ssw2mcpl_app.c: Store NPS?

[willend]

Hi @EsbenKlinkby, @tkittel

For bookkeeping and handling correct weight normalisation etc., @ebknudsen and I suggest to:

• If not already implemented, let ssw2mcpl_app.c store the MCNP input deck as a binary blob (if file provided)

• Put the NPS information from the MCNP deck into e.g. a comment field in the resulting MCPL file

• This would allow easier correct normalisation of e.g. ESS moderator performance on the McStas side, irrespective of MCNP run stats. @ebknudsen reports he will happily assist... ;-)

Peter

[tkittel]

So for the first point, this is already both supported and mentioned in the paper.

For the second point, this is actually a rather complicated issue, due to the potential for file merging (e.g. putting a comment with "NPS was 2000000" would either prevent file merging when two files has different NPS or lead to a merged file with a misleading/wrong comment when two files have identical NPS). However, it is an issue which continues to surface in one form or another, so it is definitely worthwhile to figure out a solution to this. Perhaps we should hold a brain storming session in-person one day to discuss the pros and cons of various ways to address this. It might be we need a third place for custom information in the header, in some sort of "stat" section, where statistics might be "named" numbers associated with a merge characteristics (such as "add" or "augment" or whatever).

[willend]

Excellent for the first point 👍 - As you may have noticed from the formulation I did have a recollection this might be the case... ;-)

Also agreed for point 2: Complicated, potential merge issues, brainstorm needed! :-)

[MilanKlausz]

+1 for any option to store comment/metadata that is summed when multiple files are merged. In short, the output of my Geant4 simulation is an MCPL file, and for later processing I need the information of how many initial neurons were simulated to produce that file. I'm launching a lot of Geant4 simulations, and merging the resulting MCPL files, and right now without any "additive metadata field", I have to do the bookkeeping of how many neutrons were simulated in total for my merged MCPL fle. (I don't know the total number beforehand, as I might need to do additional simulations later on to acquire better statistics.)

[tkittel]

More than 6 years later @MilanKlausz ? ;-)

But yeah, it makes sense. I am wondering if we could keep it simple and have this supported in the current comment fields, by simply having a particular format like "ADDITIVIVESTATISTICS::<somecommenthere>::<number>".

Given that this have to be implemented in C-code, however, and also make the files compatible for merges despite different <number>'s, it would be a non-trivial change. But I imagine I can do it in <6 years ;-)