1. 1 Hadronic Event Shape Variables in pp collision at 7 TeV Introduction Data-set and Event Selection Comparison of Basic Jet Objects in Data and MC Event Shape Variables in Data and MC Systematics and Sensitivity Final Results Summary PAS : QCD-10-013, CMS AN-2010/100 and CMS AN-2010/127

2. 2 Introduction Variables are defined in terms of four momenta in the transverse plane, in analogy to e + e − collider, Banfi, Salam, Zanderighi, JHEP 0408 (2004) 62 Central Transverse Thrust Central Thrust Minor Experiment : Normalised Event Shape variables are expected to be robust against jet energy scale uncertainties and jet energy resolution effects Theory : Calculations of Event Shape variables are carried out in perturbative QCD Event Shape variables can be used to distinguish different models of QCD multijet production Possibility with large statistics: measurement of α s

3. 3 Central Transverse Thrust Plotted in the natural logarithm, log τ = log(1−T)

4. 4 MC samples /MinBias/Spring10-START3X V26A 357ReReco-v2/GEN-SIM-RECO /MinBias TuneD6T 7TeV-pythia6/Spring10-START3X V26B-v1/GEN-SIM-RECO /MinBias TuneP0 7TeV-pythia6/Spring10-START3X V26B-v1/GEN-SIM-RECO /MinBias 7TeV-pythia8/Spring10-START3X V26B-v1/GEN-SIM-RECO /QCD Ptxx/Summer10-START36 V9 S09-v1/GEN-SIM-RECODEBUG, where range of pˆT from15 GeV/c to kinematic limit (D6T sample, default MC). /QCD Ptyy-herwig/Summer10-START36 V9 S09-v1/GEN-SIM-RECO, where range of pˆT from15 GeV/c to kinematic limit. /QCD Pt-yytozz 7TeV-pythia8/Summer10-START36 V10 S09-v1/GEN-SIM- RECO, where yytozz are the range of pˆT from 15 GeV/c to kinematic limit. /QCDxxJets Ptyytozz-alpgen/Summer10-START36 V9 S09-v1/GEN-SIM-RECO, where xx = 2,3,4 and 5, yytozz are the range of pˆT ranges from 40 GeV/c to kinematics limit. /QCD Ptyytozz-madgraph/Summer10-START36 V9 S09-v1/GEN-SIM-RECO, where yytozz are the range of pˆT from 50 GeV/c to 500 GeV/c. CMSSW_3_6_1 and Spring10 jet corrections

5. 5 Data and MC event samples Data : 1. /MinimumBias/Commissioning10-SD JetMETTau-Jun14thSkim v1/RECO (run# 134630 to 135802) 2. /JetMETTau/Run2010A-Jun14thReReco v2/RECO ( 135821 to 137028) 3. /JetMETTau/Run2010A-PromptReco-v4/RECO (run# 137437-139790) 4. /JetMETTauMonitor/Run2010A-PromptReco-v4/RECO (run# 137437-139790) 5. /JetMETTau/Run2010A-Jul16thReReco-v1/RECO (run# 139779 to 140160) Cert 132440-137028 7TeV June14thReReco Collisions10 JSON v2.txt, Cert 132440-139790 7TeV StreamExpress Collisions10 JSON.txt and Cert 139779-1340159 7TeV July16thReReco Collisions10 JSON.txt Dataset Luminiosty (nb−1) Total HLT Jet30U HLT Jet15U Run2010A-Jun14thReReco v2 4.92 4.92 4.92 SD JetMETTau-Jun14thSkim 6.64 6.64 6.64 Run2010A-PromptReco-v4 65.53 65.53 4.71 Run2010A-Jul16thReReco-v1 124.98 36.60 2.99 Upto Run 139790 77.09 77.10 16.26 Upto Run 140159 202.07 113.70 19.25

6. 6 Event selection Technical trigger : LHC clock (BPTX) and veto on Beam Halo !(36||37||38||39) High Level : HLT_Jet15 / HLT_Jet30 Vertex: at least one primary vertex with |Δz| 4 Scraping event : less than ten tracks or more than 25% of the tracks with HighPurity Jet Cleaning (next slide)

7. 7 Jet algorithms (AntiKt-5) CaloJet PFJet JetTPT Tracker Jet Selection of Jets : (JTF recommendation) Used all four types of jets Jet selection : Loose JetId cuts for CaloJet and JetJPT ( n90hits>1, emEnergyFraction>0.01, fHPD<0.98) PF jet : NeutralEM(Had)EnergyFraction<1.0 – In barrel (|η| 0, chargedEMEnergyFraction 0 For all types : At least two objects in jet Out of all jets with |η|<2.6, two leading jets should be within |η|<1.3 Leading two jets should pass JetID criteria and P T1 > 90 GeV/c / 60GeV/c Event shapes are calculated with all jets within |η| 30 GeV/c

8. 8 Trigger criteria JetMETTauMonitor_Run2010a_May27thRereco No matching of Reco and HLT jet, requires one HLT_Jet15 object In analysis, Pt of leading jet is > 60 GeV/c also analysis with >90 GeV/c Correction for trigger efficiency in Tracker jet Data |η|<2.6

9. 9 Event Shape variables (Prel)

10. 10 Comparison of Basic objects : ΔΦ of leading Jets With |η|<1.3 MC is normalised to total numbers of entries in data Obviously Alpgen and Pythia8 have bias TrackerJet PFJet JetJPT CaloJet

11. 11 Discrepancy in Alpgen sample Mangano : Alpgen should not looks that different CMS collegues : Alpgen and Madgraph should be similar Wrong matching efficiency values on production twiki, which have been corrected now. Thanks to the Generator group (esp. Fabian Stoeckli) for the quick response. sample Old effi New effi 2j_40_120 0.64 0.571 2j_120_280 0.25 0.279 2j_280_500 0.24 0.204 2j_500_5000 0.23 0.192 3j_40_120 0.17 0.299 3j_120_280 0.21 0.207 3j_280_500 0.20 0.144 3j_500_5000 0.17 0.134 Ratio of cross-sections after matching changed for lowest 3j/2j_Pt40-120 – sample from previously 0.18 to now 0.35 due to new evaluated efficiencies Similarly gamma+1-4jet sample also had wrong weight factor (physics- validation/775.html) Nobody has noticed this in last six months !!!!!!!!!!!!

12. 12 Comparison of Basic objects : ΔΦ of leading Jets Discrepancy in Alpgen reduces also close to madGraph, but now main difference in coming from Pythia8 TrackerJet PFJet JetJPT CaloJet

13. 13 Comparison of Basic objects : ΔPt of leading Jets Distribution in tracker jets are different from others. Combination of charge and neutral hadrons are different in different models. TrackerJet PFJet JetJPT CaloJet

14. 14 Comparison of Basic objects : Pt2 sin(ΔΦ)/Pt1 of leading Jets Deviation in Alpgen and Madgraph are opposite to Pythia8. Simplue guess : Alpgen/Madgraph, Thrust value is larger, opposite in Pythia8 TrackerJet PFJet JetJPT CaloJet

15. 15 Comparison of event shape variables Data with only statistical error MC samples – Pythia6 – Pythia8 – Herwig++ – Alpgen – Madgraph D6T and P0 tuning of Pythia6 are consistent with each other Herwig+Jimmy is also consistent with Herwig++

16. 16 Systematic uncertainty and Sensitivity Systematic uncertainty: – Jet Energy and Position Resolution (on MC) – Jet Energy Scale (on Data) – Eta dependent Jet Energy Scale (on Data) Sensitivity : – Jet Types – Jet Algorithms – Underlying event – Initial State Radiation – Final State Radiation

17. 17 Effect of Jet Energy Scale on Event Shape Variables (±5%) : An example Most of the sample is common, more (less) events are accepted with increase (decrease) in JES 3% uncertainty Effect of eta dependent scale uncertainty is negligible ThrustMinor

18. 18 Recommendations: - 10 % uncertainty in the sigma of the jet energy resolution curves - compare the distributions with the default sigma, 1.1×σ and 0.9×σ Jet Energy and position Resolution Uncertainty The jet energy resolution uncertainty leads to deviations within 2-4% over most of the range

19. 19 Comparison of various Jet types in data Thrust Minor Difference in Different types, is it same in MC too ? All are normalised to total numbers of entries in Calo objects

20. 20 Effects of Jet constituents/Algorithm on the comparison of Data and MC Double ratio : (Type1/Type2) Data / (Type1/Type2) MC Consistent with ONE, but need more statistics Only one (maximum) error in Data and MC are considered

21. 21 Effects of Jet Pt/eta criteria Double ratio : (Sel1/Sel2) Data / (Sel1/Sel2) MC Consistent with ONE, but need more statisticsConsistent with one, but need to look with larger statistics In PAS

22. 22 Stability over time

23. 23 Results Yellow Error band contains systematic and statistical uncertainties on data and MC, the black error bar shows the statistical error on Data only Pythia and Herwig++ are close to the data, Alpgen,Madgraph and Pythia8 show large discrepancies

24. 24 Thrust Minor and Y23 Pythia8 differs in Thrust/Minor, but not in Y23, splitting of jets into two looks fine in pythia8. Very sensitive to α S Minor Y23

25. 25 Comparison with two and three jet events With only two jet events, Alpgen/Madgraph and Pythai8 differ from data, which was predicted from simple Pt2 sin(ΔΦ)/Pt1 distribution 2-jet 3-jet

26. 26 Comparison with Pt> 45 and 60 GeV on all jets Same as 30 GeV criteia 45GeV 60GeV

27. 27 Central Thrust and Thrust Minor with 60GeV crit (HLT_Jet15U) Same as 90GeV/c criteria MinorThrust

28. 28 Summary First measurements of central event shape variables from pp collision data at cm energy of 7 TeV Several types of jets are used in the analysis : CaloJet, JetJPT, PFJet, TrackerJet Systematic uncertainty in the measurements studied due to – Jet energy scale (constant as well as η dependent scale) – Jet energy and position resolution for MC predictions Measurements are compared with several QCD inspired models – Reasonable agreement with different tunes PYTHIA6 and HERWIG(++) – ALPGEN, MADGRAPH and PYTHIA8, with default CMS parameter tunes, show significant discrepancies with data – Talked with different Generator people about this discrepancy

29. 29 Central Thrust and Thrust Minor with only two jet events Same as all events MinorThrust

30. 30 Comparison of Basic objects : P T of leading jets Looks fine

31. 31 Comparison of Basic objects : Pesudorapidity A dip in Data around |eta|=1.3

32. 32 Comparison of Basic objects : Azimuthal angle A periodicity is observed, less in tracker jet

33. 33 JetJPT trigger With matching Trigger jets within Δr < 0.15

34. 34 Systematic/Sensitivity But used only JER, JES, JES(η) V01-09-01-09 CondFormats/JetMETObjects V00-02-13 CommonTools/RecoAlgos V03-28-04 DataFormats/JetReco V04-03-05 RecoJets/JetProducers

35. 35 Effect of inter calibration of calo towers (4%) Most of the sample is common Average shift is ~2% Thrust Minor

36. 36 Effect of Jet cleaning criteria, loose vs tight Calo : JPT : + fHHP < 0.95 and EMfrac<0.9 PF : + EM(HAD)energy fraction < 0.9 Rejection : 6.0/3.6/0.9 for CaloJet, JetJPT, PFJet Thrust Minor Samples differ by 3% only, effect <1% effect

37. 37 Comparison of various Jet Algorithms Observe a variation, which is expected due to different algorithms Thrust Minor

38. 38 Trigger criteria of Tracker Jet