Bug in calculation the fraction

[heavyionanalysis]

Bug in calculation the fraction

[kreisl]

Could you write a short example which reproduces this problem? It makes it easier to check and then I can add it to the unit tests to make sure this case is covered.

[mam-mih-val]

Here's the small root macro reproducing the problem.
https://github.com/mam-mih-val/qn_tools_test

[mam-mih-val]

Upd: Inf/nan is on Viktor's fork version. On current version of this repo result histo ("flow") is not filled at all.

[kreisl]

@mam-mih-val can you try with the branch kreisl:projectiontest. It only appears if we have empty bins in the merging. It should fix the issues you are seeing? Please take note to merge bins after dividing we need to set the numerator as observable, to know which weight to take? In your example https://github.com/mam-mih-val/qn_tools_test i think you need to edit the lines like so:

Qn::DataContainerStats flow = *container_un_qn;
for (auto & bin : flow) {
  bin.SetWeights(Qn::Stats::Weights::OBSERVABLE);
  bin.ResetBits(Qn::Stats::CORRELATEDERRORS);
}
flow = flow / resolution;

Somehow the correlated errors don't work that's why i disabled them. maybe we need to look in a separate test into this. Please let me know if it worked.

[mam-mih-val]

Yeah, the fix you offered works. But as I can remember, there was not such a bug in earlier versions, was it?

[kreisl]

Yes this could be at some point I cleaned up the logic. I think this is not perfect yet. We should discuss again all possible cases and make sure we have them all covered.