first oscillation results from the miniboone experiment

4/19/20074/19/2007Jonathan LinkJonathan Link

First Oscillation Results from the First Oscillation Results from the MiniBooNE ExperimentMiniBooNE Experiment

Jonathan LinkJonathan Link

Virginia Polytechnic Institute & State UniversityVirginia Polytechnic Institute & State University

April 19April 19thth 2007 2007


University of Alabama Los Alamos National LaboratoryBucknell University Louisiana State UniversityUniversity of Cincinnati University of MichiganUniversity of Colorado Princeton UniversityColumbia University Saint Mary’s University of MinnesotaEmbry Riddle University Virginia Polytechnic InstituteFermi National Accelerator Laboratory Western Illinois UniversityIndiana University Yale University

The MiniBooNE CollaborationThe MiniBooNE Collaboration


The LSND ExperimentThe LSND Experiment

LSND took data from 1993-98

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+ νμ

e+ νμνeνe p e+ n

n+H→D+γ

Neutrino oscillation probability:

P(νμ → ν

e) = sin22θ sin2(1.27Δm2L/E)


The LSND SignalThe LSND SignalLSND found an excess of νe events in a νμ beam

They found 87.9 ± 22.4 ± 6.0 events over expectation.

With an oscillation probability of (0.264 ± 0.067 ± 0.045)%.

Decay in flight analysis (e) oscillationprobability of (0.10 ± 0.16 ± 0.04) %


Why is this Result Interesting?Why is this Result Interesting?

LEP found that there are only 3 light neutrinos that interact weakly

ν3

ν2ν1

m22

m12

|m32| = m1

2 ± m22

Three neutrinos allow only 2 independent m2 scales

But there are experimental results at 3 different m2 scales

Existing Neutrino Oscillation Data


What Does it Mean?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 m2 scales

Best fits to the data require at least two sterile

The Usual 3ν Model

(mas

s)2


Keep L/E the same while changing systematics, neutrino energy and event signature

P(νμ → ν

e) = sin22θ sin2(1.27Δm2L/E)

Booster

K+

target and horn detectordirt decay region absorber

primary beam tertiary beamsecondary beam

(protons) (mesons) (neutrinos)

e

Order of magnitudehigher energy (~500 MeV)

than LSND (~30 MeV)

Order of magnitudelonger baseline (~500 m)

than LSND (~30 m)

The MiniBooNE Design StrategyThe MiniBooNE Design Strategy


The Neutrino BeamThe Neutrino Beam


Booster TargetHall

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) thatincreases the flux by 6

These results correspond to (5.580.12) 1020 POT

The Neutrino BeamThe Neutrino Beam


HARP (CERN) 5% λ Beryllium target 8.9 GeV proton beam momentum

Modeling Production of PionsModeling Production of Pions

HARP collaboration,hep-ex/0702024

Data are fit to a Sanford-Wangparameterization.


K+ Data from 10 - 24 GeV.Uses a Feynman ScalingParameterization.

data ─ pointsdash ─ total error (fit parameterization)

Modeling Production of KaonsModeling Production of Kaons

K0 data are also parameterized.

In situ measurementof K+ from LMCagrees within errorswith parameterization


μ+→ e+ νμνe

K→ πeνe

K+→ μ+νμ

π+→ μ+νμ

Antineutrino content: 6%

Neutrino Flux from GEANT4 SimulationNeutrino Flux from GEANT4 Simulation

“Intrinsic” νe + νe sources:

μ+ → e+ νμνe (52%)

K+ → π0 e+ νe (29%)K0 → π e νe (14%)

Other ( 5%)

νe/ νμ = 0.5%


Stability of RunningStability of Running

Observed andexpected eventsper minute

Full ν Run Neutrino interactionsper proton-on-target.


Events in the DetectorEvents in the Detector


The MiniBooNE DetectorThe 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


Detected photons from• Prompt light (Cherenkov)• Late light (scintillation, fluorescence)

in a 3:1 ratio for ~1

Attenuation length: >20 m @ 400 nmWe have developed

39-parameter“Optical Model”

based on internal calibrationand external measurement

Optical and PMT Response ModelsOptical and PMT Response Models

Data/MC overlay


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 νμ charged current (CC) interactions with a characteristic two “subevent” structure from stopped μ decay

μ

e

Ta

nk

Hit

s

ExampleEvent

The Beam WindowThe Beam Window


All events in beam window

Veto Hits <6 removes through-going cosmic rays

This leaves “Michel electrons” (ee) and beam events

Tank Hits > 200(effectively energy)removes Michel electrons, which havea 52 MeV endpoint

Progressively introducing cuts on the beam window:

Event Timing Structure with Simple Clean-up CutsEvent Timing Structure with Simple Clean-up Cuts


Calibration Systems and SourcesCalibration Systems and Sources


From the NUANCE event generator D. Casper, NPS, 112 (2002) 161

Event neutrino energy (GeV)

Predicted Event Rates Before CutsPredicted Event Rates Before Cuts


CCQE are 39% of all events

Events are “clean” (few particles)Energy of the neutrino

can be reconstructed

Reconstructed from: Scattering angle Visible energy (Evisible)

An oscillation signal is an excess of νe events as a function of EνQE

Charged Current Quasi-elasticCharged Current Quasi-elastic

θ

μ or e

Beamν

N N’


Model describes CCQE νμ data well

Kinetic Energy of muon

From Q2 fits to MB νμ CCQE data: MA

eff ─ Effective axial mass Elo

SF ─ Pauli blocking parameter

From electron scattering data: Eb ─ Binding energy pf ─ Fermi momentum

data/MC~1across all

angles & energyafter fit

Neutrino Interaction Parameters in NUANCENeutrino Interaction Parameters in NUANCE


N

π0

N

NC π0

The π0 decays to 2 photons,which can look “electron-like” mimicking the signal...

<1% of π0 contribute to background.

NΔ

π+

N

μ+

25%

8%

CC π+

Easy to tag due to 3 subevents.Not a substantial background to the oscillation analysis.

Events Producing PionsEvents Producing Pions

The Δ also decays to a single photonwith ~0.5% probability

ν

ν

ν

Δ


Particles Produced in the Detector and PIDParticles Produced in the Detector and PID

Muons: Produced in most CC events.Usually 2 subevent or exiting.

Electrons:Tag for νμνe CC quasi-elastic (QE) signal. 1 subevent

π0s:Can form a background if onephoton is weak or exits tank.In NC case, 1 subevent.


Two Independent AnalysesTwo Independent Analyses


The goal of both analyses is tominimize background while maximizing signal efficiency.

“Signal range” is approximately300 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”)

0 1 2 3 Neutrino Energy (GeV)

Arb

itra

ry U

nits

Signal Region

Example Signals:m2=0.4 eV2

m2=0.7 eV2

m2=1.0 eV2

Analysis ObjectivesAnalysis Objectives


MiniBooNE is searchingfor 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-idwas available for all events.

Open Data for StudiesOpen Data for Studies


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-cutsAnalysis Pre-cuts


Uses detailed, direct reconstruction of particle tracks,and ratio of fit likelihoods for the various event type hypotheses (μ, e and π0) to identify particles.

Philosophy:

This algorithm was found to have the bettersensitivity to νμνe appearance.Therefore, before unblinding,this was the algorithm chosen for the “primary result”

Analysis 1: “Track-Based” (TB) AnalysisAnalysis 1: “Track-Based” (TB) Analysis


log(Le/L)>0 favors electron-like hypothesis

Note: photon conversions are electron-like.This does not separate e/π0.

Separation is clean at high energies where muon-like events are long.

Analysis cut was chosento maximize the νμ νe sensitivity

νe CCQE

νμ CCQEMC

Rejecting Muon-like Events Using log(LRejecting Muon-like Events Using log(Lee/L/Lμμ))


MC

Cuts were chosen to maximize e sensitivity

Using a mass cut Using log(Le/L)

NC π0

νe CCQE

Rejecting Rejecting ππ00 Events Events

NC π0

νe CCQE


BLI

ND

Testing e-Testing e-ππ0 0 Separation with DataSeparation with Data

1 subeventlog(Le/L)>0 (e-like)log(Le/L)<0 (π-like)mass>50 (high mass)

Signal region

eπ0

Invariant Masse π0

BLIND

Monte Carlo π0 only

log(Le/L)

γγ invariant mass

Sideband Region hasχ2 Probability of 69%


Efficiency:

Log(Le/L) + Log(Le/L) + invariant mass

Simulated Backgrounds After Cuts

Summary of Track Based Analysis CutsSummary of Track Based Analysis Cuts

“Precuts” +


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)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


Errors, Constraints and SensitivityErrors, Constraints and Sensitivity


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 Sources of ννee Background Background


Tying the νe background and signal predictionto the νμ flux constrains this analysis to a strict

νμ νe appearance-only search

Data/MC Track Based: 1.32 ± 0.26 Boosted Decision Tree: 1.22 ± 0.29

BDT

PredictNormalization& energy dependenceof both backgroundand signal

From the νμ CCQE

events

Determining the Misidentification Background RatesDetermining the Misidentification Background Rates


μ → e νμ νe

π → μ νμ Measure the νμ fluxKinematics allows

connection to the π flux

Eπ (GeV)

E = 0.43 E

Once the π flux is known,the μ flux is determined

ννμμ Constraint on Intrinsic Constraint on Intrinsic ννee from Beam from Beam μμ++ Decay Decay

Eν (

GeV

)


KK++ and Kand K00 Decay Backgrounds Decay Backgrounds

At high energies, well above the signal range, νμ and νe-like events are largely from neutrinos from kaon decay.

The events in these high energy bins are used to constrain the νe background from kaon decay.

signalrange


ππ00 Production Constrained with MiniBooNE Data Production Constrained with MiniBooNE Data

Because this constrains the Δ resonance rate, it also constrains the rate of Δ→Nγ

Reweighting improvesagreement in other

variables, e.g.

This reduces the erroron predictedmis-identified π0s


External events (sometimes referred to a “Dirt” events) are from ν interactions outside of the detector: Ndata/NMC = 0.99 ± 0.15

EnhancedBackgroundCuts

Cosmic Rays: Measured from out-of-beam data: 2.1 ± 0.5 events

External Sources of BackgroundExternal Sources of Background

Event Type of External Eventsafter PID cuts


Summary of Predicted Background RatesSummary of Predicted Background Rates

Process # of Events

νe from μ decay 132

νe from K+ decay 71

νe from KL0 decay 23

νe from π decay 3

CC quasi-elastic 10

Elastic scattering 7

NC π0 62

NC Δ → Nγ 20

External (“Dirt”) events 17

Other misID 13

Beam unrelated 2

Total Background 360

Bea

m

Intr

insi

c ν e

ν μ M

isid

entif

icat

ion


Handling Uncertainties in the AnalysesHandling Uncertainties in the Analyses

For a given source of uncertainty,

Errors on a wide rangeof parameters

in the underlying model

For a given source of uncertainty,

Errors in bins of reconstructed E

QE

and information on the correlationsbetween bins

What we begin with... ... what we need


Optical Model UncertaintiesOptical 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”


Using Multisims to convert from parameter uncertainties to errors in E

QE bins:

For each error source,“Multisims” are generated within the allowed variationsby reweighting the standard Monte Carlo.In the case of the optical model, hit-level simulations are used.

Number of events passing cuts in 500 < EQE <600 MeV

1000 multisims forK+ production

70 multisims for the Optical Model

standard MC

Multisims Used to Convert Uncertainties to ErrorsMultisims Used to Convert Uncertainties to Errors


Where:N is number of events passing cuts MC is standard monte carlorepresents a given multisimM is the total number of multisimsi,j are E

QE bins

Error Matrix Elements:

Total error matrixis sum from each source.

TB: νe-only total error matrixBDT: νμ-νe total error matrix

CVjj

MCViiij NNNN

ME

1

1 MC MC

BDT

Form Large Error Matrix for Uncertainties and CorrelationsForm Large Error Matrix for Uncertainties and Correlations


Sensitivity of the Two AnalysesSensitivity 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


The Initial ResultsThe Initial Results


Box Opening ProcedureBox 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 EQE in data and Monte Carlo, returning the fit parameters.

At this point, the box is open.

Progress cautiously, in a

step-wise fashion


Step 1 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 Evisible fit was only 1%

This probability was sufficiently low to merit further consideration

12 variables are tested for Track Based46 variables are tested for Boosted Decision Tree


In the Track Based analysis

• We re-examined our background estimatesusing 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 Step 1


Return the 2 of the data/MC comparison for a set of diagnostic variables

Parameters of the oscillation fit were not returned.

TB (EQE>475 MeV)

BDT2 probabilities returned:

12 variables 46 variables

Step 1 ReduxStep 1 Redux


Open up the plots from step 1 for inspection

Examples ofwhat we saw:

MC contains fitted signal at an unknown level

Evisible

TB (EQE>475 MeV) BDT

fitted energy (MeV)

Evisible

2 Prob= 28%

2 Prob= 59%

Step 2 Step 2


Report the 2 for a fit to EQE across full energy range

TB (EQE>475 MeV) 2 Probability of fit: 99%

BDT analysis 2 Probability of fit: 52%

Leading to...

Open the box...

Step 3 Step 3

Step 4Step 4

Counting experiment within 475<EQE<1250 MeV

380 events observed and 358 19 (stat) 35 (sys) events expeced

Signal significance:

0.55 σ

The Track-based νμνe Appearance-only Result:


Error bars are only the diagonal elements oferror matrix.

Fit errors for E > 475 MeV:Normalization 9.6%Energy scale: 2.3%

Best Fit (dashed): (sin22, m2) = (0.001, 4 eV2)

Data are in good agreement with background prediction.

Track Based Energy Dependent Fit ResultTrack Based Energy Dependent Fit Result


Under the hypothesis of νμ νe appearance-only as the oscillation mechanism

Energy fit: 475<EQE<3000 MeV

2 probability, null hypothesis: 93%

Oscillation Limits from the Track Based AnalysisOscillation Limits from the Track Based Analysis


Boosted Decision Tree Result

Counting experiment within 300<EνQE<1600 MeV

971 events observed and 1070 33 (stat) 225 (sys) events expected


Boosted Decision Tree EBoosted Decision Tree EννQEQE data/MC Comparison data/MC Comparison

Error bars are statistical and systematic(diagonals of matrix)

data - predicted (no osc) error

(sidebands used for constraint not shown)


Boosted Decision Tree analysis shows no evidence for νμ νe appearance-only oscillations.

The two independent analyses are in good agreement.

Oscillation Limits from the BDT AnalysisOscillation Limits from the BDT Analysis

solid: Track baseddashed: Boosted Decision Tree


96 ± 17 ± 20 eventsabove expectation,in the range of 300<Eν

QE<475MeV

Deviation: 3.7σ (preliminary)

to E>475 MeV

Background-subtracted

What About the Low Energy Region (E<475 MeV)?What About the Low Energy Region (E<475 MeV)?


Best Fit (dashed): (sin22, m2) = (1.0, 0.03 eV2)2 Probability: 18%

Fit to the > 300 MeV range:

Examples in LSND allowedrange

Are the Excess Events Consistent with Oscillations?Are the Excess Events Consistent with Oscillations?


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


A MiniBooNE-LSND Compatibility TestA MiniBooNE-LSND Compatibility Test

• For each m2, determine the MB and LSND measurement:

zMB zMB, zLSND zLSND where z = sin2(2) and z is the 1 error

• For each m2, form 2 between MB and LSND measurement

• Find z0 that minimizes 2

(weighted average of two measurements) and this gives 2min

• Find probability of 2min for 1 dof;

this is the joint compatibility probability for this m2


0.001

0.010

0.100

0.25 0.75 1.25 1.75 2.25 2.75

dm2 (eV2)

Jo

int

MB

-LS

ND

Pro

b (

1d

of)

2% Compatibility

Max

imum

Joi

nt P

roba

bilit

y

m2 (eV2)

MiniBooNE is incompatible with a νμνe appearance only interpretation of LSND

at 98% CL

A MiniBooNE-LSND Compatibility TestA MiniBooNE-LSND Compatibility Test


Future PlansFuture 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:

νμ CCQE productionπ0 productionMiniBooNE-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.


Today I’ve shown an analysis of MiniBooNE data within a νμνe appearance-only context

ConclusionsConclusions

The observed reconstructed energy distribution is inconsistent with oscillations in a νμνe appearance-only model

The limit set is compatible with LSND at a level of 2% or less.


Within the energy range defined by this oscillation analysis, the event rate is consistent with background.

The observed low energy deviation is under investigation.

ConclusionsConclusions


QuestionsQuestions


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


Each event is characterized by 7 reconstructed variables:

vertex (x,y,z), time, energy, and direction ()(Ux, Uy, Uz).

Resolutions: vertex: 22 cm

direction: 2.8 energy: 11%

CCQE events

2 subeventsVeto Hits<6Tank Hits>200


Fundamental information from PMTsAnalysis Hit Position Charge Hit Timing variablesEnergy

Time sequence

Event shape

Physics

Step 1:Convert the “Fundamental information”

into “Analysis Variables”

“Physics” = 0 mass, EQE, etc.


Resolutions:vertex: 24 cmdirection: 3.8ºenergy 14%

Examples of “Analysis Variables”

Reconstructed quantities which are inputs to EQE

CCQE CCQE

UZ = coszEvisible


Step 2: Reduce Analysis Variables to a Single PID Variable

“A procedure that combines many weak classifiersto form a powerful committee”

Boosted Decision Trees

hit level(charge, time,

position)

analysis variables

One singlePID “score”

Byron P. Roe, et al., NIM A543 (2005) 577.


(sequential series of cuts based on MC study)

A Decision Tree

(Nsignal/Nbkgd)

30,245/16,305

9755/23695

20455/3417 9790/12888

1906/11828 7849/11867

signal-likebkgd-like

bkgd-like sig-like

sig-likebkgd-like

etc.

This tree is one of many possibilities...

Variable 1

Variable 2

Variable 3


A set of decision trees can be developed,each re-weighting the events to enhance identification of backgrounds misidentifiedby 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”

negative positiveBackground-

like signal-like


BDT cuts on PID score as a function of energy.We can define a “sideband” just outside of the signal region


BDT Efficiency and backgrounds after cuts:

Analysis cuts on PID score as a function of Energy

signal

background

Efficiency after precuts


Other Single Photon Sources

From Efrosinin, hep-ph/0609169, 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 events @ 95% CL


Flux from +/+ decay 6.2 / 4.3 √ √ Flux from K+ decay 3.3 / 1.0 √ √ Flux from K0 decay 1.5 / 0.4 √ √ Target and beam models 2.8 / 1.3 √

-cross section 12.3 / 10.5 √ √ NC 0 yield 1.8 / 1.5 √

External interactions (“Dirt”) 0.8 / 3.4 √ Optical model 6.1 / 10.5 √ √ DAQ electronics model 7.5 / 10.8 √

Source of UncertaintyOn e background

Checked or Constrained by MB data

Furtherreduced by

tyinge to

Track Based/Boosted Decision Treeerror in %


MAQE, elo

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 MA

coh, coh ±25% Nrate function of mom + 7% BF

EB, pF 9 MeV, 30 MeVs 10% MA

1 25% MA

N 40% DIS 25%

determined fromMiniBooNE QE data

determined fromMiniBooNE NC 0 data

Example: Cross Section Uncertainties

determined from other experiments

(Many are common to and e and cancel in the fit)


How the Constraints Enter

TB: Reweight MC prediction to match measured νμ result (accounting for systematic error correlations)

Two Approaches

Systematic (and statistical) uncertainties are included in (Mij)-1

BDT: Include the correlations of νμ to νe in the error matrix:

(i,j are bins of EQE)


1) There are various waysto present limits:

• Single sided raster scan(historically used, presented here)

• Global scan• Unified approach

(most recent method)

2) This result must befolded into an LSND-Karmenjoint analysis.

We will present a full joint analysis soon.

Two points on interpreting our limit

Church, et al., PRD 66, 013001

first oscillation results from the miniboone experiment

Documents

k0 e e

excess of e events

event signaturep e

neutrino flux

neutrino energy

jonathan linkk data

oscillation results

mthan lsnd