1. Top quark physics at CMS  From Tevatron to LHC  Reconstruction of the top quark events  Precision measurements : m t,  tT,  t, W helicity, spin,...  Discovery potential Standard Model : single-top, couplings,... Beyond Standard Model : resonances, FCNC,...  Calibration sample : jet energy scale, b-tag efficiency,... Jorgen D’Hondt (Vrije Universiteit Brussel) Roberto Tenchini (Pisa) on behalf of the CMS Collaboration CTEQ workshop, Michigan (US), 14-15 May 2007

2. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)2 From Tevatron/LEP to LHC Obtaining orders in magnitude in both the integrated luminosity and the energy, we will collect a huge amount of Standard Model benchmarks channels. ~10 9 events/10fb -1 W (200 per second) ~10 8 events/10fb -1 Z (50 per second) ~10 7 events/10fb -1 tt (1 per second) These can be used as control/calibration samples for searches beyond the Standard Model, but can also be used to scrutinize even further the Standard Model. pile-up becomes an important issue (10 fb -1 = 1 year of LHC running at low luminosity 10 33, hence by end 2009) rates@10 33

3. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)3 Top Quark Production at the LHC ~87 %  NLO cross-section  NLO = 833 pb  ~8M events/10fb -1 Some references (not a complete list!): (top pairs) N.Nason et al. Nucl.Phys. B303 (1988) 607, S.Catani et al. Nucl.Phys. B478 (1996) 273, M.Beneke et al. hep-ph/0003033, N.Kidonakis and R.Vogt, Phys.Rev. D68 (2003) 114014, W.Bernreuther et al. Nucl.Phys. B690 (2004) 81-137 (single-top) T.Stelzer et al. Phys.Rev. D56 (1997) 5919, M.C.Smith and S.Willenbrock Phys.Rev. D54 (1996) 6696, T.M.Tait Phys.Rev. D61 (2000) 034001 10 tt pairs per day @ Tevatron  1 tt pair per second @ LHC qq →tt : 85% gg→tt : 87% ? ? single-top @ Tevatron  30 single-tops per minute @ LHC p p t t p p t X  NLO = 6.6 pb  NLO = 153 pb  NLO = 60 pb  top &  anti-top not equal  NLO = 4.1 pb  NLO = 90 pb  NLO = 60 pb  top production  anti-top production s-channel t-channel associated tW  NLO (total) = 373 pb  ~3.7M events/10fb -1 top pairs single-top

4. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)4 Topics in top quark physics at the LHC TOP  ≈ 0.4 10 -24 s BOTTOM W L PRODUCTION Cross section Spin-correlations Resonances X  tt Fourth generation t’ New physics (SUSY) Flavour physics (FCNC) PROPERTIES Mass (matter vs. anti-matter) Charge Life-time and width Spin DECAY Charged Higgs W helicity Anomalous couplings CKM matrix elements Calibration sample !! kinematic fit (m W ) jets b-tagging trigger missing energy This data will extend the Tevatron precision reach and allow new possible topics.

5. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)5 Current status of Simulation and Reconstruction Still some time before real data… hence all based on Monte Carlo simulation  Main generators used : PYTHIA 6.2 or AlpGen (Leading-Order) ð does not include many features present in dedicated generators ð Next-to-Leading-Order applied where needed ( eg. single-top production )  Simulation: studies based on both fast and full simulation ð ‘fast’ simulation: generated objects smeared to mimic the detector ð ‘full’ simulation: detector responds simulated with GEANT-4 ð large Data-Challenge efforts have been made to provide dedicated GEANT-4 simulation (created ~100M simulated events, ~1Mb/event)  pile-up collisions added in low-luminosity settings ( 2.10 33 cm -2 s -1 )  Reconstruction based on documented advanced tools  Results illustrate what we can learn from CMS data upon arrival  References: talk based on notes/information available on CMS  http://cms.cern.ch/iCMS (‘full’ simulation results from P-TDR) http://cms.cern.ch/iCMS

6. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)6 Top Quark Pair Selection  Fully hadronic channel : 3.7M evnt /10fb -1 QCD multijet (2→2 parton processes) → apply also topological cuts at least 6-jets p T >30-40 GeV, b-tags≥2 : S/B~1/9,  ~2.7%  Lepton + jets channel (lepton = e/  ) : 2.5M evnt /10fb -1 W+jets→lv+jets, Z+jets→ll+jets, WW→lv+jets, WZ→lv+jets, ZZ→ll+jets before selection we have S/B ~ 10 -5 p T lepton >20 GeV, kinematic fit, p T jet >30 GeV, b-tags≥2 : S/B ~ 27,  ~6.3%  Di-lepton channel : 0.4M evnt /10fb -1 Drell-Yan processes, Z+jets→ll+jets, WW+jets, bb → Z-mass cuts p T >20 GeV, E T miss >40 GeV, p T jets >20 GeV, b-tags=2 : S/B~5.5,  ~5%  main backgrounds in lepton+jet and di-lepton channel from ttbar other decays

7. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)7 Top Quark Pair Production Cross-Section  Shape analysis (identical á la D0 in Phys.Lett. B626 (2005) 45 ) ð combine the topological/kinematic information of several event observables ð not as powerful as at the Tevatron ( CMS Note 2006/064 & ATLAS Eur.Phys.J.C39(2005)63 )  Simple counting experiment  possible due to high S/N after selection  di-lepton : (e/  )0.9 (stat) ± 11 (syst) ± 3 (lumi) % (10fb -1 ) (  )1.3 (stat) ± 16 (syst) ± 3 (lumi) % (10fb -1 )  lepton+jet : (e/  )0.4 (stat) ± 9.2 (syst) ± 3 (lumi) % (10fb -1 )  fully hadronic :3 (stat) ± 20 (syst) ± 3 (lumi) % (10fb -1 ) AlpGen (W+jets) PYTHIA (tt) CMS Note 2006/064 normalized Phys.Lett B626 (2005) 45

8. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)8 Top Quark Pair Production Cross-Section  Breakdown of total uncertainty ( CMS Note 2006/064 ) ð example in the semi-leptonic muon channel ð dominated by uncertainty on b-tagging efficiency which is conservative when assuming 2% (rather than 5%)  total uncertainty ~ 7% (10fb -1 ) ð 2-3 GeV uncertainty on m t is feasible via cross section measurement conservative → Tevatron ~2% conservative

9. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)9 Properties: Top Quark Mass TOP  ≈ 0.4 10 -24 s BOTTOM W L Most important parameter is the top quark mass (m t ), to be estimated with an accuracy of around  m t ~ 1 GeV/c 2. ð fully hadronic channel: background rejection ð semi-leptonic channel: kinematic fits ð di-leptonic channel: neutrinos

10. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)10 ‘golden channel’ CMS Note 2006/066  Three top quark mass estimators are investigated: 1. Gaussian fit on reconstructed mass spectrum → m t Simple 2. convolution with Gaussian parametrized ideogram → m t ParamIdeo 3. convolution with full scanned ideogram → m t FullIdeo  A likelihood variable is constructed reflecting the probability for signal → P sign  Jet combinations are ordered according to a likelihood variable → P comb  A kinematic fit is applied ( CMS Note 2006/023 ) forcing the W boson mass  The Ideogram is convoluted with a theoretical template  Maximum likelihood gives the estimated top quark mass  4 th estimator: m t FullIdeo but IterCone, MidPoint & k T should give same jet direction  Breit-Wigner Monte Carlo parametrized can be fixed in kinematic fit

11. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)11 ‘golden channel’ CMS Note 2006/066 Properties of the estimators : spectrum with kinematic fit

12. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)12 ‘golden channel’ CMS Note 2006/066 Main uncertainy arises from heavy quark jet energy scale. (1.5%) (2%) (5% On-Off) Pile-Up: could be reduced to 10% and overlaps with JES, hence 5% taken of On-Off JES: could reach 1.5% in well understood detector range (barrel) + energy flow (to be seen as a benchmark for the jet calibration methods) B-tagging: 2% could be reached according to Tevatron experience  1 GeV uncertainty reachable with a good detector understanding

13. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)13 Top quark mass: other methods  From t  l + J/  + X decays : → 100 fb -1 gives after selection ~ 1,000 signal events (S/B > 100) ü the large mass of the J/  induces a strong correlation with the top mass ü easier to identify (extremely clean sample) M BR(overall in tt) ~ 5.3 x 10 -5 hep-ph/9912320 1GeV @ 20fb -1 CMS Note 2006/058 ‘full’ simulation First ‘full’ simulation study

14. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)14 Top quark mass: other methods CMS Note 2006/058  no jet related systematics  2 GeV uncertainty after 2 years of LHC running  conservative estimate as most of the systematic uncertainties are related to the theoretical modeling of the events  after a better understanding of these phenomena a 1 GeV uncertainty is feasible THEORETICAL EXPERIMENTAL

15. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)15 CMS notes [1] ATLAS Eur.Phys.J. C39 (2005) 63-90 [2] LHC Yellow Report on Standard Model Physics [3] Œ Single-lepton channel (Full-Analysis) [1,2] 100~1000  High p T single-lepton sample (Jet-Analysis) [3] 250~1000 Ž High p T single-lepton sample (Cluster-Analysis) [3] 150~1500  Single-lepton channel (Continuous jet algorithm) [2] 100~1000  Di-lepton channel (Jet-Analysis with m lb ) [3] 900~1300 ‘ Di-lepton channel (Energy-Analysis) [3] 400~2000 ’ Di-lepton channel (Tri-lepton events) [3] 1000~1500 “ Di-lepton channel (Full-reconstruction) [1,2] 300~1300 ” From t  l + J/  + X decays [1,2,3] 1000<1000 • High p T fully-hadronic channel [1,2] 180>3500 Stat.Unc. MeV Syst.Unc. MeV ð combining all those results could lead to a more precise measurement (correlations to be estimated !!) (correlations to be estimated !!) ð systematic effects do not necessary overlap between analyses Expectation : top mass determination better than 1 GeV after understanding the detector Top quark mass: combination !?

16. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)16 Top versus anti-top Measuring spin correlations is something different compared to observing spin correlations. The second one could be easier. The top quark does not loose its spin properties before it is decaying. Hence in the decay of top quark pairs, angular correlations should be present. results (CMS, 10fb -1 ) : good agreement with TopRex input Main systematics are related to the jet energy scale and the b-tagging CMS Note 2006/111

17. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)17 Single-top production Never observed ? Each channel sensitive to different signals heavy W’→ s-channel FCNC→ t-channel H ± → Wt-channel Also directly related to |V tb | to percent level (s-channel preferred, t-channel dominated by PDF scale uncertainties of ~10%)

18. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)18 Single-top production: t-channel  Production channel with largest cross section  Optimized event selection (  NLO ~ 243pb)  MET > 40 GeV  b-jet: p T >35GeV, |  |<2.5  light-jet: p T >40GeV, |  |>2.5 (forward)  topological cuts: m rec (top), m T (W)  Expected number of events (10fb -1 )  signal= 2389  tt= 1188  W+jets= 597 ( CompHEP, TopReX, MadGraph )  QCD= negligible ( using factorisation of  )  Resulting S/B~1.34  Estimate of cross section (10fb -1 ) CMS Note 2006/084 3 0% theory, JES, b-tagging

19. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)19 Single-top production: Wt-channel  Two production channels ( CMS Note submitted )  di-lepton → p T  /e >20 GeV, p T j1 >60 GeV & p T j2 >20 GeV ( veto others ), MET>20 GeV  semi-lepton → p T e >30 GeV & p T  >20 GeV, 1 strong b-tag & 2 non-b-jets p T >35 GeV ( veto others ), MET>40 GeV.... jet pairing via likelihood variable  Reject jets with a bad reconstruction quality (fakes) → likelihood variable 10fb -1 di-leptonsemi-lepton tW di-lepton562- tW semi-lepton-1699 tt14377624 WW65- t-channel<20351 s-channel-14 Wbb-10 W2j-130 W3j-539 W4j-80 multi-jet<10508 S/B0.370.18  Control regions for background  reduce tt bck uncertainty 60%→negligible  Results (10fb -1 )   (di-lepton) 8.8 ( stat ) ± 23.9 ( syst ) ± 9.9 ( MC stat ) %   (semi-lepton) 7.4 ( stat ) ± 17.7 ( syst ) ± 15.2 ( MC stat ) %  Main uncertainty from jet energy scale and b-tagging (will improve with better detector understanding) CMS Note 2006/086

20. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)20 Top quarks for calibration: Jet Energy Scale Œ Rescale each jet with relative energy shift  C (rescaling |p| to keep jetmass invariant)  Remake/refit the obtained W mass spectrum → m W (  C) from fit Ž Solve the simple equation m W (  C|data) = M W PDG → best estimate for  C  Compare this measured shift with the true shift from Monte Carlo information (for well matched jet-parton couples (  R<0.2) one can determine the average  C) Result :  C meas = -14.96 ± 0.26 % (  C true = -14.53 %) with 5.41fb -1 mWmW  C(%) E rec /E gen true shift  C true measured shift  C meas scaling to 1fb -1 incl. electron channel : 0.4% used space-angle match (~1% effects with  R) L = 0.33fb -1 Pile-up (On/Off) = 3% CMS Note 2006/025

21. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)21 Top quarks for calibration: b-tag efficiency The b-tagging efficiency can be measured to 4% in barrel and 5% in endcaps. Crucial for a correct estimation of the selection efficiency of Higgs decays to bb. Optimize the jet pairing efficiency via mass constraints in kinematic fits. Select a very pure b-jet sample on which the b-tag algorithms can be applied and the efficiency estimated. CMS Note 2006/013

22. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)22 Comparing the direct and indirect values of m H Ultimate test of the Standard Model To give m t and m W equal weight :  m W = 0.7 10 -2  m t Goal of LHC experiments :  m t < 1 GeV  m W < 15 MeV   m H /m H < 25 % After a discovery one can use EW measurements to differentiate between SM or MSSM Higgs bosons W boson mass references: CMS Note 2006/061 ATLAS ATL-PHYS-PUB-2006-007

23. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)23 Conclusions and outlook  Potential of top quark physics with CMS has been demonstrated ð CMS published a Physics-TDR vol.1 on detector & reconstruction tools ð and simulation results from CMS (Physics-TDR vol.2) ð based on dedicated reconstruction & analysis tools (incl. trigger) ð applied in realistic state-of-the-art settings of the detector ð talk is based on hundreds of pages of public (or almost public) notes  From Tevatron to LHC ð cross sections of signal and background are favourable ð for many topics LHC has the potential to go significantly beyond Tevatron  Outlook... prepare for data taking ð top quarks are crucial for calibration and commissioning at start-up and beyond ð jet calibration and b-tagging performance can be derived from data via top quarks ð over the next year prepare to extract top quark out of this first data P thanks to all CMS collaborators who contributed to these Top Quark studies

24. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)24 Extra’s Some extra informative slides :  lepton reconstruction: resolution  jet reconstruction  b-tagging performance  missing transverse energy  kinematic fit performance  trigger  dependency of cross section on top quark mass  some distributions of event selection observables  semi-leptonic: other top mass estimators  di-lepton events: other top mass estimators  top mass and W boson mass  breakdown uncertainty W polarization  more possibilities beyond the Standard Model

25. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)25 Event reconstruction: lepton reconstruction  Good resolution is needed for triggering (cfr. turn-on curve) ð good performance of tracking, calorimeter and muon systems ð muon p T resolution ~1.5% ( CMS P-TDR vol.1 ) ð electron p T resolution ~1% ( CMS P-TDR vol.1 )  Resolution checked with tt events ð lepton+jet ( CMS Note 2006/024)  Lepton isolation ( several lepton candidates per event ) ð combine several observables per lepton candidate ð likelihood ratio variable ð correct identification of t  l b lepton in 99% of the selected events  effect of magnetic field  (  ) ~ |  | lepton+jet (tt) CMS Note 2006/024

26. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)26 Event reconstruction: jets  Jet clustering : mostly Iterative Cone R=0.5 (optimization eg. Les Houches ’05) ð calorimeter towers above E & E T thresholds (added to online zero suppression)  JES calibration : via method which will be applied on data (to the % level) ð  +jet and Z+jet events ( CMS Note 2006/042, ATLAS P-TDR ): based on p T balans ð W  jj events ( CMS Note 2006/025, ATLAS P-TDR ): from well known W boson mass  Pile-up and underlying event treatments ð (¤) reject jets not attached to primary vertex ( CMS Note submitted ) ð underlying event measurements ( CMS Note 2006/067 ) #jets per event (¤) after primary vertex constraint lepton+jet (tt) CMS Note 2006/066 CMS P-TDR (vol.1)

27. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)27 Event reconstruction: b-tagging  Combined secondary vertex b-tagging algorithm ( CMS Note 2006/014 ) ð based on the combination of several topological and kinematical observables CMS Note 2006/064 60% uds c g 10% 2% <1% CMS P-TDR (vol.1)  General performance plot ð QCD jets 50<E T <80 GeV in barrel |  |<1.4  distribution of discriminator in lepton+jet events lepton+jet (tt)

28. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)28 Event reconstruction: missing E T  Missing transverse energy challenging due to multiple pile-up collisions ð also magnet field will influence resolution (photons versus pions) ð but very good hermiticity and granularity  Missing E T performs well at the LHC but rather hard cuts are needed Resolution of 20 GeV for E T miss ~ 40GeV CMS P-TDR (vol.1) inclusive tt before jet correction after jet correction single-top & tt CMS Note 2006/035 40 GeV QCD multi-jet

29. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)29 Event reconstruction: kinematic fit ~70% purity To obtain the same precision without the fit, one needs 5 times more data (the bias wrt 175GeV is reduced) Using kinematic fit techniques (applying Least-Square methods with Lagrange multipliers) mass constraints can be enforced to the reconstructed event topology.  in the t→Wb decay constrain the W boson mass to its precisely measured value CMS Note 2006/023

30. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)30 Triggering  CMS L1 & HLT triggers applied on full simulation ð CMS Collaboration ‘DAQ and High-Level Trigger’ TDR Œ fully hadronic: inclusive jet trigger  ~17% (QCD reduced to 23 Hz, S/B~1/300)  b-jet stream improvement of 15% on efficiency  semi-leptonic: single-lepton triggers ( p T e >26GeV, |  | 19GeV, |  |<2.1 )   (  )~62% and  (e)~48% Ž di-leptonic: also di-lepton triggers (p T e > 12/12 GeV and p T  > 7/7 GeV)   (  )~88% and  (ee)~77% Œ single-top:. OR. of single-lepton triggers and e+jet (p T e >19GeV and p T jet >45GeV) ®  (W→lv) in s-channel: 24%25%7%38% ®  (W→lv) in t-channel:25%25%7%43% single-  single-e e+jetcombined  Thresholds and rates to be updated in Physics-TDR vol.2 → changes in algorithms and improved understanding of the background → single-lepton are the dominant streams (rate(e)~23.5Hz and rate(  )~25.8Hz) (numbers for L = 2.10 33 cm -2 s -1 )

31. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)31 Top Quark Pair Production Cross-Section Sensitive to top mass  Sensitive to top mass :  ~ 5  m t /m t  5% on  gives 2 GeV on m t Cross-section sensitive to renormalisation and factorisation scale, and to the choice of PDF (Parton Density Function) systematics dominated by the uncertainty on the luminosity ATLAS ATL-PHYS-PUB-2005-024

32. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)32 Top Quark Pair Selection: distributions Fully hadronic channel: Neural Network output combining the information of several topological obervables (scaled to effective cross section) CMS Note 2006/077 Di-leptonic channel: Mass window around the Z boson mass rejects basically all Z+jet events

33. CTEQ, Michigan, May 2007Jorgen D'Hondt (Vrije Universiteit Brussel)33 Flavour Physics : like-sign top quark pairs at 30fb -1 a pp  tt cross section of 1pb becomes visible as a 5  effect on the ratio R  Di-lepton top quark pairs have a clear topology  2 b-jet and 2 isolated leptons with a different charge, selected with a large S/N  exploit the performance of the lepton isolation criteria ( CMS Note 2006/024 )  Motivation for this search  FCNC (in SM suppressed, Z’ bosons in Topcolor assisted Technicolor (TC2))  from top- and techni-pion in TC2 models  in MSSM from for example gluino pairs  Variable to be measured is R = N ++,- - / N +-  ratio of events with a pair of leptons with same and different electric charge  most of the systematics cancel in the ratio CMS Note 2006/065