Rare Decays at Tevatron

1
Rare Decays at Tevatron
Topics 1)B?mm search at Tevatron 2)
b?sll- at Tevatron

• Marco Rescigno
• INFN/Roma
• for the CDF and DØ coll.
• CKM06, Nagoya
• Dec 14th 2006

2
Tevatron
3
Tevatron Luminosity
gt2 fb-1 delivered!
1.5 fb-1 good data on tape now
Analysis shown today 0.7 1 fb-1
4
CDF DØ detectors
DO

• Good muon coverage and triggering
• DØ hlt2.2
• CDF hlt1
• Good momentum resolution, tracking
• CDFs(MB)25 MeV/c2
• reduce combinatorial and Bd?hh ? Bs search
  window contamination
• Good Vertexing
• CDF L00 (rinner1.4cm)
• DØ L0 upgrade (rinner1.6cm)

CDF
5
Triggers and Data Sample

• CDF
• di-muon triggered data
• Two separate search channels
• Central/central muons
• (CMU-CMU)
• Central/forward muons
• (CMU-CMX)
• CMU hlt0.6, CMX 0.6 lt h lt1
• pT(m)gt1.5 GeV/c
• 780 pb-1 Bs(Bd )?mm limit
• 920 pb-1 Search for Bu,d,s? mmh
• DØ
• First 300 pb-1 di-muon triggered data
• with box opened ? limit
• 400 pb-1 data still blinded
• Combined sensitivity for 700 pb-1 of recorded
  data (300 pb-1 400 pb-1 )

S/B is expected to be extremely small.
Effective bkg rejection is the key to this
analysis!!
6
Trigger _at_ high lumi?

• Clever and Clever triggering as inst. Luminosity
  increases
• Keep high purity triggers alive at high
  luminosity
• Full use of available bandwidth at lower inst.
  Luminosity with dynamically adjusted prescales
• Hardware upgrade on the level 1 track trigger
  processor
• reduce fake rate by reconstructing track segments
  also in the stereo layer

7
B?mm- search at Tevatron
8
Motivation

• Standard Model prediction very suppressed

(Buchalla Buras, Misiak Urban)

• Sizeable New Physics enhancement predicted in
  many scenarios, e.g. high tanb SUSY

Any signal _at_ Tevatron would be New Physics !
9
Strategy

• Blind optimization using signal Monte Carlo
  sample and sideband data
• Normalize to known B?J/?K

• Reconstruct Normalization mode in the same data,
  applying same criteria ? reduce systematics
• Only ratio of efficiency matters
• Evaluate expected background and then open the
  box and calculate BR or limit

10
Preselection Cuts

• CDF
• pT(m)gt2.0 (2.2) GeV/c CMU (CMX)
• pT(Bs cand.)gt4.0 GeV/c
• y(Bs) lt 1
• 4.669 lt mmm lt 5.969 GeV/c2
• muon quality cuts
• good vertex
• 3D displacement L3D between primary and secondary
  vertex
• ?(L3D)lt150 mm
• proper decay length 0 lt l lt 0.3cm

• Pre-selection DØ
• pT(m)gt2.5 GeV/c
• h(m) lt 2
• pT(Bs cand)gt5.0 GeV/c
• 4.5 lt mmm lt 7.0 GeV/c2
• muon quality cuts
• good di-muon vertex

38k events after pre-selection
300 pb-1
11
B?mm signal discrimination

• mm- mass
• 2.5s mass window (60 MeV/c2)
• B vertex displacement
• CDF ?
• D0 ?
• Isolation (Iso)
• (fraction of B?mm pT within DR(Dh2Df2)1/21
  cone)
• pointing (Da)
• (angle between Bs momentum and decay axis)

12
Search Optimization _at_ CDF

• CDF construct a lilkelihood ratio LR using l, Da,
  Iso

• Optimize LR cut on the expected a-priori 90 C.L.
  limit
• Background PDF from data sidebands
  4.669ltMmmlt5.169 GeV/c2 U 5.469ltMmmlt5.969

13
Search Optimization _at_ CDF

• Optimal value LR gt 0.99
• Signal efficiency e(Bs)? 35
• Background Rejection O(103)

• LR distributions in signal and sidebands match
• A posteriori check...

14
Search optimization _at_ DØ

• Similar efficiency and background rejection
• e(Bs)? 35

• Optimize cuts on three discriminating variables
• Pointing angle
• 2D decay length significance
• Isolation
• Maximize S/(1sqrt(B))
• B from sidebands

15
Background prediction

• Combinatorics linear extrapolation from
  sidebands into 60 MeV/c2 signal window for CDF
  ( 180 _at_ D0)

• Cross check predictions using independent
  background enriched samples
• Same Sign di-muons
• Opposite Sign di-muons with Lxylt0
• Fake muons

• B?hh expected signal convoluted with muon fake
  rate (CDF)

16
Examining Signal Boxes
Experiment Experiment B0s search B0s search B0d search B0d search
Experiment Experiment Expected Obs. Expected Obs.
CDF 780 pb-1 CMU-CMU 0.880.30 1 1.860.34 2
CDF 780 pb-1 CMU-CMX 0.390.21 0 0.590.21 0
D0 Old Data 4.31.2 4
D0 New Data 2.20.7 -
17
Normalization B?J/?K

• Need B?J/?K yield to extract limits
• CDF (780 pb-1) 4200 (CMU-CMU) 1550 (CMU-CMX)
• D0 (400 pb-1) 900

18
Results

• No signal found
• CDF Bs limit (780 pb-1)
• BR(Bs?mm) lt 8 10-8 (10) _at_ 90 (95)C.L.
• DØ average expected limit (700 pb-1)
• BR(Bs?mm) lt 19 10-8 (23) _at_ 90 (95)C.L.
• CDF Bd limit (780 pb-1), world best
• BR(Bd?mm) lt 2.3 10-8 (3) _at_ 90 (95) C.L.
• compare Babar (hep-ex/0408096, 110 fb-1 )
• BR(Bd?mm) lt 8.3 10-8 _at_ 90 C.L.

19
Impact on New Physics

• Current constraints from Dms and Bs?mm are
  differently effective in the new physics
  parameter phase space
• Improved limits on Bs?mm can further constraint
  SUSY at large tanb

Foster,Okumura,Roszkowski Phys.Lett. B641 (2006)
452
20
Tevatron Expected Reach

• Based on current analysis
• might be conservative
• Can exclude region of low 10-8 with full Run II
  statistics
• Significantly improved analysis will appear soon

21
Area of Improvement

• Improved muon selection based on additional
  information
• Energy deposition in the calorimeter
• dE/dx in the drift chamber
• Significant reduction in fakes expected.
• Neural Net based final discriminant with
  additional background suppression power
• Use the 2-dimensional dimuon mass-discriminant
  plane to evaluate signal/limit
• Stay tuned for an updated results at winter
  conferences

• Current muon fake rate
• Determined for Kaon and pions of each charge from
  high statistics D?D0?K-p sample

22
b?sll- decays at Tevatron
23
Goals

• Sensitive to New Physics (Rates and Asymmetries)
• Bd and B modes established at B-factories
• BR(B ?mm K)0.340.19-0.14 x 10-6 (PDG 06)
• BR(Bd ?mm K)1.220.38-0.32 x 10-6 (PDG 06)
• Re-establish signals in Tevatron data and
  discover unseen Bs?mmf decays
• BR(Bs ?mmf)1.6x10-6 C. Geng and C. Liu, J. Phys.
  G 29, 1103 (2003)
• CDF new results with 0.92 fb-1
• D0 published a BR(Bs?mmf) limit with 0.4 fb-1
• PRD 74, 031107 (2006)

24
Strategy

• Similar to the B?mm case
• Normalize signal to analogous B?J/?h (J/??mm)
  decays
• Blind optimization
• Exclude J/? and ? region
• Sideband data for optimization and background
  estimate
• Monte Carlo and data for efficiency ratios with
  normalization mode

25
Signal Selection Optimization

• CDF/DØ similar analysis
• CDF optimize Nsig / sqrt(NsigNbkg)
• DØ optimize Nsig / (1 sqrt(Nbkg) )

26
Bu,d Results

• For all modes
• pT(B)gt4.0 GeV/c
• pT(h)gt1.0 GeV/c
• mKp-mKlt50 MeV/c2
• mKk-mflt10 MeV/c2

• Counting events in 2s window around B mass,
  excesses seen in all modes
• Background from 3-9 s sideband extrapolated to
  signal window
• Fit shown for illustration purpose

27
Bs Results

• CDF (920 pb-1)
• 11 candidates found
• 3.51.5 expected background
• 2.4 s significance
• DØ (400 pb-1)
• 0 observed
• 1.60.6 expected

28
BR (B?mmh)

• Good agreement similar uncertainty with
• Babar PRD 73, 092001 (2006) (208 fb-1 ? 10
  mmK, 15mmK0)
• Belle hep-ex/0410006 (250 fb-1 ? 40 mmK,
  40mmK0)
• BR(B?mmK) 0.72 0.15(stat.)
  0.05(sys.)x10-6 (45 ev.)
• BR(B0?mmK) 0.82 0.31(stat.)
  0.10(sys.)x10-6 (20 ev.)
• BR(Bs?mmf) lt2.4 x10-6 _at_ 90 C.L.
• 1.16 0.56(stat.)
  0.42(sys.)x10-6
• Improve upon DØ limit (400 pb-1) BR(Bs?mmf) lt 3.3
  x 10-6 _at_ 90 C.L.

29
Summary

• CDF/DØ analyzed 800 pb-1 of Run II data
  searching for B?mm signal
• Current limits in the 10-8 territory
• No major obstacle in pushing down limits with
  increasing exposure
• Significantly improved analysis with gt1 fb-1 data
  sample
• Constraining more more New Physics
• CDF/DØ entering the b?sll arena
• New solid B?mmK signal from CDF
• A 2.4 s excess in the Bs?mmf reported from CDF
• U.L. limit close to SM prediction
• 1.5 fb-1 on tape more to come

30
BACKUP
31
MB vs Likelihood Ratio
32
D0 SENSITIVITY FOR 700 pb-1

• Obtain a sensitivity (w/o unblinding) w/o
  changing the analysis
• Combine old Limit with obtained sensitivity

(400 pb-1)
Cut Values changed only slightly!
Expect 2.2 0.7 background events
33
Signal mmm Spectra

• Mass spectra not corrected for efficiency/backgrou
  nd
• Resonance veto
• 2.9ltmmmlt3.2 (J/?)
• 3.6ltmmmlt3.75 (?)
• D (D,Ds) any 2(3) track combination within 25
  MeV of PDG mass

34
NORMALIZATION MODES
Apply similar pre-selection requirements as B?mm
analysis
450 pb-1
Clean samples of norm events
35
CDF trigger architecture

• Crossing 396 ns 2.5 MHz
• Level 1 hardware
• Calorimeter, Muon, Tracks
• 30kHz (reduction x100)
• Level 2 hardware CPU
• Cal cluster, Silicon tracks
• 900 Hz (reduction x60)
• Level 3 Linux PC farm
• Offline quantities
• 100 Hz (reduction x5)