Search for New Physics through Rare B Decays S. Uozumi for the CDF-II collaboration Feb-21th Lake Louis Winter Institute 2012 NP?

CDMS II Rouzbeh Allahverdi, Bhaskar Dutta, Yudi Santoso arXiv: Excluded by 1)Rare B decay b  s  2)No CDM candidate 3)Muon magnetic moment a b c B s   for NP (e.g.SUSY) B s   for NP (e.g.SUSY) SM Expectation Br enhancement By New Physics JHEP 1009 (2010)106 Gaugeno mass (GeV) Slepton/sqqark mass (GeV)

3 B s   Method : Relative normalization counting Count signal region events : 5.169<M  <5.469 GeV –corresponds to  6  σ m, where σ m ≈24MeV Sideband regions: additional +0.5 GeV –Used to understand background MC simulation of B s and B 0 →      mass peaks – 9.7 fb-1 of whole CDF Run-II dataset is used – Measure the rate of B s → μ + μ - decays relative to B  J/  K + – Apply same sample pre-selection criteria – B s → μ + μ - sample is highly purified with Neural Network event selection Dominant Backgrounds Sequential semi-leptonic decay: b → cμ - X → μ + μ - X Double semileptonic decay: bb → μ + μ - X Continuum μ + μ -, μ + fake, fake+fake Peaking Background in signal region (B  hh)

Control sample 4 Roughly selected with pre-selection cuts same with one for Bs   pre- selection Control sample is used not only Br. Normalization, but also various checks on BG modeling. Opposite-sign  B decay length < 0 (OS-) Same-sign  B decay length > 0 (SS+) Same-sign  B decay length < 0 (SS-) One  is tagged as fake  B decay length > 0 (FM+) Prediction from B d  J/  K + sample Observed in B s   control sample

Further BG discrimination by Neural Network NN input variables 3D pointing angle Isolation Proper decay length Proper decay length sig. P T (Bs) P T (  ) Etc  Multi-variable analysis : Neural Network Unbiased optimization based on MC signal and data sidebands Output (0 ~ 1) Discriminating variables Input nodes NeuroBayes with 14 discriminating parameters strongly distinguish signal and BG.

6 BG shape modeling from M  Sideband Events are separated in each NN output bins In each bin, continuum BG is estimated from SB polinomial fitting < 5 GeV is rejected from the fit to avoid B   + neutrinos events)

7 Di-muon mass for B d / B s signal region Completely consistent with BG For B d   channel. On B s   channel, slight excess can be seen in high NN value events.

8 Summary on Br (Bs -> mumu) limit p-values assuming B s background only : 0.94% B s SM+background : 7.1% Again with SM prediction SM prediction

9 b  s  for New Physics NP? Strongly suppressed in SM due to 2 nd order weak interaction New physics in a loop can change Br. Fraction and shape of those event Many channels with different spectator available to explore b  s  decays Tools to go beyond SM : Br fraction of several B baryons into b  s  process Direction, angle of outgoing particles … may differ depending on intermediate new physics model

10 b  s  Br measurement - Again Relative normalization counting - Starts with di-muon trigger datasets with 6.8 fb -1 Control sample events are pre-selected by kinematical cuts Signal candidates are selected by Neural Network in addition to pre-selection for further purification Ratio of Br. Fraction of signal and control sample is determined by Signal event Yield (w/ NN cut) Control sample yield (only with pre-selection ) Pre-selection Efficiencies ratio NN cut efficiency From PDG

11 B(Λ b 0 →Λμ + μ - ) =[1.73±0.42(stat)±0.55(syst)]×10 -6 B(B s 0 →φμ + μ - ) =[1.47±0.24±0.46]×10 -6 B(B + →K + μ + μ - ) =[0.46±0.04±0.02]×10 -6 B(B 0 →K *0 μ + μ - ) =[1.02±0.10±0.06]×10 -6 B(B 0 →K 0 μ + μ - ) =[0.32±0.10±0.02]×10 -6 B(B + →K *+ μ + μ - ) =[0.95±0.32±0.08]×10 -6 World’s first observation !! Other b  s  channels have comparably large statistics with B factories. Reconstructed b  s  events and Br. fractions

12 Event-shape (A FB ) analysis for b  s  decay Now we have 6.8 fb -1 from CDF Run II !! There are various event-shape parameters to access beyond the SM (arXiv: )arXiv: Most interesting parameter is FB asymmetry of q 2 =M  2 compared with muon polar angle on B rest frame :

13 A FB analysis with 6.8 fb -1 B→K * μ + μ - (simultaneous fit of K *0 and K *+ channels) B + →K + μ + μ - Belle result CDF provides b  s  A FB measurement comparable with B factories!

B rare decays are excellent stage to search for New Physics effect. Using full of 9.7 fb-1 of CDF Run-II data, we observed : while SM predicts Summary & Conclusions Various b  s  decay channels are observed : B(Λ b 0 →Λμ + μ - ) =[1.73±0.42(stat)±0.55(syst)]×10 -6 B(B s 0 →φμ + μ - ) =[1.47±0.24±0.46]×10 -6 B(B + →K + μ + μ - ) =[0.46±0.04±0.02]×10 -6 B(B 0 →K *0 μ + μ - ) =[1.02±0.10±0.06]×10 -6 B(B 0 →K 0 μ + μ - ) =[0.32±0.10±0.02]×10 -6 B(B + →K *+ μ + μ - ) =[0.95±0.32±0.08]×10 -6 … world’s first!! Observed A FB is comparable with B factories, and may imply New Physics effect.

15 Backups

16 SM Diagrams Penguin Diagram Box Diagram 2HDM Penguin Diagram SM Expectation NP Expectation Br enhancement Rare decay : FCNCs, forbidden at tree level Powerful Probe to New Physics JHEP 1009 (2010)106

The CDF Detector Silicon Vertex Detector (precise position measurement) Tracking Chamber (momentum measurement) 1.4 Tesla Solenoidal magnet Calorimeter (red and blue) (energy measurement) Muon Detector (yellow & blue) General purpose collider detector for many physics programme : B physics QCD / Top quark properties Electroweak physics (W, Z bosoms) Higgs / Exotic physics

 = TOF Silicon vertex (SVX-II) detector Central tracker drift chamber Time-of-flight (TOF) system EM and HAD calorimeters Muon detector Front-end, DAQ and trigger electronics The CDF Detector

19 New Neural Network Neural Network Input Variables 3D proper decay length Isolation Pointing angle (  3d ) Lower |p T (  )| 3D proper decay length significance Larger |d 0 (  )| Smaller |d 0 (  )| Smaller |d 0 (  )| significance Larger |d 0 (  )| significance Vertex Fitting  2 Decay length (L 3d ) 2D pointing angle (  2d ) L xy significance |d 0 (Bs)| New 14 variable NN to increase S/B Carefully chose input variables to avoid bias for di-muon mass shape Impact parameter

20 B s (B 0 ) Signal vs. Background From B. Casey, ICHEP2010 K p b c K BsBs   c p p b b   p p BsBs   b p p   b c Signal sequential double B→KK

21 J/  region analysis region entries / 10 MeV/c 2 Trigger Data collected using dimuon trigger “CC”: –2 central muons “CMU”, |  |<0.6, –p T >1.5 GeV –2.7<M  <6.0 GeV –p T(  ) + p T(  ) >4 GeV “CF”: –one central, one forward muon “CMX”, 0.6<|  |<1.0 –p T >2 GeV –other cuts same as above Trigger efficiency same for muons from J/  or B s (for muon of a given p T )

22 Two-body B→ hh decays where h produces a fake muon can contribute to the background fake muons dominated by  +,  -,K +, K - fake rates are determined separately using D*-tagged D → K -  + events Estimate contribution to signal region by: take acceptance, M hh, p T (h) from MC samples. Normalizations derived from known branching fractions convolute p T (h) with p T and luminosity-dependent  -fake rates. Double fake rate ~0.04% 2) Background from two-body hadronic B decays

23 Current experimental results from CDF & LHC

24 Example of D 0 peaks in one bin of p T, used to extract a p T and luminosity- dependent fake rate for K + and K - Fake rates from D*-tagged D 0 → K -  + events K-K- K+K+ Kaons passing muon selection: K-K- K+K+

25 “ Smoking gun” of some Flavor Violating NP models: –ratio BR(B s →      BR(B 0 →  highly inform ative about whether NP violates flavor significantly or not –clear correlation between CP violating mixing pha se from B s → J/  and BR(B s →      Important complementarity with direct se arches at Tevatron and LHC –Indirect searches can access even higher mass s cales than LHC COM energies New bounds on BR(B 0 →      and BR(B s →      are of crucial importance, and are a top priority at the Tevatron and LHC. Altmannshofer, Buras, Gori, Paradisi, Straub, Nucl.Phys.B830:17-94,2010 Probing New Physics 

26 Plenary talk A.Buras, Beauty 2011: Probing New Physics 

27 Ensemble of background-only pseudo-experiments is used to determine a p-value for a given hypothesis Determination of the p-value Log Likelihood Distribution of pseudo-experiments for background-only hypothesis for B 0 →      signal window for each pseudo-experiment, we do two fits and form the log-likelihood ratio in the denominator, the “signal” is fixed to zero (I.e. we assume background only), and in the numerator s floats L(h|x) is the product of Poisson probabilities over all NN and mass bins systematic uncertainties included as nuisance parameters, modeled as Gaussian. Result: the p-value for the background-only hypothesis is 23.3% with