1. Update of the Fiducial calibration study in 2km WC detector
2km pre-meeting 7 July, 2006
G. Mitsuka
T2K-2km working group

2. Contents Motivation Fiducial PMT geometry
Criteria to identify the vertex in/out FV
Fiducial PMT simulation in 2KM Water Cherenkov detector
Summary
(From now on, Fiducial PMT is shortened to FVPMT)

3. Motivation  Need to implement new fiducial calibration system
In the WC detector, fiducial volume is expected as one of the large systematic errors (difficult to calibrate fiducial volume)
 Need to implement new fiducial calibration system
An idea of FV calibration
Set the array of PMTs(FVPMT) around FV boundary
Vertex is identified as in/out by FVPMT signals
In order to check whether FV calibration will works well or not
Performance test of the candidate PMT for FVPMT(2inch PMT : Hamamatsu R9869) is carried out in the 1kton detector  R9869 fulfills requirements
Estimate the efficiency of FV calibration and error by using 2km simulation with FVPMTs
Update

4. How to set the array of FVPMT
Requirements
Avoid the effect of shadow of FVPMT
Identify the FV boundary with high resolution
Solutions
2inch PMTs are set with a distance of 50cm
FVPMT is set on the FV boundary
R9869 with water proof

5. FVPMT geometry Z direction R direction (Top,Bottom) IFPMT 50cm OFPMT
FV boundary
IFPMT
50cm
50cm
OFPMT
Fiducial volume
Inward-Facing PMT = IFPMT Outward-Facing PMT = OFPMT
Front & End-cap 104PMTs + Barrel 360PMTs = Total 464PMTs

6. Criteria for in/out identification(1)
FVout event L
Vertex is identified as FVin/out by FVPMT hits  FVout requires OFPMT hits  FVin requires IFPMT hits
Hit due to scattering or reflection is cut by PMT-facing direction and m direction
OFPMT Hit
FVin event
OFPMT No-hit
FV boundary
IFPMT Hit

7. Estimation of efficiency and error
Estimate the FV calibration efficiency and error by using 2km WC simulation with FVPMTs
Neutrino events are simulated with taking account of the neutrino flux at 2km position
These events are independently reconstructed of FVPMT by 2km official reconstruction software
Sample events are 1-ring m-like event (~170000)

8. Event reconstruction with FVPMT
Reconstruction performance seems to be good even with FVPMT

9. FV identification efficiency (Z dir)
ID as FVin(ID)
ID as FVout(ID)
X,Y,Z is fitted position
Events in R(=sqrt(X2+Y2)) < 150cm are selected  except for less efficient region (FV barrel edge)
Efficiency curve shows sharp shift at FV boundary

10. FV identification efficiency (R dir)
Events in -300<Z<100cm are selected
Efficiency curve shows sharp shift at FV boundary
ID as FVin(ID)
ID as FVout(ID)
Efficiency summary
FVin(fit)
FVout(fit)
FVin(ID)
83.2%
16.8%
FVout(ID)
17.6%
82.4%

11. Additional study
FV calibration efficiency and error with 464’s FVPMT were estimated  predict that FV calibration works well
For the cost reason, the number of FVPMTs is required to be much small  search for the alternative configuration
Assuming the half number of FVPMTs, check the efficiency and error

12. How to decrease FVPMT
FVPMTs set on the endcap and frontcap are very important to calibrate the FV edge, because m of CCQE interaction are almost forward scattered
Try to remove barrel FVPMTs ( next page)
zbs and ntuple files are tentatively same as full FVPMT version, skip barrel FVPMT when reading ntuple

13. What FVPMT is skipped Type 1 Type 2 Full464 – skip240 = 224 PMTs used
used PMT
skipped PMT
Type 1
Type 2
Y(cm)
Y(cm)
Z(cm)
Z(cm)
Full464 – skip240 = 224 PMTs used
464 – 210 = 254 PMTs used

14. Calibration efficiency (Zdir:(x2+y2)1/2<150cm)
FVin(ID) original / modified
FVout(ID) original / modified
Type 1
Type 2

15. Calibration efficiency (Rdir:-300<z<100cm)
FVin(ID) original / modified
FVout(ID) original / modified
Type 1
Type 2
Worse efficiency in/out FV
Same level inside FV, worse outside FV

16. Efficiency as a function of En
FV out (full) FV in (full) FV out (type2) FV in (type2)
Efficiency
80% of all events
En(MeV)

17. Efficiency summary
Black : Full FVPMT
Blue : type1
Red : type2
FVin (id)
FVout (id)
FVin (true)
83.2 / 77.5 / 80.9
16.8 / 22.6 / 19.1
FVout (true)
17.7 / 17.8 / 18.8
82.4 / 82.2 / 81.2
(unit is %)
type2 keeps 81% efficiency inside FV, although type1 is 78%
According to the efficiency curve, type2 could be realistic

18. Shadow effects of FVPMT (1)
(w/o FVPMT is normalized by the number of events in FV)
Ring-counting
PID
w/o FVPMT
w/ FVPMT
w/o FVPMT
w/ FVPMT
Events
Events
Excess appeared, but not affect Single-ring
Likelihood
Likelihood
According to the comparison of Ring-counting and PID, FVPMT doesn’t seem to affect 1-ring m-like events

19. Shadow effects of FVPMT (2)
Hatched type requires identification as FVin
Efficiency of QE,Non-QE and NC in 1-ring m-like events
Type
w/o FVPMT
True
Full
type1
type2 QE
75.6
71.5
72.6
71.1
71.8
Non-QE
22.9
24.1
23.3
24.8 NC
1.5
4.3
4.0
4.1
(unit is %)
Efficiency of miss-PID in 1-ring m-like QE events
Type
w/o FVPMT
True
Full
type1
type2
miss-PID
1.02
0.80
0.68
0.72
0.71
(unit is %)

20. Error of efficiency In all types, error is smaller than 1%
e : Efficiency of FV identification
Real data MC
Table of errors of MC efficiency
Type
Full
type1
type2
Q(+/-30%), Nhit(+/-10%)
0.1%
0.6%
0.4%
Non-QE(+/-20%), NC(+/-30%)
0.2%
Ring-counting(+/-5%)
<0.1%
PID(+/-4%)
Energy-scale
Total
0.3%
0.7%
In all types, error is smaller than 1%

21. Summary
Checked the efficiency and error by using fitted 2km MC  predict FV calibration works well
Vertex of 1-ring m-like events are correctly identified with efficiency > 80%
Error of identification efficiency is less than 0.5%
Searched for alternative FVPMT setup  Even if the half number of FVPMTs, keep the efficiency of 80%

22. Backup ….

23. miss-identified event
FVin events miss-identified as FVout
Outwrd-going event whose vertex is near from FV boundary  OFPMT hit around exit point
LowE event whose vertex is almost center of FV
 Neither PMT hits
FVout events miss-identified as FVin
Inward-going event whose vertex is near from FV boundary
 IFPMT hits, but OFPMT no-hit

24. Slope of Efficiency curve in FVq1
q1 + q2 = p/2 q1 > q2 q2
OFPMT hits due to scattering or reflection  identified as FVout event
cos(q2) is generally large  Hit is ignored by angle cut cos(q2) is generally small  Hit is taken into account in the calibration criteria Z

25. Systematic error of FV
DATA MC
FV syst. error

26. Introduction
Reconstruction is applied for generated MC~170000events(shown at last meeting)
FV identification criteria is improved
Estimate the error of FV calibration

27. Summary Improved criteria is applied for fitted MC
　Events are correctly identified with efficiency > 80%
　Error of identification efficiency is 0.7%
In preparation for the commitment of improved routines
Studying the configuration using lower number of PMTs as keeping efficiency and error

28. Summary
Checked the efficiency and error by assuming the half number of FVPMTs
type2 keeps 81% efficiency in/out FV
Error of efficiency is less than 0.6% in all geometries
type2 could be realistic geometry

29. Error of Identification efficiency
Systematic error of FV
NinID : #events in FV
e : Efficiency of FV identification
Real data MC
Error of identification efficiency
　Calibration system (Q:+/-20%,Nhit:+/-10%)  0.07%
　Neutrino interaction (nonQE:+/-20%,NC: +/-30%)  0.33%
　Reconstruction (PID:+/-4%,RingCounting:+/-5%)  0.33%
　Energy scale (+/-2%)  0.11%
Error of efficiency = 0.48%

30. Error of efficiency NinID : #events in FV
e : Efficiency of FV identification
Real data MC
Table of errors of MC efficiency
Type
Full
type1
type2
Q(+/-30%), Nhit(+/-10%)
0.07%
0.49%
0.32%
Non-QE(+/-20%), NC(+/-30%)
0.33%
0.20%
0.11%
Ring-counting(+/-5%)
<0.01%
0.12%
PID(+/-4%)
0.02%
0.01%
Energy-scale
0.19%
0.17%
Total
0.48%
0.57%
0.40%

31. Criteria for in/out identification(1)
VTX-Cross point = LVC
Fitted vertex Cross point
FVout event
FVout event L
OFPMT
Hits
IFPMT
OFPMT Hit
L - LVC
FVin event
IFPMT
FVin event
Hits
OFPMT No-hit
FV boundary
IFPMT Hit
L - LVC
If OFPMT hit (L-Lvc < 0)  ID as FVout else if IFPMT hit != 0  ID as FVin else(i.e. No-IFPMT hit)  ID as FVout

32. Criteria for in/out identification(2) Merit of FVPMT facing both sides
Backward extrapolated cross point exists && VTX-Corss(forward) > VTX-Cross(Backward)
 Hits around backward cross point are taken into account in order to correct miss-fitted event
miss-fitted FVout event
OFPMT Hit
miss-fitted FVout event True vertex(★) is FVout Fitted vertex(●) is FV in  OFPMT hit can correct FVin to FVout
miss-fitted FVin event
True vertex(★) is FVin Fitted vertex(●) is FVout  OFPMT hit can correct FVin to FVout
miss-fitted FVin event
IFPMT Hit

33. Criteria for in/out identification(3) Hits by scattering or reflection
OFPMT hits due to the scattering or reflection  Identified as FVout by these OFPMT hits
In order to avoid such miss-ID
cosq(PMT dir~particle dir) > 0.8  ignore hit q
OFPMT Hit

34. Ring counting likelihood (check)
Full
Type1
Type2
Efficiency
dlfct