130 likes | 242 Views
Missing ET H1-style versus Local- hadronic style calibration. Work on comparisons between both calibration schemes Event-by-event correlated MET_Final (i.e. H1-style calibrated MET MET_CorrTopo + MET_Cryo + MET_MuonBoy ) with MET_LocHadTopoObj + MET_MuonBoy (unless otherwise stated)
E N D
Missing ET H1-style versus Local-hadronic style calibration • Work on comparisons between both calibration schemes • Event-by-event correlated MET_Final (i.e. H1-style calibrated MET MET_CorrTopo + MET_Cryo + MET_MuonBoy) with MET_LocHadTopoObj + MET_MuonBoy (unless otherwise stated) • Did this for the rel 15.1.0 (dijet and ttbar) samples produced for the hadronic calibration workshop (mc08.1050xx.Jx_pythia_jetjet.recon.ESD.e344_s479_r635) D.Berge
J5 correlations All events Require true MET>50 GeV Require true MET>100 GeV 487 events per 10pb-1 beyond 50 GeV difference LC tends to be larger LC tends to be larger True MET >100 GeV, LC TopoObj versus H1 CorrTopo+Cryo Event-by-event difference, no true MET cut SumET correlation all events D.Berge
J6 correlations All events Require true MET>50 GeV Require true MET>100 GeV 82 events per 10pb-1 beyond 50 GeV difference LC tends to be larger LC tends to be larger True MET >100 GeV, LC TopoObj versus H1 CorrTopo+Cryo Event-by-event difference, no true MET cut SumET correlation all events D.Berge
J7 correlations All events Require true MET>50 GeV Require true MET>100 GeV 2 events (20%!) per 10pb-1 beyond 50 GeV difference True MET >100 GeV, LC TopoObj versus H1 CorrTopo+Cryo Event-by-event difference, no true MET cut SumET correlation all events D.Berge
Ttbar semi-leptonic correlations All events Require true MET>50 GeV Require true MET>100 GeV 11 events per 10pb-1 beyond 50 GeV difference True MET >100 GeV, LC TopoObj versus H1 CorrTopo+Cryo Event-by-event difference, no true MET cut SumET correlation all events D.Berge
Tails – local hadronic versus H1 missing Et J5 All events, MET resolution • Asymmetric tail towards positive (rec-true) more pronounced for LC • Beyond 100 GeVDMET, 60% more events in LC scheme • 852 events for LC versus 527 for H1 • For 10pb-1: 231 events for LC, 143 events for H1 D.Berge
Tails – local hadronic versus H1 missing Et J6 All events, MET resolution • Asymmetric tail towards positive (rec-true) more pronounced for LC • Beyond 100 GeVDMET, 60% more events in LC scheme • 12816 events for LC versus 8077 for H1 • For 10pb-1: 49 events for LC, 31 events for H1 D.Berge
Tails – local hadronic versus H1 missing Et J7 All events, MET resolution • Asymmetric tail towards positive (rec-true) more pronounced for LC • Beyond 100 GeVDMET, 18% more events in LC scheme • 57919 events for LC versus 49000 for H1 • For 10pb-1: 1.9 events for LC, 1.6 events for H1 D.Berge
Tails versus Jet eta – local hadronic versus H1 missing Et J5 • Statistics not great • +ve / -ve asymmetry apparent, found to be correlated with beamspotz position Rebinned, no log scale D.Berge
Tails versus Jet eta – local hadronic versus H1 missing Et J6 • Statement for J6 in Portugal was that LC behaves better in the cracks and at eta=0 • This was a mistake, the +ve / -ve asymmetry and a scaling mistake fooled me • Instead the barrel-to-extended-barrel transition around fabs(eta)=1 a bit more problematic • Also here clearly visible that the overall number of events in tails is increased in LC Wiggles at Deta=0.1 <=> Tile granularity Rebinned, no log scale D.Berge
Tails versus Jet eta – local hadronic versus H1 missing Et J7 Wiggles at Deta=0.1 <=> Tile granularity • Here indeed LC better at central eta=0, worse at fabs(eta) = 1.5 • Note how the periodic feature due to a slight jet reconstruction bias towards single Tiles remains after cutting on MET Rebinned, no log scale D.Berge
Summary • The correlation of H1 and local hadronic calibration for the very same event (!) has in the best case quite some spread, in the worst case (J7) there isn’t very much correlation! • J7 is notably bad, at least when looking at all events (requiring true MET in the event improves the situation) • For all samples shown here, there are events where one calibration scheme sees a lot of MET (say >100 GeV), the other much less (say <50 GeV) • This is really the application of calibration weights (compared MET_LocHadTopoObj to MET_CorrTopo(+MET_Cryo) to check) • SumET is not as problematic, both schemes see a correlated amount transverse energy in the event • The local hadronic calibration is consistently larger, both in MET and SumET • SumET correlation plot for ttbar is quite asymmetric, to be understood… • Need to check whether certain topologies are problematic, or in general where large differences come form • As for the tails in general, LC reconstructs larger MET values, correspondingly has larger tails • And also slightly worse resolution in general, see backup D.Berge
Backup – H1 versus LC resolution • Consistent with previous releases • Departure at high SumEt due to constant term in jet resolution • LC tends to be a bit worth over the full range • But both calibration schemes have very similar performance D.Berge