Neutrino interaction systematic errors in MINOS and NOvA Mayly - - PowerPoint PPT Presentation

Neutrino interaction systematic errors in MINOS and NOvA

Mayly Sanchez

Iowa State University Argonne National Laboratory

Nufact 2012 - Williamsburg, VA July 24, 2012

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS and NOvA in a nutshell

• Produce a high intensity beam
• f muon neutrinos at Fermilab.
• Measure background at the

Near Detector and use it to predict the Far Detector spectrum.

• Many uncertainties are

expected to cancel.

• Similar detectors ensure that

neutrino interaction related uncertainties largely cancel.

• If neutrinos oscillate we will
• bserve a distortion in the data

at the Far Detector at either site.

← long baseline → 1st/2nd generation

NO!A Far Detector MINOS Far Detector 810 km 735 km 2

See more general MINOS and NOvA talks on Friday

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

The MINOS detectors

• Functionally identical: Near and Far detectors
• Octogonal steel planes (2.54cm thick ~1.44X0)
• Alternating with planes of scintillator strips (4.12cm wide, Moliere rad ~3.7cm).
• Near (ND): ~ 1kton, 282 steel squashed octagons. Partially instrumented.
• Far (FD): 5.4 kton, 486 (8m/octagon) fully instrumented planes.

Near Far

3

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Neutrino event topologies in MINOS

To select νe CC we focus on finding compact showers.

MC events µ- ! Reducible Background !µ CC Event NC Event !0 e- “Irreducible” Background Signal !e CC Event NC Event

4

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS Monte Carlo

• MC tuned to external bubble

chamber data for hadronization models.

• Tuning focused in the following

quantities:

• Charged/neutral pion

multiplicity and dispersion.

• Forward/backward

fragments.

• Fragmentation functions.
• Transverse momentum.
• Transverse momentum still too

low in forward hemisphere.

• Model at lower W2 is an

extrapolation.

Neutrino data taken with the Near Detector was used to correct the Far Detector expectation. Region of interest: 1 - 15 GeV2 in W2

5

• T. Yang et al., Eur. Phys. Jour. C 63:1-10 (2009)

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS electron neutrino selection

2009 the early days

• Initially the ND predicted

backgrounds were 20% higher than observed in data.

• Hadronization and final state

interactions uncertainties give rise to large uncertainties in ND prediction.

• External data is sparse in our

region of interest.

• Strong background

suppression, since we select tails of BG distributions.

• Improvements to nuclear

rescattering model in MC reduced data/MC discrepancies in current analyses.

6

ANN

0.2 0.4 0.6 0.8 1

Events

1000 2000 3000 4000 5000

Total Data Total MC

MINOS PRELIMINARY Near Detector

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS electron neutrino selection

• Initially the ND predicted

backgrounds were 20% higher than observed in data.

• Hadronization and final state

interactions uncertainties give rise to large uncertainties in ND prediction.

• External data is sparse in our

region of interest.

• Strong background

suppression, since we select tails of BG distributions.

• Improvements to nuclear

rescattering model in MC reduced data/MC discrepancies in current analyses.

ANN

0.2 0.4 0.6 0.8 1

Events

1000 2000 3000 4000 5000

Total Data Total MC

MINOS PRELIMINARY Near Detector Reconstructed Energy (GeV)

1 2 3 4 5 6 7 8 9

POT

19

Events/GeV/10

1000 2000

Total Data Total MC

ANN-selected

MINOS PRELIMINARY Near Detector

2010

7

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

• The Near Detector νe selected NC and

νμ CC background components are corrected by the Far/Near MC ratio.

• Far/Near ratio accounts for geometry,

fiducial volume ratio, intensity, detector differences and oscillations.

• Data-driven background

decomposition techniques allow us to treat each component separately.

• The signal νe and the ντ CC from νμ
• scillations are corrected using the

extrapolation of the νμ CC spectrum.

Predicting the FD background

~10-4 expected from geometry and fiducial volume ratio alone

Reconstructed Energy (GeV)

1 2 3 4 5 6 7 8

)

• 4

Far/Near Ratio (x10

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

CC

!

• MINOS PRELIMINARY
• Reconstructed Energy (GeV)

1 2 3 4 5 6 7 8

)

• 4

Far/Near Ratio (x10

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

NC

Monte Carlo

• Reconstructed Energy (GeV)

1 2 3 4 5 6 7 8

)

• 4

Far/Near Ratio (x10

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

CC

e

• beam

8

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS systematic errors

• For the main background components the hadronization model systematic is

corrected to about 4%, while intranuclear and cross sections are down to 1%

• r less.
• More recent analyses have these below 2.5%.
• To study the systematic errors, we generate special MC with the

modified parameter in ND and FD. Using this modified MC for extrapolation and calculate the difference with the standard results.

MINOS 2010 9

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

The NOvA detectors

• 14 kton Far Detector (~3x MINOS).
• >70% active detector.
• 360,000 detector cells read by APDs.
• 0.3 kton Near Detector
• 18,000 cells/channels.
• Each plane just 0.15 X0. Great for e- vs !0 .

!"#"$ %#%&$ '#(&$

!"#$%&'()#"*'%+,#-

.%/0$&&"$,0(1/23+ 4"5'&'02+3(63/7/02 8/9'#(1,,: ),(;!<(='">,?+

! " # % ) $

!"#$%&'&()*+(,'--.(/--'&(01#2( 334(-1#'$.(56(.,178--9#5$ 17.#$%:'7#'&(01#2 !;.21<17=(/>'$(97&(?)@.

Far Detector 14 kton 928 layers

Near Detector 0.3 kton 206 layers

!"#$%&'("%) &*%+,&-.&/$00)

• .1'"2$0&345

Prototype installed on the surface and taking data!

10

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Neutrino event topologies in NOvA

*+,$-./012)13140+)5,06)7)819)3,+,:.1;

"(A(' $(A(' #B(A('

</= $->4 ?@)+6>51A =A>0>4

!)$101A !)$101A

@,&61.)'B

Signal Background Background

11

NOvA uses GENIE for neutrino interactions, same tuning as MINOS.

Monte Carlo events.

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Signal N C CC b e a m

NOvA electron neutrino selection

• Several electron neutrino

identification techniques are being developed.

• Performance already

comparable to the NOvA TDR.

• PID re-optimized for large θ13.
• Method used for this study is

based on library matching algorithm (a la MINOS).

• Current implementation is

promising and has room for improvement.

• An alternate method based on

transverse/longitudinal likelihoods

• f the shower energy profile will

be shown in the poster session.

Signal:Bkg PID>0.7

100:40

PID>0.9

70:15

12

PID

0.5 1

Events

5 10 15

=0.1

13

• 2

2

POT FHC, sin

21

10 ! 1.8

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Signal NC beam C C

NOvA electron neutrino selection

13

PID

0.5 1

Events

5 10 15

=0.1

13

• 2

2

POT FHC, sin

21

10 ! 1.8

Signal:Bkg PID>0.7

100:40

PID>0.9

70:15

• Several electron neutrino

identification techniques are being developed.

• Performance already

comparable to the NOvA TDR.

• PID re-optimized for large θ13.
• Method used for this study is

based on library matching algorithm (a la MINOS).

• Current implementation is

promising and has room for improvement.

• An alternate method based on

transverse/longitudinal likelihoods

• f the shower energy profile will

be shown in the poster session.

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Signal:Bkg PID>0.7

100:40

PID>0.9

70:15

Signal NC beam C C

NOvA electron neutrino selection

14

PID

0.5 1

Events

5 10 15

=0.1

13

• 2

2

POT FHC, sin

21

10 ! 1.8

• Several electron neutrino

identification techniques are being developed.

• Performance already

comparable to the NOvA TDR.

• PID re-optimized for large θ13.
• Method used for this study is

based on library matching algorithm (a la MINOS).

• Current implementation is

promising and has room for improvement.

• An alternate method based on

transverse/longitudinal likelihoods

• f the shower energy profile will

be shown in the poster session.

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Hadronic y

0.2 0.4 0.6 0.8 1

Selection Efficiency (%)

20 40 60

signal

e

signal

e

• Hadronic y

0.2 0.4 0.6 0.8 1

Selection Efficiency (%)

0.5 1 1.5 2 2.5

!

• CC !
• CC

s

• Number of

1 2 3 4 5

Selection Efficiency (%)

1 2 3

Neutral currents Neutral currents

NOvA electron neutrino selection

• The electron neutrino selection

prefers low hadronic y νe signal events and high hadronic y νμ CC background events.

• For neutral current events the

main background arises from events with one or more !0.

15

PID>0.7 PID>0.9

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Studying systematic errors in NOvA

• The neutrino interaction systematic errors are modified in this study:
• Cross-section: MA(QE) and MA(RES) varied by ± 20%.
• Hadronization model changes:
• The !0 selection probability in the hadronization model

changed by ± 33%.

• Change in average Pt resulting in broader showers.
• Re-weighting Pt and Xf distributions of hadron distribution.
• Intranuclear formation zone changed by ± 50%.
• These systematics should mostly cancel, however they can be

affected by Far/Near detector differences.

• We expect the most significant of them to be:

energy spectra, light levels and event energy containment.

16

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Energy spectra differences

• Neutrino energy depends on angle with respect

to original meson direction and energy.

• Angular distributions different between neutrinos

seen at Near and Far detectors.

17

Far Detector Near Detector

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Light level differences

• The Far and Near Detectors are different sizes.
• Events in the Far Detector can have different detection

efficiency depending on distance to readout.

max FD max ND

18

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

X (cm)

• 200
• 150
• 100
• 50

50 100 150 200 0.2 0.4 0.6 0.8 1

X (cm)

• 200
• 150
• 100
• 50

50 100 150 200 0.2 0.4 0.6 0.8 1

Event containment differences

• In the NOvA Near Detector 82-87% of neutrino events are
• contained. Also Up to 10% of the NC lose a !0.
• We do not expect these effects to be present in the Far Detector.

NC 1-2 GeV νe 1-2 GeV

19

Energy νe CC νμ CC NC NC w/lost !0

1-2 GeV 85 ± 1% 59 ± 1% 87 ± 2% 10 ± 2% 2-3 GeV 85 ± 1% 48 ± 1% 82 ± 3% 8 ± 2%

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

• As in MINOS we plan to predict the event rate at each energy bin by

correcting the expected Monte Carlo rate using the ratio of data to Monte Carlo in the Near Detector.

• Far/Near ratio accounts for geometry, fiducial volume ratio, intensity,

detector differences and oscillations.

• The Far/Near ratios for muon and electron neutrinos in the beam before

selection are shown below.

20

Far/Near νμ Far/Near νe

Predicting the FD background

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

• The Near Detector νe selected NC

and νμ CC background components are corrected by the Far/Near MC ratio.

• Data (modified Monte Carlo in

this case) and MC are compared in the ND. The ratio is used to modify the prediction int he FD.

• The prediction is expected to be a

closer match to the Data in the FD having been corrected.

Reco Energy (GeV) 1 2 3 POT

20

10 ! Events / 0.5 GeV / 18 2 4 6 8

Reco Energy (GeV) 1 2 3 4 FD Events / ND Events 0.0001 0.0002 0.0003 All

e

"

µ

" NC

Predicting the FD background

~10-4 expected from geometry and fiducial volume ratio alone 21

Reco Energy (GeV) 1 2 3 POT

20

10 ! Events / 0.5 GeV / 18 50 100 150

3

10 !

MC “data” MC “data” Pred ND FD

Ratios and energy distributions for MA(QE) changed by +30%

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Expected NOvA systematic errors

• We evaluated a set of neutrino interaction

systematic uncertainties on the background for electron neutrino appearance in NOvA.

• The largest systematic error arises from

the Pt and Xf changes at 5%.

• All other errors are within 3% for

background, currently limited by the statistics of the study.

• For the signal the largest uncertainties

correspond to the cross section systematics.

• These are expected to be corrected using

the extrapolation of the νμ CC spectrum from the Near Detector to less than 1%.

• All others systematics on the signal are also

within the statistics of the study.

22

Reco Energy (GeV) 1 2 3 POT

20

10 ! Events / 0.5 GeV / 18 2 4 6 8

Background

Reco Energy (GeV) 1 2 3 POT

20

10 ! Events / 0.5 GeV / 18 10 20

Signal MC “data” Pred “data” MC

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Effects on NOvA sensitivities

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of hierarchy determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

A hierarchy determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr Shown with and without 10% background errors

• Neutrino interaction systematics on background for NOvA are well

within the required 10% overall systematics.

• In this study we estimated them to be within 3%, with the largest
• ne possibly reaching 5%.

23

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Effects on NOvA sensitivities

• Neutrino interaction systematics on background for NOvA are well

within the required 10% overall systematics.

• In this study we estimated them to be within 3%, with the largest
• ne possibly reaching 5%.

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of CPv determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0

A CPv determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr Shown with and without 10% background errors

24

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Effects on NOvA sensitivities

• Neutrino interaction systematic uncertainties on the signal for NOvA are

expected to be very small.

• Other systematic effects unrelated to neutrino interactions are under

consideration.

• We do not expect these systematic errors to be an issue in our first analysis.

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of CPv determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0

A CPv determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: signal systematic errors at 0%, 3%, and 6%

25

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Conclusions

• The two detector technique allows significant cancellation of neutrino interaction

(cross-section, hadronization model and intranuclear rescattering) systematic errors.

• In MINOS first analyses, it resulted in a reduction of systematic errors from 20% to

about 4% demonstrating the power of this technique.

• In the most recent analyses these uncertainties are below 2.5%.
• In a preliminary study for NOvA, we have determined these systematic uncertainties

are small and cancel to less than 3% (limited by the statistics of the study).

• Current studies are based on models and studies done for MINOS, tuning and

studies with other models within GENIE are expected as part of the NOvA program.

• Hadronization and intranuclear model uncertainties such as those on the Pt and Xf

hadron distributions are expected to be the larger systematic errors in this class.

• Cross section systematic uncertainties on the electron neutrino signal can be

corrected down to 1% using muon neutrino charge current interactions. Other systematic errors will dominate the signal.

• This level of systematic errors are well within the tolerance of NOvA sensitivities to

mass hierarchy, CP violation and the octant determination.

26

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Backup

27

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

MINOS Monte Carlo

• MC tuned to external bubble

chamber data for hadronization (or fragmentation) models.

• Region of interest in mass of

hadronic system: 1 - 15 GeV2 in W2

• r 1 - 4 GeV in W.
• However, data available is for

relatively higher energy than the region of interest for this analysis.

Good agreement in W2 for forward and backward hemispheres

Also agreement for !0 vs charged hadron multiplicities for different W2 28

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

Other Far/Near differences

• Far/Near spectrum different due to beamline geometry and oscillations in the Far.
• Readout patterns:
• Light level differences due to differences in fiber length.
• Multiplexing in the Far (8 fibers per PMT pixel).
• Partial (one-sided) readout in the Near.
• Photomultipliers (M64 in Near Detector, M16 in Far):
• Different gains/front end electronics.
• Different crosstalk patterns (also related to readout patterns).
• Neutrino intensity:
• higher rates in the Near Detector thus faster readout.
• Relative energy calibration.

MINOS detectors are very similar, however there are small differences:

These considerations affect the Far/Near ratio and are considered in the extrapolation uncertainties

29

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of hierarchy determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

A hierarchy determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 20%

Effects on NOvA sensitivities

• Neutrino interaction systematics on background for NOvA are well

within the required 10% overall systematics.

• In this study we estimated them to within 3%, with the largest one

possibly reaching 5%.

30

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of CPv determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0

A CPv determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 20%

Effects on NOvA sensitivities

• Neutrino interaction systematics on background for NOvA are well

within the required 10% overall systematics.

• In this study we estimated them to within 3%, with the largest one

possibly reaching 5%.

31

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of octant determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

A octant determination for different exposures and systematic errors $ NO

>0

2

m % , ° >45

23

& )=0.97,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 20%

Effects on NOvA sensitivities

• Neutrino interaction systematics on background for NOvA are well

within the required 10% overall systematics.

• In this study we estimated them to within 3%, with the largest one

possibly reaching 5%.

32

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of hierarchy determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

A hierarchy determination for different exposures and systematic errors $ NO

>0

2

m % )=1.00,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: signal systematic errors at 0%, 4%, and 8%

Effects on NOvA sensitivities

• Neutrino interaction systematic uncertainties on the signal for NOvA are

expected to be very small.

• Other systematic effects unrelated to neutrino interactions are under

consideration.

• We do not expect these systematic errors to be an issue in our first analysis.

33

Mayly Sanchez - ISU/ANL NuFact 2012 - 07/24/12

) ! / (2 " 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ) # significance of octant determination ( 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

A octant determination for different exposures and systematic errors $ NO

>0

2

m % , ° >45

23

& )=0.97,

23

& (2

2

)=0.095, sin

13

& (2

2

sin 3+3 yr ) $ + $ 1+1 yr ( 5+5 yr For each: signal systematic errors at 0%, 4%, and 8%

Effects on NOvA sensitivities

• Neutrino interaction systematic uncertainties on the signal for NOvA are

expected to be very small.

• Other systematic effects unrelated to neutrino interactions are under

consideration.

• We do not expect these systematic errors to be an issue in our first analysis.

34