Search for the rare decay!" → $%&&̅ at J-PARC KOTO experimentSatoshi Shinohara (Osaka Univ.),on behalf of the KOTO collaboration
KAON201910-13 September 2019 University of Perugia (Italy)
2019/9/10 KAON2019 1
Contents• Introduction• Results of 2015 physics run• 2016~2018 data analysis-Single event sensitivity-BG estimation
2019/9/10 KAON2019 2
!" → $%&&̅ decay• Direct CPV
• FCNC : highly suppressed decay-BR (SM) : 3�10-11
• Small theoretical uncertainty�~2%�→ Good probe for new physics search
2019/9/10
10-8
10-9
10-10
10-11
direct limit (KOTO 2015 result, Phys.Rev.Lett. 122 (2019) no.2, 021802)3.0�10-9 (90% C.L.)
SM prediction3.0�10-11
indirect limit (Grossman-Nir bound)1.5�10-9
&
&̅
Ws d
t
Z
KAON2019 3
!" → $%&&̅ decay• Direct CPV
• FCNC : highly suppressed decay-BR (SM) : 3�10-11
• Small theoretical uncertainty�~2%�→ Good probe for new physics search
2019/9/10
10-8
10-9
10-10
10-11
indirect limit (Grossman-Nir bound)1.5�10-9
&
&̅
Ws d
t
Z
A.J.Buras, et al. JHEP11(2014)121 M.Tanimoto, K.Yamamoto PTEP 2016 (2016) no.12, 123B02
NP ?
direct limit (KOTO 2015 result, Phys.Rev.Lett. 122 (2019) no.2, 021802)3.0�10-9 (90% C.L.)
KAON2019 4
KOTO experiment
2019/9/10
J-PARC @ Tokai village
KAON2019 5
Signal
2019/9/10
KL
π0->2γ
!!̅ (undetectable)
#$ → &'!!̅ : &' → 2) + nothing
CsI calorimeter(reconstruct &' from 2))
Hermetic veto detectors
Au target
30 GeV/c
proton
pencil beam
(KL, n, γ)
KAON2019 6
collimator +
sweeping magnet
Background
2019/9/10
KL
π0->2γ
!!̅ (undetectable)
ex.) #$ → 2'( → 4*
CsI calorimeter(reconstruct '( from 2*)
Hermetic veto detectors
30 GeV/cproton
pencil beam(KL, n, γ)
KAON2019 7
π0->2γAu target
collimator + sweeping magnet
Reconstruction
2019/9/10
KL
π0->2γ
!!̅decay region(vacuum)
π0 decay Z position
π0 Pt(transverse
momentum)
pencil beam(KL, n, γ)
KAON2019 8
Missing transverse momentum → finite π0 pt
CsI
signal region on the Pt vs. Z plane(blinded analysis)
CsI calorimeter
Dec2012
Dec2013
Dec2014
Jan2016
Dec2016
Dec2017
Dec2018
Acc
umul
ated
P.O
.T.
0102030405060708090
1001810×
Bea
m P
ower
(kW
)
0
20
40
60
80
100
History of data taking
2019/9/10
2013 data(PTEP 2017
021C01)
Operation
of Hadron
facility was
stopped.
2015 data
(Phys. Rev.
Lett. 122,
021802)
2016, 2017, 2018 data
KAON2019 9
2019 data
(covered in another talk)
Detector and DAQ upgrade in 2016-2018• Barrel detector was upgraded in 2016.4-13.5X0 → 18.5X0
-!" → 2%& BG : ×1/3
• DAQ upgrade (covered in poster session)
-Online cluster counting (2017~)-Higher DAQ live time ratio
• ~80% (2015 @ 42kW beam) → ~99% (2018 @ 51kW beam)
2019/9/10
!" → 2%&
Inner barrel (IB)
online clustering map
Energy deposit map of CsI
KAON2019 10
cluster
Result of 2015 physics run
2019/9/10 KAON2019 11
observed
expectation
contour : !" → $%&&̅ (MC)
Phys. Rev. Lett. 122, 021802
• Single event sensitivity :
1.30 ± 0.01-./. ± 0.14-1-. ×1034-No event in the signal region
• Upper limit (90% C.L.) :
Br(56 → 789:9) < 3.0�10-9
�10 improvement from
previous limit (KEK E391a)
Toward 2016-2018 data analysis
• To suppress hadron cluster BG is important-Need more understanding-Need more reduction
2019/9/10 KAON2019 12
BG source No. events!" → $%$&$' 0.05 � 0.02!" → 2$' 0.02 � 0.02Other !" decays 0.03 � 0.01Hadron cluster 0.24 � 0.17Upstream $' 0.04 � 0.03CV ) 0.04 � 0.02Total 0.42 � 0.18
Neutron related BG
KL BG
2015 run background summary table
neutron
Hadron cluster BG
CSI
Phys. Rev. Lett. 122, 021802
Toward 2016-2018 data analysis
• Higher beam power
→ larger signal loss due to the accidental hit
(76% loss in 2015 analysis)
• To achieve better sensitivity
→ Narrower veto window
-Need BG study due to the narrower veto window
2019/9/10 KAON2019 13
study from 2015 data
2018 June, 51kW
2016~2018 data analysisSingle event sensitivityBG analysis
2019/9/10 KAON2019 14
2016~2018 data analysisSingle event sensitivityBG analysis
2019/9/10 KAON2019 15
Single event sensitivity (S.E.S)
• S.E.S = (KL yield)�(signal acceptance)
2019/9/10 KAON2019 16
reconstructed KL mass distributionfrom !" → 2%& analysis
KL yield (@ beam exit) : 7.1�1012
(increase by a factor of 1.5 from 2015 data)
Preliminary
(covered in poster session)
Single event sensitivity (S.E.S)• S.E.S = (KL yield)�(signal acceptance)
• Signal box : determined by S/N• S.E.S : 6.9�10-10 (S.E.S2015 : 1.3�10-9)
2019/9/10 KAON2019 17
signal acceptance : 2.0�10-4
!" → $%&&̅ MC blind regionsignal region
2015 signal box
2016 ~ 2018 signal box
signal acceptance : + 6% (from 2015 signal box)
Preliminary
2016~2018 data analysisSingle event sensitivityBG analysis
2019/9/10 KAON2019 18
Hadron cluster BG
2019/9/10 KAON2019 19
n nn
Al target(10 mm thickness)
Physics run Special run
Physics data Control sample
n n
preliminary preliminary
Hadron cluster BG update from 2015 (statistics)
2019/9/10 KAON2019 20
0
50
100
150
200
250
300
RUNAll
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500
1163
3945 0
28375
821632010
82407(33753)
RUNAll
�~9
control sample (2015) control sample (2016~2018)
preliminary
The collection efficiency of control sample was improved by DAQ upgrade
Hadron cluster BG update from 2015 (cut)
2019/9/10 KAON2019 21
1. Cluster shape cut with deep learning-S/N :�~2
from 2015
2. Pulse shape discrimination with Fourier transformation-S/N :�~1.8
from 2015
Hadron cluster BG update from 2015 (estimation)
• Found contamination in the control sample-Developed more reliable estimation method-reduction : �~27 with the new method
2019/9/10 KAON2019 22
(10 mm thickness) (10 mm thickness)
Hadron cluster BG
2019/9/10 KAON2019 23
0.02 ±0.00
0.00 ±0.00
0.00 ±0.00
0.01 ±0.00
0.00 ±0.00
0.01 ±0.00
Preliminary
Extended signal region to downstream from this result
Upstream-π0 BG
• Halo-neutron hits upstream detector- γ + γ
• small visible energy shifted reconstructed Z position
-n + γ• various reconstructed Z position
-Shrink signal box of upstream Z• #BGupstream-π0 : 0.00 ± 0.00
2019/9/10 KAON2019 24
n
π0->2γ
CsI
Preliminary
CV-η BG
• Neutron hits Charged Veto (CV) and generates η- reconstruction :
assumed π0 mass-mη ~ 4mπ
0
→ reconstructed Z position : shifted in the signal region
#BGη = 0.03 ± 0.01
2019/9/10 KAON2019 25
η->2γn
CV (front)Z = 5862 [mm]
CSI Z = 6168 [mm]
CsI
Preliminary
Overlapped pulse BG • Overlapped pulse shifts measured time-Narrower veto window
• recover signal acceptance• increase BG from overlapped pulse
2019/9/10 KAON2019 26
clock [8ns]
example of overlapped waveform
puls
e he
ight
[a.u
.]
veto window
Overlapped pulse BG
2019/9/10 KAON2019 27
• Overlapped pulse discriminator-Pulse shape information : Fourier transformation- χ2 : compared with single hit template
Preliminary
w/o new methodw/ new method
χ2 distribution of upstream detector
Overlapped pulse BG• Narrower veto window + overlapped pulse discriminator
2019/9/10 KAON2019 28
Overlapped pulse BG from !" → 3%& MC analysis
Overlapped pulse BG from !" → %'( MC analysis
Narrower veto window : Recovered 10% signal acceptanceNo remaining event due to overlapped pulse
PreliminaryPreliminary
!" → $%$&$' BG• Neural net with deep leaning for !" → $%$&$' BG-Similar behavior as $' Pt and Z cut-S/N : improved even after applying Pt and Z cut which is
optimized !" → $%$&$' BG
2019/9/10 KAON2019 29
!" → $'((̅ efficiency !" → $%$&$' reductionPreliminary
Preliminary
Rec. π0 Z [mm] Rec. π0 Z [mm]
Rec
. π0
Pt [M
eV]
Rec
. π0
Pt [M
eV]
!" → $%$&$' BG• Applying neural net with deep leaning cut-#BG!" → $%$&$' : 0.02 → <0.02 (90% C.L.)- signal acceptance : 90%
2019/9/10 KAON2019 30
2.05 ±0.14
0.02 ±0.01
0.65 ±0.08(including blind region)
0.05 ±0.02
0.01 ±0.01
<0.02
before Neural net cut after Neural net cut
observedexpectation of !" → $%$&$' BG
PreliminaryPreliminary
!" → 2%&, !" → 2' BG
2019/9/10 KAON2019 31
Result from !" → 2%& MC study Result from !" → 2' MC study
PreliminaryPreliminary
BG summary
2019/9/10 KAON2019 32
0
1
2
3
4
5
6
7
&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
462 0 0
0
00473.1 ±4.2 0.08 ± 0.06 0.00 ± 0.00
0.18± 0.09
0.00± 0.00
0.07± 0.05
0.05 ± 0.02
#BGKLpi0pi0 <0.18
KLpi+pi-pi0 <0.02
KL3pi0 (overlapped pulse)
<0.04
Ke3 (overlapped pulse) <0.09
KL2gamma 0.00 � 0.00
Upstream π0 0.00�0.00
Hadron cluster 0.02 �0.00
CV-pi0 <0.10
CV-eta 0.03�0.01
Total 0.05±0.02
S.E.S : 6.9�10-10
observedexpectation
Preliminary
Preliminary
0.33 ± 0.11 (including signal region)
0
1
2
3
4
5
6
7
&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
2019/9/10 KAON2019 33
Preliminary
0
1
2
3
4
5
6
7
&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.922
2019/9/10 KAON2019 34
Preliminary
unblinded in the end of Aug. 2019
2019/9/10 KAON2019 35
0
1
2
3
4
5
6
7
&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.749
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.749
upstream detector
On-timing peak is shifted by large pulse
on-time
waveform of the event
measured time
0.08 ± 0.06
0.18± 0.09
473.1 ±4.2
0.00± 0.00
0.07± 0.05
0.05�0.02
expectation
Prelimin
ary
Preliminary
clock [8ns]AD
C [c
ount
s]
0.33�0.11
2019/9/10 KAON2019 36
0
1
2
3
4
5
6
7
&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.749
Pi0RecZ2000 3000 4000 5000 6000
Pi0Pt
050100150200250300350400450500&&MinCSDDLVal>0.985&&MaxThetaChisq<4.5&&MinWFourierLH>0.5&&EtaCSD>0.91&&KLpipipi0DLVal>0.749
#BGKL3pi0 (overlapped pulse) <0.04
Ke3 (overlapped pulse) <0.09
BG estimation related overlapped pulse
• Underestimated the BG from overlapped pulse?
• Checking the properties of the other candidates
• Did we miss other background sources?• planning to reevaluate
other BG sources
0.08 ± 0.06
0.18± 0.09
0.00± 0.00
0.07± 0.05
expectation
Preliminary
Prelimin
ary
0.05�0.02
0.33�0.11
473.1 ±4.2
Summary• !" → $%&&̅ at J-PARC KOTO experiment
• S.E.S : 6.9�10-10
• BG estimation : 0.05 ± 0.02
• Opened the box
→ 4 candidates in the signal region
Event properties:
One event : overlapped waveform
Checking other candidates carefully
2019/9/10 KAON2019 37
END
2019/9/10 KAON2019 38