1. Measuring the t-tbar Cross-Section in the Dilepton Channel at CDF* J. Incandela for C. Mills Jan. 17, 2008 DOE Site Visit UC Santa Barbara * PhD Thesis. PRD in preparation

2. Top Dilepton Selection 3) Missing transverse energy (“MET”) 1) One well-identified high-E T lepton  electron or muon  Triggers the event in the data 4) Two or more jets 2 Note that we do not require a b tag (this was done later by T. Speltzer (Toronto) 2) One “track lepton” = high-p T isolated track ®doesn’t distinguish e/  /  ®Increased acceptance CDF top cross section measurement

3. Main Backgrounds NB: tt lepton+jets with one jet faking a lepton is also an important background q    / Z Makes 2 Leptons Has MET Drell-Yan Large cross-section Needs mis-measured ET (false MET), extra jets q W/Z q q / q’ g W q l+l+ l-l- W/Z q q q Diboson Small cross-section Real MET Needs extra jets l W+jets Large cross-section Real MET + 1 lepton Needs extra jets, one misidentified as 2 nd lepton 3 CDF top cross section measurement

4. Skiing to the Cross-Section N obs Number of candidate events in the data  L dt Integrated luminosity A Acceptance = Number selected / Number generated B Background: Estimated number of events from sources other than top 4 CDF top cross section measurement

5. MET Selection MET based on calorimetry –Drastically reduces backgrounds which do not contain neutrinos Corrections performed for: –Isolated muon p T –E T significantly less than isolated track p T –Jet energy corrections Threshold 25 GeV 5 CDF top cross section measurement

6. Selecting Against Background 1) Z Veto: Require MET > 40 GeV if dilepton invariant mass 76 < M < 106 ~96% of top events pass ~50% of Drell-Yan ee/  pass 2) Opposite-sign lepton ~100% efficient for top (and most backgrounds) ~ 60% efficient for W+jets with a fake lepton 6 CDF top cross section measurement

7. Selecting Against Background 3) Delta phi cuts* –Z  ee/  : MET points at mis-measured object –Require that no object points directly at MET ~86% efficient for top ~30%for Z  ee/  tight lepton track lepton jets* 5o5o 5o5o 25 o *exempt event if MET > 50 GeV 7 CDF top cross section measurement

8. Acceptance by the Numbers Acceptance x Branching Ratio = 0.87±0.06% –Get 0.35% using full lepton selection on both leptons Systematics –Lepton Identification (1.6%) Identifying leptons in the messy top event environment. –Parton Distribution Functions (0.5%) –Jet Energy Scale (1.3%) Affects ≥2 jets requirement –Initial/Final State Radiation (1.7%) Affects jet counting –Monte Carlo Generator (1.5%) Total: 3.1 % CDF top cross section measurement 8

9. Z  & Diboson Backgrounds Can’t isolate in data so use MC simulation: –cross-sections well known Measure acceptance MC, as for signal –Jet multiplicity taken “pure” Z + jets data sample Invert cross-section formula: N =   A   L dt Systematic uncertainties: –Lepton ID, jet energies and production q    / Z ++ -- e,     /K q g g 9 CDF top cross section measurement q W W l+l+ l-l- q W Z q q’ g l l’l’ ’ g l g q Z Z q’ g l l g g

10. Z/  *  ee/  : The Basic Problems Z/  *  ee/  events pass selection if: 1.One (or more) objects in the event is mis-measured enough to generate large MET 2.And there are additional jets But we do not trust the MC to do either 1. or 2. correctly. How do we correct it? –Additional jets from a “pure” Z + jets data sample –For MET, what do we do? 10 CDF top cross section measurement

11. Z/  *  ee/  : Established Method Data with missing E T but close to Z mass (any number of jets) 1.Select dilepton sample in the data: Count events with invariant mass inside the Z resonance with large MET Subtract (small) contributions from other sources 2.Select dilepton sample in MC simulation: Find fractions f i with i jets (correct these to match data) and ratios R i = Number of events with i jets outside Z peak with MET > 25 over number with i jets inside Z peak Use ratios R i from 2. to multiply times data event counts from 1. to obtain estimated DY in data with i jets outside Z peak 11 CDF top cross section measurement Total Systematic 25% Sample statistics (20%) MC modeling of MET (13.5%)

12. Background from Fake Leptons W+jets with a fake lepton  Want to know: probability of jets associated with a W to be identified as the second lepton – i.e. an isolated track This is a true isolated track, but of hadronic, not leptonic, origin Basic question: parton (q or g) fragmentation/ hadronization N charged tracks in a jet - or - ? q q / q’ g W q q l 12 CDF top cross section measurement

13. Signs of Trouble Fake rates usually derived from jet triggers: –CDF: “jetXX” = trigger with at least one jet with E T > XX GeV Fake rates derived from different trigger samples don’t match! 13 CDF top cross section measurement

14. q and g jets MC: Quark jets more likely to fake a lepton: –Multijets dominated by gluon jets –Leading jet in W/Z/  +jets tends to be quark jet 14 CDF top cross section measurement

15. q vs. g jets: Evidence from Data 15 CDF top cross section measurement

16. A Smarter Fake Rate from  +jets Q. What makes W+jets and  +jets different? A. W is massive Require  with E= 80 GeV q q / q’ g  / W 16 CDF top cross section measurement

17. Test in Data 80 GeV  +jets sample predicts Z+jets fakes 1 jet events: Predict 100 ± 13 observe 101 2 jet events: Predict 28 ± 2 observe 26 ≥3 jet events: Predict 12 ± 1 observe 13 ≥2 jet events: Predict 40 ± 2 observe 39 17 CDF top cross section measurement

18. Opposite-charge requirement, but fake rate has no sign info Fake leptons a fluke of fragmentation –Might think 50/50 chance for opposite sign But, recall the leading diagram –Quark has the opposite charge! W+jets opposite-sign fraction –Measured in fakes-enhanced subset of zero jet data –Dependence on number of jets from simulation 1 jet: 67%2 jet: 63%≥3 jet: 59% Opposite-Sign Fakes q(+2/3) q’(-1/3) g W+W+ 18 CDF top cross section measurement

19. Fakes from Top Top “Lepton + jets” channel: Fake rate approx. same as for W+jets –Find 23% of fake lepton events are from top Estimate 67% are opposite-sign Total systematic on all fakes ~20% –(dominated by largest discrepancy seen in MC tests – 1 jet bin case) CDF top cross section measurement 19 1) One legitimate  for MET 2) Typically four jets 3) Have ud, or cs quarks from W decay  high fake rate

20. Signal, Background, and Data Compare predicted and observed event counts in 1 fb -1 (uncertainties include statistical and systematic components): Cross-section at M t = 175 with opposite-sign events with ≥ 2 jets: Compare 6.7 -0.9 +0.7 (theoretical prediction for M t = 175 GeV) CDF Run II Preliminary 1.1 fb-1 0 jets1 jet≥ 2 jets Diboson104 ± 921.2 ± 1.85.7 ± 0.5 Z/  *  ee 72 ± 1625.9 ± 6.17.8 ± 2.2 Z/  *  19 ± 58.9 ± 2.73.4 ± 1.2 Z/  *  36 ± 326.5 ± 2.57.3 ± 0.9 fakes244 ± 4676.8 ± 14.629.9 ± 5.9 tot. background475 ± 52159.2 ± 16.954.0 ± 6.6 top (  = 6.7 pb) 1 ± 017.3 ± 0.660.5 ± 1.9 total predicted476 ± 52176.5 ± 17.0114.5 ± 7.0 observed443187129 20 CDF top cross section measurement

21. Summary Great acceptance e, ,  –Few detector requirements Measured cross-section in 1.01 fb -1 of CDF data –  = 8.3 ± 1.3 (stat.) ± 0.7 (sys.) ± 0.5 (lumi.) pb –Cross-checks all ok CDF top cross section measurement 21

22. Additional Material 22 CDF top cross section measurement

23. Cross-Section vs. Top Mass Remeasure acceptance –Higher mass  more energetic decay products Uncertainties are statistical ± systematic Implied mass between 180-185 GeV/c 2 CDF Run II Preliminary 1fb-1 mass (GeV/c 2 ) theory  (pb) acceptance measured  (pb) 1707.8 -1.0 +0.9.69 %10.2 ± 1.5 ± 0.9 1756.7 -0.9 +0.7.85 %8.3 ± 1.3 ± 0.7 1805.8 -0.8 + 0.6 1.01%6.9 ± 1.1 ± 0.6 23 CDF top cross section measurement

24. A Bit of History Run II –Goal: high-acceptance analyses –19 candidates in 200 pb -1, measured 7.0 -2.4 +2.9 pb (6.7 -0.9 +0.7 pb predicted for  s = 1.96 TeV) Run I  9 dilepton candidates in 109 pb -1  Measured 8.2 -3.4 +4.4 pb (5.2 -0.7 +0.4 pb predicted for  s = 1.8 TeV)  High, but with huge uncertainty, and odd kinematic features  Speculation: new physics? 24 CDF top cross section measurement

25. “Track Leptons” 1.Acceptance for  leptons  85% of decay modes have 1 charged track  No distinction between lepton flavors (e, , or  ) 2.Don’t require fiducial regions of calorimetry, muon detectors 3.Rejection of jets using isolation –  p T of tracks in cone small relative to candidate p T tracking EM cal hadronic cal muon electron tau (hadronic) muon jet 25 CDF top cross section measurement

26. Z/  *  ee/  : Established Method Monte Carlo: Use to get fraction of DY outside Z peak 26 CDF top cross section measurement

27. Z/  *  ee/  Systematic Statistical uncertainties ~20% Modeling of extra jets –Data/MC scale factor uncertainty 5.5% How well does the Monte Carlo model events with false MET? Look at Z events in data, MC –Require lepton or MET pointing at jet What fraction of events are above MET cut (25 GeV) in data? (1.7%) Where would you have to place cut in MC to have the same fraction above the cutoff? (24 GeV) Remeasure Monte Carlo ratios R i –Rederive Drell-Yan estimate ≥2 jet bin: number drops by 13.5% Total 25% CDF top cross section measurement 27 all events MET points at a jet 25 WW, ttbar etc

28. Fakes: Established Method From jet trigger data Fake rate definition: Number of isolated tracks Total number of jets Function of jet E T,  28 CDF top cross section measurement

29. Fakes: Established Method How many W (real lepton) + fake lepton +  2 jet events? From jet trigger data 29 CDF top cross section measurement

30. Questioning the Assumptions Use Monte Carlo to check the basic assumption: –Do jets in pure-QCD events have the same fake rates as jets in W+jets events? Answer: no. 30 CDF top cross section measurement

31. Why different fractions? Parton Distribution Functions (PDFs) –What’s carrying the momentum in the proton? 1960 GeV available to make particles at the Tevatron –Multijet,  +jets, W+jets at relatively low x (~100 GeV) –Incoming gluons preferred x = momentum fraction f p (x) = probability to find parton p with x gluons up down strange x 31 CDF top cross section measurement

32. Why different fake rates? FR (quark jets) >> FR (gluon jets) Quarks and gluons have to make colorless objects Cartoon of fragmentation: quark gluon hadrons anti-red 32 CDF top cross section measurement

33. Why different fractions? Gluons dominate at relatively low x in parton distribution functions (PDFs) q q / q’ g  / W g g g W/Z or prompt photon production: Multijet production: Leading 2  2 diagrams (highly schematic) g 33 CDF top cross section measurement

34. Test in Simulated Data Identical procedure to data: 80 GeV  +jets predicts W+jets fakes 1 jet events: Predict 5470 ± 150 observe 4480 2 jet events: Predict 1332 ± 200 observe 1050 ≥3 jet events: Predict 304 ± 67 observe 226 ≥2 jet events: Predict 1640 ± 210 observe 1270 34 CDF top cross section measurement

35. Fake Leptons from Top What is the fake rate? –Approx. same as for W+jets Estimate 67% are opposite-sign? What fraction of our “W+jets” are from top? –Predict from simulation Start with top xsec 6.7pb (175 GeV), measure cross-section, iterate until changes are small –Find 23% of fake lepton events are from top CDF top cross section measurement 35 predict 518 ± 45 observe 424

36. W + fake + jets Systematics Statistical: “error bars” on fake rate (6%) Systematic on absolute fake rate (18%) Use the largest discrepancy in MC tests i.e. for 1 jet bin Opposite-sign (5%) –Statistical uncertainty 36 CDF top cross section measurement

37. Check: “tight-tight” Require track lepton also matched to tight lepton Candidate sample is subset of candidates in default analysis Background estimates use same methods 0 jets1 jet≥ 2 jets diboson45.9 ± 2.89.4 ± 0.62.6 ± 0.2 Drell-Yan41.4 ± 9.621.4 ± 4.26.9 ± 1.6 fakes15.8 ± 3.34.8 ± 1.01.6 ± 0.3 background103.1 ± 10.535.6 ± 4.311.2 ± 1.7 top (  = 6.7 pb) 0.5 ± 0.07.2 ± 0.226.2 ± 0.7 total predicted103.6 ± 10.542.7 ± 4.337.4 ± 1.9 observed794535 37 CDF top cross section measurement

38. Check: 15/15, 25/25 Move track lepton p T and jet E T threshold both to 15 or 25 GeV Scale factors and background estimation techniques unchanged 15,15 20,2025,25 background109 ± 1354 ± 732 ± 4 top (  = 6.7 pb) 74 ± 261 ± 249 ± 2 total predicted183 ± 14115 ± 780 ± 4 observed18912987 38 CDF top cross section measurement