Overlap removal in Hgg MC samples

[mhl0116]

Implement the overlap removal method studied in this presentation originally dedicated to ttH analysis: https://indico.cern.ch/event/754291/contributions/3133922/attachments/1713144/2762621/ttH_gg_overlapRemove_20180910.pdf

There is an old PR #1083 implementing the method in TTHLeptonicTagger, which can be generalized so that this option is available to other tags.

[shervin86]

It makes sense to me to modify the gen match types in: https://github.com/cms-analysis/flashgg/blob/dev_legacy_runII/DataFormats/interface/Photon.h#L29 one might have:

• kFake
• kPromptME
• kPromptHad

with modifications here to assign the flags: https://github.com/cms-analysis/flashgg/blob/4e958a7c7997466be6c95115734527577d9d88d8/MicroAOD/interface/PhotonMCUtils.h

The not nice part of the job is to make sure that in the framework the usage of kPrompt is replaced correctly with (kPromptME or kPromptHad). There should be somewhere a method that given a photon returs isPrompt() as the or of the two flags to be backward compatible with the previous definition of kPrompt.

[shervin86]

Have a look at the header of the GenParticle, there are already several methods that tells if the particle comes from the hardscattering or not, etc...

https://github.com/cms-sw/cmssw/blob/982c595daa9fdde39b7a5925f706b457b178607d/DataFormats/HepMCCandidate/interface/GenParticle.h

[sam-may]

Hi @shervin86 , related to the overlap removal (in terms of implementation), I discussed your other suggestion with @mhl0116 : want to store the mother particle of each photon so that we can study photon origins (does it come from a quark, W, etc.).

Hualin already needs to calculate the mother ID of each photon to perform the overlap removal, so he will also store that information as a property of each photon. Then, we can remake microAOD with these changes and make tables of photon origins for MC samples as you suggest.