Study of Halo ! " $% Background in the KOTO experiment Yuya - - PowerPoint PPT Presentation

Study of Halo !" → $% Background in the KOTO experiment

Yuya Noichi

1

Year End Presentation 2019

2019/12/23 2019 Year End Presentation

2

KOTO Experiment

• Search for !" → $%&'

& decay

() undetectable

2+Nothing (4, veto)

KL

CsI Calorimeter

& ' &

veto counter

)

Proton Au Target Signal

2019/12/23 2019 Year End Presentation

3

Halo !" → $% Background

• PT of Halo KL must be measured correctly because it

may cause Halo &' → 2) background. ) ) CsI )

KL

) CsI * + +

KL

!" → ,-.* . halo !" → $%

• If beam halo KL decay for 2), it may be confused with

&' → /01* 1 signal.

2019/12/23 2019 Year End Presentation

z z

4

COE(Center of Energy)

• COE can be calculated from CsI information; Hit position and energy
• f each gamma.

!"#$ = ∑(()*!+ ,+) ∑ ,+ ."#$ = ∑(()*.

+ ,+)

∑ ,+

COE

/

012

z 34

• This can be indicator of halo KL because COE represents the arrival

point of KL for the case angle between PT and PZ is small.

2019/12/23 2019 Year End Presentation

5

Motivation

• Conventional reconstruction method assumes mass of each

2gamma is !"# and vertex position is on z axis.

• Transverse position of KL was determined by the interpolation

between target and COE.

• However, assumption that KL fly from target to COE may not be

correct because KL can scatter at the downstream of target. We want to develop new method without using this assumption.

2019/12/23 2019 Year End Presentation

6

New Method

!"($, &, ') =

*+

, -,.,/ 0123

"

4+ "

+

*,

, -,.,/ 0123

"

4, "

+

*6

, -,.,/ 0123

"

46 "

Decay position of KL may be determined by minimizing χ2 of following formula.

78 78 78

z

Target KL

($, &, ')

I checked the precision of this new method by using toy simulation.

2019/12/23 2019 Year End Presentation

7

!" → $%& Toy Simulation

z

KL

z[m]

BH width (150 mm) θ : Fixed (Pt - Pz angle)

P = ( Pt, Pz )

+,

• + +/
• 400 mm

CsI

|P|: spectrum measured at the beam exit

θ is fixed, KL does not go through the beam hole

• 01 → 334 decay

2019/12/23

8

Event Selection Criteria

• 6 gamma positions are all inside the fiducial region of

the CsI

• Minimum gamma energy > 150 MeV
• Minimum 2 gamma distance > 150 mm
• 0 < Decay z < CsI (6168 mm)

Using this toy simulation, I checked whether this new method can reconstruct the decay position.

2019/12/23 2019 Year End Presentation

9

True information of Toy Simulation

hist9 Entries 10000 Mean 4072 Std Dev 824.6 1000 2000 3000 4000 5000 6000 Z position (mm) 50 100 150 200 250 300 350 hist9 Entries 10000 Mean 4072 Std Dev 824.6

True z position

We want to reconstruct these vertex positions

600

• 400
• 200
• 200

400 600

X position (mm)

600

• 400
• 200
• 200

400 600

Y position (mm) hist1 Entries 10000 Mean x 0.6744

• Mean y

0.8278 Std Dev x 223.2 Std Dev y 223.4 2 4 6 8 10 12 14

hist1

Entries 10000 Mean x 0.6744

• Mean y

0.8278 Std Dev x 223.2 Std Dev y 223.4

True X,Y position

True vertex Z [mm]

# of Events

True vertex X [mm] True vertex Y [mm]

2019/12/23 2019 Year End Presentation

600

• 400
• 200
• 200

400 600 600

• 400
• 200
• 200

400 600

hist2

• Mean y

2 4 6 8 10 12 14 16

hist2

• Mean y

new reconstructed X,Y

hist3

• 5.841e

• 1.513e

200

• 150
• 100
• 50
• 50 100 150 200

1000 2000 3000 4000 5000 6000 7000 8000 9000

hist3

• 5.841e

• 1.513e

new reconstructed Z - true Z

hist11

100

• 80
• 60
• 40
• 20
• 0 20 40 60 80 100

2000 4000 6000 8000 10000

hist11

new reconstructed r - true r

Reconstructed Vertex position(1)

10

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events NEW NEW NEW conventional conventional conventional

• Energy and position resolution of CsI... neither considered

conventional conventional conventional

2019/12/23 2019 Year End Presentation

# of Events # of Events

good reconstruction

11

• Rec. vertex Y [mm]
• Rec. vertex X [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events NEW NEW NEW conventional conventional conventional

• Energy and position resolution of CsI... both considered

Comparison of new and conventional

2019/12/23 2019 Year End Presentation

larger difference can not see the image

1σ

12

Chi2 distribution

Rec vertex X [mm] Rec vertex Y [mm]

1 sigma(ΔChi2 = 2.3) contour (for example 1 event) position resolution ~300mm ... not small

• The failure of reconstruction using new method is due

to this bad position resolution.

2019/12/23 2019 Year End Presentation

!

z

! ! ! ! !

Reason for bad resolution

Constraint surface

• Constraint from 6 gamma looks like sphere, and sum of

them become too shallow to minimize.

! !

CsI

2019/12/23 2019 Year End Presentation

14

Summary and Next

• Halo KL cause !" → 2% background, so we need to measure

halo KL correctly.

• I tried to develop new reconstruction method minimizing chi2

function of reconstructed mass of &' .

• Position resolution of new method was not enough to

reconstruct vertex position correctly.

• To know whether we need to develop more reconstruction

method, I’m checking discrepancy of COE between data and MC.

2019/12/23 2019 Year End Presentation

Backup

15 2019/12/23 2019 Year End Presentation

600

• 400
• 200
• 200

400 600 600

• 400
• 200
• 200

400 600

hist2

• Mean y

2 4 6 8 10 12 14 16

hist2

• Mean y

new reconstructed X,Y

hist3

• 5.841e

• 1.513e

200

• 150
• 100
• 50
• 50 100 150 200

1000 2000 3000 4000 5000 6000 7000 8000 9000

hist3

• 5.841e

• 1.513e

new reconstructed Z - true Z

hist11

100

• 80
• 60
• 40
• 20
• 0 20 40 60 80 100

2000 4000 6000 8000 10000

hist11

new reconstructed r - true r

Reconstructed Vertex position(1)

16

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events NEW NEW NEW conventional conventional conventional

• Energy and position resolution of CsI... neither considered

conventional conventional conventional

2019/12/23 2019 Year End Presentation

# of Events # of Events

good reconstruction

17

• Rec. vertex Y [mm]
• Rec. vertex X [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events

• Rec. vertex X [mm]
• Rec. vertex Y [mm]

Rec Z – True Z [mm] Rec R – True R [mm]

# of Events # of Events NEW NEW NEW conventional conventional conventional

• Energy and position resolution of CsI... both considered

Reconstructed Vertex position(2)

2019/12/23 2019 Year End Presentation

18

Sigma of M2

!M2 =

#$%('(,*(,+(,'%,*%,+%) #'(

!'(+

#$%('(,*(,+(,'%,*%,+%) #*(

!*(+ ... +

#$%('(,*(,+(,'%,*%,+%) #+%

!+% #$%('(,*(,+(,'%,*%,+%) #'(

= $% '(-. ,*(,+(,'%,*%,+% /$% '( ,*(,+(,'%,*%,+%

.

2019/12/23 2019 Year End Presentation

Chi2 distribution

19 2019/12/23 2019 Year End Presentation

20

KL Generation in Toy Simulation(2)

z

KL

z[m]

BH width

P = ( Pt, Pz )

%&

' + %) '

• CsI

Decay Position

|P|=

[1]Nakagiri-san’s Dr.thesis https://www-he.scphys.kyoto-u.ac.jp/theses/doctor/nakagiri_dt.pdf

[1]

Generation

2019/12/23 2019 Year End Presentation

21

KL Generation in Toy Simulation(3)

z

KL

z[m]

BH width

• CsI

Decay Position

Flying Distance

= &'() ∗ − log(012345[0,1])

P

2019/12/23 2019 Year End Presentation

22

MINUIT Detail

• used “MIGRAD” minimizer (also checked HESSE,MINOS)
• used some limits in order to prevent the parameter

from taking on unphysical values e.g.) 0 < z < CsI, -1000< x,y <1000

• fitting step width of (x,y,z)... (10,10,100)
• max call ... 1000 times (500~ not chenged)

2019/12/23 2019 Year End Presentation

23

Problem of Conventional Reconstruction Method

z !"#$%#& True Scatter point

'()

Target True vertex point Reconstructed vertex point

24

Resolution of CsI

25

Only Position Resolution of CsI

26

Only Energy Resolution of CsI

27

Conventional method

First, we reconstruct a !0 assuming mass of "" is #!0and vertex position is on z axis.

(from shimizu-san’s slide)