Adding new example21 which uses G4VTrackingManager #266
Conversation
|
Can one of the admins verify this patch? |
|
@WitekPokorski is there a way to enable/disable the integration, to be able to run the same app with G4 only? |
|
Yes, to run without AdePT you add the --no_AdePT argument when you call the example, like this: /example21 -m example21.mac --no_AdePT |
examples/Example21/CMakeLists.txt
Outdated
| if(NOT Geant4_flush_FOUND) | ||
| message(STATUS "Disabling example21 (needs G4VFastSimulationModel::Flush interface available starting with geant4-11-00-ref-01)") | ||
| return() | ||
| endif() |
There was a problem hiding this comment.
This example doesn't use G4VFastSimulationModel::Flush but G4VTrackingManager, which is available since Geant4 11.0
|
|
||
| //....oooOO0OOooo........oooOO0OOooo........oooOO0OOooo........oooOO0OOooo...... | ||
|
|
||
| void AdePTTrackingManager::BuildPhysicsTable(const G4ParticleDefinition &part) { |
There was a problem hiding this comment.
This function is called once per particle type (e-/+/gamma); it needs a protection to only initialize once.
examples/Example21/example21.cpp
Outdated
| AdePTTrackingManager* trackingManager = nullptr; | ||
|
|
||
| if(useAdePT){ | ||
| trackingManager = new AdePTTrackingManager; |
There was a problem hiding this comment.
The example currently only works with one thread. The reason is that we need one tracking manager per thread (the same way we had one fast simulation model per thread)...
There was a problem hiding this comment.
Argh... I new it was too easy... ;-) OK, I won't be able to look into it this week. If anybody has some time to touch this code, please go ahead. Otherwise, I am back next week...
* Reduced the size of the scoring arrays to just the number of sensitive PVols * G4 Physical Volume ID to Hit mapping in geant4 * VecGeom Placed Volume ID to Hit mapping + VecGeom PV to Geant4 PV ID mapping in AdePT * Removed unused code * Fixes for initialization issues in multithreaded mode
Modified scoring on CPU and GPU
New example21 demonstrates integration of AdePT into Geant4 using the G4VTrackingManager instead of Fast Simulation hook. In this case all e+,-, gammas are immediately handed over to AdePT (in the entire detector). No particles ever come back from AdePT. It seems to run OK with example21.mac, but certainly further polishing may be needed (and further bugs may be present due to limited testing so far).