HLT_DoubleMediumDeepTauPFTauHPS35_eta2p1

7.HLT_DoubleMediumDeepTauPFTauHPS35_eta2p1

L1 seed: pDoublePuppiTau52_52

• pDoublePuppiTau52_52 : select two Taus (CL2Taus) with $p_T>52$ GeV - Table

HLT filters:

• hltPreDoublePFTauHPS:
  prescaler with offset = cms.uint32(0)

  • NOTE: Offset is set to 0.

• hltHpsSelectedPFTausTrackFinding:
  cut = cms.string('pt > 0'), discriminator = cms.InputTag("hltHpsPFTauTrackFindingDiscriminator"), selectionCut = cms.double(0.5), src = cms.InputTag("hltHpsPFTauProducer")

Looks for a track matching the Tau within $\Delta R = 0.5$

• hltHpsPFTauTrack:
  select single HPS Tau (hltHpsPFTauProducer) with no $p_T$ cut, with $|\eta|<2.5$ and triggerType=84.

  • NOTE: The minimum eta value is correct: when it is negative, the condition is not applied.

  • NOTE: Are we selecting only one tau? The path indicates two taus.

• hltHpsSelectedPFTausMediumDitauWPDeepTau:
  cut = cms.string('pt > 35 && abs(eta) < 2.1'), discriminator = cms.InputTag("hltHpsPFTauDeepTauProducer","VSjet"), workingPoints = cms.vstring('double t1 = 0.649, t2 = 0.441, t3 = 0.05, x1 = 35, x2 = 100, x3 = 300; if (pt <= x1) return t1; if (pt >= x3) return t3; if (pt < x2) return (t2 - t1) / (x2 - x1) * (pt - x1) + t1; return (t3 - t2) / (x3 - x2) * (pt - x2) + t2;'), src = cms.InputTag("hltHpsPFTauProducer")

$p_T$ value	DeepTauVSjet Medium WP
$p_T\leq35$ GeV	0.649
$35<p_T<100$ GeV	0.649 + (0.441 - 0.649) / (100 - 35) * ($p_T$ - 35)
$100p_T<300$ GeV	0.441 + (0.05 - 0.441) / (300 - 100) * ($p_T$ - 100)
$p_T\geq300$ GeV	0.05

• NOTE: Why is the Medium WP at a lower level than the Loose WP for $p_T$ around 100 GeV ?

• hltHpsDoublePFTau35MediumDitauWPDeepTau:
  select two HPS Taus (hltHpsSelectedPFTausMediumDitauWPDeepTau) with $p_T>35$ GeV, with $|\eta|<2.1$ and triggerType=84

  • NOTE: The maximum eta value goes from 2.5 to 2.1, since the tau reconstruction in Phase 1 is based on tracking information. This should be changed for Phase 2 to extend the taus up to the HGCAL coverage.