1. Top Quark Pair Production Cross Section using the ATLAS Detector at the LHC P. Skubic (On behalf of the ATLAS collaboration) April 29, 2014 P. Skubic - OU, ATLAS

2. Outline Introduction: –Top Production and decay Production Cross section – inclusive cross sections 7 TeV 8 TeV – differential cross section at 7 TeV Summary P. Skubic - OU, ATLAS

3. Top quark pair production at LHC ≈ 85/87/90% @ 7/8/14 TeV ≈ 15/13/10% @ 7/8/14 TeV Integrated Luminosity top pair events produced L int =5 fb -1 @ 7 TeV835,000 L int =20 fb -1 @ 8 TeV4,760,000 Top pair σ[pb]@LHC E cm =7 TeVE cm =8 TeV (Scales+PDF) 3P. Skubic - OU, ATLAS The large mass of the top quark results in large coupling to the Higgs and possibly to new physics processes. Theory: Full NNLO+NNLL with contributions from: Baernreuther, Czakon, Mitov arXiv:1204.5201 Czakon, Mitov arXiv:1207.0236 Czakon, Mitov arXiv:1210.6832 Czakon, Fiedler, Mitov arXiv:1303.6254

4. Finding top quark pair events  Decay: weak interaction : t  wb (~ 100 %)  Final state: from the W decays All jets 46% e/μ jets 34% Dilepton (e/μ) 6% τ’s 14%  Gives several handles for identification e/μ/τ from W decays b-jets Missing transverse energy from neutrino Each must be understood with high precision P. Skubic - OU, ATLAS b-tagging performance : characterized by b-tagging efficiency (probability to identify a b-jet as such) and rejection of non-b-jets. A typical working point at 70% efficiency using the MV1 tagger has a rejection factor of about 140.

5. Typical top pair e-µ dilepton candidate with two b-jets Typical event selection Requirements: p Te > 25 GeV p Tµ > 25 GeV E tmiss > 45 GeV |η cl |< 2.47 5P. Skubic - OU, ATLAS

6. 7 TeV dilepton cross section Measurement with/with out b-tagging JHEP 1205 (2012) 059 Simple counting analysis Since comparatively clean signal Stat. Error: ~3% Sys. Error: ~8% (JES, lepton SF, fakes) Lumi. Error: ~5% 6P. Skubic - OU, ATLAS

7. 8 TeV single lepton cross section ATLAS-CONF-2012-149 Lepton(e/μ) + jets 8 TeV Multivariate technique used with b- tagging to separate tt signal from backgrounds Variables used in Likelihood are lepton η and aplanarity transform (exp(-8A)) Dominant systematics include: MC modeling of signal (11%) and Jet/MET reconstruction and calibration (~6%) 7P. Skubic - OU, ATLAS Good agreement with theory.

8. Inclusive dilepton cross section 8 Events with at least two jets Require opposite sign (OS) eµ with exactly 1 or 2 b-tagged jets tt purity: 89% 96% 8 TeV ATLAS-CONF-2013-097 8P. Skubic - OU, ATLAS

9. Inclusive dilepton cross section (con’t) e µ pTpT Bkg: Single top (Wt) (from simul.), Data-driven fake leptons (extrapolated from same sign lepton sample), Z + jets (extrapolated from Z  µµ sample) Good data-MC agreement and signal/background ratio 9P. Skubic - OU, ATLAS η

10. Inclusive dilepton cross section (con’t) Primary systematic errors: Lumi ~ 3.1%, E beam ~ 1.7%, tt modelling ~ 1.5%, Electron ID/isol ~ 1.4% Simultaneous fit for cross section and efficiency to select, reconstruct, and b-tag a jet in 1-b-tag and 2-b-tag samples in order to minimize jet and b-tag syst 10P. Skubic - OU, ATLAS Good agreement with theory.

11. Inclusive cross-section summary 8 TeV summary 7 TeV summary/history Measurements are in good agreement with predictions P. Skubic - OU, ATLAS

12. Top quark pair production vs center of mass energy Good agreement with predictions at several values of E cm P. Skubic - OU, ATLAS

13. Require 1 isolated e, µ; E T miss > 30 GeV, m T W > 35 GeV, ≥ 4 jets, ≥ 1 b-tag Reconstruct with kinematic likelihood fit: (m t, m W constraint) with cut on quality of fit Unfold d(N-N bkg )/dX to full phase space: (regularized unfolding, linearity tests), scale with L and Combine (e,µ) + jets channels with minimal covariance estimator including correlations Propagate syst. Uncertainties through unfolding: Modify migration matrix and acceptances, correct data Top quark pair production – differential cross-section ATLAS-CONF-2013-099 Compare to MC simulations and selected theoretical calculations. 13 P. Skubic - OU, ATLAS

14. Backgrounds: W + jets ( data-driven: normalize pre-tag with W + /W - asymmetry; extrapolate b-tag prob. from 2-jet-bin); fake leptons (data-driven method); Single top, dibosons (from MC) ATLAS-CONF-2013-099 dN/dp T,top (con’t) 14P. Skubic - OU, ATLAS

15. (con’t) Compare with MC, NLO & approx NNLO p T,top spectrum is softer than most predictions for p T,top > 200 GeV ATLAS-CONF-2013-099 15P. Skubic - OU, ATLAS

16. (con’t) Data show sensitivity to PDF with Some preference for HeraPDF Compare data with NLO QCD using FCFM with different PDF sets 16P. Skubic - OU, ATLAS

17. Summary Top production measurements are in precision era - Pair production cross section uncertainty O(5%) level at LHC compared to ~4% prediction uncertainty (NNLO+NNLL) - Differential cross-sections now measured with 10%-20% relative uncertainties Most top physics measurements systematics dominated - Work is on going for full run-1 LHC samples Run 2 @ : new kinematic phase space to be explored with ~ factor 3 enhanced cross section – Higher statistics inclusive, exclusive, and differential cross section measurements – Fiducial measurements New physics decaying into top quark (pairs) not yet seen - Large machinery developed looking into many signatures, reusable in 2015 - 17P. Skubic - OU, ATLAS

18. Backup Slides P. Skubic - OU, ATLAS

19.  Variables chosen (based on the optimization w.r.t. stat+JES error) : lepton η : ttbar more central lepton q : ttbar symmetric, W+jets asymmetric aplanarity : ttbar more isotropic transformed to e -8xA for uniformity; Aplanarity defined: 1.5x smallest eigenvalue of momentum tensor Lepton + Jets: Analysis (No Btag) 19P. Skubic - OU, ATLAS

20. Lepton + Jets: Analysis Strategy 20  Measurement strategy (multivariate) : exploits the difference in kinematic distributions of signal and background events.  Projective Likelihood (LH) is used: to separate signal from bkg (both analysis with and without b-tag)  Discriminant constructed from multiple variables  MC signal and background models these variables for building LH discriminant  Fit the likelihood discriminant distribution in data by sum of two “templates”, signal and bkg, and get the 20P. Skubic - OU, ATLAS

21. Lepton + Jets: Background Estimate  Backgrounds:  W+jets backgrounds  Shape is determined by MC  Normalization from fit  Small Bkgd (Z+jets, diboson, single top)  Shape from MC  Normalization from NLO calculation  QCD multijet (Fake lepton)  Due to mis-ID of lepton, not well modeled in simulation  Used (for example) matrix method for μ channel  anti-electron for e channel 21 P. Skubic - OU, ATLAS

22. QCD MULTIJET : MUON CHANNEL  Dominated by b-jets or c-jets producing muons  Background in signal region can be estimated by using matrix method :  : from data – Z decay  : Control regions (loose the standard criteria)  These are applied to the signal region  Uncertainty : 30 % P.Skubic – OU, ATLAS Isolated muons from W decays QCD muon from jets ε fake ε real Standard muon selection ε real ε fake  Apply b-tagging to get the estimate after b-tagging 22P. Skubic - OU, ATLAS

23. Top pair production cross-section – 35 pb -1 Measurement without use of b-tag [Phys. Lett. B711(2012) 244-263] Multivariate analysis in e/μ + 3,≥4 jets Background: W+ jets, Multijet, WW/WZ/ZZ, single top Lepton charge, lepton η, aplanarity – Stat. Error ~10 % – Syst Error ~11% (JES 5%, bkg modelling ~ 4.0%, IFSR 6%) 23P. Skubic - OU, ATLAS

24. Top pair production cross-section – 35 pb -1 Measurement with use of b-tag [Phys. Lett. B711(2012) 244-263] Multivariate analysis in e/μ + 3,4,≥5 jets Lepton charge, lepton η, aplanarity transform (exp(-8A)), b-tag weight – Stat. Error ~6 % – Syst Error ~9.7% (JES 4%, bkg modeling ~ 4.0%, IFSR 5%) 24P. Skubic - OU, ATLAS

25. Top pair production cross-section – 0.70 fb -1 Measurement without use of b-tag [ATLAS-CONF-2011-121] Multivariate analysis in e/μ + 3,4,≥5 jets Background: W+ jets, Multijet, WW/WZ/ZZ, single top lepton η, aplanarity transform (exp(-8A)), leading jet p T, and H T,3p transform (exp(-4H T,3p ) – Stat. Error ~4 % – Syst Error ~9.0% (JES 4%, signal modeling ~ 5.0%, IFSR 3.0%) 25P. Skubic - OU, ATLAS

26. Top quark pair production –  lepton channels  + jets EPJC 73 3 (2012) 2328 signal n track for τ had candidates after all selection cuts 7 TeV  + lepton Phys. Lett. B 717 (2012) 89-108 26P. Skubic - OU, ATLAS

27. LHC: A Top producer  Run (2010 – 2011)  2x10 32 cm -2 sec -1 (instantaneous lumi)  3.6x10 33 cm -2 sec -1  Run (2012)  7.7x10 33 cm -2 sec -1 0.048 fb -1 @7 TeV 5.6 fb -1 @7 TeV ~23 fb -1 @ 8 TeV Cumulative Lumi Vs time 27P. Skubic - OU, ATLAS

28. THE ATLAS EXPERIMENT Muon Detector Tile Calorimeter LAr Calorimeter Toroid Magnets Vertex & Tracker Trigger system to record online interesting events(collisions every 25/50 ns) 28P. Skubic - OU, ATLAS

29. PIXEL DETECTOR This is high-granularity silicon detector Layout (Oklahoma group was involved in the Pixel detector construction) : 1744 modules located on 3 layers with both barrel and end cap disk geometry 80 million channels Low occupancy (1,000 trk/event at LHC design luminosity) Pixel is close to the intense LHC collision, it is radiation hard, and has an excellent spatial resolution (10 μm * 115 μm ) Because of its fantastic spatial resolution, pixel detector plays a unique role in the identification of b-quark jets or b-tagging. b-quark jet identification plays a central role in many searches of new physics and top quark physics Commissioning : Large testing activity during Integration : Connectivity test : Last chance to repair before Installation inside ATLAS ! 29P. Skubic - OU, ATLAS

30. B-TAGGING PRINCIPLES  b-tagging : identification of b-jets (jets originating from b- quarks) crucial for many physics channels (top quarks, SM and MSSM Higgs, SUSY)  b-tagging algorithms in ATLAS : two main approaches SV based : search for a secondary vertex inside the jet: signed decay length significance : S(Lxy) = Lxy/  (Lxy). IP based : count tracks with large impact parameter significance (IPS): Impact parameter significance : S(IP) = d 0 /  (d 0 ) (complimentary) : look for soft leptons inside the jet JetFitter: takes into account track & vertex info, energy fraction of charged tracks, S(L xy ) in a neural net MV1: combination of all above methods (default) b-tagging performance : characterized by b-tagging efficiency (probability to identify a b-jet as such) and rejection of non-b- jets. A typical working point at 70% efficiency using the MV1 tagger has a rejection factor of about 140. 30P. Skubic - OU, ATLAS