Different results from same input on 2 computers

[Andrea_Missaglia]

Dear FLUKA experts,
I ran the same simulation (same input and flair files) on 2 different PCs: proton irradiation of a TISSUEIC target, sorrounded by a LYSO detector, so that the chosen scoring card is DETECT.
The results are the following:

Unfortunately, the 2 spectra are quite different, with an average relative error between the 2 results of 20%. The error was computed through MATLAB as:

where counts_PC_i with i=1,2 is the third column of the DETECT output.

I wonder if it is just a matter of FLUKA version, which is different for the 2 computers, as one can see below:

1. PC 1
   

2. PC 2
   

Anyway, if it is not the case, what could the be the source of this discrepancy?

Thanks in advance and have a nice day,
Andrea

Input and flair files of the simulation:

• Input:
  prova_simple_f.inp (1.7 KB)
• Flair:
  prova_simple_f.flair (2.6 KB)

[cesc (Francesc Salvat Pujol)]

Dear Andrea,

Thanks for reporting this unexpected behavior.

We have looked in detail at the issue and indeed spotted a spurious setting in the pre-equilibrium stage of FLUKA’s nuclear inelastic interaction model. In your particular case, this produced a visible effect in the DETECT spectrum due to a moderate increase in the natO(p,g) yield with respect to version 4-1.1.

The fix for this bug will be shipped with an upcoming minor release (4-2.2). Note that with the latter you will get closer to the 4-1.1 result, but not exactly overlapping, due to an additional model refinement that will be indicated in the RELEASE-NOTES.

With kind regards,

Cesc

PS: in your problem (150 MeV p on water) it may make sense to pass a PHYSICS card with SDUM=COALESCE.