Range of low-energy muons changed between 11.0.p0 and 11.2.p01

nberger · May 23, 2024, 1:18pm

In the simulation of the Mu3e experiment we see that the range of muons at p=27.5 MeV/c in Mylar is reduced by about 100 um when changing from 11.0.p0 to 11.2.p01. We have reproduced this on several linux machines using different versions of gcc. Our muon physics setup is posted below. Is this expected by some change in the models/tables? Note that the old simulations (11.0. and earlier) agree well with measurements at PSI.
Best regards
Nik

    else if(particleName == "mu+"
         || particleName == "mu-"
         || boost::starts_with(particleName, "mu+/")
    ) {
        helper->RegisterProcess(getScatteringProcess(msmodel), particle);

        //helper->RegisterProcess(new G4MuMultipleScattering, particle);
        helper->RegisterProcess(new G4MuIonisation, particle);
        helper->RegisterProcess(new G4MuBremsstrahlung, particle);
        helper->RegisterProcess(new G4MuPairProduction, particle);

        auto muonNuclearProcess = new G4MuonNuclearProcess();
        muonNuclearProcess->RegisterMe(new G4MuonVDNuclearModel);
        helper->RegisterProcess(muonNuclearProcess, particle);
    }

civanch · May 23, 2024, 2:47pm

Hello, Nik,

nothing was expected to be change for range of muons.

Please, try to do few simple checks: 1) compare properties of Mylar in these two versions, criticals are density and mean ionisation potential; 2) infor us what is used as a multiple scattering model; 3) try to use $G4INSTALL/examples/extended/electromagnetic/TestEm5, use G4_MYLAR as a target and reproduce the effect (this run very fast, more time will be needed to switch the G4 version than to run).

After these checks we may decide if there is a subject for Bugzilla bug report.

VI

nberger · May 27, 2024, 11:14am

Hi,
thanks for the quick response. I see no difference in the properties of Mylar between the two versions. We use the Urban model for multiple scattering. I now ran TestEm5 with Mylar and emstandard_opt0 as physics and also there see a difference in range - 825 um in 11.0.0 to 623 um in 11.2.1. I have attached the output of both runs, I hope this helps with further diagnosis.
Thanks
Nik

output_11_0.txt (75.3 KB)
output_11_2_1.txt (73.5 KB)

nberger · June 11, 2024, 10:08am

Is there any other input you need? How would I go about filing a bug report?
Many thanks
Nik

nberger · July 1, 2024, 11:12am

Hi again, is there any progress or something I can do?

nberger · September 3, 2024, 3:25pm

Hi again, I now ran into a second group that encountered the same problem, this seems to be affecting quite a few low energy experiments. What can we do to address this bug?

christophe.rappold · September 12, 2024, 7:56pm

Hello everyone,

I would like to notify that we encounter a similar problem of the mu+ range in the nuclear emulsion plate at low kinetic energy:

For some research in nuclear emulsions, we try to reproduce the published results of the measured range of mu+ obtained from pi+ decay at rest (see the uploaded plot with reference). The mu+ has a kinetic energy of 4.1205 MeV, it is fixed from the 2-body decay of pi+. The range that was measured by Barkas et al. is 600 ± 27 um.
In our simulation code based on geant4 version 11.2.1, with all the different physics lists that can be set from the physicsListFactory (main + EM options), we get ranges of the mu+ of around 492.2 ± 17 um. The material is from the geant4 database “G4_PHOTO_EMULSION” that corresponds to the emulsion plate related to the Barkas et al. published results.

So we also see a missing ~100 um in range of mu+, and this from a different field.

Where could this shift come from ?

It is critical for our research.

maire1 · September 13, 2024, 5:05pm

Could you repeat your simulation with cut = 2.5 um ?
Please, keep us informed. Thanks, Michel

christophe.rappold · September 13, 2024, 6:39pm

We already tried with or without SetUserLimit into the logicalVolume (the emulsion), from very small (0.1 um) to bigger (half of the expect range) to even bigger.
Same results : we are missing this ~100 um.

Today I tried with the geant4 version 11.0.3: and I have a correct range of 608 ± 19 um, compatible with the Barkas et al. results. With the exact same conditions of the previous post.

So any idea of the problem in the current geant4 version ?

maire1 · September 14, 2024, 1:16pm

UserLimit is not the same thing as ProductionCut.
If you use UI commands, it is enough to do :
/run/setCut 2.5 um

christophe.rappold · September 14, 2024, 2:34pm

Yes, I did not realized that you were talking about ProductionCut.

So in the macro file to start a batch simulation, I added the line you suggested.
The simulations finished : range of mu+ from the pi+ decay at rest : 509 ± 17 um.

It did change from 492 um to 509 um (still within 1 sigma), that should be 600 um.

The current version of geant4 with the productionCut has still 100 um of range missing for mu+, compared to the older version 11.0.03.

maire1 · September 16, 2024, 7:15am

I have run TestEm0 and TestEm1.
Both confirms your observation : a range of ~500 um instead of 600 um.

christophe.mac.txt (516 Bytes)

maire1 · September 19, 2024, 4:44pm

@christophe.rappold @nberger @civanch
1- I notice that this ‘change’ happens for mu+, but not for mu-
2- It is visible with TestEm0 (eg. EmCalculator). So, multiple scattering is not concerned.