Nue Cross Section Measurement Mark Hartz Kavli IPMU (WPI), - - PowerPoint PPT Presentation

Nue Cross Section Measurement

Mark Hartz Kavli IPMU (WPI), University of Tokyo and TRIUMF

NuPRISM Reco

Goals

2

• Using the intrinsic electron (anti)neutrino component, measure the

ratios:

• The purity improves with larger off-axis angle:


• We can use linear combinations to match the fluxes:

Off-axis angle (º) νe Flux 
 0.3-0.9 GeV νμ Flux 0.3-5.0 GeV Ratio νe/νμ 2.5 1.24E+15 2.46E+17 0.507% 3.0 1.14E+15 1.90E+17 0.600% 3.5 1.00E+15 1.47E+17 0.679% 4.0 8.65E+14 1.14E+17 0.760%

Off-axis Angle (degrees) 1 1.5 2 2.5 3 3.5 4 Fitted Coefficient 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

(GeV)

ν

E 0.5 1 1.5 2 2.5 3 (Arb. Norm.) σ × Φ 1000 2000 3000 4000 5000 6000 7000 8000

6

10 ×

) ° (2.5-4.0

e

ν PRISM ν Linear Combo.

µ

ν PRISM ν

NuPRISM Reco

Showing Today

3

• Updated electron neutrino selection in neutrino mode
• Re-optimized ToWall, DWall and PID likelihood ratio cuts
• Re-calculate systematic and statistical error estimates
• Discussion on how to proceed

NuPRISM Reco

Monte Carlo

4

• T2K 13bv3.2 neutrino flux model
• NEUT 5.1.4.2 with RFG, MA=1.2 GeV
• WCSim simulation of the NuPRISM inner detector
• ID radius = 3 m
• ID height = 10 m
• 8 inch diameter PMTs with 40% photo-coverage
• Center of ID positioned at 2.5, 2.92, 3.34 and 3.76 degrees off-axis
• 2.28e21 POT simulated at each position
• fiTQun single ring and pi0 reconstruction of simulated events

NuPRISM Reco

Event Pre-selection

5

• Before optimizing fiducial volume and particle ID cuts
• Cuts on true quantities:
• No OD information, so cut muons with mom (MeV/c)>2.0*ToWall (cm)
• No multi-ring reconstruction, so cut events with >1 ring with “true”

visible energy above 30 MeV

• Reconstructed quantities:
• fq1rpcflg[0][1]==0 (contained event)
• fqnse==0 (no decay electron)
• fq1rmom[0][1]>100. (momentum>100 MeV/c)
• Preliminary PID cuts for fiducial volume optimization for Yoshida-san

(next slide)

NuPRISM Reco

PID Pre-selection

6

p<375: Lratio>-30+0.3*p p>375: Lratio>270-0.34*p mass<133 && 
 Lratio<120-mass*0.9

• T. Yoshida

NuPRISM Reco

Procedure

7

• Apply the pre-selection on the previous two slides
• Optimize the dWall and toWall cuts using S/sqrt(S+B+σsyst2)*
• Signal = CC interactions of nu_e
• BGND = CC interactions of nu_mu + NC interactions
• Remove the PID cuts and apply the optimized dWall and toWall cuts
• Redo the optimization of the PID cuts
• Redo the optimization of the fiducial volume cuts
• Evaluate the signal statistics and purity with the optimized cuts

*Formula includes background systematic error in the denominator

NuPRISM Reco

dWall/toWall Distributions

8

• Events with toWall<100 cm are generally not reconstructed as

contained events

• NC background events are at low dWall (second photon has small

toWall?)

• CC-numu background events are at low toWall (poor sampling of ring)

dWall (cm) 50 100 150 200 250 300 toWall (cm) 50 100 150 200 250 300 Events 5 10 15 20 25 30 35 40

• CC

e

ν

dWall (cm) 50 100 150 200 250 300 toWall (cm) 50 100 150 200 250 300 Events 10 20 30 40 50 60

• CC

µ

ν

dWall (cm) 50 100 150 200 250 300 toWall (cm) 50 100 150 200 250 300 Events 10 20 30 40 50 60 70

NC

NuPRISM Reco

dWall/toWall Cut Optimization

9

• Optimized on S/sqrt(S+B+(0.05*BNC)2+(0.05*BCC)2)
• Most optimal point in cut space is dWall>135 cm and toWall>280 cm
• Old optimization was dWall>150 cm and toWall>260 cm
• Will re-optimize after optimizing the PID cuts

dWall Cut (cm) 50 100 150 200 250 300 toWall Cut (cm) 50 100 150 200 250 300 S+B S/ 5 10 15 20 25 30 35 40 Cut Optimization

NuPRISM Reco

Le/Lmu Cut Optimization

10

• Apply all cuts except for Le/Lmu cut
• Optimized on S/sqrt(S+Bcc+(0.05*BCC)2) for each bin in electron momentum
• Cut lines are:

(MeV/c)

e

p 200 400 600 800 1000 Cut

ratio

L

• 1000
• 500

500 1000 S+B S/ 6 8 10 12 14 16 18 20 22 24

Electrons Muons

p<325: Lratio>80+0.277*p p>325: Lratio>252-0.252*p

NuPRISM Reco

)

2

(MeV/c

π

m 50 100 150 200 Cut

ratio

L 500 1000 S+B S/ 5 10 15 20 25 30 35 40

Lpi0/Le Cut Optimization

11

• Apply all cuts except for Le/Lmu cut
• Optimized on S/sqrt(S+BNC+(0.05*BNC)2) for each bin in electron momentum
• Cut lines are:

Electrons Pi0

mass<90 && Lratio<130-0.95*mass

NuPRISM Reco

dWall/toWall Cut Re-optimization

12

• Optimized on S/sqrt(S+B+(0.05*BNC)2+(0.05*BCC)2)
• Change cut slightly dWall>135 cm and toWall>275 cm

dWall Cut (cm) 50 100 150 200 250 300 toWall Cut (cm) 50 100 150 200 250 300 S+B S/ 5 10 15 20 25 30 35 40 45 50 Cut Optimization

NuPRISM Reco

Background and Signal Error Estimate

13

• The NCpi0 and CC-numu backgrounds should be constrained by high

statistics samples in NuPRISM

• Assume we can achieve a 5% uncertainty on the background

prediction

• The NCgamma background cannot be directly constrained. Assume a

50% error coming mainly from the cross section and flux models

• Assume we can achieve a 1% error on the modeling of the signal

efficiency

NuPRISM Reco

POT Weighted Signal (Old Cuts/MC Stats)

14

• Weighting to 1.5e21 neutrino mode POT for each off-axis position

between 2.5 and 4.0 degrees Purity for Erec<1.2 GeV = 73% Nue Signal for Erec<1.2 GeV =3184 Expect improvements:

• OD veto on exiting particles
• More optimization of PMT 


granularity

• Larger ID size?

(MeV)

rec

E 500 1000 1500 2000 Events/(200 MeV) 500 1000 1500

• CC

e

ν π NC γ NC

• CC

µ

ν

1-Ring e Candidates

NuPRISM Reco

(MeV)

rec

E 500 1000 1500 2000 Fractional Error 0.05 0.1 0.15 0.2 0.25

Total Statistical (S-B) Background Systematics Signal Efficiency )

µ

ν /

e

ν Flux (

(MeV)

rec

E 500 1000 1500 2000 Fractional Error 0.05 0.1 0.15 0.2 0.25

Total Statistical (S-B) Background Systematics Signal Efficiency )

µ

ν /

e

ν Flux (

Total Error Size (Old Cuts/MC Stats)

15

• The estimated error for 200 MeV wide bins is shown on the left
• For an estimate of the error on the overall rate, I divided into two bins of

1 GeV (right)

• Current estimate is ~5% error in the sub-GeV region

200 MeV bins 1 GeV bins

NuPRISM Reco

(MeV)

rec

E 500 1000 1500 2000 Events/(200 MeV) 500 1000 1500

• CC

e

ν π NC γ NC

• CC

µ

ν

1-Ring e Candidates

POT Weighted Signal (New Cuts/MC Stats)

16

• Weighting to 1.5e21 neutrino mode POT for each off-axis position

between 2.5 and 4.0 degrees

Purity for Erec<1.2 GeV = 71(73)% Nue Signal for Erec<1.2 GeV =3501(3184)

20 inch PMT Results

NuPRISM Reco

(MeV)

rec

E 500 1000 1500 2000 Fractional Error 0.05 0.1 0.15 0.2 0.25

Total Statistical (S-B) Background Systematics Signal Efficiency )

µ

ν /

e

ν Flux (

(MeV)

rec

E 500 1000 1500 2000 Fractional Error 0.05 0.1 0.15 0.2 0.25

Total Statistical (S-B) Background Systematics Signal Efficiency )

µ

ν /

e

ν Flux (

Old vs New Total Error Size

17

• The total uncertainty below 1 GeV is only slightly reduced with cut
• ptimization on large MC statistics
• Configuration changes are likely needed to get any significant

improvments

New Old

NuPRISM Reco

What’s Next

18

• For the T2K phase II studies, run some 2.5-4.0 degree MC with 4 m ID

radius

• For the summer PAC meeting, run MC for the full off-axis angle range

with the following configurations:

• 8 inch PMT, 3 m radius ID (2.5-4.0 degrees already exist)
• 8 inch PMT, 4 m radius ID
• 3 inch PMT, 4 m radius ID
• 3 inch PMT, 3 m radius ID
• Do full error propagation once MC is available for the full off-axis angle

range

• Antineutrino analysis if there is time

NuPRISM Reco 19

Extra Slides

NuPRISM Reco

Flux Error Estimate

20

Flux error estimates were previously calculated by selecting Nue candidates based on truth information Since the error will be applied after the background subtraction, we can apply the same error to our reconstructed events

(GeV)

ν

E 0.5 1 1.5 2 Fractional Error 0.05 0.1 0.15

Nominal Flux Error 1/2 Had. Int. Error

Flux Error

µ

ν

σ /

e

ν

σ

Saved as covariance matrix for application to the 
 signal after background 
 subtraction

NuPRISM Reco

Cut Optimization

21

• Start with ln(Le/Lμ) cut set at >150
• Just eyeballed. Numu background is small compared to NC

background.

• Optimize DWall and ToWall cuts on the NCpi0 background and Signal

with events having pi0mass<60 MeV/c2

• Optimize cut line in ln(Lπ0/Le) plane

NuPRISM Reco

DWall/ToWall Optimization

22

• Set cut at DWall>150 cm, ToWall>260 cm
• May be able to improve the optimization with more stats (some of the

structure is probably from statistical fluctuations)

ToWall (cm) 50 100 150 200 250 300 DWall (cm) 50 100 150 200 250 300

0.5 1 1.5 2 2.5

S+B S/

NuPRISM Reco

X Intercept 20 40 60 80 100 120 140 160 180 200 Y Intercept 50 100 150 200 250 300 350 400 450 500

2 3 4 5 6 7 8

S+B S/

Pi0 Cut Optimization

23

• Allow the X and Y intercept points to vary
• Above X intercept, cut is for likelihood ratio less than 0

NuPRISM Reco

MC for Nue Selection

24

• Mark Scott has generated files some files with OAA>2.5 degrees:
• 3.76 degrees: 5e19 POT
• 3.34 degrees: 5e19 POT
• 2.92 degrees: 1e19 POT
• 2.5 degrees: 4e19 POT
• For most off-axis bins expect ~30 times the exposure during T2K-II
• 40% photo-coverage and 8 inch PMTs

NuPRISM Reco

Nue Candiate Reconstructed Energy

25

(GeV)

rec

E 500 1000 1500 2000 Events/(200 MeV) 10 20 30

• CC

e

ν π NC γ NC

• CC

µ

ν

1-Ring e Candidates

• Good purity above 400 MeV
• Numu background is not so negligible. Should optimize the cut.

NuPRISM Reco

Error Estimate

26

• Statistical Error: assume 30 times the exposure from the current MC

stats

• Background systematic errors:
• 5% for NCpi0
• 50% for NCgamma
• 5% for numu-CC
• Have generated flux errors, but haven’t estimate flux errors yet because

they have to be evaluated on the ratio to the CC-numu

NuPRISM Reco

Uncertainty on Bgnd Subtracted Candidates

27

• The statistical errors between 400 and 1000 MeV are <5% in each bin
• Potential improvements from larger ID, better PMT granularity to enlarge

fiducial volume

• Background systematic errors are low in the region of interest
• How important is the 200-400 MeV bin?
• Will look into evaluating as an error on the SK prediction

(MeV)

rec

E 500 1000 1500 2000 Fractional Error 0.05 0.1 0.15 0.2

NuPRISM Reco

Things to do (Before PAC?)

28

• Re-optimize the muon/electron separation
• Estimate the error on the SK event rate from the uncertainties estimated here
• Assume 100% correlation for systematic errors
• Update MC stats as they become available
• Update the flux error estimated using selected events (may not be ready for

the PAC)

• Estimate the uncertainty on the 1-Ring e candidates measured at 2.5

degrees off-axis (direct measuring of intrinsic+NC background)

• The first three items should be ready by Wednesday morning on Japan