1. Oscillations B s The XXXVIIth Rencontres de Moriond
QCD AND HIGH ENERGY HADRONIC INTERACTIONS
Les Arcs, March
Oscillations B s
Results from CDF, SLD and the LEP experiments
New ALEPH results for these Winter Conferences
Combinations from BOSC working group
(many thanks to O. Schneider)
Conclusions
CERN-EP/
Submitted to EPJ
Title
D. Abbaneo

2. Time-dependent asymmetry between “mixed” and “unmixed” decays
Phenomenology
Example:
(LEP-like numbers)
sp/p = 0.15
sL= 250 mm
p = 32 GeV/c
Signal only
No mistag
Time-dependent asymmetry between “mixed” and “unmixed” decays
Oscillating term damped by the exponential decay and proper time resolution effects
st = (m/p) sL  (sp/p) t
D. Abbaneo

3. The basic strategy Initial state Final stateIP Kf
Bs candidate K
S.V.
Final state
Choose a selection criterion
determines directly the final state tag method
and signal fraction
Opposite hemisphere
P.V. charge, S.V. charge, Jet charges, Kaon charge, Lepton charge & kinematics
Same hemisphere
P.V. charge, Kf charge
Overall
F-B asymmetry (very SLD!)
h  LEP
h  SLD
Tag initial and final state
total mistag h = hi ( 1-hf ) + hf ( 1-hi )
Measure proper time
resolution: st = (m/p) sL  (sp/p) t
Maximum likelihood fit to Dms
(or to the oscillation amplitude with fixed frequency)
D. Abbaneo

4. Selection methods Statistics Fully inclusive Delphi, SLD
FS tag from dipole or NN
Very high statistics ( N  LEP, N  SLD )
“Natural” signal purity ( fs = 10%)
Semi-inclusive Aleph, Delphi, Opal, CDF, SLD
Inclusive l, f-l, K
Still high statistics ( N  104 LEP, N  SLD )
Signal purity “natural” or slightly enhanced.
Semi-exclusive Aleph, Delphi, Opal, SLD
Ds-h, Ds-l
Lower statistics ( N  102 LEP, N  SLD )
Signal purity up to 60%, controlled from data.
Fully exclusive Aleph, Delphi
Little statistics ( N  50 events )
Purity  50%, excellent proper time resolution!
Signal purity
Resolution
D. Abbaneo

5. Crucial for inclusive analyses
The oscillation fit
For every event:
Key point
sL, sp
sL, sp, fj, Pj estimated event-by-event
Each event is treated as a single experiment!
Resolution, bias corrections and pull corrections vary WIDELY as a function of the event topology.
Careful event-by-event treatment is essential especially for the vertexing.
The statistical power at high frequency depends crucially on a careful description of the sample.
Crucial for inclusive analyses
D. Abbaneo

6. The amplitude method F (sp,sL)  exp(-st w )
P(t) = ½ Ge-Gt [ 1  cos(Dmst) ] P(t) = ½ Ge-Gt [ 1  A cos(wt) ]
Fit for the amplitude of an oscillation with a given frequency*
Set a limit by combining analyses unable to resolve oscillations:
A =0 for Dms » w
A =1 for Dms = w
All w values for which
A sA < 1
are excluded
at 95% C.L.
sA -1 (w)  N½ fs (1-2h) F (sp,sL)
F (sp,sL)  exp(-st w )
For w  Dms, the shape A (w) depends on
the details of the analysis, and can be
calculated analytically** in simple cases
* Moser and Roussarie NIM A 384 (1997) 491
** Abbaneo and Boix JHEP 08 (1999) 004
D. Abbaneo

7. The three ALEPH analyses
Released as preliminary in now improved and submitted for publication
Initial state IP Kf
Bs candidate K
S.V.
Final state
Fully reconstructed
candidates
Ds - lepton
final states
Inclusive lepton
final states
“topological”
c meson l+ B n s p+ B
Ds- s l+ B n
Ds- s
Old Ds – hadron analysis dropped
D. Abbaneo

8. Fully reconstructed candidates
80 candidates (32 in the main peak)
11 with purity > 80%
23 with purity > 50%
All
B+ control sample
g, p0 recovery
D. Abbaneo

9. (D -l) candidates: the channels
D+K+K-p+
Over 300 candidates
Fraction of the resonant component estimated from the data
NN to discriminate resonant background
D. Abbaneo

10. Decay length for the inclusive l analysis
Basic idea:
Reconstruct inclusively a “D track” and fit it with the lepton
charm
lepton
Define a “B track” passing through the PV,
with angular uncertainties parameterized from the simulation
charm
photon
Look for photons around the D track with m(D,g)<1.8 GeV
Improve the D direction
charm
lepton
Fit the D track, the B track and the lepton
to find the B decay vertex
D. Abbaneo

11. Event-by-event treatment for the incl l analysis
Divide events in classes according to topological properties
[vertex classes]
m(D), cosq(l,D), ntr(D), c2(D), c2(B)
For each class:
Final state tag
Initial state tag
D. Abbaneo

12. Event-by-event treatment for the incl l analysis
Sample composition
Decay length uncertainty from the vertex fit [data]
Decay length bias correction and pull correction
Momentum uncertainty and bias correction
Parameterized as a function
of event properties
D. Abbaneo

13. Building a reliable analysis
Control reliability of parameterizations [MC]
Sample of Bs events
in simulated hadronic events
Simulated hadronic events
(1.6 x data stat)
Control reliability of simulation
Use control samples to check IS tag, decay length and momentum reconstruction
D. Abbaneo

14. The estimate of the systematic uncertainties
For each source considered, both a change of the
amplitude and of its statistical error are observed
Calculate the total error taking into account both effects
by means of toy experiments
D. Abbaneo

15. The three analysis results
Exclusive
Ds-lepton
Inclusive lepton
Obs. limit: ms>2.4 ps-1
Exp. limit: ms>0.3 ps-1
ms>7.2 ps-1
ms>7.4 ps-1 (was 6.6 ps-1)
ms>11.4 ps-1 (was 9.5 ps-1)
ms>14.0 ps-1 (was 9.8 ps-1)
2.1
2.5
1.2
D. Abbaneo

16. Fully inclusive final states
Good at SLD
Hard at LEP IP Kf
“topological”
c vertex
b vertex 
Final state tag based on charge flow between b and c vertices
B  D - s +
dq = |QD- QB||VD-VB|
 
Charge dipole
Needs excellent vertexing capability!
@ SLD:
sL = 72 mm (60%) & 265 mm
sp/p = 0.07 (60%) & 0.21
N = 8.5K decays
(4  inclusive lepton)
D. Abbaneo

17. SLD privileged environment for Bs oscillation hunt
SLD versus LEP
SLD privileged environment for Bs oscillation hunt
LEP SLD
4M qq events <400K qq events
per experiment
sL  250 mm (core) sL  mm (core)
Initial State Tag Initial State Tag
heff  25% heff  10-15%
(beam polarization)
Best analyses: Best analyses:
Ds lepton Inclusive lepton
Inclusive lepton Fully inclusive
D. Abbaneo

18. Amplitude error reduction
The world combination
Combined limit
Dms > % CL
Expected limit Dms = 19.3 ps-1
Now Summer99
@ 10 ps
@ 15 ps
@ 20 ps
Amplitude error reduction
D. Abbaneo

19. Comparison by experiment
Sensitivities
Weights in the combination
D. Abbaneo

20. Conclusions
Bs oscillation searches have substantially improved over the last two years
in the absence of new data!
(e.g. the uncertainty at w  20 ps-1 was reduced by a factor  2)
Some improvements can still be expected, but not as significant
We know that Dms > % CL
and we have a (mild) indication of a signal at w between 16 and 20 ps-1
The burden and the pleasure of continuing the hunt goes now to CDF and D0
D. Abbaneo