Transcript Slide 1

Tania Moulik
(Kansas University)
presented by
Andrei Nomerotski
(Fermilab/Oxford)
33rd International Conference in
High Energy Physics
(Jul 26th – Aug 2nd, Moscow, Russia)
1
B mixing
Mass eigenstates are a
mixture of flavor
eigenstates:
Dominant Diagram for the
transition :


 q B  p B 
BH  q B  p B
BL
BH and BL have a
different mass and may
have different decay
width.
Dm = MH – ML = 2|M12| ,
DG = GH - GL = 2|G12|
Time evolution follows the
Schrodinger equation
d  B( t)   M11  iΓ11
i
 


B
(
t)
dt 
  M 21  iΓ 21
M12  iΓ12  B( t) 


M 22  iΓ 22  B ( t) 
2
“Opposite sign”
In an Ideal Scenario..
Oscillations with amplitude = 1.0 and
Frequency = Dms.
“Same sign”
N OS  N SS
Ai (t ) 
N OS  N SS
3
DZero Detector
Spectrometer : Fiber and Silicon Trackers in 2 T Solenoid
Energy Flow : Fine segmentation liquid Ar Calorimeter and
Preshower
Muons : 3 layer system & absorber in Toroidal field
Hermetic : Excellent coverage of Tracking, Calorimeter and
Muon Systems
SMT H-disks
SMT F-disks SMT barrels
4
Analysis outline
X
Identify e/m.
PT (e/m) > 2.0
| h| (e/m) < 1.0/2.0
μ(e)
B
Signal Selection
μ+/e+
Bs0  Bs0
D-S
n
p-
φ
Look for tracks displaced from primary vertex in
same jet as m/electron
 Two tracks should form a vertex and be
consistent with f mass (fp  K Kp) or K*
mass (K*K  KKp)
 KKp invariant mass should be consistent
with Ds mass
K+
K-
5
Signal Selection
X
μ(e)
B
μ (e)+
B B
0
s
0
s
D-S
ν
πφ
K+
K-
Muons were selected by triggers without lifetime
bias
= no online/offline Impact Parameter cuts
Trigger muon can be used as tag muon : gives
access to eDs sample with enhanced tagging purity
6
Signal Selection
X
μ(e)
Eff=30%
B
Bs0  Bs0
μ+
PV
D-S
LT(DS)
ν
πφ
K-
K+
Ds lifetime is used to have non-zero selection
efficiency at Interaction Point
Bs can decay at IP and be reconstructed
7
Effect of Neutrino
Need to correct Decay Length
for relativistic contraction 
need to know Bs momentum
Can estimate Bs momentum from
MC (through so called k-factor)
at expense of additional
uncertainty
Dk/k uncertainty causes
additional smearing of
oscillations
Only few first periods are
useful for semileptonic channels
Sensitivity at DL=0 is crucial
All above represents the main
difference wrt hadronic
channels
200 micron
8
# of periods
Flavor Tagging and dilution calibration
Identify flavor of reconstructed BS candidate using
information from B decay in opposite hemisphere.
Ds
a) Lepton Tag :
Use semileptonic b decay :
Charge of electron/muon identifies b flavor
e / m
Bs
n
b) Secondary Vertex Tag :
m
Search for secondary vertex on opposite
Side and loop over tracks assoc. to SV.
cos f (l, Bs) < 0.8
c) Event charge Tag:
Secondary Vertex
All tracks opposide to rec. B
9
Dilution in Δmd measurement
Combine all tagging variables using likelihood ratios
Bd oscillation measurement with combined tagger
Dmd= 0.5010.030±0.016ps-1
Combined dilution: εD2=2.48±0.21±0.08 %
Input for Bs
measurement
10
Bs decay samples after flavor tagging
NBs( fp  m) = 5601 102
NBs(fp + e) = 1012  62
(Muon tagged)
NBs(K*K + m) = 2997  146
BsDs mn X
Ds fp
BsDs mn X
Ds  K*K
BsDs e n X
Ds fp
11
K*K Fit Components
Difficult mode due to K* natural width and mass
resolution – larger errors wrt fp mode
Ds  K *0 K  (signal)
D   Kpp or
D   K *0p ( K *0  K p  )
c  K p  P  (reflection)
D   K *0 K  ( K *0  K p  )
(Cabibbo suppressed)
12
Results ofKx the Lifetime Fit
p snos / osc ( x)

K
c Bs

e
c Bs
 0.5  1  D cos Δm s  Kx / c 
From a fit to signal and background region:
cBs (mm)
cbkg (mm)
BsDs e n X, Ds  fp
44429
64518
BsDs m n X, Ds  K*K
40722
54910
Decay Mode
BsDs m n X, Ds  fp
BsDs mn X
Ds  K*K
4049
6276
BsDs e n X
Ds fp
13
Amplitude Method
Asymmetry  cosDmS t 
Amplitude fit = Fourier analysis + Maximum likelihood fit
often used in oscillation measurements
A  D  cosDmst 
Need to know dilution (from Δmd analysis)
If A=1, the Δm’s is a measurement of Bs oscillation
frequency, otherwise A=0
14
Cross-check on BdXμD±(fp)
Amplitude Scan
DØ Run II Preliminary
EXACTLY the same sample & tagger
Amplitude Scan shows Bd oscillations
at correct place  no lifetime bias
with correct amplitude  correct dilution calibration
Same results for two other modes
15
Measure Resolution Using Data
Ultimately Dms sensitivity is limited by decay length
resolution – very important issue
Use J/ψ→μμ sample
Fit pull distribution for J/ψ Proper Decay Length with 2 Gaussians
Resolution Scale Factor is 1.0 for 72% of the events and 1.8 for the
rest
Cross-checked by several other methods
μ
DØ Run II Preliminary
J/ψ vertex
PV
μ
L±σL
16
Amplitude Scan of BsXμDs(fp)
Deviation of the amplitude at 19 ps-1
2.5σ from 0  1% probability
1.6σ from 1  10% probability
17
Log Likelihood Scan
In agreement with the amplitude scan
Systematic
 Resolution
 K-factor variation
 BR (BsmDsX)
 VPDL model
 BR (BsDsDs)
Have no sensitivity
above 22 ps-1
17 < Dms < 21 ps-1 @ 90% CL assuming Gaussian errors
Most probable value of Dms = 19 ps-1
18
Interpretation
Results of ensemble tests:
DZero result :
15%
0
80%
17 21
5%
Dms(ps-1)
Combined with World (before CDF measurement):
5%
0
90%
17 21
5%
Dms(ps-1)
19
Impact on the Unitarity Triangle
Before
BS mixing
20
Impact on the Unitarity Triangle
With D0
21
Impact on the Unitarity Triangle
With CDF
22
“Golden” Events for Visualization
DØ Run II Preliminary
Period of oscillations @ 19ps-1
Weigh events using
Signif  Fsig M fp , log10 y  D  e
# of periods

Dms  2

2
23
Can We See Bs Oscillations By Eye ?
Weighted asymmetry
This plot does not
represent full statistical
power of our data
# of periods
24
More Amplitude Scans
New results : Amplitude scans from two
additional modes
BsDs (fp e n X
Ds fp
BsDs mn X
Ds  K*K
25
Combination
Amplitude is centred at 1 now, smaller errors
Likelihood scan confirms 90% CL Dms limits: 17-21 ps-1
Data with randomized tagger : 8% probability to have a
fluctuation (5% before for mfp mode)
26
Detailed ensemble tests in progress
Outlook
Add Same Side Tagging
Add hadronic modes triggering on tag muon
Add more data (4-8 fb-1 in next 3 years)
with improved detector – additional layer of
silicon between beampipe and Silicon Tracker
(Layer0) – better impact parameter resolution
Layer0 has been successfully installed in April 2006
• S/N = 18:1 & no pickup noise
• First 50 pb-1 of data on tape, first tracks have been
reconstructed
27
Summary
Established upper and lower limits on Dms using
Bs  Ds fp mn X mode
Analysis published in PRL 97 (2006) 021802
Combined with two other channels
Bs  Ds K*K mn X
Bs  Ds fp en X
considerable improvement in sensitivity
14.1  16.5 ps-1, no improvement for Dms interval
Looking forward to a larger dataset with improved vertex
detection
If Dms is indeed below 19 ps-1 expect a robust
measurement with the extended dataset
28
BACKUP SLIDES
29
Bu+
B0
Bs0
Bc+
Matter
b
b
b
b
u
d
s
c
Anti-Matter
B Mesons
b
b
b
b
u
d
s
c
30
CKM matrix and B mixing
Why are we interested to study B meson oscillations
 d   Vud Vus Vub  d 
 
 
 s    Vcd Vcs Vcb  s 
 
 b   V
   td Vts Vtb  b 
 1  l2
l


l
1  l2
 3
2
A
l
(


i
h
)

A
l

*
ud ub
*
cd cb
Wolfenstein parametrisation
- expansion in l.
l  sin c  0.2265 0.002
Al3 (   ih ) 
029
 A  0.80100..018
Al2

2


(
1

l
2) 

1
 h  (1  l2 2)h
*
td tb
V V V V V V  0
complex
Vub | Vub | ei
Vtd | Vtd | ei
VudVub* VtdVtb*


1
*
*
VcdVcb VcdVcb
31
B Mixing
In general, probability for unmixed and mixed decays Pu,m(B)  Pu,m(B).
In limit, G12 << M12 (DG << DM) (Standard model estimate and confirmed
by data), the two are equal.
e t / 
p( B  B) 
(1  cos Dmt)
2
e t / 
p( B  B ) 
(1  cos Dmt)
2
~ 103
~ 10-4 for Bs system
~ 10-3 for Bd system
32
Constraing the CKM Matrix from Dms
2

m
2  t
Dmd 
m
m
F
2 b t  2
6p
 mW
GF2

2
2
*
hQCD B f B VtbVtd
Bd
d


CDF+D0 (2006) Dms inputs
And similar expression for
Dms
x2
2
Dms
M Bs f Bs
BBs Vts

2
Dmd M Bd f Bd
BBd Vtd
2
Vts  Vcb
x  1.24 0.040.06
from Lattice QCD calculations)
Ratio suffers from lower theoretical
Uncertainties – strong constraint Vtd
33
Excellent Tevatron Performance
Run II Integrated Luminosity
19 April 2002 - 22 February 2006
2.0
1.9
1.8
1.7
1.41
1.6
Luminosity (fb-1)
1.5
1.4
1.19
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
Delivered
0.3
0.2
Recorded
0.1
0.0
Apr-02
Jul-02
Oct-02
Jan-03
Apr-03
Jul-03
Oct-03
Jan-04
Apr-04
Jul-04
Oct-04
Jan-05
Apr-05
Jul-05
Oct-05
Jan-06
Data sample corresponding to over 1 fb-1 of the
integrated luminosity used for the Bs mixing analysis
Full dataset is ready (85-90% DAQ efficiency)
Apr-06
34
Jul-06
Muon Triggers
Limitation of data recording. Triggers are needed to select useful
physics decay modes. 396 ns bunch crossing rate ~ 2.5 MHz  ~50 Hz
for data to be recorded.
Single inclusive muon Trigger:
|η|<2.0, pT > 3,4,5 GeV
Muon + track match at Level 1
Prescaled or turned off depending on inst. lumi.
We have B physics triggers at all lumi’s
Extra tracks at medium lumi’s
Impact parameter requirements
Associated invariant mass
Track selections at Level 3
Dimuon Trigger : other muon for flavor
tagging
e.g. at 50·10-30 cm-2s-1, L3 trigger rate :
20 Hz of unbiased single μ
1.5 Hz of IP+μ
2 Hz of di-μ
No rate problem at L1/L2
35
μfp Sample
Opposite-side flavor
tagging
μD±: 7,422
μDs: 26,710
μD±: 1,519
Tagging efficiency 21.9±0.7%
μDs: 5,601±102
36
check Using BdXμD±(fp)
The Amplitude Scan shows Bd oscillations at 0.5 ps-1
no lifetime bias
(A=1) : correct dilution calibration
37
Detector Effects
flavor tagging power,
background
Decay length
momentum

resolution
resolution
p)/p = ? %
l = ?
1


SD
2
2
( Dms t ) 2

2
e
S
SB
SM prediction - Dms ~ 20 ps-1
Trying to measure :
Tosc~0.3 X 10-12 s !
38
Sample Composition
Estimate using MC simulation, PDG Br’s, Evtgen exclusive
Br’s
Signal: 85.6%
39
Flavor tag Dilution calibration
Bd mixing measurement using Bd  D* m n X, D* D0 p, D0 K
p, and evaluate dilution in various diution bins. Follows similar
analysis outline as Bs mixing.
Form measured asymmetry in 7 bins in visible proper decay
length (xM) – Count OS and SS events (compare charge of
reconstructed muon with tagger decision)
N OS  N SS
Ai ( x ) 
N OS  N SS
M
Fit the c2: c 2 (Dm, D )
7


i 1
( Ai  Aie (Dm, D )) 2
 2 ( Ai )
Also include B+ D0 m n X decay asymmetry.
40
Dilution calibration : Results
For final fit, bin the tag variable |d| in 5 bins and do a simultaneuos fit
c2 (i) where i=1,5. Parameters of the fit : Dm, fcc, 5 Dd, 5 Du = 12
Increasing dilution
B+
Increasing dilution
B0
Dm  0.506 0.020 stat.) ps-1
D2 = (2.48  0.21) (%) (stat.)
  19.9 0.2 % stat.
41
Individual Taggers performance
Tagger
 %
Muon
6.61  0.12
D (%)
D2 (%)
0.473  0.027
1.48  0.17(stat)
Electron 1.83  0.07 0.341  0.058
0.21  0.07 (stat)
SV
Total
OST
2.77  0.08 0.424  0.048
0.50  0.11 (stat)
2.19  0.22 (stat)
11.14  0.15
Note :
To evaluate the individual tagger performance |dpr| > 0.3
This cut was not imposed for final combined tagger.
Final eD2 is higher.
42
Likelihood minimization to get Dms
Minimize
f 
1  F

sig
candidates
fi  p
xM
xM , x
M
 2 ln f
f  F f 
i ,bg

, d pr p
 xM
sig
p
dpr
i , sig
p
Mf p
p
 log1 0 y
Form Probability Density Functions (PDF) for each source
Dilution Calibration
(From Dmd measurement)
Signal selection function (y)
dpr
43
Bs Signal and background
Signal PDF:

/ osc
M
nos / osc
p nos
(
x
,

,
d
)

dK
f
(
K
)

(
x
)
p
( x, d pr , K )  g ( x)
M
j
pr
j
j
M
s
x
Background PDF composed of long-lived and prompt
components – Evaluated from a lifetime fit.
Long Lived Background – Described by exponential
convoluted with a gaussian resolution function.
Non-sensitive to the tagging
Non-oscillating
Oscillating with Δmd frequency
Prompt Background – Gaussian distribution with
resolution as fit parameter.
44
Combined flavor tag algorithm
Combine individual tag informations to tag the event.
Get tag on opposite side and construct PDF’s for variables
discriminating b (m ) and b (m+) (Use B+  D0 m n X decays in data)
Discriminating variables (xi):
Electron/Muon
SV Tagger
more pure
more pure
45
Ensemble Tests
Using data
Simulate Δms=∞ by randomizing the sign of flavour
tagging
Probability to observe Δlog(L)>1.9 (as deep as ours) in
the range 16 < Δms < 22 ps-1 is 3.8%
5% using lower edge of syst. uncertainties band
Using MC
Probability to observe Δlog(L)>1.9 for the true Δms=19
ps-1 in the range 17 < Δms < 21 ps-1 is 15%
Many more parameterized MC cross-checks
performed – all consistent with above
46