Constraining systematic errors using the T2K near detector K Mahn, NuFACT Kendall Mahn For the T2K collaboration 2012/07/241

The Tokai-to-Kamioka (T2K) experiment T2K sends a ~1 GeV   beam across Japan to study neutrino oscillations  Measure  μ disappearance ( Δm 2 32,  23 )  Deficit of   rate and distortion of   energy spectrum  Measure  e appearance (  13 )  Excess of  e rate over background  Subleading terms depend on mass hierarchy,  CP Far detector: Super-Kamiokande located near Kamioka Beam source and near detectors: J-PARC accelerator complex located in Tokai-mura 2012/07/242K Mahn, NuFACT  Precision measurements depend upon understanding of systematic errors

Neutrino interactions at T2K K Mahn, NuFACT W CCQE l ν p n Z ν N N’ Δ π ν NC π Most interactions are Charged Current Quasi Elastic  Neutrino flavor determined from flavor of outgoing lepton i.e. e for  e,  μ for    Llwellyn Smith base model, with Smith-Moniz relativistic Fermi Gas representation of nucleus W l ν N N’ Δ π CC π 2012/07/243  /E ( cm 2 /GeV) Additional interactions important for analysis are CCπ and NCπ (single pion production)  Rein-Seghal resonance model E  ( GeV)

Off-axis near detectors: ND280 P0D ECAL Barrel ECAL K Mahn, NuFACT Suite of near detectors sit within UA1 (B=0.2T, 850 tons) magnet Measure unoscillated CC   rate in Tracker:  Neutrinos interact on FGDs (carbon target)  Measure lepton angle, momentum and flavor with TPCs  Separate selection into CCQE-like (CCQE) and all other CC (CCnQE)  CC ν e rates also measured in P0D, Tracker μ νμνμ TPC2 FGD1 Downstream ECAL 2012/07/244 NIM A 624, 591 (2010) p? or π? ND280 CCnQE   candidate

Off-axis near detectors: ND280 (ND) P0D ECAL Barrel ECAL K Mahn, NuFACT μ νμνμ TPC2 FGD1 Downstream ECAL 2012/07/245 NIM A 624, 591 (2010) W CCQE   p n W N N’ Δ π CC π +   Suite of near detectors sit within UA1 (B=0.2T, 850 tons) magnet Measure unoscillated CC   rate in Tracker:  Neutrinos interact on FGDs (carbon target)  Measure lepton angle, momentum and flavor with TPCs  Separate selection into CCQE-like (CCQE) and all other CC (CCnQE)  CC ν e rates also measured in P0D, Tracker

Far detector: Super-Kamiokande (SK) K Mahn, NuFACT  22.5kton fiducial mass Cherenkov detector (water target)  11,129 inner region PMTs  1885 outer veto region PMTs  Select  e events from ring shape and topology 2012/07/246  e appearance analysis Signal: CC  e   to  e oscillation Background: CC  e Irreducible beam  e Background: NC   Mimics CC  e

Neutrino flux prediction Oscillation analysis strategy 2012/07/24 7 K Mahn, NuFACT Neutrino cross section prediction Neutrino cross section prediction Neutrino event rate constraint using ND CCQE and CCnQE   data samples Neutrino event rate constraint using ND CCQE and CCnQE   data samples Neutrino event rate at SK (  , ν e ), with rate correction and reduced uncertainties from ND Neutrino event rate at SK (  , ν e ), with rate correction and reduced uncertainties from ND Determine oscillation parameters   disappearance: Δm 2 32,  23 ν e appearance:  13 Determine oscillation parameters   disappearance: Δm 2 32,  23 ν e appearance:  13 Φ (E ν )σ (E ν ) N = Φ x σ x ε Stepwise approach:  Strong constraint where flux, cross section is same at SK, ND  Weaker constraint where acceptance, efficiency is different

Off-axis (narrow band) neutrino beam FLUKA/Geant3 beam simulation Unoscillated flux at SK:    from π +  K decay  ~1%   from  K decay Neutrino flux prediction K Mahn, NuFACT Prediction and uncertainties determined by external or in-situ measurements  π, K production from NA61 experiment Phys.Rev.C 84, (2011) Phys.Rev.C 85, (2012) Details in Wed joint WG1-2-3 session, S. Murphy 2012/07/248

Neutrino flux at ND and SK K Mahn, NuFACT2012/07/249 ND samples represent   flux    from π decay: CCQE, CCnQE samples    from K decay: CCnQE sample

Neutrino flux at ND and SK K Mahn, NuFACT2012/07/2410 SK signal and NC background events come from   flux directlymeasured at ND p-θ of pions which produce   in ND p-θ of pions which produce   from   oscillations in SK

Neutrino flux at ND and SK K Mahn, NuFACT2012/07/2411 CC background from beam   is strongly correlated with   flux at ND: 11 p-θ of pions which produce   in ND p-θ of pions which produce muons which decay to   π+π+ ++    

Neutrino interaction uncertainties K Mahn, NuFACT Cross section model (NEUT, GENIE) composed of:  Base model prediction  Final state interaction model (FSI) 2012/07/2412 NC π + to NC π 0 Cross section model uncertainties set from fits to MiniBooNE data (E  ~ 1 GeV) for signal and background (CCQE, CC1π and NCπ 0 ) interactions  Single pion (CC and NC) interaction datasets fit simultaneously (details in Thurs WG2 session by P. Rodrigues)  SciBooNE, K2K datasets used as cross check ν p ν π0π0 π+π+ Δ ++ p n p 16 O

Neutrino interactions at ND and SK K Mahn, NuFACT CCQE and CC1π are the largest interaction mode in ND, SK samples Caveats:  Acceptance: ND sample is forward going (small angle, low Q 2 )  External data covers larger Q 2 (MiniBooNE, 4π Cherenkov detector)  Target: ND selection is C, SK is O  C-O model dependent uncertainties included 2012/07/2413

Neutrino interactions at ND and SK K Mahn, NuFACT2012/07/2414  Indirect constraint on NC (1π 0 ) through CC1π in ND measurement  Additional ND selection of NCπ 0 with P0D detector  No MiniBooNE-based tuning applied here Data NCπ 0 signal + background Background only

Neutrino flux term ND280 constraint fit pieces 2012/07/24 15 K Mahn, NuFACT Neutrino cross section term Flux parameterization, f  Normalization on E ν bins  11 bins for   flux Cross section parameterization, x  Combination of normalization and model parameters  Used reweighting techniques MAQE (GeV)Axial mass (QE) MARES (GeV)Axial mass (1π) QE1 0<E ν <1.5 GeVNormalization QE2 1.5<E ν <3.5 GeVNormalization QE3 E ν >3.5 GeVNormalization CCRES1 E ν <2.5 GeVNormalization CCRES2 E ν >2.5 GeVNormalization NC1π 0 Normalization pF (MeV/c)Fermi momentum Spectral FunctionModel comparison CC otherNormalization

Neutrino flux term ND280 constraint fit pieces 2012/07/24 16 K Mahn, NuFACT Neutrino cross section term ND280 constraint Detector systematics, d Normalization on recon momentum, angle bins Compare CCQE, CCnQE sample data to prediction assuming a given flux, cross section pars Binned in p μ -cos(θ μ )

Neutrino flux term ND280 constraint fit pieces 2012/07/24 17 K Mahn, NuFACT Neutrino cross section term ND280 constraint Minimize likelihood and propagate constrained parameters Minimize likelihood and propagate constrained parameters Determine oscillation parameters Marginalize over uncorrelated cross section, detector parameters Fit for flux and shared ND/SK xsec parameters (f’, x’) Fit correlates f’ and x’ Oscillation parameters, o   disappearance: Δm 2 32,  23 ν e appearance:  13 Apply tuned f’, x’ parameters to SK prediction

ND280 likelihood 2012/07/24 18 K Mahn, NuFACT Neutrino flux term Neutrino xsec term ND280 constraint

ND280 likelihood 2012/07/24 19 K Mahn, NuFACT Likelihood function, with Poisson statistics Fit CCQE, CCnQE p μ -θ μ distribution (20x2 bins) Sensitive to to rate (Φ x σ ) changes:

ND280 likelihood 2012/07/24 20 K Mahn, NuFACT Prior constraint terms for flux, cross section parameters  V f and V x are covariance matricies  Determined using in-situ and external datasets: beam monitors, NA61, MiniBooNE

ND280 likelihood 2012/07/24 21 K Mahn, NuFACT Prior constraint likelihood terms for detector systematic errors  Also includes uncertainties (e.g. FSI) which could not be otherwise easily parameterized  Determined from control samples, calibration data, and external pion scattering data

Detector systematic errors 2012/07/24 22 K Mahn, NuFACT Fractional systematic uncertainty for vs. momentum Currently statistics dominated ND CCQE sample 0.94<cos(θ μ ) <1

ND measurement results K Mahn, NuFACT2012/07/24 23 Distribution of ND samples Δχ 2 = 29.1 (p-value based on pseudo-experiments) Data without ND information with ND information

Effect of ND measurement on ν e signal, background  Rate of  e signal and backgrounds without ND measurement and with ND measurement  Uncertainty envelope from constrained flux, cross section parameters  Includes correlation between flux and cross section at ND, SK 2012/07/24 24 K Mahn, NuFACT Signal: CC  e Background: CC  e Background: NC   % change to event rate

Total systematic uncertainty for ν e appearance K Mahn, NuFACT2012/07/2425 Background# events beam  e +  e 1.73   +   (mainly NC) background 1.31 osc through  total:3.22±0.43(sys) Signal (ν μ to ν e osc)# 2 2θ 13 =0.1,δcp=07.81 Uncertaintiesν e bkrdν e sig+bkrd ν flux+xsec (constrained by ND280) ±8.7%±5.7% ν xsec (unconstrained by ND280) ±5.9%±7.5% Far detector±7.7%±3.9% Total±13.4%±10.3%   Δ m 2 32 =2.4 x eV 2, sin 2 2  23 =1.0 Event predictions normalized by ND280 Determine sin 2 2  13 from  e candidates’ rate and kinematic distributions More details in Fri plenary session (C. McGrew) No ND measurement26%22%

Total systematic uncertainty for ν e appearance K Mahn, NuFACT2012/07/2426 Background# events beam  e +  e 1.73   +   (mainly NC) background 1.31 osc through  total:3.22±0.43(sys) Signal (ν μ to ν e osc)# 2 2θ 13 =0.1,δcp=07.81 Uncertaintiesν e bkrdν e sig+bkrd ν flux+xsec (constrained by ND280) ±8.7%±5.7% ν xsec (unconstrained by ND280) ±5.9%±7.5% Far detector±7.7%±3.9% Total±13.4%±10.3%   Δ m 2 32 =2.4 x eV 2, sin 2 2  23 =1.0 Event predictions normalized by ND280 Reduction in total uncertainty from:  Reduced individual uncertainties (e.g. far detector and flux uncertainties)  Improved use of ND information Determine sin 2 2  13 from  e candidates’ rate and kinematic distributions More details in Fri plenary session (C. McGrew) 2011 (normalization- only) analysis 23%18%

Summary K Mahn, NuFACT Precision measurements of neutrino mixing parameters requires increased control over systematic uncertainties:    disappearance ( Δm 2 32,  23 ) and  e appearance (  13 ) For the T2K  e appearance analysis, the systematic errors are constrained to ~10% on the far detector expected  e rate using data from the near detector  Improvement since 2011 result (~22%)    disappearance results will be updated soon with the improved near detector constraint Future T2K analyses will extend the current ND measurement to include:  More control samples (CC1π selection)  Backward going tracks for increased acceptance  Carbon and water target comparisons 2012/07/2427

Backup slides K Mahn, NuFACT2012/07/2428

Core of analysis: Flux parameterization 2012/07/24 K Mahn, NuFACT Flux parameterization: f i Normalization on E  bin i for SK and ND samples Current binning for T2K flux:    : , , , , , , , , , , GeV (11 bins each ND and SK samples, index 0-10 and 11-21)    : , GeV (2 bins, SK sample, index 22-23)   e : , , , , , , GeV (7 bins, index 24-30)   e : , GeV (2 bins, SK sample, index 31-32) Neutrino flux prediction 29

Flux parameterization 2012/07/24 30 K Mahn, NuFACT Correlations in flux covariance are shared hadron production uncertainties Flux covariance built from measurements of beam or external data (e.g. NA61) Correlations between flux bins ND   SK   SK  e Neutrino flux prediction Flux parameterization: f i Normalization on E  bin i for SK and ND samples

T2K neutrino flux uncertainties K Mahn, NuFACT312012/07/24   flux fractional uncertainties

Cross section parameterization 2012/07/24 32 K Mahn, NuFACT Cross section parameterization: x k Model parameters: MAQE and MARES (modify Q 2 distribution of QE and resonant 1pi cross sections) Fermi momentum (pF) provides low Q2 handle, and is target dependant (C vs. O) Spectral function – RFG model- model difference is also target dependant Normalizations provide overall scaling independent of Q 2 on a particular interaction Apply cross section to observables at ND, SK using reweighting techniques Parameter value, uncertainty is extrapolated to SK sample Parameter value, uncertainty is determined from MiniBooNE single pion samples

ND CCQE and CCnQE ν μ samples K Mahn, NuFACT CC inclusive   candidate selection: 1.Require no tracks in TPC1 (veto  interactions upstream of FGDs) 2.Select events which originate in FGD1 fiducial volume 3.Use the highest momentum, negative TPC2 track 4.Select μ from TPC dE/dx information to determine flavor of neutrino 2012/07/24 33 Separate sample into two subsamples, CCQE enhanced and CCnQE enhanced CCQE enhanced: 1 TPC-FGD matched track, no decay electron in FGD1 CCnQE enhanced: all other CC inclusive Dataset shown here: 1.43 x POT Prediction includes NA61, MiniBooNE (external) data tuning

Flux parameters change after ND measurement 2012/07/24 34 K Mahn, NuFACT   flux parameters at ND   flux parameters at SK   flux parameters at SK  Flux parameters without ND measurement and with ND measurement

Covariance 2012/07/24 35 K Mahn, NuFACT  ,  e fluxes are correlated according to external data and how flux at ND, SK is shared Correlations between ND, SK shared cross section parameters determined from external data fits   and cross section parameters prior to ND measurement

Covariance post-ND measurement 2012/07/24 36 K Mahn, NuFACT Flux and shared cross section become correlated   and cross section parameters after ND measurement

ν μ disappearance results K Mahn, NuFACT  Disappearance distorts energy spectrum and rate of   candidates  31 events pass   selection with expected for no oscillation Reconstructed energy E  (GeV) 2012/07/2437 MINOS: Phys. Rev. Lett. 101, (2008) Super-K: Phys.Rev.D71: (2005) Summary of uncertainties ν μ signal Δm 2 23 =2.4 x eV 2 sin 2 2θ 23 =1.0 ν flux±4.8% ν interactions % Near detector % Far detector±10.3% Total % Phys. Rev. D 85, (R) (2012)

Canada TRIUMF U of Alberta U of B Columbia U of Regina U of Toronto U of Victoria U Winnipeg York U France CEA Saclay IPN Lyon LLR E Poly LPNHE-Paris Germany RWTH Aachen U Italy INFN Bari INFN Roma Napoli U Padova U Japan ICRR Kamioka ICRR RCCN KEK Kobe U Kyoto U Miyagi U of Ed Osaka City U U of Tokyo Korea Chonnam Nat’l U Dongshin U Seoul Nat’l U Poland NCBJ IFJ PAN T U Warsaw U of Silesia Warsaw U Wroclaw U Russia INR Switzerland Bern ETH Zurich U of Geneva UK U of Oxford Imperial C London Lancaster U Queen Mary U of L Sheffield U STFC/RAL STFC/Daresbury U of Liverpool U of Warwick USA Boston U Colorado State U Duke U Louisiana State U Stony Brook U U of California, Irvine U of Colorado U of Pittsburgh U of Rochester U of Washington The T2K Collaboration Spain IFIC, Valencia IFAE, Barcelona K Mahn, NuFACT Italy INFN Bari INFN Roma Napoli U Padova U 2012/07/2438

TPC dE/dx particle ID Energy loss of the particle (dE/dx) can be used to separate particle type dE/dx resolution for MIPs is 8% Probability for a muon between 0.2 and 1.0 GeV to be identified using dE/dx as an electron is less than 0.2% PID “pull” variable K Mahn, NuFACT2012/07/2439

ND280 beam ν e rate cross-check (I) 4/11/1240 Select  e candidates at ND280 with TPC PID to check rate of intrinsic beam  e Additional backgrounds to  e selection, measured via control samples   misidentified as e  e from photon conversion (photons emitted in   interactions in FGD and other subdetectors) Ratio of observed  e /   events is consistent with untuned prediction N(  e )/ N(   ) = R(e:  ) = 1.0% ± 0.7% (statistics) ± 0.3% (systematics) R(e: , data) / R(e: , MC) = 0.6 ± 0.4 (statistics) ± 0.2 (systematics) N(  e )/ N(   ) = R(e:  ) = 1.0% ± 0.7% (statistics) ± 0.3% (systematics) R(e: , data) / R(e: , MC) = 0.6 ± 0.4 (statistics) ± 0.2 (systematics) Improvements to the analysis:  Improved rejection of backgrounds with ECals  More data: 2.88 x POT shown here  e selection candidates vs. momentum K Mahn, Columbia HEP seminar

ND280 beam ν e rate cross-check (II) 4/11/1241 Select high energy CC  e candidates within the P0D:  Reconstructed track matched in x,y with vertex in FV consistent with an single EM shower (reject π 0 mutiple photon showers and muons)  Primary backgrounds are HE π 0 events Consistent with current untuned MC: data-bkrd(MC)/sig(MC)= R = 1.19 ± 0.15(statistics) ± 0.26 (systematics)  e selection candidates vs. energy, assuming QE K Mahn, Columbia HEP seminar