Impossible Run Time

Hi, I’ve been building linear accelerator to simulate radiotherapy in geant4.
I used GDML format to create the geometry, and successfully ran for the first 2e9 events then stored its value (dose deposits in certain organs) with about 22 hours of runtime.
But the thing is, I need to run approximately 7e13 events (for 1 monitor unit) which impossible in my setup to do simulation with that much time (approximately need 4583 days to run).

The solutions I tried to shorten the runtime were, disabling visualization and using multithread. Any suggestions to this? or can I just multiply the value I got earlier to match number of events I want to achieve?

yes, in general you can scale the number of events to match the situation you want to simulate - you just need to make sure that the number of events is large enough for the statistics you want to achieve.

very powerful methods for increasing effective performance are biasing techniques

thank you weller, I’ve read the documentation, and if I get it correctly, I need to do a good sampling to my simulation, and eventually I can multiply values from my sample to match number of event I want to achieve, right?


with biasing on top, you might achieve good sampling with fewer events (or better sampling with the same number of events).

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.