proton beam impinging on a beryllium target (500 mm long, 2 mm high, 160 mm wide) and scored the particles leaving the target volume, normalised per incident proton. I used the default cuts and processes in FTFP_BERT; I can share the code if useful.
I tested 10.7.1, 11.0.4, 11.1.3 and 11.4.2. Versions 10.7 and 11.0 agree with each other, and 11.1 and 11.4 agree with each other, so in the attachment I show only the 11.0 vs 11.1 comparison. The drops are energy dependent and reach up to ~40%.
I saw in the changelog of 11.1 that changes were made to the hadronic physics to better describe NA61 data, but the change in meson production seems large. I checked that the list of relevant process is the same when changing version, but I might have missed something.
Is this change expected? And if so, is 11.1 expected to agree better with reference data (like Atherton p+Be in the North Area) than 11.0 for meson production in this channel?
no, the ~40% drops in meson production for 400 GeV/c proton on Be in G4 11.1 was not expected/predicted !
The cause for this is, very likely, what you have correctly guessed, i.e. the change in the Fritiof (FTF) string model to better describe NA61/SHINE data.
The hope is that this improvement will also bring a better description of other hadron-nuclear interactions, including the one relevant for the North Area.
However, we have seen cases in the past in which this hope was not fulfilled, unfortunately !
Therefore, my suggestion would be to run FLUKA-Cern (there is an example in examples/extended/hadronic/FlukaCern/, introduced in Geant4 11.2, which shows how to run Fortran FLUKA-Cern hadronic physics in a Geant4 application) using the same set-up you have been using for comparing the different Geant4 versions to see whether G4 11.0 or 11.1 is closer to FLUKA.
If it is the version Geant4 11.0 which closer to FLUKA, I can try to roll back some of the relevant FTF classes so that you can run a more recent version of Geant4 (e.g. 11.4.2) with the same behavior as in Geant4 11.0, at least for the meson production in 400 GeV/c proton on Be.
Thanks a lot for your feedback, and please keep us update!
I see. I tried running it using the binaries from FLUKA 4.5.2 and using Geant4 11.4.2. I would say that in general both differ quite much in the production from FLUKA. Pion production seems a bit closer to FLUKA before the change in Geant4 11.1, but I’m not sure how conclusive this can be. I tried both using only the hadron physics from FLUKA and trying to compile a physics list with the other processes from FTFP_BERT, but I don’t see any difference:
