Latest results from NOvA Rencontres de Moriond March 5, 8 Chris Backhouse University College London
The NOvA experiment νµ disappearance symmetries in neutrino mixing νe appearance neutrino mass ordering CP-violation Future / 9
NOvA,ft view νµ beam from Fermilab, L Detector 8km away in MN Smaller detector onsite to measure flux before oscillations νµ νµ ν µ ν µ νµ νe ν µ ν e Precision measurements of and θ m Determine the mass hierarchy Search for sin δcp 6= / 9
NuM performance Week ending : Monday March 8.4 Mean RMS 49.4.446.5 This week World s highest power neutrino beam 7kW design power since June 6 5 protons / pulse. Apr-May 5.5..5. 8 Protons per week ( 8) 5 Low Energy antineutrinos Low Energy neutrinos Higher Energies NOvA neutrinos NOvA antineutrinos 6.5. 4 5 5 8 6 Total Protons ( ) 4 5 5/5/ 6// 8/7/ // /9/ Date /4/ 4// 6/6/ 8// 5 5 5 5 4 45 5 Protons per pulse (E) Week ending : Monday March 8.9 Mean RMS 67.98 87.68.8 This week.7 Apr-May 5.6.5.4.... 4 5 6 7 Beam power (kw) These results use data from Feb 6 4 to Feb 7 Beam power ramping up, detector under construction at start 8.85 POT equivalent, about.5 years of nominal running / 9
Detector technology 64% liquid scintillator by mass 4 6cm resolution, two views for D reco. 44, channels in 4 kton FD, on surface ton ND, underground at FNAL 4 / 9
Event topologies Very good granularity, especially considering scale X = 8cm (6 cell depths, cell widths) 5 / 9
What s new? 5% more data than previous results ( 6 9 POT) Data-driven flux estimates from MNERvA Retuned cross-section model Detector sim. improvements Using CVN computer vision classifier for all analyses Analysis improvements Resolution binning for νµ Peripheral sample for νe NOvA Simulation Light Models: Normalized to Minimum onization 6 Birks-Chou Model Photons/cm 7 Birks Model + Cherenkov Light Proton KE 9 MeV βγ Phys. Rev. Lett. 8 (7) 58 and 8 Phys. Rev. D94 (6) 95 6 / 9
Nuclear correlations (Mark Hartz) ND data reveals some data/mc disagreement in Ehad spectrum nter-nucleon correlations a hot topic in neutrino xsecs currently Evidence for extra MEC component from NOvA, MNERvA, etc We pick the model that best matches our data, but allow a lot of freedom in the shape of the energy transfer distribution 7 / 9
Nuclear correlations (Mark Hartz) ND data reveals some data/mc disagreement in Ehad spectrum nter-nucleon correlations a hot topic in neutrino xsecs currently Evidence for extra MEC component from NOvA, MNERvA, etc We pick the model that best matches our data, but allow a lot of freedom in the shape of the energy transfer distribution 7 / 9
Nuclear correlations (Mark Hartz) ND data reveals some data/mc disagreement in Ehad spectrum nter-nucleon correlations a hot topic in neutrino xsecs currently Evidence for extra MEC component from NOvA, MNERvA, etc We pick the model that best matches our data, but allow a lot of freedom in the shape of the energy transfer distribution 7 / 9
Selecting muon neutrinos Now using same convolutional neural net ( CVN ) as νe analysis Also have to reject cosmic rays, use containment, dir. and size Factor 5 from µs spill window vs Hz beam, 7 from cuts 9% pure FD νµ CC sample, % higher efficiency than prev. sel. Particularly at lower energies important to resolve osc. dip 8 / 9
νµ energy estimation Estimate energy of selected events to trace out osc. structure Known muon de/dx Eµ = f (Ltrk ) k Ltrk Hadronic part of the event estimated calorimetrically Eν = f (Ltrk ) + Ehad 9 / 9
νµ resolution bins ~6% ~8% ~% ~% Bin into 4 equal quantiles by hadronic energy fraction Energy resolution varies from 6% to % between bins / 9
νµ resolution bins NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. Data Prediction -σ syst. range Beam bkg. Cosmic bkg. Events/. GeV 4 Quantile best resolution 6 5 Events/. GeV 5 4 NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. Data Prediction -σ syst. range Beam bkg. Cosmic bkg. Quantile 4 Reconstructed Neutrino Energy (GeV) 5 4 Reconstructed Neutrino Energy (GeV) NOvA Preliminary 5 NOvA Preliminary 6 NOvA Normal Hierarchy 8.85 POT-equiv. NOvA Normal Hierarchy 8.85 POT-equiv. Data Prediction -σ syst. range Beam bkg. Cosmic bkg. Quantile Data Prediction -σ syst. range Beam bkg. Cosmic bkg. 5 Events/. GeV Events/. GeV 4 Quantile 4 4 worst resolution 4 Reconstructed Neutrino Energy (GeV) 5 4 5 Reconstructed Neutrino Energy (GeV) Bin into 4 equal quantiles by hadronic energy fraction Energy resolution varies from 6% to % between bins / 9
νµ disappearance results Expect 76 FD νµ CC events with no oscillation Observe 6 (inc..4 beam bkg. and 5.8 cosmic) / 9
νµ disappearance results Expect 76 FD νµ CC events with no oscillation Observe 6 (inc..4 beam bkg. and 5.8 cosmic) / 9
νµ disappearance results m (-ev) NOvA Preliminary.6.4. σ. σ.4 sinθ Observe 6 (inc..4 beam bkg. and 5.8 cosmic) m (-ev) Expect 76 FD νµ CC events with no oscillation Best Fit NOvA Preliminary NH.5.4 σ.6.7.6.8 σ. σ.4 σ.5 sin θ H.6.7 m = (.44 ±.8) ev (NH) sin θ =.56+.4. or.48+.4.4 / 9
νµ disappearance results NOvA Preliminary Normal Hierarchy 9% C.L.. NOvA 8.85 POT-equiv. TK 6 m (- ev) MNOS 4.8.6.4. Joint analysis Expect 76 FD νµ CC events with no oscillation Observe 6 (inc..4 beam bkg. and 5.8 cosmic).4.5 sinθ.6.7 m = (.44 ±.8) ev (NH) sin θ =.56+.4. or.48+.4.4 / 9
Principle of the νe measurement To first order, NOvA measures P(νµ νe ) and P(ν µ ν e ) at GeV These depend differently ) and δ on sign( m CP. Pµe nverted Hierarchy Normal Hierarchy.8.6 π/ π.4 π/ π/ π...4 π/.6.8. Pµe / 9
Principle of the νe measurement To first order, NOvA measures P(νµ νe ) and P(ν µ ν e ) at GeV These depend differently ) and δ on sign( m CP. Pµe.8.6 π/ π/ Measuring Pµe today Ultimately constrain to some region of this space 8 POT FHC +8 POT RHC π.4 nverted Hierarchy Normal Hierarchy NH δ CP=π/ σ σ π/ π...4 π/.6.8. Pµe / 9
Principle of the νe measurement Measuring Pµe today Ultimately constrain to some region of this space.6.5.6.4.5.4. P also sin θ.6 8 POT FHC +8 POT RHC θ.8.4 nverted Hierarchy Normal Hierarchy NH δ CP=π/ σ σ These depend differently ) and δ on sign( m CP. si n To first order, NOvA measures P(νµ νe ) and P(ν µ ν e ) at GeV Pµe..4.6.8. Pµe / 9
Use ND data to predict three FD background components Beam νe CC NC νµ CC Data/MC Events / 8.9 POT Making FD bkg prediction.5.5.4..8.6 Ereco (GeV) 4 5 Can separate statistically: νe /νµ share common π + /K + ancestors µ in νµ CC events leaves decay electron Beam νe %, NC %, νµ CC % Extrapolate components for FD prediction / 9
νe peripheral sample NOvA Preliminary 8.8.6.6.5.4.4 Efficiency (%) Cosmic Rejection BDT.7. NOvA Simulation Core Core+Peripheral NOvA FD 6 4..9.9.94 CVN.96.98 True Visible Energy (GeV) 4 5 e selection DATA FD data broken down by PD value (low high purity) Events that fail containment and cosmic rejection cuts given a second chance Require high CVN score plus specialized cosmic rejection BDT Equivalent to 6% more exposure Basic Quality cuts Preselection cuts Cosmic Rejection cuts no no Peripheral Preselection CVN PD cut Low PD Mid. PD {z Core sample CVN and BDT cut High PD } Peripheral bin {z } Peripheral sample 4 / 9
Event count expectations More NH π > 45 Less H π < 45 Signal prediction NH π H π for θ = 45 ±9% syst. 48 NOvA Simulation Total events expected P(νµ νe ) Hie. δcp θ sinθ=.8 sinθ=.4-.6 NOvA FD 8 8.85 POT eq. 6 4 π NH: m=+.44 -ev H: m=.48 -ev π π π δ CP Background components Total bkg NC beam νe νµ CC.5 6.6 7.. ντ CC. cosmics 4.9 ±% syst. Essentially independent of oscillation parameters 5 / 9
νe appearance results FD data Best Fit prediction Total Background Cosmic Background 5 NOvA Preliminary 6 4 Data (±σ) 5 4 4 sinθ=.8 sinθ=.4-.6 NOvA FD 8 8.85 POT eq. Total events High PD Core Mid. PD Peripheral Events / 8.85 POT-equiv NOvA Preliminary Low PD 4 π NH: m=+.44 -ev H: m=.48 -ev π π π δ CP Reconstructed Neutrino Energy (GeV) Observe 66 events passing νe selection On.5 background Towards the higher end of expectations 6 / 9
νe fit results NOvA Preliminary.7 Joint fit from νµ and νe spectra Constrain θ to reactor avg. sin θ =.8 ±.5.5.4. σ Prefer NH and (weakly) δcp π/ σ π.7 σ Best Fit NOvA Preliminary NH π δcp π π.6 sinθ sinθ.6.5.4. σ σ π σ π δcp H π π 7 / 9
νe fit results Joint fit from νµ and νe spectra Constrain θ to reactor avg. sin θ =.8 ±.5 Prefer NH and (weakly) δcp π/ H disfavoured at σ level NOvA Preliminary 5 Significance (σ) 4 NOvA FD 8.85 POT equiv. Normal hierarchy nverted hierarchy π π δcp π π 7 / 9
NOvA future sensitivity Normal δcp=π/, sin θ=.5 Total events - antineutrino mode NOvA Simulation NOvA FD 9.49 POT (ν) 8. POT (ν) 5 sin θ=.8 sinθ=.47,.56 H Δm=.5 -ev 5 δcp= δcp= π δcp= π/ δcp= π/ 4 6 NOvA Simulation NOvA joint νe+νµ Hierarchy 4 CPV All projected beam intensity and analysis improvements 6 8 4 Year (ncluding incremental beam improvements to 9kW by ) 7 best fit prediction Total events - neutrino mode 5 NH Δm=+.45 -ev Significance (σ= Δχ) Δm=.45 -ev, sinθ=.8 Currently favoured values avoid ambiguous region Will release large sample ( 7 POT) of antineutrino data in June 4σ hierarchy measurement by end of experiment? 8 / 9 8
Conclusion Muon neutrino disappearance now compatible with maximal Very competitive measurement of m νe appearance favours NH, δcp π/ H at δcp = π/ disfavoured at >σ, approaching σ H rejection Syst. reductions from testbeam this year Opening large sample of antineutrinos at Neutrino 8 Stay tuned! 9 / 9
Thank you!
Backup
Current world knowledge c s e iδ c s s c U = c s iδ s c s e c θ 45 ±.5 ev θ 8.5 m δcp =? ντ θ 7.5 5 ev ν νe νµ ν ν mass m matm m matm ν ν m Normal Hierarchy ν nverted Hierarchy / 9
Making neutrinos GeV protons from Main njector Strike graphite target Produce mainly π ± and K ± Focused by two magnetic horns Allow us to select charge sign for a neutrino or antineutrino beam 675m decay-pipe: π + µ+ + νµ Muons absorbed by rock / 9
Principle of the νµ measurement Separate νµ CC interactions from backgrounds Long muon track with distinctive de/dx easy to spot Extrapolate observed ND spectrum to make FD unosc. prediction Measure shape of νµ deficit in the FD Two flavor approx. works well here m L Pµµ sin θ sin 4E θ 45 almost all νµ expected to disappear at oscillation max..8 P(νµ νµ ).6 sinθ.4. m 4 5 Neutrino energy (GeV) 4 / 9
Mixing patterns ντ ν ν Only a small fraction of νe in ν i (sin θ ) The remainder is split 5/5 νµ /ντ (sin θ ) Accident? Or a sign of underlying structure? s θ exactly f not, is it... mass νe νµ ν matm 45? < 45 ( ν i more ντ, like the quarks) > 45 ( ν i more νµ, unlike quarks) ν ν m ν 5 / 9
Extrapolation procedure Translate ND observations to true energy Transport to far detector and oscillate Smear back to reco energy Cosmics prediction from out-of-time data 6 / 9
Extrapolation procedure Translate ND observations to true energy Transport to far detector and oscillate Smear back to reco energy Cosmics prediction from out-of-time data 6 / 9
νµ systematics Abs. Calibration Abs. Calibration Rel. Calibration Cross Sections Cross Sections Normalization Value of δcp Abs. Eµ Scale Normalization Rel. Calibration Abs. Eµ Scale Neutrino Flux Scintillation Model Rel. Eµ Scale Neutrino Flux Scintillation Model Rel. Eµ Scale Value of δcp Total syst. error Total syst. error Statistical error Statistical error.5.5 Uncertainty on Δm ( - ev) 5 5 Uncertainty on sinθ ( -) Evaluate systematics by replacing nominal MC by shifted versions Hard work here means we re still stats limited Calibration and cross-section (MEC) systematics largest 7 / 9
νe systematics ν Cross Sections ν Cross Sections Normalization Normalization Calibration Calibration Beam Beam Detector Response Detector Response Total syst. error Total syst. error Statistical error Statistical error Background Uncertainty (%) Signal Uncertainty (%) Dominated by statistics and then cross sections (MEC shape) 8 / 9
Changes from previous result NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. 68% C.L. 9% C.L. 99% C.L. PRL.8.58. m (- ev).8.6.4. Best fit Joint analysis (.56,.45 -ev).4.5 sinθ.6.7 Phys. Rev. Lett. 8 (7) 58 9 / 9
Changes from previous result New simulation Some effect from decreased Eres h7 MeVi shift in energies expect (observe).5 () events migrating out of dip region NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. 68% C.L. 9% C.L. 99% C.L. PRL.8.58. m (- ev).8.6.4. Best fit Joint analysis (.56,.45 -ev).4.5 sinθ.6.7 Events Selected in Both Analyses NOvA Preliminary σ Hadronic Energy Scale Uncertainty 8 6 4 5 5 Fractional Change in Reconstructed Hadronic Energy (%) Phys. Rev. Lett. 8 (7) 58 9 / 9
Changes from previous result New simulation Some effect from decreased Eres h7 MeVi shift in energies expect (observe).5 () events migrating out of dip region NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. 68% C.L. 9% C.L. 99% C.L. PRL.8.58. m (- ev).8.6.4. Best fit Joint analysis (.56,.45 ev ) -.4.5 sinθ.6.7 Events Selected in Both Analyses NOvA Preliminary σ Neutrino Energy Scale Uncertainty 5 5 5 5 5 Fractional Change in Reconstructed Neutrino Energy (%) Phys. Rev. Lett. 8 (7) 58 9 / 9
Changes from previous result New simulation New selection and analysis NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. 68% C.L. 9% C.L. 99% C.L. PRL.8.58. m (- ev).8 Some effect from decreased Eres h7 MeVi shift in energies expect (observe).5 () events migrating out of dip region 5% of mock experiments have a larger change, mostly driven by low selection overlap (especially cosmics).6.4. Best fit Joint analysis (.56,.45 -ev).4.5 sinθ.6.7 Phys. Rev. Lett. 8 (7) 58 9 / 9
Changes from previous result New simulation New selection and analysis NOvA Preliminary NOvA Normal Hierarchy 8.85 POT-equiv. 68% C.L. 9% C.L. 99% C.L. PRL.8.58. m (- ev).8.6 Best fit Joint analysis (.56,.45 ev ) -.4.5 sinθ.6 5% of mock experiments have a larger change, mostly driven by low selection overlap (especially cosmics) New data.4. Some effect from decreased Eres h7 MeVi shift in energies expect (observe).5 () events migrating out of dip region.7 New.8 POT of data prefers maximal mixing Phys. Rev. Lett. 8 (7) 58 9 / 9
Principle of the νe measurement Separate νe CC interactions from beam backgrounds Harder problem than νµ CC selection Evaluate remaining backgrounds in ND ntrinsic beam νe Neutral currents νµ CC mostly oscillates away An excess in the FD is the sign of νµ νe oscillations Pµe sin θ sin θ sin θ only 8.5 degrees, most νµ go to ντ instead Sensitive to mass ordering ( hierarchy ), δcp and θ octant L m 4E )) + f (δ + f (sign( m CP ) / 9
Why hierarchy? s the electron-like state lightest? i.e. Does the pattern of the masses match the charged leptons? Are neutrinos Majorana particles (ν = ν )? Observation of νββ would be proof they are mpact of H determination: lack of νββ implies Dirac nature / 9
Matter effects Electrons in the Earth drag on the electron neutrino states Sign of the effect opposite for antineutrinos and for NH/H ν νe νµ ντ ν ν mass matm m matm ν ν m Normal Hierarchy ν nverted Hierarchy / 9
Neutrino/antineutrino symmetry Does P(νµ νe ) = P(ν µ ν e )? nsight into fundamental symmetries of the lepton sector CP violation described by oscillation parameter δcp Why is the universe not equal parts matter and antimatter? Need ppb early universe asymm. Existing CP-violation insufficient Leptogenesis : generate ν/ν imbalance, transfer to baryons Require neutrino appearance experiment to discover / 9
Sample composition High PD 5 4 Beam Bkg with an Electron Beam Bkg with a π (KE >.5 GeV) Beam Bkg without EM Activity Cosmic Bkg NC Cosmic νe CC Core signal νe CC beam νe CC NC νµ CC ντ CC νe CC cosmic νµ CC 5 Mid. PD Peripheral Events / 8.85 POT-equiv NOvA Preliminary Low PD 4 4 Reconstructed Neutrino Energy (GeV) 4 6 Number of Bkg Events Break spectrum down into PD bins (low to high purity) Plus additional peripheral sample Backgrounds predominantly have EM activity: π γγ or intrinsic beam νe 8 4 / 9
θ NOvA Preliminary π D Gauss. σ σ σ Reactor 68%C.L. δcp π π π H.. sinθ. 5 / 9
TK latest νe ν e Obs 89 7 Exp(no-CPV) 67 9 Used (4.7 + 7.6) POT Approved to 7.8 POT Proposal for by 6 Mark Hartz colloquium @ KEK, Aug 7 6 / 9
Event reconstruction First cluster hits in space and time Start with -point Hough transform Line-crossing are vertex seeds ElasticArms finds vertex Fuzzy k -means clustering forms prongs νµ analysis uses a Kalman filter to reconstruct any muon track 7 / 9
Event reconstruction First cluster hits in space and time Start with -point Hough transform Line-crossing are vertex seeds ElasticArms finds vertex Fuzzy k -means clustering forms prongs νµ analysis uses a Kalman filter to reconstruct any muon track 7 / 9
Calibration and energy scale Response varies substantially along cell due to light atten. Use cosmic ray muons as a standard candle to calibrate, channels individually Use de/dx near the end of stopping muon to set abs. scale 8 / 9
Calibration and energy scale Response varies substantially along cell due to light atten. Use cosmic ray muons as a standard candle to calibrate, channels individually Use de/dx near the end of stopping muon to set abs. scale NOνA Preliminary ND,.66 POT 6 Simulated ν µ CC Simulated Background Multiple calibration x-checks Beam muon de/dx Michel energy spectrum π mass peak Hadronic energy/hit Take 5% abs. and rel. errors on energy scale Data Events 4....4 Average Hadronic Energy Per Hit (GeV) 8 / 9
ND decomposition beam νe p to ND νµ νe π+ Target µ+ νµ e+ Low-E νµ and νe trace back to the same π + ancestors All Ancestors Other Ancestor KL Ancestor K+ Ancestor π+ Ancestor 4 5 5 True Neutrino Energy (GeV) All Ancestors Other Ancestor KL Ancestor K+ Ancestor π+ Ancestor Events / Bin / 6E POT Events / Bin / 6E POT 6 5 5 4 6 5 8 5 True Neutrino Energy (GeV) 9 / 9
ND decomposition Michels NOvA Preliminary NOvA Preliminary.75 < CVN <.87 Data MC νµ CC MC NC MC Beam νe 5 Number of Michels + Events /.7 POT Events /.7 POT 5.87 < CVN <.95.95 < CVN < Data MC Beam νe MC νµ CC MC NC 4 4 4 5 Reconstructed neutrino energy (GeV) νµ CC background events have Michel electron from muon decay Also produced in νe CC and NC by pions, but νµ have more Fit observed Nmichel spectrum in each bin by varying components νe and NC near-degenerate, fix νe to parent-reweight estimate 4 / 9
ND decomposition Michels NOvA Preliminary NOvA Preliminary.75 < CVN <.87 Data MC νµ CC MC NC MC Beam νe 5 Number of Michels + Events /.7 POT Events /.7 POT 5.87 < CVN <.95.95 < CVN < Data Beam νe νµ CC NC Uncorrected MC 4 4 4 5 Reconstructed neutrino energy (GeV) νµ CC background events have Michel electron from muon decay Also produced in νe CC and NC by pions, but νµ have more Fit observed Nmichel spectrum in each bin by varying components νe and NC near-degenerate, fix νe to parent-reweight estimate 4 / 9
νe selection efficiency MRE EM showers should be well modelled Any νe signal efficiency differences coming from the hadronic side? Remove muon from clear νµ CC events in ND, replace with simulated shower O(%) efficiency difference to select MRE data/mc events 4 / 9
νe selection efficiency MRE EM showers should be well modelled Any νe signal efficiency differences coming from the hadronic side? Remove muon from clear νµ CC events in ND, replace with simulated shower O(%) efficiency difference to select MRE data/mc events 4 / 9
νe selection efficiency MRE EM showers should be well modelled Any νe signal efficiency differences coming from the hadronic side? Remove muon from clear νµ CC events in ND, replace with simulated shower O(%) efficiency difference to select MRE data/mc events 4 / 9
νe selection efficiency EM activity Brem shower Find FD data cosmic rays w/ brems 4 / 9
νe selection efficiency EM activity Brem shower Find FD data cosmic rays w/ brems Remove µ leaving pure EM activity Run through PD in data and MC Very good agreement 4 Events νe analysis core preselected MRBrem events Cosmics Data Cosmics MC νe signal MC 8 6 4..4.6.8 CVN νe selector 4 / 9
Principle of the NCs Where do those νµ go? Do any oscillate to a sterile state? (νs ) NC spectrum unaffected by oscillations among active flavours Select NC events in ND, extrapolate to FD prediction Count NC events in FD, compare to prediction =.5 ev, rapid osc in FD, minimal in ND Fix m4. Neutrino Energy (GeV) Neutrino Energy (GeV) ND.8.6.4.8 -Flavor Prob. m4 =.5 ev m4 =.5 ev m4 = 5 ev.6.4.. FD - P(νµ νs). - - L/E (km/gev) 4 / 9
Sterile neutrinos.5 FD Data Total -flavor Prediction Syst. Uncertainty. CC Background Cosmic Background.5 NOvA 6.5 POT-equiv. m =.44 - ev NOvA 68% C.L. θ = 8.5, θ = 45 NOvA 9% C.L. m4 =.5 ev SK 9% C.L. * cecube-dc 9% C.L. ** µ4 U Events /.5 GeV 5 5 PRD 9, 59 (5) * PRD 95, (7) **..5 5 Calorimetric Energy (GeV) 4. Are all the disappearing νµ going to νe or ντ? Might some fraction be oscillating to a a 4th, sterile, state? Would expect a depletion of NC events at FD Expect 8.5 ± 9.7(stat) ± 9.4(syst) see 95 Set limits on Uµ4 and Uτ 4 Phys. Rev. D 96, 76 (7). U..4 τ4 44 / 9