ttH stitching

[camendola]

In 2017/2018, we have ttH datasets:

• ttHtoBB
• ttHtoNonBB
• ttHtoTauTau

ttHtoTauTau and ttHtoNonBB are most likely not orthogonal (ttHtoTauTau events already contained in ttHtoNonBB). To increase the statistics for ttHtoTauTau events, they should be removed when occurring in ttHtoNonBB so that we can add the ttHtoTauTau dataset.

[dzuolo]

Ciao @camendola I remember that Konstantin suggested a solution during Friday 8 May solution. We should get in touch with him :)

[camendola]

Ciao Davide, in the skimmer we already have gen information in place for other processes, can take inspiration from those. For example, we already check the gen info for HH events and identify the gen tau pair. In the ttHToNonBB case, we only need to look for the only H in the event and check its daughters. If they're taus, they should be trashed. Of course the xs for the ttHToNonBB also needs to be corrected to take into account that we get rid of HTauTau decays. This is how Konstantin said it's done in Pisa.

Of course one can do a more sophisticated stitching, as we do for the DY, but I think it's overkill.

Anyway it's not a high priority issue, it's here to keep it in mind!

[francescobrivio]

I have a proposed solution (to be tested): https://gist.github.com/francescobrivio/39eb040802500fcd24240bf23ffec0ff

It should be applied in the skimNtuple_201*.cpp, for example here.

[camendola]

Makes sense to me, I'm just not sure that "isFirst" is enough to identify H (actually I would have thought that "isLast" is needed)

[francescobrivio]

Makes sense to me, I'm just not sure that "isFirst" is enough to identify H (actually I would have thought that "isLast" is needed)

Indeed I'm not sure either, I copied the code frome here: https://github.com/LLRCMS/KLUBAnalysis/blob/VBF_legacy/test/skimNtuple2016.cpp#L1772-L1793 where both "isFirst" and "isLast" are checked, but in the end only "isFirst" is used.

But as you suggest, maybe "isLast" is more correct because we need to find the last higgs in the decay chain and check wheter it decays to taus or not. In the end I think this depends on whether the higgs can have copy/clones in the decay chain. I will try to dump the gen particles for some events to check.

[francescobrivio]

Ok, apparently there are more copies of the higgs in the decay chain, so probably "isLast" is the right choice, as @camendola suggests. I will update the gist code

[francescobrivio]

In #122 the solution proposed in this previous comment is implemented. Explanation of the code:

• A loop on gen particles is done
• The gen Higgs is identified
• The decay mode of the gen Higgs is assessed using the genpart_HZDecayMode branch
• If the decay mode is <= 6 the event is discarded. See this enum in LLRHiggsTauTau for an explanation of the Higgs decay modes

Once #122 is merged we can close this issue.

[francescobrivio]

Solved by merge of #122 . Closing this issue