Hello to all.
I have to perform this simulation:
geom_detect_performance_Proton.inp (32.6 KB)
That involves producing and tracking the optical photons in an optical fibre. The problem has to do with the fact that the simulation is taking about 1.5 seconds per primary. QUESTION: I would like to know if there is a way to decrease the computation time.
Thank you in advance 
.
Kind regards,
Duarte.
Question 2: Let’s suppose I try to run this simulation on different folders, each with less primaries.
If I give the indication that I want FLUKA to run a different cycle on every folder ( e,g.: folder 1 rfluka -M 1 filename.inp and folder 2: folder 1 rfluka -M 2 filename.inp) , is the RNG seed going to be different for each cycle or that does not work?
Thank you.
Duarte
2. You cannot start a run with a cycle different from (larger than) 1, unless the previous one has been already run in the same folder. The solution is to run in each folder an input with a different RANDOMIZe seed (WHAT(2)), starting always from cycle 1.
1. Your computation time is likely linked to your geometry definition, in particular the three regions featuring one single zone with plenty of (subtracted) bodies. Redefine them in a revised manner, as the union of several zones, each involving a reduced number of bodies (instead of subtracting 200 bodies from a big container, subtract 10-20 bodies from a suitably smaller container, and join all the resulting zones in the same region, considering that zones can overlap).
That is what I am doing at the moment. Regarding the multiple folders.
Thank you for the insight on the run time.
Kind regards,
Duarte