in our application we are simulating a sampling calorimeter. We have noticed some differences in the results obtained by using Geant4 10.06.p03 and 11.02.01. In particular, the distribution of the times of the energy depositions shows a large tail in 11.02.01 compared to 10.06.p03:
The distribution of the energy depositions in the calorimetric channels show a discrepancy in the lower energy bin (see EL1.pdf). The larger tail in the time distribution is due to these depositions: in fact, excluding energy depositions lower than 1 MeV we get the distribution shown in the plot time_logy.pdf.
I wanted to stress that we are considering true energy deposits, without including optical simulation. The physics list used is QGSP_BERT_HP.
As far as we understand from the release notes, there should be no significant difference between the outputs obtained with the two Geant4 versions.
Any suggestions and feedback on things we may have missed are very well welcome.
it is likely that the weird behavior that you observed with QGSP_BERT_HP in Geant4 version 11.2 is due to a bug (not yet fixed even in the latest patch 11.2.2).
In all Geant4 versions before 11.2, the “_HP” behavior regarding low-energy (< 20 MeV) neutrons is the same regardless of the physics list considered, e.g. QGSP_BERT_HP, FTFP_BERT_HP, QGSP_BIC_HP, Shielding, etc.
In Geant4 11.2 (and the same will be in the coming release 11.3, expected for December 6th), we have introduced some “experimental” changes that affect the behavior of “_HP” only for the physics list QGSP_BERT_HP, but not the others HP-based physics lists. This was described - not very clearly indeed - in the following sentence of the Release Notes of Geant4 11.2 :
" G4NeutronHPCaptureXS, G4NeutronHPElasticXS, G4NeutronHPInelasticXS, G4NeutronHPFissionXS, G4CrossSectionHP, G4NeutronFissionVI, G4NeutronRadCaptureHP: updated models and cross-sections which are included only in the physics list QGSP_BERT_HP"
The choice of the QGSP_BERT_HP was motivated by our (likely wrong) assumption that this physics list is the least used between the HP-based physics lists.
In conclusion, for Geant4 11.2, we recommend to use a different HP-based physics list than QGSP_BERT_HP - for instance FTFP_BERT_HP.
Just for curiosity, we would be very much interested if you could redo the plot with the coming release Geant4 11.3, with QGSP_BERT_HP, to see whether you still see the same wrong behavior as in 11.2 - same improvements have been made…
Thanks for your report and apologies for the missing warning regarding the use of QGSP_BERT_HP in Geant4 11.2 !
