Effect of t42 algorithm on jets Emmanuel Busato LPNHE Paris Noise jets suppression with t42 Effect of t42 on real jets Jet quality cuts efficiencies Energy lost by good jets Reconstruction efficiency in Monte Carlo Resolution in Monte Carlo
1. Noise jets suppression with t42 Data sample : alljets skim from top group = 4JT10 + 4JCCB + HT>100 Color code : t42 no t42 Number of jets/event Big difference
t42 and new seeding go in the same direction remove fake jets Effect of new seeding Jets found in both cases Jets not found with new seeding mean = 0.43 mean = 2.22 Effect of t42 algorithm t42 and new seeding go in the same direction remove fake jets
2. Effect of t42 on real jets Jet quality cuts efficiencies Data sample : stripped thumbnails of jet-photon events from JES : reco version : p13.06.01 triggers : all single electromagnetic triggers in CMT-9.50 At least one EM object with |id|=10,11 and pT > 4.0 GeV Additional cuts : Photon candidate selection : Hmx8<20 Isolation<0.15 EM fraction>0.9 in fiducial no track match (jet-photon)>3.0 Method : use photonjet events, apply back to back requirement to maximize real jets over fakes apply quality cuts on these jets
Jets properties t42 no t42 EMF CHF f90 n90
Efficiencies for each cut individually Total 0.914 0.008 0.929 0.008 (Statistical error only) Higher efficiency with t42 Quality cuts may be reconsidered
Energy lost by good jets the same jet is less energetic with t42 Data (alljets skim from top group) 735 MeV difference for the same jet need new JES correction MC (single top events) : with noise without noise t42 cuts real energy
Reconstruction efficiency in MC (qcd pT20) Method : match particle jets to calorimeter jets with R 0.5 reco efficiency = pT particle jets found in calo / pT all particle jets Efficiency lower with t42 when pTjet 8 GeV. Gain a lot of efficiency by lowering the threshold to 6 GeV What about noise jets with this new threshold ?
Data (alljets skim) New low pT good jets (reco efficiency higher). New jets found with [t42, pTjet 6 GeV] in second bump may also be low pT good jets. Even if they are fake, situation better than in the [not42, pTjet 8 GeV] case.
Energy resolution in MC (qcd pT20) Method : match particle jets to calorimeter jets with R 0.5 E / E = (E particle jets – E cal jets ) / E particle jet distribution in several pT bin Fit with a gaussian in each pT bin.
Result resolution better at low energy better fit after t42
Conclusion t42 algorithm allows to cut out fake jets without degrading real jets reconstruction. We propose to set it in killing mode in the reconstruction in p14.04.00