Created attachment 803 [details] muons stop vs layer for v10.7.p03 and v11.1 There seem to be a big difference in muon energy loss between Geant4 v10.7.p0n and v11+ when using default standard EM option. One way the difference manifests itself is in the number of muons which stop in a collection of thin aluminum layers. The attached plot shows overlayed histograms which show the numbers of muons which stop (are captured) in a given layer in a setup where where there are 20 G4_Al 0.1mm thick layers placed every 20mm in G4_Vacuum and 100k 40MeV momentum muons are sent straight at the first layer. The solid dark blue histogram corresponds to Geant4 v10.7.p03 and QGSP_BERT physics list. The red histogram corresponds to v11.1 and QGSP_BERT. The black diamonds correspond to v10.7.p03 and QGSP_BERT_EMZ. The cyan open cross markers correspond to v11.1 and QGSP_BERT_EMZ. Another way one can see the effect is to look at the kinetic energy of the muons exiting the layers. The two to be attached scatterplots show this for Geant4 v10.7.p03 and v11.1 using QGSP_BERT physics list.
Created attachment 804 [details] kinetic energy of muons exiting layers for v10.7.p03 Adding scatterplot showing kinetic energy of muons exiting the layers for v10.7.p03
Created attachment 805 [details] kinetic energy of muons exiting layers for v11.1
As a complementary information, there was a change from 10.7 to 11.0 in G4VMscModel::GetTransportMeanFreePath, introducing an additional pFactor (density correction with respect to the base material) in the MSC mean free path calculation - please compare https://geant4.kek.jp/lxr/source/processes/electromagnetic/utils/include/G4VMscModel.hh?v=11.0#L412 and https://geant4.kek.jp/lxr/source/processes/electromagnetic/utils/include/G4VMscModel.hh?v=10.7.p4#L405
Hello Krzysztof, thank you very much for this problem report. This problem is not obvious and history of how this happens is not trivial. In summary, mu-, pi-, and K- dEdx, Range, and InverseRange tables are using linear interpolation instead of spline interpolation. This provides a small numerical effect on tracking. The most sensitive energy for this problem for mu- ~50 MeV/c. I am working on the fix, which will be propagated to patches to 11.0 and 11.1, in a mean time, the best recommendation would be to increase number of bins in tables: /process/em/binsPerDecade 20 This should fix the problem. Vladimir
Hello, the problem is fixed and the fix will be available in the next public version and patches to previous versions of Geant4. VI
