| Summary: | Identical initial random seed does not produce identical events probably due to G4LowEnergyGammaConversion | ||
|---|---|---|---|
| Product: | Geant4 | Reporter: | David Jaffe <djaffe> |
| Component: | processes/electromagnetic/lowenergy | Assignee: | Maria.Grazia.Pia |
| Status: | CLOSED INVALID | ||
| Severity: | normal | CC: | djaffe |
| Priority: | P5 | ||
| Version: | 8.1 | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Attachments: | modified routines of TestEm1 example, input macro and random engine state, and output log file | ||
There is no preStepKinEnergy variable in G4LowEnergyGammaConversion, nor such string appears in the whole electromagnetic/lowenergy package. All G4LowEnergyGammaConversion data members are properly initialized in the class constructor. Therefore, there is no way to establish any connection between your random number generation problem and G4LowEnergyGammaConversion. Maria Grazia Pia |
Created attachment 8 [details] modified routines of TestEm1 example, input macro and random engine state, and output log file Non-identical events are produced for identical initial states of the random number engine when using electromagnetic low energy processes. I suspect the problem is lack of proper initialization of preStepKinEnergy and related quantities in the G4LowEnergyGammaConversion process. This problem was first noted within the framework of the Daya Bay experiment using Geant4.8.2.p01 and successfully reproduced using the extended/TestEm1 example using Geant4.8.1. The file out016.log in the attachment contains the track history for 4 events generated with the same initial state of the random number engine. The initial event differs from the 3 subsequent events. The 3 subsequent events are all identical. Studies within the Daya Bay framework indicate that preStepKinEnergy for G4LowEnergyGammaConversion contains 'stale' information from the previous event upon first invocation in a subsequent event and is therefore probably not being initialized properly.