Neutrino interaction systematic errors in MINOS and NOvA

Neutrino interaction systematic errors in MINOS and NOvA Mayly Sanchez Iowa State University Argonne National Laboratory Nufact 01 - Williamsburg, VA July 4, 01

MINOS and NOvA in a nutshell Produce a high intensity beam of 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 observe a distortion in the data at the Far Detector at either site. NO!A Far Detector 810 km MINOS Far Detector 735 km 1st/nd generation long baseline See more general MINOS and NOvA talks on Friday Mayly Sanchez - ISU/ANL

The MINOS detectors Functionally identical: Near and Far detectors Octogonal steel planes (.54cm thick ~1.44X 0 ) Alternating with planes of scintillator strips (4.1cm wide, Moliere rad ~3.7cm). Near (ND): ~ 1kton, 8 steel squashed octagons. Partially instrumented. Far (FD): 5.4 kton, 486 (8m/octagon) fully instrumented planes. Near Far Mayly Sanchez - ISU/ANL 3

Neutrino event topologies in MINOS To select ν e CC we focus on finding compact showers. MC events Signal! e CC Event NC Event e - Irreducible Background! 0! µ CC Event NC Event Reducible Background µ -! Mayly Sanchez - ISU/ANL 4

MINOS Monte Carlo Region of interest: 1-15 GeV in W 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 W is an extrapolation. T. Yang et al., Eur. Phys. Jour. C 63:1-10 (009) Neutrino data taken with the Near Detector was used to correct the Far Detector expectation. Mayly Sanchez - ISU/ANL 5

MINOS electron neutrino selection Initially the ND predicted backgrounds were 0% 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. Events 5000 4000 3000 Near Detector MINOS PRELIMINARY the early days 000 1000 0 0. 0.4 0.6 0.8 1 ANN Total Data Total MC 009 Mayly Sanchez - ISU/ANL 6

MINOS electron neutrino selection Initially the ND predicted backgrounds were 0% 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. Events POT 19 Events/GeV/10 5000 4000 3000 Near Detector MINOS PRELIMINARY Total Data Total MC 000 1000 0 0. 0.4 0.6 0.8 1 1000 ANN Near Detector MINOS PRELIMINARY ANN-selected 000 Total Data Total MC 010 0 1 3 4 5 6 7 8 9 Reconstructed Energy (GeV) Mayly Sanchez - ISU/ANL 7

Predicting the FD background The Near Detector ν e selected NC and ν μ CC background components are corrected by the Far/Near MC ratio. ~10-4 expected from geometry and fiducial volume ratio alone 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 ν e and the ν τ CC from ν μ oscillations are corrected using the extrapolation of the ν μ CC spectrum. Far/Near Ratio (x10-4 ) Far/Near Ratio (x10-4 ) Far/Near Ratio (x10-4 ) 3.5 3.0.5 MINOS PRELIMINARY! CC.0 1.5 1.0 0.5 0.0 1 3 4 5 6 7 8 3.5 3.0.5 Reconstructed Energy (GeV) Monte Carlo NC.0 1.5 1.0 0.5 0.0 1 3 4 5 6 7 8 3.5 3.0.5 Reconstructed Energy (GeV) beam e CC.0 1.5 1.0 0.5 0.0 1 3 4 5 6 7 8 Reconstructed Energy (GeV) Mayly Sanchez - ISU/ANL 8

MINOS systematic errors 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 010 For the main background components the hadronization model systematic is corrected to about 4%, while intranuclear and cross sections are down to 1% or less. More recent analyses have these below.5%. Mayly Sanchez - ISU/ANL 9

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 X 0. Great for e - vs! 0. -.1'"$0&345!"#$%&'("%) &*%+,&-.&/$00) ),(;!<(='">,?+!"#$%&'&()*+(,'--.(/--'&(01#( 334(-1#'$.(56(.,178--9#5$ 17.#$%:'7#'&(01#!;.1<17=(/>'$(97&(?)@. Far Detector 14 kton 98 layers.%/0$&&"$,0(1/3+!"#"$!"#%)$!"#$%&'()#"*'%+,#- Near Detector 0.3 kton 06 layers Prototype installed on the surface and taking data! 4"5'&'0+3(63/7/0 8/9'#(1,,: '#(&$ Mayly Sanchez - ISU/ANL 10 %#%&$

!)$101A Neutrino event topologies in NOvA Monte Carlo events.!)$101a =A>0>4 "(A(' $->4 @,&61.)' B Background Signal $(A('?@)+6>51A # B (A(' </= Background *+,$-./01)13140+)5,06)7)819)3,+,:.1; NOvA uses GENIE for neutrino interactions, same tuning as MINOS. Mayly Sanchez - ISU/ANL 11

NOvA electron neutrino selection Several electron neutrino identification techniques are being developed. Events 15 1 1.8!10 POT FHC, sin 13 =0.1 Performance already comparable to the NOvA TDR. 10 PID re-optimized for large θ 13. Method used for this study is based on library matching algorithm (a la MINOS). Current implementation is promising and has room for improvement. 5 0 0 0.5 1 Signal:Bkg Signal PID An alternate method based on transverse/longitudinal likelihoods of the shower energy profile will be shown in the poster session. PID>0.7 100:40 PID>0.9 70:15 NC CC beam Mayly Sanchez - ISU/ANL 1

NOvA electron neutrino selection Several electron neutrino identification techniques are being developed. Events 15 1 1.8!10 POT FHC, sin 13 =0.1 Performance already comparable to the NOvA TDR. 10 PID re-optimized for large θ 13. Method used for this study is based on library matching algorithm (a la MINOS). Current implementation is promising and has room for improvement. 5 0 0 0.5 1 Signal:Bkg Signal PID An alternate method based on transverse/longitudinal likelihoods of the shower energy profile will be shown in the poster session. PID>0.7 100:40 PID>0.9 70:15 NC beam CC Mayly Sanchez - ISU/ANL 13

NOvA electron neutrino selection Several electron neutrino identification techniques are being developed. Events 15 1 1.8!10 POT FHC, sin 13 =0.1 Performance already comparable to the NOvA TDR. 10 PID re-optimized for large θ 13. Method used for this study is based on library matching algorithm (a la MINOS). Current implementation is promising and has room for improvement. 5 0 0 0.5 1 Signal:Bkg Signal PID An alternate method based on transverse/longitudinal likelihoods of the shower energy profile will be shown in the poster session. PID>0.7 100:40 PID>0.9 70:15 NC beam CC Mayly Sanchez - ISU/ANL 14

NOvA electron neutrino selection The electron neutrino selection prefers low hadronic y ν e signal events and high hadronic y ν μ CC background events. For neutral current events the main background arises from events with one or more! 0. Selection Efficiency (%) 60 40 0 signal e 0 0 0. 0.4 0.6 0.8 1 Hadronic y Selection Efficiency (%) 3 Neutral currents Selection Efficiency (%).5 1.5 1 PID>0.7 PID>0.9 CC! 1 0.5 0 0 1 3 4 5 Number of 0 s 0 0 0. 0.4 0.6 0.8 1 Hadronic y Mayly Sanchez - ISU/ANL 15

Studying systematic errors in NOvA The neutrino interaction systematic errors are modified in this study: Cross-section: M A (QE) and M A (RES) varied by ± 0%. Hadronization model changes: The! 0 selection probability in the hadronization model changed by ± 33%. Change in average P t resulting in broader showers. Re-weighting P t and X f 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. Mayly Sanchez - ISU/ANL 16

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. Far Detector Near Detector Mayly Sanchez - ISU/ANL 17

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 ND max FD Mayly Sanchez - ISU/ANL 18

Event containment differences In the NOvA Near Detector 8-87% 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. 1 1 0.8 0.8 0.6 0.6 0.4 0. ν e 1- GeV 0.4 0. NC 1- GeV 0-00 -150-100 -50 0 50 100 150 00 X (cm) 0-00 -150-100 -50 0 50 100 150 00 X (cm) Energy ν e CC ν μ CC NC NC w/lost! 0 1- GeV 85 ± 1% 59 ± 1% 87 ± % 10 ± % -3 GeV 85 ± 1% 48 ± 1% 8 ± 3% 8 ± % Mayly Sanchez - ISU/ANL 19

Predicting the FD background 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. Far/Near ν μ Far/Near ν e Mayly Sanchez - ISU/ANL 0

Predicting the FD background The Near Detector ν e selected NC and ν μ CC background components are corrected by the Far/Near MC ratio. FD Events / ND Events 0.0003 0.000 0.0001 All " e " µ NC ~10-4 expected from geometry and fiducial volume ratio alone 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. POT 0 10 Events / 0.5 GeV / 18! 150 100 50 0 0 1 3 4 Reco Energy (GeV) 0 0 1 3 Reco Energy (GeV) 0 0 1 3 Reco Energy (GeV) Mayly Sanchez - ISU/ANL 1 3!10 data MC ND POT 0 10 Events / 0.5 GeV / 18! 8 6 4 data MC Pred FD Ratios and energy distributions for M A (QE) changed by +30%

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 P t and X f 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 ν μ CC spectrum from the Near Detector to less than 1%. All others systematics on the signal are also within the statistics of the study. POT 0 POT Events / 0.5 GeV / 18! 10 0 Events / 0.5 GeV / 18! 10 8 6 4 Background 0 0 1 3 Reco Energy (GeV) 0 10 data MC Pred Signal data MC 0 0 1 3 Reco Energy (GeV) Mayly Sanchez - ISU/ANL

Effects on NOvA sensitivities significance of hierarchy determination (#) NO$A hierarchy determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 4.5 4.0 3.5 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr Shown with and without 10% background errors 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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 one possibly reaching 5%. Mayly Sanchez - ISU/ANL 3

Effects on NOvA sensitivities significance of CPv determination (#) NO$A CPv determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr Shown with and without 10% background errors 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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 one possibly reaching 5%. Mayly Sanchez - ISU/ANL 4

Effects on NOvA sensitivities significance of CPv determination (#) NO$A CPv determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: signal systematic errors at 0%, 3%, and 6% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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. Mayly Sanchez - ISU/ANL 5

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 0% to about 4% demonstrating the power of this technique. In the most recent analyses these uncertainties are below.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 P t and X f 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. Mayly Sanchez - ISU/ANL 6

Backup Mayly Sanchez - ISU/ANL 7

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 GeV in W or 1-4 GeV in W. However, data available is for relatively higher energy than the region of interest for this analysis. Good agreement in W for forward and backward hemispheres Also agreement for! 0 vs charged hadron multiplicities for different W Mayly Sanchez - ISU/ANL 8

Other Far/Near differences MINOS detectors are very similar, however there are small 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. These considerations affect the Far/Near ratio and are considered in the extrapolation uncertainties Mayly Sanchez - ISU/ANL 9

Effects on NOvA sensitivities significance of hierarchy determination (#) NO$A hierarchy determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 4.5 4.0 3.5 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 0% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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%. Mayly Sanchez - ISU/ANL 30

Effects on NOvA sensitivities significance of CPv determination (#) NO$A CPv determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 0% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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%. Mayly Sanchez - ISU/ANL 31

Effects on NOvA sensitivities significance of octant determination (#) NO$A octant determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=0.97, & 3 >45, %m >0 5.0 4.5 4.0 3.5 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: BG syst. error at 0%, 5%, 10%, 15%, 0% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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%. Mayly Sanchez - ISU/ANL 3

Effects on NOvA sensitivities significance of hierarchy determination (#) NO$A hierarchy determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=1.00, %m >0 4.5 4.0 3.5 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: signal systematic errors at 0%, 4%, and 8% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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. Mayly Sanchez - ISU/ANL 33

Effects on NOvA sensitivities significance of octant determination (#) NO$A octant determination for different exposures and systematic errors sin (& 13 )=0.095, sin (& 3 )=0.97, & 3 >45, %m >0 5.0 4.5 4.0 3.5 3.0.5.0 1.5 1.0 0.5 1+1 yr ($ + $) 3+3 yr 5+5 yr For each: signal systematic errors at 0%, 4%, and 8% 0.0 0 0.1 0. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 " / (!) 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. Mayly Sanchez - ISU/ANL 34