Skip to content

Design a HHtobbyy workflow

Below is the list of HHtobbgg workflow features, added for v1 :

  • Use of ParticleNet AK4 tagger for b-jet energy and tagging. This ParticleNet simultaneously performs jet flavour tagging, tau tagging as well as flavour-aware jet energy regression. Have a look at these slides.

    • Use of Pnet instead of DeepJet for b-tagging

    • there are also ParticleTransformer based tagger scores saved in the NanoAOD. After asking the BTV experts regarding which one to use : should be "Jet_btagRobustParTAK4B"

  • Save first N (maybe 6) jets with preselection , otherwise pad with zeros:

    • Four vector, ParticleNet scores, Deep Jet scores

  • For ttH killer:

    • Save 2 leading electrons and muons each
    • PuppiMET info

  • Addition of truth information for the conditional classifiers: mX for X->HH, mY for Y->bb or Y->gg, kt, kl, CV, C2V, etc.

    • Add kt, kl, c2,CV, C2V, C3, BulkGraviton_M (mX), Radion_M (mX)
    • Add mY info when the sample request is submitted

  • Addition of fatjets, i.e. at least AK8 or even AK15, along with all the relevant b-tagging info

    • Preselection: pT>250 GeV, dR>0.8 from photons
    • Order by PNet scores
    • Save first 4 fatjets
    • Add SubJets and GenJetAK8 matched to FatJets

Below are the recent developments , planned or already merged :

  • ttH killer : Save other higher-level variables as required

  • VBF-jets selection implementation

  • Addition of other gen variables to optimize the reconstruction (ME level)

    • gen_mHH (for nonres reco test)
    • 4 momenta of the two generated Higgs
    • 4 momenta of the vector bosons (if it exists)
    • 4 momenta of gen_top (if it exists)
    • gen-matching forJets, Fatjets, Higgs, Leptons

  • ParticleNet btag count: nBTight, nBMedium, nBLoose

  • Selection of one dijet OR one fat jet

  • b-jet energy: ParticleNet correction factors to be applied to the raw jet pT are already available in in NanoAOD v12 onwards

  • b-tagging: ParticleNet tagger info is also available in NanoAOD v12 onwards. Some info about the scale factors are available here --> To do : add b-tagging SF corrections and systematics MR

  • Currently, a cut is applied on the invariant mass of the dijet system (70 GeV < dijet_mass < 190 GeV). The workflow has to be adapted for resonant analysis (X->YH->bbgg), where the mass of Y can be beyond the range [70, 190] GeV : We will implement a tag to differentiate between resonant and non-resonant

  • Cut Flow implementation MR

  • n_electron, n_muons

  • Flag to know which jets were chosen for the dijet pair

  • Pre-selection changes see this presentation

Suggestions :

  • Maybe useful to explicitly name the variables of diphoton system as "diphoton_{variable}". But this might get inconsistent with the other workflows which has just {variable}. Have to decide to change it or not.

  • Increase the number of jets registered

References:

Edited by Elise Anne Marie Jourd'Huy