vhbb / cmssw

CMS Offline Software
cms-sw.github.io/cmssw
4 stars 5 forks source link

Clean iso leptons pt>100 from ak08 jets #584

Closed degrutto closed 7 years ago

degrutto commented 7 years ago

Currently we do not clean AK08 from isolated high pt leptons (as for example we do for AK04 jets), so that for example for boosted WH or ttbar we end up counting as an AK08 also the isolated leptons

for example running on 10K ttbar events you can see tree->Scan("vLeptons_pt[0]:vLeptons_eta[0]:vLeptons_phi[0]:FatjetAK08ungroomed_pt[0]:FatjetAK08ungroomed_eta[0]:FatjetAK08ungroomed_phi[0]:FatjetAK08ungroomed_pt[1]:FatjetAK08ungroomed_eta[1]:FatjetAK08ungroomed_phi[1]", "(Vtype==2 || Vtype==3) && vLeptons_pt[0]>200")


while after this cleaning

tree->Scan("vLeptons_pt[0]:vLeptons_eta[0]:vLeptons_phi[0]:FatjetAK08ungroomed_pt[0]:FatjetAK08ungroomed_eta[0]:FatjetAK08ungroomed_phi[0]:FatjetAK08ungroomed_pt[1]:FatjetAK08ungroomed_eta[1]:FatjetAK08ungroomed_phi[1]", "(Vtype==2 || Vtype==3) && vLeptons_pt[0]>200")


@gkasieczka @cvernier @arizzi @jpata hope you like the code, please also suggest other checks we could do, or for example if we want at all to clean, and also to clean CA15

cvernier commented 7 years ago

@degrutto how much is changing the efficiency for H-signal? probably we should try to compare efficiency for an HH>4b sample.

arizzi commented 7 years ago

need experts answer before merging....

arizzi commented 7 years ago

plus it doesn't merge anymore

cvernier commented 7 years ago

I'd like to see the efficiency for H-signal, in an another sample : HH>4b and see that the signal efficiency there is unchanged. Also lower pt threshold should be tested too.

arizzi commented 7 years ago

@cvernier can't you check how many selected leptons we have in our signal events? aren't we removing those events anyhow to reduce ttbar?

cvernier commented 7 years ago

Indeed for fully hadronic final state it should be the same. I just want to check that the signal efficiency is unchanged as expected.

cvernier commented 7 years ago

I check it's fully efficient for HH>4b.