Problem 2126

Summary:	G4mplIonisationWithDeltaModel produces too many delta electrons
Product:	Geant4	Reporter:	Doris Kim <dorisykim>
Component:	processes/electromagnetic	Assignee:	Vladimir.Ivantchenko
Status:	RESOLVED FIXED
Severity:	normal	CC:	daren.sawkey
Priority:	P4
Version:	10.1
Hardware:	All
OS:	All

Description Doris Kim 2019-02-04 08:39:08 CET

According to the header document in

G4mplIonisationWithDeltaModel.cc

S.P. Ahlen and K. Kinoshita, Phys. Rev. D26 (1982) 2347

is consulted to determine the low energy behavior. However, it's difficult to find the corresponding formula in the codes. It looks like this class is generating too many delta electrons.

Comment 1 Vladimir.Ivantchenko 2019-02-05 09:32:21 CET

Hello,

can you, please, point to some other publication with delta-electron cross section?

VI

Comment 2 Doris Kim 2019-02-06 13:31:25 CET

Dear Vladimir,

     I do not have separate reference. But I'm going to send an e-mail to Prof. Kinoshita asking for more details on her paper, such as intended energy range for delta ray electron production.

     Regards,
     Doris

Comment 3 Vladimir.Ivantchenko 2019-02-06 15:39:55 CET

Dear Doris,

please, have in mind that we use "restricted" cross section - integral from cut kinetic energy to max energy transfer. In that case, scattering on free electron is a good approach. Formulas in that case are simpler. We have no interest to full cross section, only for electrons which may provide a separate hits.

best regards,
Vladimir

Comment 4 Doris Kim 2019-02-14 01:41:32 CET

Dear Vladimir,

     Thank you for the comment. So the code assumes free electrons, not bound electrons. This is a simpler approach indeed.

     By the way, we just found that ATLAS used a modified version of your code. The entire ATALS code set is public now. We are going to decipher what is the difference between their code and the default one, which may or may not show why Belle II is seeing many hits in the pixel detector.

     Regards,
     Doris

Comment 5 Vladimir.Ivantchenko 2019-02-26 11:25:52 CET

Dear Doris,

is it possible to add a copy of the class used by ATLAS? 

Discussion with theorists may take long time. In a mean time, it would be useful to compare the class in the Geant4 distribution and one used by ATLAS.

Vladimir

Comment 6 Doris Kim 2019-02-26 16:09:02 CET

Dear Vladimir,

    Thanks for sending us a follow-up question. Here is the address of the ATLAS' codes, which is public now:

https://gitlab.cern.ch/atlas/athena/tree/master/Simulation/G4Extensions/Monopole/src

My people adopted this set of codes to the Belle II library and found that it generated reasonable amount of delta rays. I'd like to check what is the differences between the ATLAS codes, the official Geant4 code, and our version of codes, to figure out how that happenes But right now I got caught up with other urgent matters.

     Best Regards,
     Doris

Comment 7 Vladimir.Ivantchenko 2019-03-01 19:33:53 CET

Hello,

I have checked ATLAS model and current model in Geant4 10.4. For Silicon target the results are identical. There difference between models is in multi-threaded mode of the current code and more correct low-energy approximation for other targets, different from Silicon.

In the code a problem was identified: there is no low limit for delta-electron production threshold. In the current model such low-limit should be introduced.
Modified code is proposed for the development version of Geant4 and will be available with the next patch to Geant4 10.5.

Vladimir

Comment 8 Doris Kim 2019-03-07 02:54:52 CET

Hello,

    I apologize for late reply. It took for a while for me to recover the bugzilla login password.

    I appreciate that you have checked the ATLAS code against the official Geant4 code. I will convey your result to my monopole colleagues. And we have to keep in mind that there will be more to come with the Geant4 10.5 patch version on the monopole library.

     Regards,
     Doris

Comment 9 Vladimir.Ivantchenko 2019-04-08 16:28:05 CEST

Hello,

the fix will be in the next public release and in the patch for 10.5. If any question, please, reopen this report.

VI