φ3(γ) measurement by B0→[KS0π+π−]D K*0 at Belle
Kentaro Negishi 19th February 2015 Lake Louise
1
Lake Louise Winter Ins/tute
Introduc?on
• CKM (Cabibbo-‐Kobayashi-‐Maskawa) matrix – The quark mixing matrix, which is unitary.
19th Feb. '15 LLWI’15 K.Negishi
2
• Unitary triangle
!2/"
!1/#!3/$
|VudV!ub|
|VcdV!cb|
|VtdV!tb|
|VcdV!cb|
Complex phase
(0,0) (1,0)
(ρ, η) _ _
Introduc?on
• CKM (Cabibbo-‐Kobayashi-‐Maskawa) matrix – The quark mixing matrix, which is unitary.
19th Feb. '15 LLWI’15 K.Negishi
3
3q
2q
2q
dm6
K¡
K¡sm6 & dm6
ubV
1qsin 2
(excl. at CL > 0.95) < 0
1qsol. w/ cos 2
2q
1q
3q
l-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
d
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7ex
clud
ed a
rea
has
CL
> 0.
95
Summer 14
CKMf i t t e r
• Unitary triangle
Complex phase
CKMfiCer 2014 Summer
φ3 Measurement
B0 → [ f ]D K*0
19th Feb. '15 LLWI’15 K.Negishi
4
– Access φ3 with interference D0K*0 and D0K*0 decays. Weak Int. phase Strong Int. phase Amp.
Difference between D0K*0 and D0K*0 φ3 δS rS
W!
b
d d
B0sK"0
u
cD0Vub
W!
b
d d
B0sK"0
c
uD0→ f → f
(No penguin)
rS is crucial parameter in φ3 measurement. (Expected to be ~0.3.)
_ _
_ _ _
Ø B flavor can be decided by the charge of K from K*0. Br(K*0 → K+π−) = 2/3
_
Dalitz Analysis Method
• Measure B0/B0 asymmetry across Dalitz plot. – D is required to decay in to three body like KS0π+π−.
19th Feb. '15 LLWI’15 K.Negishi
5
• Sensi/vity to φ3 in interference term. – |f(m2
+, m2−)| from flavor-‐tagged D*+→D0π+ events.
– Phase difference(δD) between D0/D0 from Charm-‐Factory.
_
_
D0 _
D0
←
B
m+2 m+
2
m−2
m−2
Model-‐Independent Dalitz 19th Feb. '15 LLWI’15 K.Negishi
6
[A. Giri, Y. Grossman, A. Soffer, J. Zupan, PRD 68, 054018 (2003)]
where observables
i
-i
Number of events in D0-‐plot : Ki
Number of events in B-‐plot :
From Charm-‐Factory
Belle Experiments KEKB accelerator
• Asymmetric energy collision – (8.0 v.s. 3.5 GeV)
• 10.58 GeV center of mass energy at Υ(4S) resonance; It is suitable for BB produc/on.
• 772 × 106 BB pair
• Charged par/cle momentum (σpt/pt(%) = 0.19pt + 0.30β)
• Good par/cle iden/fica/on ((K/π) Eff. ~90%, Fake ~10%)
• Good vertex resolu/on (~50 µm)
19th Feb. '15 LLWI’15 K.Negishi
7
Belle detector
○
Signal and Backgrounds 19th Feb. '15 LLWI’15 K.Negishi
8
Signal
• Light quark jets (u,d,s,c). • Random mis-‐reconstruc/on
makes fake signal candidate.
• Other B decay modes. • Difficult to dis/nguish from signal.
• D true BB BG • D fake BB BG
• 1 mis-‐PID π as K • D0ρ0
• 1mis-‐PID and 1 lost π • D0a1+
Rejec/on Use decay shape difference. Signal : spherical decay qq BG : 2 jet-‐like decay Neural network is used for mul/ variable analysis.
qq BG
BB BG
Peaking BG
B0 K*0
D
K+
KS π−
π− π+
(B flavor specific)
resonant π+
π−
e+e− → qq q = u,d,s,c
_
BB event qq event
3D Fit for Signal Extrac?on 19th Feb. '15 LLWI’15 K.Negishi
9
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
0
2
4
6
8
10
12
14
16
18
E (GeV)"6A RooPlot of " E (GeV)"6A RooPlot of "
TRANSNB-10 -8 -6 -4 -2 0 2 4 6 8 10
Even
ts /
( 0.5
)
0
10
20
30
40
50
60
"TRANSA RooPlot of "NB "TRANSA RooPlot of "NB
bcM5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29
Even
ts /
( 0.0
02 )
0
5
10
15
20
25
"bcA RooPlot of "M "bcA RooPlot of "M
C’NB
Energy difference btw. beam energy and B candidate.
Modified distribu/on of Neural network output used qq suppression.
Mass of B candidate from beam energy and B’s momentum.
BB like qq like
[GeV] [GeV]
Ayer reconstruc/on and BG rejec/on, 3-‐D fit (ΔE, C’NB, Mbc) is done without Dalitz informa/on. Each component yield is free. Shapes are fixed.
Yield is Ntotal = 44.2 (sta/s/c significance 2.8 σ), which are used for the (x,y) fit.
+ 13.3 − 12.1
Red : Signal Green : D0a1+ Blue : D fake BB Light blue : D true BB Magenta : qq
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
0
2
4
6
8
10
12
14
16
18
E (GeV)"6A RooPlot of " E (GeV)"6A RooPlot of "
TRANSNB-10 -8 -6 -4 -2 0 2 4 6 8 10
Even
ts /
( 0.5
)
0
10
20
30
40
50
60
"TRANSA RooPlot of "NB "TRANSA RooPlot of "NB
bcM5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29
Even
ts /
( 0.0
02 )
0
5
10
15
20
25
"bcA RooPlot of "M "bcA RooPlot of "M
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
0
2
4
6
8
10
12
14
16
18
20
E (GeV)"6A RooPlot of " E (GeV)"6A RooPlot of "
TRANSNB-10 -8 -6 -4 -2 0 2 4 6 8 10
Even
ts /
( 0.5
)
0
10
20
30
40
50
60
70
"TRANSA RooPlot of "NB "TRANSA RooPlot of "NB
bcM5.21 5.22 5.23 5.24 5.25 5.26 5.27 5.28 5.29
Even
ts /
( 0.0
02 )
0
5
10
15
20
25
"bcA RooPlot of "M "bcA RooPlot of "M
0
1
2
3
4
5
6
7
8
2+m0.5 1 1.5 2 2.5 3
2 -m
0.5
1
1.5
2
2.5
3
Phase binsPhase bins
From CLEO
(x,y) Fit 19th Feb. '15 LLWI’15 K.Negishi
10
…
N1.
N2.
Ki from D*+→D0π+, D0 decay
y
x
(x!, y!)
(x+, y+)
!3!3
"
rS
rS
PRD 82, 112006 (2010)
Obtain the signal number,
Obtain the signal number,
(x, y) Result BG frac/ons btw bins for each component are fixed from MC.
19th Feb. '15 LLWI’15 K.Negishi
11
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 1.4iN0Bbin -8
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
1
2
3
4
5
6
= 1.9iN0Bbin -7
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.5
1
1.5
2
2.5
3 = 0.5iN
0Bbin -6
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 1.0iN0Bbin -5
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 0.6iN0Bbin -4
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 0.3iN0Bbin -3
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.5
1
1.5
2
2.5
3 = 0.6iN
0Bbin -2
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 0.5iN0Bbin -1
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 2.0iN0Bbin +1
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 2.7iN0Bbin +2
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 2.9iN0Bbin +3
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
1
2
3
4
5
6
7
= 1.4iN0Bbin +4
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.51
1.52
2.53
3.54
4.5
= 0.5iN0Bbin +5
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.5
1
1.5
2
2.5
3 = 1.0iN
0Bbin +6
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
1
2
3
4
5
6
7
= 2.8iN0Bbin +7
E (GeV)6-0.1 0 0.1 0.2 0.3
Even
ts /
( 0.0
1333
33 )
0.5
1
1.5
2
2.5
3 = 0.9iN
0Bbin +8
x− = + 0.4 ± 0.0 y− = − 0.6 ± 0.1 x+ = + 0.1 ± 0.1 y+ = + 0.3 ± 0.1
+ 1.0 + 0.0 − 0.6 − 0.1 + 0.8 + 0.1 − 1.0 − 0.0 + 0.7 + 0.0 − 0.4 − 0.1 + 0.5 + 0.0 − 0.8 − 0.1
stat. syst. ci, si
(x, y) Result 19th Feb. '15 LLWI’15 K.Negishi
12
X-2 -1.5 -1 -0.5 0 0.5 1 1.5 2Y-2
-1.5-1
-0.50
0.51
1.52
CL
CO
MB.
0
0.2
0.4
0.6
0.8
1
X-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Y
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
x− = + 0.4 ± 0.0 y− = − 0.6 ± 0.1 x+ = + 0.1 ± 0.1 y+ = + 0.3 ± 0.1
+ 1.0 + 0.0 − 0.6 − 0.1 + 0.8 + 0.1 − 1.0 − 0.0 + 0.7 + 0.0 − 0.4 − 0.1 + 0.5 + 0.0 − 0.8 − 0.1
stat. syst. ci, si
B0
_
B0
Dot : Most probable point Line : 1σ contour
(x, y) Result 19th Feb. '15 LLWI’15 K.Negishi
13
X-2 -1.5 -1 -0.5 0 0.5 1 1.5 2Y-2
-1.5-1
-0.50
0.51
1.52
CL
CO
MB.
0
0.2
0.4
0.6
0.8
1
X-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Y
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
δS 2φ3
x− = + 0.4 ± 0.0 y− = − 0.6 ± 0.1 x+ = + 0.1 ± 0.1 y+ = + 0.3 ± 0.1
+ 1.0 + 0.0 − 0.6 − 0.1 + 0.8 + 0.1 − 1.0 − 0.0 + 0.7 + 0.0 − 0.4 − 0.1 + 0.5 + 0.0 − 0.8 − 0.1
stat. syst. ci, si
B0
_
B0
Dot : Most probable point Line : 1σ contour
(x+, y+) (B0) is 0 consistent.
Sr0 0.5 1 1.5 2
1 - C
.L.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
rS Result • rS is crucial parameter in φ3 measurement.
– φ3 uncertainty is scaled as 1/r.
19th Feb. '15 LLWI’15 K.Negishi
14
rS < 0.87 @ 68 % C.L. rD2 = (3.79±0.18)10-‐3 rD is small.
rS < 0.4 → RDK* ~ rS2
B0→[Kπ]DK*0 PRD 86 , 011101 (2012)
Blue : B0 (x−, y−) Red : B0 (x+, y+) Black : Combine
_
rS
Conclusion
• New result of B0→[KS0π+π−]DK*0 Mod.-‐Ind. Dalitz analysis.
19th Feb. '15 LLWI’15 K.Negishi
15
x− = + 0.4 ± 0.0 y− = − 0.6 ± 0.1 x+ = + 0.1 ± 0.1 y+ = + 0.3 ± 0.1
+ 1.0 + 0.0 − 0.6 − 0.1 + 0.8 + 0.1 − 1.0 − 0.0 + 0.7 + 0.0 − 0.4 − 0.1 + 0.5 + 0.0 − 0.8 − 0.1
rS < 0.87 at 68 % C.L.
φ3 measurement with neutral B is promising for Belle II!!
stat. syst. ci, si
BACK UP
19th Feb. '15 LLWI’15 K.Negishi
16
Event Selec?on
• Primary track – IP |Δr| < 5 mm, |Δz| < 5 cm
• KS reconstruc/on – NIS KS Finder
• D0 reconstruc/on – KS and opposite charge 2 π(LR(K/π) < 0.6) – |MKsππ − mD0| < 0.015 GeV
• K*0 reconstruc/on – Opposite charge K(LR(K/π) > 0.7) and π(LR(K/π) < 0.6) – |MKπ − mK*0| < 0.050 GeV
• B0 reconstruc/on – Best candidate is selected by D0 mass and B0 vertex – Δm > 0.15 GeV for real D0 BG from D*±
– |MK*0π− − mD0| > 0.04 GeV for [K+π−π−]D−[KSπ+]K*+
19th Feb. '15 LLWI’15 K.Negishi
17
qq suppression
• qq events are suppressed by using following 12 parameters as NeuroBayes inputs.
19th Feb. '15 LLWI’15 K.Negishi
18
1. LR(KSFW) 2. cosθthr 3. Δz 4. dist. DK* 5. |qr| 6. |cosθB| 7. cosθD
B 8. v1_z 9. v1_v1 10. v2_v2 11. v3_v3 12. thru_oth
NB output-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Nor
mal
ized
num
ber o
f eve
nts
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
NB output-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Nor
mal
ized
num
ber o
f eve
nts
-310
-210
-110 Red : Signal Blue : qq BG
Signal efficiency0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Bac
kgro
und
reje
ctio
n
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NB-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1
Probability
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
TransNB-10 -8 -6 -4 -2 0 2 4 6 8 10
Probability
00.010.020.030.040.050.060.070.080.090.1
NBlow = − 0.6 NBhigh = 0.9992
Red : Signal Blue : qq Green : BB
PDF 19th Feb. '15 LLWI’15 K.Negishi
19
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
0
10000
20000
30000
40000
50000 E Projection6 /ndf = 91.862r E Projection6
TRANSNB-10 -5 0 5 10
Even
ts /
( 0.5
)0
2000400060008000
1000012000140001600018000
ProjectionTRANSNB /ndf = 6.342r ProjectionTRANSNB
bcM5.22 5.24 5.26 5.28
Even
ts /
( 0.0
02 )
0
10000
20000
30000
40000
50000
60000
ProjectionbcM /ndf = 19.202r ProjectionbcM
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
0200400600800
10001200140016001800200022002400
E Projection6 /ndf = 1.682r E Projection6
TRANSNB-10 -5 0 5 10
Even
ts /
( 0.5
)
0
2000
4000
6000
8000
10000
ProjectionTRANSNB /ndf = 17.422r ProjectionTRANSNB
bcM5.22 5.24 5.26 5.28
Even
ts /
( 0.0
02 )
0
500
1000
1500
2000
2500
ProjectionbcM /ndf = 3.882r ProjectionbcM
E (GeV)6-0.1 -0.05 0 0.05 0.1 0.15
Even
ts /
( 0.0
0625
)
050
100150200250300350400450
E Projection6 /ndf = 1.722r E Projection6
TRANSNB-10 -5 0 5 10
Even
ts /
( 0.5
)
0
200
400
600
800
1000
1200
ProjectionTRANSNB /ndf = 11.672r ProjectionTRANSNB
bcM5.22 5.24 5.26 5.28
Even
ts /
( 0.0
02 )
0
50
100
150
200
250
300
350
400
ProjectionbcM /ndf = 1.482r ProjectionbcM
MCから 分布の形状を得る
ピーキング背景事象
MC MC MC
MC MC MC
MC MC MC
• シグナルは3次元(ΔE, NB’, Mbc)の分布をフィットして得る
• 同時に(コンティニュアム, BB, ピーキング)背景事象もフィットする コン
ティ
ニュ
アム
BB
イベ
ント
ΔE ΝΒ’ Mbc
ΔE ΝΒ’ Mbc
ΔE ΝΒ’ Mbc
MCから 分布の形状を得る
Dが真 Dが真
Dが偽 Dが偽
_
_ + ピ
ーキ
ング
Dπ control sample analysis 19th Feb. '15 LLWI’15 K.Negishi
20
• We obtain Dπ (x,y) consistent with previous result.
Error contours are stat. err. only. (1, 2 and 3 σ)
x− = − 0.0130 ± 0.0077 y− = + 0.0018 ± 0.0076 x+ = − 0.0169 ± 0.0083 y+ = + 0.0225 ± 0.0076
x− = − 0.0045 ± 0.0087 ± 0.0056 y− = − 0.0231 ± 0.0107 ± 0.0077 x+ = − 0.0172 ± 0.0089 ± 0.0065 y+ = + 0.0129 ± 0.0103 ± 0.0088
• Anton (Previous study, 605 �-‐1)
• KS selection • qq suppression • D0 mass selection • BGs Dalitz distributions • Cross-feed between bins • Efficiency correction.
Difference of my and previous study
• To check (x,y) fit, we use B+→Dπ+ as control sample. • B+→Dπ+ is also analyzed as control sample of
B+→DK+ (PRD 85,112014(2012))
Sta?s?cal uncertainty 19th Feb. '15 LLWI’15 K.Negishi
21
• We decide to not use normal error of likelihood distribu/on on (x,y) because of unreliable of (x,y) likelihood due to small sta/s/cs.
• To obtain sta/s/c uncertainty, we use Feldman-‐Cousin method.
Feldman-‐Cousin method 1. Generate >20,000 (x,y) fit result with toy MC at 1600 points. 2. Dist. of (x,y)obt. fit results at (x,y)true space are obtained.
( PDF(xobt.,yobt.|xtrue,ytrue) ) 3. We define confidence level as integral of the PDF
in a region Ω which sa/sfy PDF(x,y) > PDF(xmeas.,ymeas.). ( We call it “CL(xtrue,ytrue)”. )
4. Draw contours of e.g.) α = 0.393 (1σ), 0.865 (2σ) and so on.
P(xobt., yobt.)
CL(xtrue, ytrue) Fit result = (xmeas., ymeas.)
(xtrue, ytrue)
Ω
(x(-2., 2.), y(-2., 2.)) 0.1 step
(BN#564 appendix D)
Total signal number = 44.2 +13.3 −12.1
Statistic significance = 2.8 σ
Systema?c uncertainty 19th Feb. '15 LLWI’15 K.Negishi
22
We combine the uncertainty from stat. and syst. with assump/on of (x,y) 2D Gauss. for syst. err.
• Δx− = ± 0.0 • Δy− = ± 0.1
• Δx+ = ± 0.1 • Δy+ = ± 0.1
+ 0.0 − 0.1 + 0.1 − 0.0
+ 0.0 − 0.1 + 0.0 − 0.1
w/o ci,si ci,si
Discussion 19th Feb. '15 LLWI’15 K.Negishi
23
• rS は0と無矛盾 – B0→DK*0シグナル数が小さかった 44.2 ���
崩壊分岐比で Br(B0→DK*0) = (2.9 ± 0.9)×10−5
– 統計的なふらつきによる
• Belle II 実験(予定)では
+ 13.3 − 12.1
イベント数 Br(B0→DK*0) ずれ
本結果 44.2 (2.9 ± 0.9)×10-‐5
BaBar 78 (5.2 ± 1.2)×10-‐5 −1.5σ
PDG 64 (4.2 ± 0.6)×10-‐5 −1.2σ
(統計誤差が支配的)
ただし”ずれ”は 大きくない
x− = + 0.4 y− = − 0.6 ± 0.1 x+ = + 0.1 ± 0.1 y+ = + 0.3 ± 0.1
+ 1.0 + 0.0 − 0.6 − 0.1 + 0.8 − 1.0 + 0.7 − 0.4 + 0.5 − 0.8
統計 系統
50倍BB 統計誤差→ O(<0.1)
現系統誤差と同等
1. K/π識別能力が上がる →BB背景事象の抑制
2. Super-‐Charm-‐Factory →ci, siの誤差が減る
_
B0→DK*0崩壊を用いφ3測定の可能性
_
Belle II + Super-‐Charm-‐Factory Δ(x,y)stat. = ± 0.1 Δ(x,y)syst. = ± 0.1