search

mctools / mcpl

Monte Carlo Particle Lists
https://mctools.github.io/mcpl/
Other
29 stars 13 forks source link

Inconsistency of mcpl2ssw #73

Open mightyFredJ opened 1 year ago

mightyFredJ commented 1 year ago

Hi,

I have a problem with mcpl2ssw, which may arise from a bug. Apply to translations ssw2mcpl then mcpl2ssw seem to alter the inner data:

I have a MCNP simulation (mre.i) computing a tally while creating a SSW file (mre1.wssa) Then I convert the mre1.wssa file to mre1.mcpl with ssw2mcpl (ssw2mcpl mre1.wssa mre1.mcpl -n) Then I immediately go back with mcnp2ssw to create mre2.wssa (mcpl2ssw mre1.mcpl mre1.wssa mre2.wssa -s10). At this point I expect this new file to contain same data as mre1.wssa

mre1.wssa and mre2.wssa have the same size but lead to different results after a SSR run (mre.i minus SDEF minus SSW plus SSR). (of course) the SSR run using the original mre1.wssa gives the same results as the first SSW run. It's mre2.wssa which gives unexpected (wrong) results.

I can attach all these files if needed.

thanks for your work

tkittel commented 1 year ago

I can't tell for sure, but at least you should use the -s flag when invoking ssw2mcpl in this case (and remove the -s10 flag when invoking mcpl2ssw. Otherwise you can try to invoke ssw2mcpl once more, and compare the two mcpl files with mcpltool to get an idea of what might have been lost in translation.

(ps. I am on vacation so I might not be able to help more with this until ~August)

tkittel commented 1 year ago

Also, please supply details such as mcnpl version.

mightyFredJ commented 1 year ago

Thank you for your answer. I use MCPL v1.6.2 (likely the last version, downloaded a few days ago and compiled from the *fat .c files).

I tried to move the -s option but it seems to change nothing to my case.

I attach you the files used to analyze this behaviour, but feel free to ask for any complementary information (I'll be ask in july too, back in august)

mre_ssw.i.txt mre_ssr.i.txt ssw2mcpl_inspection.sh.txt ssw2mcpl_inspection.log

tkittel commented 1 year ago

Thanks. I most likely won't be able to look at this until August, but in any case I don't actually run MCNP myself. Perhaps @EsbenKlinkby can have a look (although he is most likely also on vacation).

tkittel commented 1 year ago

Btw., are you sure you expect the exact same results? I.e. there might be some sort of random number seed thing going on. Are the discrepancies you observe actually statistically significant?