Radiative & Electroweak Penguin B decays
at Belle
Yutaro Sato
Nagoya University
(Belle Collaboration)
19, Jul. 2013
Flavor Changing Neutral Current (FCNC) process
• Forbidden at tree level and allowed via loop diagrams in the SM.
• New particles may enter in the loop and alter physical observables.
Introduction 2
b
l
s l
W
t
g/Z
t
W W b
l
s
l
n
b s W
t
g
Radiative penguin decay
( bsg ) Electroweak penguin decay
( bsl+l- )
t ~
c ~ b s b s t
H-
Contents
Radiative penguin B decay (bsg)
• Search for B0p𝛬 p-g
Electroweak penguin B decay (bsl+l-)
• Forward-backward Asymmetry in inclusive BXs l+l-
• Both (preliminary) results are based on the full U(4S) data sample at Belle
and shown for the first time.
3
Time [year]
inte
grat
e lu
min
oti
sy [
fb-1
]
U(4S), 711 fb-1,
U(5S)
U(nS), n=1,2,3
Off-reso. Scan
Belle
B0p𝜦 p-g
• In B+pLg decay, mass peak of the pL system near threshold has been observed.
– Br (B+p𝛬 g) = 2.45−0.38
+0.44 ± 0.22 × 10−6
M.-Z. Wang et al. (Belle Collaboration), Phys. Rev. D. 76, 052004 (2007).
• Following hierarchy has been observed
in bs and bc baryonic transition.
– Br (B+p𝛬 p+p-) > Br (B0
p𝛬 p-) > Br (B+p𝛬 )
– Br (B+p𝛬 cp
+p-) > Br (B0p𝛬 cp
-) > Br (B+p𝛬 c)
• To check the hierarchy in the bsg transition, B0p𝛬 p-g is searched.
– It is also useful to tune the parameters in JETSET (hadronization program).
4
Theory prediction
B+p𝜦 g
Analysis method
• Main backgrounds : continuum (e+e-qq; q=u, d, s, c) events
– Likelihood ratio is constructed from event topology variables (KSFW)
and B flight direction(cosqB).
• Signals are extracted from 2D fit of Mbc and DE.
–
–
5
: beam energy @ c.m. frame
: 4-momentum of reconstructed B @ c.m. frame
ee BB continuum
「spherical」 「jet-like」
Result of B0p𝜦 p-g
• Nsig = 9.5−10.7+9.5 (statistical significance = 0.9)
– Dominant systematic source : experimentally unmeasured rare B decays
• Upper limit (90% C.L.) : Br (B0pΛ p-g ) < 6.5×10-7
– < Br (B+p𝛬 g) = 2.45−0.38
+0.44 ± 0.22 × 10−6
The expected hierarchy is not observed.
6
Total Signal
Self-cross feed Combinatorial BG
Rare B BG (BpLg ) Rare B BG (BpLp0 )
B0p𝛬 r-
B0p𝛴 0r-
B0p𝛬 p-h
B+p𝛴 0p0
B+p𝛬 h
Select events in DE [-0.4 GeV, +0.3 GeV]
Select events in Mbc [5.24 GeV, 5.29 GeV]
Forward-backward asymmetry in inclusive BXsl+l-
The bsl+l- has rich set of observables.
Forward-backward Asymmetry (AFB)
• very sensitive to the new physics
• Sensitive to three Wilson coefficients (C7, C9, C10).
Inclusive measurement
• Less theoretical uncertainty
than exclusive BK(*) l+l-.
• Experimental error is comparable with
theoretical uncertainty in exclusive
measurements.
The inclusive measurement would be
more important.
7
@ dilepton rest frame
b l+
l- s
q
Forward event b l+
l- s
q
Backward event
dilepton invariant mass squared
AFB in exclusive B K*l+l-
• The semi-inclusive reconstruction (sum-of-exclusive) method is utilized
for inclusive bsl+l-.
– Xs l+l- is reconstructed from 36(=18×2) exclusive modes.
– 20(=10×2) exclusive modes are used for the measurement of AFB.
• For (anti-)B0, only the self-tagging modes with a K± are used.
• K4p states are not used due to low expected signal yields.
• The final states not used for the measurement of AFB are removed
after the best candidate selection to suppress cross-feed.
The fraction of all Xs decays covered by 20 final states is 50 %.
Semi-inclusive reconstruction method 8
Xs = K±/KS + up to four p (at most one p0)
[ K ] : K, KS
[ Kp ] : Kp, KSp, Kp0, KSp0
[ K2p ] : K2p, KS2p, Kpp0, KSpp0
[ K3p ] : K3p, KS3p, K2pp0, KS2pp0
[ K4p ] : K4p, KS4p, K3pp0, KS3pp0
l+l- = e+e- or m+m-
Analysis method
Backgrounds
• NeuroBayes (NB) neural network is employed for backgrounds suppression.
– Semileptonic B decays
• lepton vertex separation
• missing mass
• visible energy
– Continuum events
• event topology variables
NB inputs : 23 variables in total
• Three peaking Backgrounds are considered in the final fit.
9
b b
l-
nl
X
l+
nl
X’
_ _
b
c
l-
l+
nl
X
nl
_
p K
l l
p p
2. Double miss-PID
Miss-PID
1. charmonium
p K l l
Escape veto selection
J/y, y(2S)
3. Swapped miss-PID
p K l
l p
l
Miss-PID
J/y, y(2S)
Analysis method
Signal extraction
• Divide q2 into 4 bins
• Extract AFB from simultaneous fit of Mbc with
forward/backward events in electron/muon channels.
– Dependence of reconstruction efficiency to the q2 and cosq is taken into account.
– AFBraw is transformed to AFB by the linear scaling factor.
10
AFB = a ・ AFBraw
Rec. eff. Xs m+m-
1st
2n
d
3rd
4th
J/y
vet
o
y(2
S)
vet
o
Xs e+e-
1st
2n
d
3rd
4th
J/y
vet
o
y(2
S)
vet
o
• Dominant systematic sources
– transformation from AFBraw to AFB
– estimation of peaking backgrounds
Fit result in 2nd q2 bin 11
Total Signal
Cross-feed Non-peaking B.G.
Peaking B.G.
Xs e+e- , Forward
Xs m+m-, Forward
Xs e+e- , Backward
Xs m+m-, Backward
Nsigee = 30.1±9.2 (stat)
Nsigmm = 23.9±10.5 (stat)
AFB = 0.04±0.31(stat) ±0.05(syst)
Results
• Results are consistent with the SM.
– Low q2 region ( 1st bin ) : Consistent with the SM within 1.8s (6.6 C.L.).
– High q2 region (3rd and 4th bin) : Exclude C10*C9 > 0 with 2.3s (97.9% C.L.)
Result of AFB
12
● Data ○ SM (bsll )
Total signal yields Nsig
ee = 139.9 ± 18.6 (stat) Nsig
mm = 160.8 ± 20.0 (stat)
Summary
• Radiative and electroweak penguin B decays are sensitive for physics
beyond SM.
Search for B0p𝜦 p-g
– Upper limit (90% C.L.) : Br (B0pLp-g ) < 6.5×10-7
Forward-backward asymmetry in inclusive BXs l+l-
– low q2 region : consistent with the SM within 1.8s (6.6% C.L.)
– high q2 region : exclude C10*C9 > 0 with 2.3s (97.9% C.L.)
Can be used to constrain new physics and will play important role at
precise measurement of Wilson coefficients.
13
Backup Slides
14
Operator Product Expansion
15
Current-current
QCD penguin
Magnetic operator
Vector electroweak penguin
Axial-vector electroweak penguin
Correction function
Procedure for making the correction function
1. Generate MC samples,
varying three Wilson coefficients(C7,C9,C10).
2. Calculate the AFBraw using rec. eff..
16
Parameter Range
[ A7 ] +A7 SM, -A7
SM
[A9, A10] -200% ~ +200 %
(Ai is leading term of Ci.)
Xs m
+ m-
Xs e
+ e-
1st q2 bin 2nd q2 bin 4th q2 bin 6th q2 bin
Ideal line (AFB = AFBraw) Fitted line
AFBraw
AF
B
B.G. Suppression with Neural Network
• Neural network is employed for the background suppression
17
“Function” is determined to efficiently
separate signal and background
by the “training”.
Outputs
+1 : signal-like
-1 :background-like
23 parameters are input. 1. Distance between dilepton along z axis
2. Confidence level of reconstructed B-vertex
3. Flight direction of B meson
4. Energy difference DE
5. Missing mass
6. Visible energy
7. 17 event shape parameters based on Legendre polynomials
Fit result in 1st q2 bin 18
Total Signal
Cross-feed Non-peaking B.G.
Peaking B.G.
Xs e+e- , Forward Xs e
+e- , Backward
Xs m+m-, Forward Xs m
+m-, Backward
Nsigee = 45.7±10.9 (stat)
Nsigmm = 43.4±9.2 (stat)
AFB = 0.34±0.24(stat) ±0.02(syst)
Results
Fit result in 3rd q2 bin 19
Total Signal
Cross-feed Non-peaking B.G.
Peaking B.G.
Xs e+e- , Forward Xs e
+e- , Backward
Xs m+m-, Forward Xs m
+m-, Backward
Nsigee = 25.0±7.0 (stat)
Nsigmm = 30.7±9.9 (stat)
AFB = 0.28±0.21(stat) ±0.01(syst)
Results
Fit result in 4th q2 bin 20
Total Signal
Cross-feed Non-peaking B.G.
Peaking B.G.
Xs e+e- , Forward Xs e
+e- , Backward
Xs m+m-, Forward Xs m
+m-, Backward
Nsigee = 39.2±9.6 (stat)
Nsigmm = 62.9±10.4 (stat)
AFB = 0.28±0.15(stat) ±0.01(syst)
Results