1. 1 LBNL RPM 15 February 2007 Probing the Physics Frontier with Rare B Decays at CDF Cheng-Ju S. Lin (Fermilab)

2. 2 Interesting time in particle physics with many exciting questions: - Origin of EW symmetry breaking - Nature of cosmological dark matter - Nature of dark energy - + … To get a consistent picture would require physics beyond the Standard Model Tevatron could potentially uncover those mysteries. We can: - Look for things directly (e.g. production of new particles) - Look for deviations from the Standard Model predictions !!!

3. 3 Gold Mine for Heavy Flavor Physics Mixing: B s, B d, D 0 Lifetimes:  b, B s, B c, B +, B d … New particles: X(3872), X b, Cascade(b) … Mass measurements: B c,  b, B s, … Rare decays: B s     , B  K*      D 0     , … Production properties:  (b),  (J/  ),  (D 0 ), … CP Violation: Acp(B  hh), Acp(D 0  K  ), … B and D Branching ratios SURPRISES!? Focus of today’s talk

4. 4 Indirect Search of New Physics : B s      Solid prediction from the Standard Model (SM) In the SM, the decay of B s   +  - is heavily suppressed B d   is further suppressed by another factor of ~20 SM prediction is well below the sensitivity of current generation of experiments  no observation yet New physics could significantly enhance the branching ratio  Any signal would be a clear indication of NP SM prediction  ~ a few decays per 1 billion B s produced B s = bsbs

5. 5  ’ i23  i22   b s R-parity violating SUSY - MSSM: Br(B  ) is proportional to tan 6 . BR could be as large as ~100 times the SM prediction - Tree level diagram is allowed in R-parity violating (RPV) SUSY models. Possible to observe decay even for low value of tan. Some Scenarios of NP Either discovery or null result could shed light on the the structure of new physics !!!

6. 6 Collide proton (p) and anti-proton (p) at the c.m. energy of ~2 TeV Chicago Tevatron CDF D0 p p TEVATRON Collider Lots of B s produced in the collision debris ! Record luminosity: ~270E30 cm -2 s -1 (design 300E30)

7. 7 Integrated Luminosity Close to 2fb -1 of data collected by CDF Analyses presenting today use from 780pb -1 to ~1fb -1 of data

8. 8 Flavor Creation (annihilation) qb q b Flavor Creation (gluon fusion) b g g b Flavor Excitation q q b g b Gluon Splitting b g g g b b’s produced via strong interaction decay via weak interaction Tevatron is great for heavy flavor: Enormous b production cross-section, x1000 times larger than e + e - B factories All B species are produced (B 0, B +, B s,  b,  b, etc…) However, Inelastic (QCD) background is about x1000 larger than b cross-section Online triggering and reconstruction is a challenge: collision rate ~1MHz  tape writing limit ~100Hz Heavy Flavor Physics in Hadron Environment

9. 9 CDF II Detector Significant detector, trigger, and DAQ upgrades in Run II

10. 10 Looking for B s   with CDF Detector Look for Bs production and decay vertices Measure muon track momentum and charge CMU: ||<0.6 CMX 0.6<||<1.0

11. 11 CDF has implemented a 3-tier trigger Level-1 is a synchronous hardware trigger - Can process one event every 132ns - Input rate = 1.7MHz (396ns 36x36 bunches) L1A rate ~ 30KHz (limited by L2) Level-2 is a combination of hardware and software trigger (asynchronous) - Average Level-2 processing time is ~30  s - L2A rate ~1KHz (limited by event-builder) Level-3 is purely a software trigger - Massive PC farm - L3A rate ~ 100Hz (limited by tape writing) Data reduction rate (L1+L2+L3)  1 : 17000 CDF Trigger: Lifeline of B Physics Program

12. 12  + (e + )  - (e - ) IP (1) Dimuon trigger: For triggering on J/  and rare B decays (e.g. Bs   and B   X) (3) Lepton+Displaced Track(SVT): For triggering on semileptonic B decays. (2) Two-track trigger (SVT): For triggering on hadronic B and charm decays. Both tracks are required to have an impact parameter d 0 > 120  m. (D 0  , B  e , etc…) X } d0d0 X IP X } d0d0  + (e + ) Three Classes of B Physics Triggers at CDF

13. 13 CDF is the first hadron collider experiment to be able to trigger on fully hadronic B events Level-2 SVT Trigger SVT links drift chamber tracks from Level-1 with silicon hits to compute the impact parameter of the track. SVT impact parameter (  m) -600 -300 0 300 600 SVT d0 resolution is ~ 47  m (35  m beamline  33  m resol). SVT revolutionized B and Charm physics at CDF. Track impact parameter Silicon Vertex Tracker (SVT)

14. SVT Performance Our physics program is dictated by what triggers we have In Run I, B  hh physics was a fantasy In Run II with SVT, we are making world class measurements First observations of: B s  K + K - B s  K +  -  b  p  -  b  pK + Did I mention Bs mixing too?

15. 15 Keeping trigger rates under control is a constant battle !!! Main issue: trigger rate blows up rapidly vs. luminosity For illustration, the following rare B dimuon triggers alone: - CMU-CMU pT>1.5GeV - CMU-CMX pT>1.5GeV would take up the entire level-2 trigger bandwidth at L=200E30 cm -2 s -1 In the latest trigger table, there are more than 150 level-2 triggers that need to co-exist RARE B TRIGGERS AT TEVATRON ~540 Hz @ 200 E30 ~320 Hz @ 200 E30 CDF L2 Dimuon Trigger Cross Section

16. 16 KEEPING RARE B TRIGGERS ALIVE Handles to control rates: - Tighter selection cuts (e.g. pT of muon) - Apply prescales (DPS, FPS, UPS, etc.) - Improving trigger algorithm - Upgrading trigger hardware We’ve been using a combination of all four handles to control the trigger rate  trading efficiency for purity It’s been a great challenge keeping B and high pT triggers alive at Tevatron It’ll be an even greater challenge at the LHC !! Non-optimal for rare searches

17. 17 Using 780pb-1 of dimuon trigger data: CMU-CMU trigger CMU-CMX trigger Use this inclusive sample to search for: B s   +  - (Mass B s =5.37GeV) B d   +  - Even if BR is x10 the SM value, only expect a hand full of signal events in the signal region Signal region is swamped with various kinds of background: both from SM processes and detector effect (fake muons) B   Data Sample Effective background rejection is the key to this analysis!! Search region

18. 18 Analysis Overview Motto: reduce background and keep signal eff high Step 1: pre-selection cuts to reject obvious background Step 2: optimization (need to know signal efficiency and expected background) Step 3: reconstruct B +  J/  K + normalization mode (take into account Br of B  J/  K and J/    : >> 100million B + ) Step 4: open the box  compute branching ratio or set limit

19. 19 Pre-Selection cuts: –4.669 < m  < 5.969 GeV/c 2 –p T (  )>2.0 (2.2) GeV/c CMU (CMX) –p T (B s cand.)>4.0 GeV/c –Track, muon and vertex quality cuts –3D displacement L 3D between primary and secondary vertex CDF Pre-selection Bkg substantially reduced but still sizeable at this stage

20. 20 Background Rejection: B s   Decay Characteristics B s -mass: M Bs =5.37 GeV B s -flight distance 2 muons point to a “common vertex” Background rejection cuts: muons add up to B s mass muons are isolated and have common vertex vertex is well separated from beam-spot No other tracks from this vertex pp ++ --

21. 21 –  +  - mass (E  energy, P  momentum) –B vertex displacement: –Isolation (Iso): (fraction of p T from B   within  R=(  2 +  2 ) 1/2 cone of 1) –“pointing (  )”: (angle between B s momentum and decay axis) B      Signal vs Background Discrimination ++ -- L 3D primary vertex di-muon vertex P T (  ) L 3D For each di-muon candidate, we compute the 4 variables: M Iso 

22. Distribution of the Discriminating Variables Blue = B s   signal events from simulation (Monte Carlo) Black = expected bkg distributions Using  and Iso to construct a new variable, likelihood ratio: P s(b) i is the probability distribution function for signal (background) for variable i, with i loops over, , iso

23. 23 Likelihood Ratio (L R ) Distribution Di-muon events with large L R (near 1) are more likely to be signal Events with L R near 0 are more likely to be background Signal distribution is based on simulation

24. 24 Background Estimate Use control samples to cross-check bkg estimate Assume linear background shape extrapolate # of background events in sidebands to signal region (± 60 MeV signal window) 1.) OS- : opposite-charge dimuon,  < 0 2.) SS+ : same-charge dimuon,  > 0 3.) SS- : same-charge dimuon, < 0 4.) FM : fake muon sample (at least one leg failed muon stub chi2 cut)

25. 25 L R CMU-CMU CMU-CMX cut pred obsv prob pred obsv prob >0.50 489±12 483 41% 351 ± 10 338 27% 28% OS- >0.90 62 ± 4 73 12% 56 ± 4 63 22% 7% >0.99 4.8 ± 1.2 9 8% 3.9±1.1 8 7% 2% >0.50 5.4 ± 1.3 4 40% 3.3 ± 1.0 2 39% 27% SS+ >0.90 <0.10 0 - 0.9 ± 0.5 0 43% 43% >0.99 <0.10 0 - <0.10 0 - - >0.50 6.6 ± 1.4 7 49% 4.2 ± 1.1 5 41% 40% SS- >0.90 0.6 ± 0.4 1 45% 0.3 ± 0.3 0 70% 57% >0.99 <0.10 0 - <0.10 0 - - >0.50 188±8 159 3% 33±3 37 29% 7% FM >0.90 34±3 24 7% 6±1 5 46% 6% >0.99 4.5 ±1.0 9 6% 0.6±0.4 0 55% 12% Using a wider ± 150 MeV signal window for cross-check (Probability factor in uncertainty on Poisson mean) Combined prob Cross Check BKG Estimate in Control Samples

26. 26 B  hh Background CDF signal region is also contaminated by B  h + h - (e.g. B  K + K -, K +  ,     ) K   muon fake rates measured from data using D* sample Convolute fake rates with expected B  h + h - distributions to to obtain B  hh bkg Total bkg = B  hh + combinatorial Decay B  hh Background Combinatoric Background Total Background BsBs 0.19±0.061.08±0.361.27±0.36 BdBd 1.37±0.161.08±0.362.45±0.39 L R > 0.99

27. 27  LH (B s ) cut CMU-CMU CMU-CMX L R >0.90 (70+/-1)% (66+/-1)% L R >0.92 (67+/-1)% (65+/-1)% L R >0.95 (61+/-1)% (60+/-1)% L R >0.98 (48+/-1)% (48+/-1)% L R >0.99 (38+/-1)% (39+/-1)% determined from B s   MC MC modeling checked by comparing  LH (B + ) between MC and sideband subtracted Data (stat uncertainties only) Likelihood Ratio Efficiency for Bs Signal Optimize analysis based on a-priori expected upper limit  L R >0.99 !!!

28. 28 Now Look in the B s and B d Signal Windows B s Branching Ratio Limit: Br(B s   )<1.0×10 -7 @ 95%CL Br(B s   )<0.8×10 -7 @ 90%CL B d Branching Ratio Limit: Br(B d   )<3.0×10 -8 @ 95%CL Number of observed events in the signal box is consistent with bkg expectation  set branching ratio limit B s  : Observed 1 candidate Expect ~1.3 background events B d  : Observed 2 candidates Expect ~2.5 background events Best limits in the world, but still no hints of new physics

29. 29 CDF B s ->  176 pb -1 7.5×10 -7 Published DØ B s ->  240 pb -1 5.1×10 -7 Published DØ B s ->  300 pb -1 4.0×10 -7 Prelim. DØ  700 pb -1 Prelim. Sensitivity CDF B s ->  364 pb -1 2.0×10 -7 Published CDF B s ->  780 pb -1 1.0×10 -7 Prelim. Branching Ratio Limits Evolution of limits (in 95%CL): World’s best limits Babar B d ->  111 fb -1 8.3×10 -8 Published CDF B d ->  364 pb -1 4.9×10 -8 Published CDF B d ->  780 pb -1 3.0×10 -8 Prelim. 90% CL

30. R. Dermisek et al., JHEP 0304 (2003) 037 SO(10) Grand Unification Model tan(  )~50 constrained by unification of Yukawa couplings Pink regions are excluded by either theory or experiments Green region is the WMAP preferred region Blue dashed line is the Br(Bs   ) contour Light blue region excluded by old Bs   analysis R. Dermisek et al., hep-ph/0507233 (2005) Red arrows indicate exclusion from this result!!! Remaining white region is still not excluded by experiment

31. SUSY General Flavor Mixing (GFM) framework J. Foster et al. Hep-ph/0604121  xy parameters quantify variations from Minimal Flavor violating assumption B  s , Bs  , and Bs mixing severely constrain non-MFV Implications on SUSY Flavor Violation

32. 32 SUSY General Flavor Mixing (GFM) framework J. Foster et al. Hep-ph/0604121 2-D scan over  xy space non-MFV SUSY phase space is severely constrained Implications on SUSY Flavor Violation tan=40

33. 33 Sneak Preview of B s   1fb -1 update is near completion Implemented NN selection to enhance signal efficiency and bkg rejection: NN contains additional variables: pT(B s ), pT(muon), etc. For a given bkg level, NN signal eff is 15-20% higher than L R !!

34. 34 Sneak Preview of B s   Apply improved muon selection and particle ID to suppress fake muons and B  hh backgrounds Instead of single-bin counting experiment, use multi-bin parameterizations: 780pb -1  1fb -1, only about 30% increase in stat Expect the sensitivity to increase by a factor of 2!!! Neural Net Output B s Signal Eff

35. 35 CDF Projection Conservative projection based on our current (780pb -1 ) performance Improvement expected from 1fb -1 analysis Achievable in RunII

36. 36 B Rare Decays B      h : B +   K + B 0   K * B s     b   Penguin or box processes in the Standard Model: Rare processes: predicted BR(B s    )=16.1x10 -7 observed at Babar, Belle   not seen s s s b s b s s   B u,d,s       K + /K*/  PRD 73, 092001 (2006) PRL 96, 251801 (2006) C. Geng and C. Liu, J. Phys. G 29, 1103 (2003)

37. 37 Probe various Wilson coefficients (b  s) - Predicted in several new physics scenarios (e.g. SUSY) - Large forward backward asymmetry in B 0   K* decay expected 37 Standard Model Need better statistics In low q 2 region CDF may be able to contribute to resolving this Flipped signs Probe of New Physics

38. 38 proper decay length ( ) significance Pointing (  ) |  B –  vtx | Isolation (Iso) 38 Discriminating Variables For B   K Using similar discriminating variables and analysis strategy as B s   search SIGNAL SIDEBAND SIGNAL SIDEBAND SIGNAL SIDEBAND

39. 39 B u,d,s       K + /K*/  Results with 1fb -1 Using similar discriminating variables and strategy as B s   analysis Signal Sideband Extrapolated fit Note: bins are counted. Gaussian is for illustration of expected width only B u       K +  B d       K*  B s      

40. 40 We see evidence for the B +   K + B 0   K* rare modes We are homing in on the B s rare mode B u,d,s       K + /K*/  Summary With 1fb -1 of data:

41. Summary Tevatron heavy flavor physics program is in full swing. What I’ve showed today is only the “tip of the iceberg” CDF could observe any modest enhancement of B s  in Run II. CDF is also on the verge of observing B s   decay For the next few years, Tevatron will continue to search for new physics with direct and indirect searches The ultimate frontier machine will be the LHC. We may finally have a first clean glimpse of the physics beyond the Standard Model Look forward to seeing what that new physics really is