Problem 2366

Summary: No transmission asymmetries when using multi-threading mode
Product: Geant4 Reporter: Jennifer Popp <albenstern.jenni>
Component: processes/electromagnetic/polarisationAssignee: Vladimir.Ivantchenko
Status: ASSIGNED ---    
Severity: normal CC: Vladimir.Ivantchenko
Priority: P4    
Version: 10.6   
Hardware: All   
OS: Linux   

Description Jennifer Popp 2021-05-05 16:01:24 CEST 
Hello,

Based on example B4a and Pol01 I was simulating the transmission of circular polarized gammas through a polarized iron block (e.g. 5 MeV gammas and an iron block of 5 cm diameter and 7.5cm thickness) . Depending on the chirality of the incident photons , or whether the polarization direction of the iron is parallel or antiparallel to the propagation direction of the photons, there should be an asymmetry.  The polarization of the iron block was set to (0,0,1) and that of the iron block to (0,0, -1/1), the z direction also being the propagation direction of the photons. 

When I used Pol01 in sequential mode, only edited the geometry to my needs, I see transmission asymmetries in the order I expect. 

Unfortunately, I started with example B4a, changed that to my needs and implemented the physicslist from Pol01. The asymmetry was statistically zero. I spend ages trying to find out why. 

When I found out that it was working when I started with the Pol01 example, I assimilated both codes until I was sure that it is the multi threading that is causing the problem. Unfortunately, I do not know what exactly is causing the problem. There are no errors thrown and I found nothing in the user documentation about it. So I do not know whether there is something wrong with the process, or whether this is rather a suggestion for an enhancement of the user documentation stating how to implement polarized materials in Multi Threading mode. 

If you need any more information or material from me, please let me know.

Kind regards,

Jennifer
Comment 1 Vladimir.Ivantchenko 2021-05-05 23:22:33 CEST 
Hello,

please, assign the problem to me.

Vladimir
Comment 2 Vladimir.Ivantchenko 2021-05-05 23:32:05 CEST 
Hello,

when making migration from sequential to multi-threaded regime several modifications should be implemented to any application, in particular in user actions and scoring. The polarisation library was not tested in multi-treated mode but I do think it is compatible. More probably, the Pol01 example requires some revision. 

VI
Comment 3 Vladimir.Ivantchenko 2021-05-08 10:47:21 CEST 
Hello,

if you really want to use polarization library in the MT mode, I would propose, that you copy Pol01 into new application Pol02 and try to migrate it from sequential to the MT mode. This will require modifications practically in all classes of the example but this was done many times. I would suggest that the example of the MT implementation 

$G4INSTALL/examples/extended/electromagnetic/TestEm5. 

The polarisation sub-library belong to EM domain of Geant4, so better making Pol02 coherently with the rest of Geant4. TestEm5 may be more complicate than you needed but it has everything inside.

Vladimir