4/19/2007Jonathan Link First Oscillation Results from the MiniBooNE Experiment Jonathan Link Virginia Polytechnic Institute & State University April 19 th 2007

4/19/2007Jonathan Link University of Alabama Los Alamos National Laboratory Bucknell University Louisiana State University University of Cincinnati University of Michigan University of Colorado Princeton University Columbia University Saint Mary’s University of Minnesota Embry Riddle University Virginia Polytechnic Institute Fermi National Accelerator Laboratory Western Illinois University Indiana University Yale University The MiniBooNE Collaboration

4/19/2007Jonathan Link The LSND Experiment LSND took data from Baseline of 30 meters Energy range of 20 to 55 MeV L/E of about 1m/MeV LSND’s Signature 2.2 MeV neutron capture Čerenkov Scintillation π +  μ + ν μ e + ν μ ν e ν e p  e + n n+H→D+γ Neutrino oscillation probability: P(ν μ → ν e ) = sin 2 2θ sin 2 (1.27Δm 2 L/E)

4/19/2007Jonathan Link The LSND Signal LSND found an excess of ν e events in a ν μ beam They found 87.9 ± 22.4 ± 6.0 events over expectation. With an oscillation probability of (0.264 ± ± 0.045)%. Decay in flight analysis (   e ) oscillation probability of (0.10 ± 0.16 ± 0.04) %

4/19/2007Jonathan Link Why is this Result Interesting? LEP found that there are only 3 light neutrinos that interact weakly ν3ν3 ν2ν2 ν1ν1 m22m22 m12m12 |  m 3 2 | =  m 1 2 ±  m 2 2 Three neutrinos allow only 2 independent  m 2 scales But there are experimental results at 3 different  m 2 scales Existing Neutrino Oscillation Data

4/19/2007Jonathan Link What Does it Mean? First, one or more of the experiments may be wrong LSND, being the leading candidate, has to be checked → MiniBooNE Otherwise, add one or more sterile neutrinos… Giving you more independent  m 2 scales Best fits to the data require at least two sterile The Usual 3ν Model (mass) 2

4/19/2007Jonathan Link Keep L/E the same while changing systematics, neutrino energy and event signature P(ν μ → ν e ) = sin 2 2θ sin 2 (1.27Δm 2 L/E) Booster K+K+ target and horndetectordirtdecay regionabsorber primary beamtertiary beamsecondary beam (protons)(mesons)(neutrinos)    e   Order of magnitude higher energy (~500 MeV) than LSND (~30 MeV) Order of magnitude longer baseline (~500 m) than LSND (~30 m) The MiniBooNE Design Strategy

4/19/2007Jonathan Link The Neutrino Beam

4/19/2007Jonathan Link Booster Target Hall MiniBooNE extracts beam from the 8 GeV Booster… delivers it to a 1.7 Be target… within a magnetic horn (2.5 kV, 174 kA) that increases the flux by  6 These results correspond to (5.58  0.12)  POT The Neutrino Beam

4/19/2007Jonathan Link HARP (CERN)  5% λ Beryllium target  8.9 GeV proton beam momentum Modeling Production of Pions HARP collaboration, hep-ex/ Data are fit to a Sanford-Wang parameterization.

4/19/2007Jonathan Link K + Data from GeV. Uses a Feynman Scaling Parameterization. data ─ points dash ─ total error (fit  parameterization) Modeling Production of Kaons K 0 data are also parameterized. In situ measurement of K + from LMC agrees within errors with parameterization

4/19/2007Jonathan Link μ + → e + ν μ ν e K→ πeν e K+→ μ+νμ K+→ μ+νμ π+→ μ+νμπ+→ μ+νμ Antineutrino content: 6% Neutrino Flux from GEANT4 Simulation “Intrinsic” ν e + ν e sources:  μ + → e + ν μ νe (52%)  K + → π 0 e + ν e (29%)  K 0 → π e ν e (14%)  Other ( 5%) ν e / ν μ = 0.5%

4/19/2007Jonathan Link Stability of Running Observed and expected events per minute Full ν Run Neutrino interactions per proton-on-target.

4/19/2007Jonathan Link Events in the Detector

4/19/2007Jonathan Link The MiniBooNE Detector 541 meters downstream of target 12 meter diameter sphere Filled with 950,000 liters of pure mineral oil — 20+ meter attenuation length Light tight inner region with 1280 photomultiplier tubes Outer veto region with 240 PMTs. Simulated with a GEANT3 Monte Carlo

4/19/2007Jonathan Link Detected photons from Prompt light (Cherenkov) Late light (scintillation, fluorescence) in a 3:1 ratio for  ~1 Attenuation length: > nm We have developed 39-parameter “Optical Model” based on internal calibration and external measurement Optical and PMT Response Models Data/MC overlay

4/19/2007Jonathan Link The 1.6 μs spill sits inside a 19.2 μs beam trigger window An unbroken time cluster of hits forms a “subevent” Most events are from ν μ charged current (CC) interactions with a characteristic two “subevent” structure from stopped μ decay μ e Tank Hits Example Event The Beam Window

4/19/2007Jonathan Link All events in beam window Veto Hits <6 removes through-going cosmic rays This leaves “Michel electrons” (   e e) and beam events Tank Hits > 200 (effectively energy) removes Michel electrons, which have a 52 MeV endpoint Progressively introducing cuts on the beam window: Event Timing Structure with Simple Clean-up Cuts

4/19/2007Jonathan Link Calibration Systems and Sources

4/19/2007Jonathan Link From the NUANCE event generator D. Casper, NPS, 112 (2002) 161 Event neutrino energy (GeV) Predicted Event Rates Before Cuts

4/19/2007Jonathan Link CCQE are 39% of all events Events are “clean” (few particles) Energy of the neutrino can be reconstructed Reconstructed from: Scattering angle Visible energy (E visible ) An oscillation signal is an excess of ν e events as a function of E ν QE Charged Current Quasi-elastic θ μ or e Beam ν N N’

4/19/2007Jonathan Link Model describes CCQE ν μ data well Kinetic Energy of muon From Q 2 fits to MB ν μ CCQE data: M A eff ─ Effective axial mass E lo SF ─ Pauli blocking parameter From electron scattering data: E b ─ Binding energy p f ─ Fermi momentum data/MC~1 across all angles & energy after fit Neutrino Interaction Parameters in NUANCE

4/19/2007Jonathan Link N π0π0 N NC π 0 The π 0 decays to 2 photons, which can look “electron-like” mimicking the signal... <1% of π 0 contribute to background. N Δ π+π+ N μ+μ+ 25% 8% CC π + Easy to tag due to 3 subevents. Not a substantial background to the oscillation analysis. Events Producing Pions The Δ also decays to a single photon with ~0.5% probability ν ν ν Δ

4/19/2007Jonathan Link Particles Produced in the Detector and PID Muons: Produced in most CC events. Usually 2 subevent or exiting. Electrons: Tag for ν μ  ν e CC quasi-elastic (QE) signal. 1 subevent π 0 s: Can form a background if one photon is weak or exits tank. In NC case, 1 subevent.

4/19/2007Jonathan Link Two Independent Analyses

4/19/2007Jonathan Link The goal of both analyses is to minimize background while maximizing signal efficiency. “Signal range” is approximately 300 MeV < E QE < 1500 MeV One can then either: look for a total excess (“counting experiment”) fit for both an excess and energy dependence (“energy fit”) Neutrino Energy (GeV) Arbitrary Units Signal Region Example Signals:  m 2 =0.4 eV 2  m 2 =0.7 eV 2  m 2 =1.0 eV 2 Analysis Objectives

4/19/2007Jonathan Link MiniBooNE is searching for a small but distinguishable event signature In order to maintain blindness, Electron-like events were sequestered, Leaving ~99% of the in-beam events available for study. Rule for cuts to sequester events: <1  signal outside of the box Low level information which did not allow particle-id was available for all events. Open Data for Studies

4/19/2007Jonathan Link Both analyses share these hit-level pre-cuts: Only 1 subevent Veto hits < 6 Tank hits > 200 And a radius precut: R<500 cm (where reconstructed radius is algorithm-dependent) Analysis Pre-cuts

4/19/2007Jonathan Link Uses detailed, direct reconstruction of particle tracks, and ratio of fit likelihoods for the various event type hypotheses (μ, e and π 0 ) to identify particles. Philosophy: This algorithm was found to have the better sensitivity to ν μ  ν e appearance. Therefore, before unblinding, this was the algorithm chosen for the “primary result” Analysis 1: “Track-Based” (TB) Analysis

4/19/2007Jonathan Link log(L e /L  )>0 favors electron-like hypothesis Note: photon conversions are electron-like. This does not separate e/π 0. Separation is clean at high energies where muon-like events are long. Analysis cut was chosen to maximize the ν μ  ν e sensitivity ν e CCQE ν μ CCQE MC Rejecting Muon-like Events Using log(L e /L μ )

4/19/2007Jonathan Link MC Cuts were chosen to maximize   e sensitivity Using a mass cutUsing log(L e /L  ) NC π 0 ν e CCQE Rejecting π 0 Events NC π 0 ν e CCQE

4/19/2007Jonathan Link BLIND Testing e-π 0 Separation with Data 1 subevent log(L e /L  )>0 (e-like) log(L e /L  )<0 (π-like) mass>50 (high mass) Signal region e π0π0 Invariant Mass e π0π0 BLIND Monte Carlo π 0 only log(L e /L  ) γγ invariant mass Sideband Region has χ 2 Probability of 69%

4/19/2007Jonathan Link Efficiency: Log(L e /L  ) + Log(L e /L  ) + invariant mass Simulated Backgrounds After Cuts Summary of Track Based Analysis Cuts “Precuts” +

4/19/2007Jonathan Link Construct a set of low-level analysis variables which are combined to give a score to each event that is used to select electron-like events. Philosophy: This algorithm represents an independent cross check of the Track Based Analysis. Analysis 2: Boosted Decision Trees (BDT) Boosted decision trees, or “boosting” is a technique for pattern recognition through training (like a neural net). In the interest of time I will focus on analysis 1 which is the primary analysis

4/19/2007Jonathan Link Errors, Constraints and Sensitivity

4/19/2007Jonathan Link We have two main categories of background:  mis-id intrinsic e Uncertainties in the background rates are among the sources of significant error in the analysis. We have handles in the MiniBooNE data for each source of BG. Sources of ν e Background

4/19/2007Jonathan Link Tying the ν e background and signal prediction to the ν μ flux constrains this analysis to a strict ν μ  ν e appearance-only search Data/MC Track Based: 1.32 ± 0.26 Boosted Decision Tree: 1.22 ± 0.29 BDT Predict Normalization & energy dependence of both background and signal From the ν μ CCQE events Determining the Misidentification Background Rates

4/19/2007Jonathan Link μ → e ν μ ν e π → μ νμπ → μ νμ Measure the ν μ flux Kinematics allows connection to the π flux E π (GeV) E = 0.43 E  Once the π flux is known, the μ flux is determined ν μ Constraint on Intrinsic ν e from Beam μ + Decay E ν (GeV)

4/19/2007Jonathan Link K + and K 0 Decay Backgrounds At high energies, well above the signal range, ν μ and ν e -like events are largely from neutrinos from kaon decay. The events in these high energy bins are used to constrain the ν e background from kaon decay. signal range

4/19/2007Jonathan Link π 0 Production Constrained with MiniBooNE Data Because this constrains the Δ resonance rate, it also constrains the rate of Δ→Nγ Reweighting improves agreement in other variables, e.g.  This reduces the error on predicted mis-identified π 0 s

4/19/2007Jonathan Link External events (sometimes referred to a “Dirt” events) are from ν interactions outside of the detector: N data /N MC = 0.99 ± 0.15 Enhanced Background Cuts Cosmic Rays: Measured from out-of-beam data: 2.1 ± 0.5 events External Sources of Background Event Type of External Events after PID cuts

4/19/2007Jonathan Link Summary of Predicted Background Rates Process# of Events ν e from μ decay132 ν e from K + decay71 ν e from K L 0 decay23 ν e from π decay3 CC quasi-elastic10 Elastic scattering7 NC π 0 62 NC Δ → Nγ20 External (“Dirt”) events17 Other misID13 Beam unrelated2 Total Background360 Beam Intrinsic ν e ν μ Misidentification

4/19/2007Jonathan Link Handling Uncertainties in the Analyses For a given source of uncertainty, Errors on a wide range of parameters in the underlying model For a given source of uncertainty, Errors in bins of reconstructed E QE and information on the correlations between bins What we begin with what we need

4/19/2007Jonathan Link Optical Model Uncertainties 39 parameters must be varied Allowed variations are set by the Michel calibration sample To understand allowed variations, we ran 70 hit-level simulations, with randomly selected parameters.  “Multisims”

4/19/2007Jonathan Link Using Multisims to convert from parameter uncertainties to errors in E QE bins: For each error source, “Multisims” are generated within the allowed variations by reweighting the standard Monte Carlo. In the case of the optical model, hit-level simulations are used. Number of events passing cuts in 500 < E QE <600 MeV 1000 multisims for K + production 70 multisims for the Optical Model standard MC Multisims Used to Convert Uncertainties to Errors

4/19/2007Jonathan Link Where: N is number of events passing cuts MC is standard monte carlo  represents a given multisim M is the total number of multisims i,j are E QE bins Error Matrix Elements: Total error matrix is sum from each source. TB: ν e -only total error matrix BDT: ν μ -ν e total error matrix MC BDT Form Large Error Matrix for Uncertainties and Correlations

4/19/2007Jonathan Link Sensitivity of the Two Analyses With all the errors calculated and before the box is opened, we can calculate the sensitivity of the two analyses. The Track-based sensitivity is better, thus this becomes the pre- determined default analysis

4/19/2007Jonathan Link The Initial Results

4/19/2007Jonathan Link Box Opening Procedure After applying all analysis cuts: 1.Fit sequestered data to an oscillation hypothesis, returning no fit parameters. Return the  2 of the data/MC comparison for a set of diagnostic variables. 2. Open up the plots from step 1. The fitted signal is unreported. Plots are chosen to be useful diagnostics, without indicating if signal was added. 3. Report the  2 for a fit to E QE, without returning fit parameters. 4.Compare E QE in data and Monte Carlo, returning the fit parameters. At this point, the box is open. Progress cautiously, in a step-wise fashion

4/19/2007Jonathan Link Step 1 Return the  2 of the data/MC comparison for a set of diagnostic variables All analysis variables were returned with good probability except... Track Based analysis  2 Probability for the E visible fit was only 1% This probability was sufficiently low to merit further consideration 12 variables are tested for Track Based 46 variables are tested for Boosted Decision Tree

4/19/2007Jonathan Link In the Track Based analysis We re-examined our background estimates using sideband studies.  We found no evidence of a problem However, knowing that backgrounds rise at low energy, We tightened the lower Energy cut for the oscillation fit E QE > 475 MeV We agreed to report events over the original full range: E QE > 300 MeV, …Step 1

4/19/2007Jonathan Link Return the  2 of the data/MC comparison for a set of diagnostic variables Parameters of the oscillation fit were not returned. TB (E QE >475 MeV) BDT  2 probabilities returned: 12 variables 46 variables Step 1 Redux

4/19/2007Jonathan Link Open up the plots from step 1 for inspection Examples of what we saw: MC contains fitted signal at an unknown level E visible TB (E QE >475 MeV) BDT fitted energy (MeV) E visible  2 Prob= 28%  2 Prob= 59% Step 2

4/19/2007Jonathan Link Report the  2 for a fit to E QE across full energy range TB (E QE >475 MeV)  2 Probability of fit: 99% BDT analysis  2 Probability of fit:52% Leading to... Open the box... Step 3 Step 4 Counting experiment within 475<E QE <1250 MeV 380 events observed and 358  19 (stat)  35 (sys) events expeced Signal significance: 0.55 σ The Track-based ν μ  ν e Appearance-only Result:

4/19/2007Jonathan Link Error bars are only the diagonal elements of error matrix. Fit errors for E > 475 MeV: Normalization 9.6% Energy scale: 2.3% Best Fit (dashed): (sin 2 2 ,  m 2 ) = (0.001, 4 eV 2 ) Data are in good agreement with background prediction. Track Based Energy Dependent Fit Result

4/19/2007Jonathan Link Under the hypothesis of ν μ  ν e appearance-only as the oscillation mechanism Energy fit: 475<E QE <3000 MeV  2 probability, null hypothesis: 93% Oscillation Limits from the Track Based Analysis

4/19/2007Jonathan Link Boosted Decision Tree Result Counting experiment within 300<E ν QE <1600 MeV 971 events observed and 1070  33 (stat)  225 (sys) events expected

4/19/2007Jonathan Link Boosted Decision Tree E ν QE data/MC Comparison Error bars are statistical and systematic (diagonals of matrix) data - predicted (no osc) error (sidebands used for constraint not shown)

4/19/2007Jonathan Link Boosted Decision Tree analysis shows no evidence for ν μ  ν e appearance-only oscillations. The two independent analyses are in good agreement. Oscillation Limits from the BDT Analysis solid: Track based dashed: Boosted Decision Tree

4/19/2007Jonathan Link 96 ± 17 ± 20 events above expectation, in the range of 300<E ν QE <475MeV Deviation: 3.7σ (preliminary) to E>475 MeV Background-subtracted What About the Low Energy Region (E<475 MeV)?

4/19/2007Jonathan Link Best Fit (dashed): (sin 2 2 ,  m 2 ) = (1.0, 0.03 eV 2 )  2 Probability: 18% Fit to the > 300 MeV range: Examples in LSND allowed range Are the Excess Events Consistent with Oscillations?

4/19/2007Jonathan Link This will be addressed by further study This discrepancy is interesting, but requires further investigation  A two-neutrino appearance-only model systematically disagrees with the shape as a function of energy.  We need to investigate non-oscillation explanations, including unexpected behavior of low energy cross sections. This will be relevant to future   e searches

4/19/2007Jonathan Link A MiniBooNE-LSND Compatibility Test For each  m 2, determine the MB and LSND measurement: z MB   z MB, z LSND   z LSND where z = sin 2 (2  ) and  z is the 1  error For each  m 2, form  2 between MB and LSND measurement Find z 0 that minimizes  2 (weighted average of two measurements) and this gives  2 min Find probability of  2 min for 1 dof; this is the joint compatibility probability for this  m 2

4/19/2007Jonathan Link Maximum Joint Probability  m 2 (eV 2 ) MiniBooNE is incompatible with a ν μ  ν e appearance only interpretation of LSND at 98% CL A MiniBooNE-LSND Compatibility Test

4/19/2007Jonathan Link Future Plans A paper on this analysis has been posted to the archive and will soon be sent to PRL. Many more papers supporting this analysis will follow, in the very near future: ν μ CCQE production π 0 production MiniBooNE-LSND-Karmen joint analysis We are pursuing further analyses of the neutrino data, including... an analysis which combines TB and BDT, more exotic models for the LSND effect. MiniBooNE is presently taking data in antineutrino mode.

4/19/2007Jonathan Link Today I’ve shown an analysis of MiniBooNE data within a ν μ  ν e appearance-only context Conclusions The observed reconstructed energy distribution is inconsistent with oscillations in a ν μ  ν e appearance-only model The limit set is compatible with LSND at a level of 2% or less.

4/19/2007Jonathan Link Within the energy range defined by this oscillation analysis, the event rate is consistent with background. The observed low energy deviation is under investigation. Conclusions

4/19/2007Jonathan Link Questions

4/19/2007Jonathan Link 10% Photocathode coverage Two types of Hamamatsu Tubes: R1408, R5912 Charge Resolution: 1.4 PE, 0.5 PE Time Resolution 1.7 ns, 1.1ns

4/19/2007Jonathan Link Each event is characterized by 7 reconstructed variables: vertex (x,y,z), time, energy, and direction (  )  (U x, U y, U z ). Resolutions: vertex: 22 cm direction: 2.8  energy: 11%  CCQE events 2 subevents Veto Hits<6 Tank Hits>200

4/19/2007Jonathan Link Fundamental information from PMTs Analysis Hit Position Charge Hit Timing variables Energy  Time sequence  Event shape  Physics  Step 1: Convert the “Fundamental information” into “Analysis Variables” “Physics” =  0 mass, E QE, etc.

4/19/2007Jonathan Link Resolutions: vertex: 24 cm direction: 3.8º energy 14% Examples of “Analysis Variables” Reconstructed quantities which are inputs to E QE  CCQE U Z = cos  z E visible

4/19/2007Jonathan Link Step 2: Reduce Analysis Variables to a Single PID Variable “A procedure that combines many weak classifiers to form a powerful committee” Boosted Decision Trees hit level (charge, time, position) analysis variables One single PID “score” Byron P. Roe, et al., NIM A543 (2005) 577.

4/19/2007Jonathan Link (sequential series of cuts based on MC study) A Decision Tree (N signal /N bkgd ) 30,245/16, / / / / /11867 signal-like bkgd-like sig-like bkgd-like etc. This tree is one of many possibilities... Variable 1 Variable 2 Variable 3

4/19/2007Jonathan Link A set of decision trees can be developed, each re-weighting the events to enhance identification of backgrounds misidentified by earlier trees (“boosting”) For each tree, the data event is assigned +1 if it is identified as signal, -1 if it is identified as background. The total for all trees is combined into a “score” negativepositive Background- like signal-like

4/19/2007Jonathan Link BDT cuts on PID score as a function of energy. We can define a “sideband” just outside of the signal region

4/19/2007Jonathan Link BDT cuts on PID score as a function of energy. We can define a “sideband” just outside of the signal region

4/19/2007Jonathan Link BDT Efficiency and backgrounds after cuts: Analysis cuts on PID score as a function of Energy signal background Efficiency after precuts

4/19/2007Jonathan Link Other Single Photon Sources From Efrosinin, hep-ph/ , calculation checked by Goldman, LANL Neutral Current: + N  + N +  Charged Current + N   + N’ +  negligible where the presence of the  leads to mis-identification Use events where the  is tagged by the michel e -, study misidentification using BDT algorithm. < 6 95% CL

4/19/2007Jonathan Link Flux from  + /  + decay 6.2 / 4.3 √ √ Flux from K + decay3.3 / 1.0 √ √ Flux from K 0 decay 1.5 / 0.4 √ √ Target and beam models2.8 / 1.3 √ -cross section 12.3 / 10.5 √ √ NC  0 yield1.8 / 1.5 √ External interactions (“Dirt”) 0.8 / 3.4 √ Optical model6.1 / 10.5 √ √ DAQ electronics model7.5 / 10.8 √ Source of Uncertainty On e background Checked or Constrained by MB data Further reduced by tying e to  Track Based /Boosted Decision Tree error in %

4/19/2007Jonathan Link M A QE, e lo sf 6%, 2% (stat + bkg only) QE  norm 10% QE  shape function of E  e /  QE  function of E NC  0 rate function of  0 mom M A coh, coh  ±25%   N  rate function of  mom + 7% BF E B, p F 9 MeV, 30 MeV  s 10% M A 1  25% M A N  40% DIS  25% determined from MiniBooNE  QE data determined from MiniBooNE  NC  0 data Example: Cross Section Uncertainties determined from other experiments (Many are common to  and e and cancel in the fit)

4/19/2007Jonathan Link How the Constraints Enter TB: Reweight MC prediction to match measured ν μ result (accounting for systematic error correlations) Two Approaches Systematic (and statistical) uncertainties are included in (M ij ) -1 BDT: Include the correlations of ν μ to ν e in the error matrix: (i,j are bins of E QE )

4/19/2007Jonathan Link 1) There are various ways to present limits: Single sided raster scan (historically used, presented here) Global scan Unified approach (most recent method) 2) This result must be folded into an LSND-Karmen joint analysis. We will present a full joint analysis soon. Two points on interpreting our limit Church, et al., PRD 66,