Fix gen jet variables, add new jet variables #138
Conversation
riga
changed the title
|
@riga : I assume this PR has been tested with the whole 2017 production right? ;) If there is nothing else you would like to add i propose we merge it. Do you agree @portalesHEP ? |
|
I was thinking of waiting for some preliminary results to be sure we'll use the new variables, but it's fine for me either way |
Yes, we ran the whole 2017 bigNtuple production with it, but we haven't looked into their distributions yet. We we're planning to do this with the small ntuples asap, and we bring them into one of the next meetings. |
|
Following up on this morning's discussions, since we'll very likely want to study ParticleNet for future analyses as well I guess it does not hurt to have the variables included already even if we end up not using it in the current analysis (although I guess the variables and framework may change in the meanwhile). So if nothing else should be added, I'll merge |
|
I agree with you Louis, if these variables are ready to be used I propose to add them! |
| _bdiscr15.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probbb")); | ||
| _bdiscr16.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probpu")); | ||
| _bdiscr17.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probcc")); | ||
| _bdiscr18.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probundef")); | ||
| _bdiscr19.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probc")); | ||
| _bdiscr20.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probb")); | ||
| _bdiscr21.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probuds")); | ||
| _bdiscr22.push_back(ijet->bDiscriminator("pfParticleNetAK4JetTags:probg")); | ||
|
|
There was a problem hiding this comment.
Just double checking: why are these called AK4 when they are used for AK8 jets?
There was a problem hiding this comment.
I learned in the talk by Raffa at our MIB meeting that ParticleNet can tag also resolved jets ;)
There was a problem hiding this comment.
Ok cool" Thanks Davide! My question was more about the "correctness" of the names of the taggers?
There was a problem hiding this comment.
I assume these are the taggers for AK4 jets, the one for AK8 have AK8 in the name...
|
One question that came to my mind: for the AK8 jets (maybe there is something similar also for AK4) is the regressed mass also available? I know it has been used in a couple of boosted HH analysis. |
|
|
I have the same recollection, I was just proposing to check if something similar was added to miniAOD |
|
This has been a while ... 😇 I just checked again and it's does not seem to be added on mini-level. |
|
Indeed this was left hanging. Can it (/should it) be merged now or are there further tests being considered? |
|
I re-checked the event content shipped with the miniAOD files and I think the code in PR should cover everything that we might need, so from our side the PR could be merged 👍 |
This PR fixes the lookup of the gen jet index per
pat::Jet, and adds a jet variables which seem helpful.Fixed indices:
New variables:
bParticleNetAK4JetTags_probbbbParticleNetAK4JetTags_probpubParticleNetAK4JetTags_probccbParticleNetAK4JetTags_probundefbParticleNetAK4JetTags_probcbParticleNetAK4JetTags_probbbParticleNetAK4JetTags_probudsbParticleNetAK4JetTags_probgak8jets_massIndependentDeepDoubleBvLJetTags_probHbbak8jets_deepDoubleBvLJetTags_probHbbak8jets_deepBoostedJetTags_probHbbak8jets_particleNetJetTags_probHbbak8jets_particleNetDiscriminatorsJetTags_HbbvsQCDEdit: Sorry, I saw only now that my editor converted places that had mixed spaces/tabs into just spaces. I can revert that if needed.