First Oscillation Results from the MiniBooNE Experiment

1
First Oscillation Results from the MiniBooNE
Experiment
Jonathan Link Virginia Polytechnic Institute
State University April 19th 2007
2
The MiniBooNE Collaboration
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 Marys University of Minnesota Embry
Riddle University Virginia Polytechnic
Institute Fermi National Accelerator Laboratory
Western Illinois University Indiana
University Yale
University
3
The LSND Experiment
LSND took data from 1993-98
Neutrino oscillation probability P(?µ ?
?e) sin22? sin2(1.27?m2L/E)
p? µ ?µ
Baseline of 30 meters Energy range of 20 to 55
MeV L/E of about 1m/MeV
?e p? e n
e ?µ?e
nH?D?
LSNDs Signature
Scintillation
Cerenkov
2.2 MeV neutron capture
4
The LSND Signal
LSND found an excess of ?e events in a ?µ 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 (nm?ne)
oscillation probability of (0.10 0.16 0.04)
5
Why is this Result Interesting?
Existing Neutrino Oscillation Data
LEP found that there are only 3 light neutrinos
that interact weakly
Three neutrinos allow only 2 independent Dm2
scales
Dm32 Dm12 Dm22
But there are experimental results at 3 different
Dm2 scales
6
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 Dm2 scales Best fits to the data
require at least two sterile
The Usual 3? Model
(mass)2
7
The MiniBooNE Design Strategy
Keep L/E the same while changing systematics,
neutrino energy and event signature
P(?µ ? ?e) sin22? sin2(1.27?m2L/E)
Order of magnitude higher energy (500 MeV) than
LSND (30 MeV)
Order of magnitude longer baseline (500 m) than
LSND (30 m)
8
The Neutrino Beam
9
The Neutrino Beam
within a magnetic horn (2.5 kV, 174 kA)
that increases the flux by ?6
Booster
Target Hall
MiniBooNE extracts beam from the 8 GeV Booster
delivers it to a 1.7 l Be target
These results correspond to (5.58?0.12) ?1020
POT
10
Modeling Production of Pions

• HARP (CERN)
• 5 ? Beryllium target
• 8.9 GeV proton beam momentum

Data are fit to a Sanford-Wang parameterization.
HARP collaboration, hep-ex/0702024
11
Modeling Production of Kaons
K Data from 10 - 24 GeV. Uses a Feynman
Scaling Parameterization.
data - points dash - total error (fit ?
parameterization)
K0 data are also parameterized.
In situ measurement of K from LMC agrees within
errors with parameterization
12
Neutrino Flux from GEANT4 Simulation

• Intrinsic ?e ?e sources
• µ ? e ?µ?e (52)
• K ? p0 e ?e (29)
• K0 ? p e ?e (14)
• Other ( 5)

p? µ?µ
K? µ?µ
µ? e ?µ?e K? pe?e
Antineutrino content 6
?e/ ?µ 0.5
13
Stability of Running
Full ? Run
Neutrino interactions per proton-on-target.
Observed and expected events per minute
14
Events in the Detector
15
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
16
Optical and PMT Response Models
Attenuation length gt20 m _at_ 400 nm
We have developed 39-parameter Optical
Model based on internal calibration and external
measurement

• Detected photons from
• Prompt light (Cherenkov)
• Late light (scintillation, fluorescence)
• in a 31 ratio for b1

Data/MC overlay
17
The Beam Window
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 ?µ
charged current (CC) interactions with a
characteristic two subevent structure from
stopped µ decay
Tank Hits
µ
Example Event
e
18
Event Timing Structure with Simple Clean-up Cuts
Progressively introducing cuts on the beam window
Veto Hits lt6 removes through-going cosmic
rays This leaves Michel electrons (m?nmnee) and
beam events
Tank Hits gt 200 (effectively energy) removes
Michel electrons, which have a 52 MeV endpoint
All events in beam window
19
Calibration Systems and Sources
20
Predicted Event Rates Before Cuts
From the NUANCE event generator
D. Casper, NPS, 112 (2002) 161
Event neutrino energy (GeV)
21
Charged Current Quasi-elastic
CCQE are 39 of all events Events are clean
(few particles) Energy of the neutrino can be
reconstructed
µ or e
?
?
Beam
N
N
Reconstructed from Scattering angle
Visible energy (Evisible)
An oscillation signal is an excess of ?e events
as a function of E?QE
22
Neutrino Interaction Parameters in NUANCE
From Q2 fits to MB ?µ CCQE data MAeff -
Effective axial mass EloSF - Pauli blocking
parameter From electron scattering data Eb
- Binding energy pf - Fermi momentum
data/MC1 across all angles energy after fit
Model describes CCQE ?µ data well
Kinetic Energy of muon
23
Events Producing Pions

CC p Easy to tag due to 3 subevents. Not a
substantial background to the oscillation
analysis.
µ
?
25
p?
?
N
N
?
NC p0 The p0 decays to 2 photons, which can look
electron-like mimicking the signal... lt1 of
p0 contribute to background.
?
8
p0
?
N
N
The ? also decays to a single photon with 0.5
probability
24
Particles Produced in the Detector and PID
Muons Produced in most CC events. Usually 2
subevent or exiting. Electrons Tag for ?µ??e CC
quasi-elastic (QE) signal. 1 subevent p0s Can
form a background if one photon is weak or exits
tank. In NC case, 1 subevent.
25
Two Independent Analyses
26
Analysis Objectives

• The goal of both analyses is to
• minimize background while
• maximizing signal efficiency.
• Signal range is approximately
• 300 MeV lt EnQE lt 1500 MeV
• One can then either
• look for a total excess
• (counting experiment)
• fit for both an excess and
• energy dependence
• (energy fit)

27
Open Data for Studies
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 lt1s signal outside of
the box Low level information which did not
allow particle-id was available for all events.
28
Analysis Pre-cuts
Both analyses share these hit-level pre-cuts
Only 1 subevent Veto hits lt 6 Tank hits gt 200
And a radius precut Rlt500 cm (where
reconstructed radius is algorithm-dependent)
29
Analysis 1 Track-Based (TB) Analysis
Philosophy
Uses detailed, direct reconstruction of particle
tracks, and ratio of fit likelihoods for the
various event type hypotheses (µ, e and p0) to
identify particles.
This algorithm was found to have the
better sensitivity to ?µ??e appearance. Therefore,
before unblinding, this was the algorithm chosen
for the primary result
30
Rejecting Muon-like Events Using log(Le/Lµ)
log(Le/Lm)gt0 favors electron-like hypothesis
Note photon conversions are electron-like. This
does not separate e/p0. Separation is clean at
high energies where muon-like events are
long. Analysis cut was chosen to maximize the
?µ ? ?e sensitivity
31
Rejecting p0 Events
Using a mass cut
Using log(Le/Lp)
?e CCQE
MC
NC p0
NC p0
?e CCQE
Cuts were chosen to maximize nm ? ne sensitivity
32
Testing e-p0 Separation with Data
1 subevent log(Le/Lm)gt0 (e-like) log(Le/Lp)lt0
(p-like) massgt50 (high mass)
Signal region
BLIND
Sideband Region has ?2 Probability of 69
33
Summary of Track Based Analysis Cuts
Precuts
Log(Le/Lm) Log(Le/Lp) invariant mass
Simulated Backgrounds After Cuts
34
Analysis 2 Boosted Decision Trees (BDT)
Boosted decision trees, or boosting is a
technique for pattern recognition through
training (like a neural net).
Philosophy
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.
This algorithm represents an independent cross
check of the Track Based Analysis.
In the interest of time I will focus on analysis
1 which is the primary analysis
35
Errors, Constraints and Sensitivity
36
Sources of ?e Background
We have two main categories of background
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.
37
Determining the Misidentification Background Rates
BDT
Normalization energy dependence of both
background and signal
Predict
From the ?µ CCQE events
Data/MC Track Based 1.32 0.26
Boosted Decision Tree 1.22 0.29
Tying the ?e background and signal prediction to
the ?µ flux constrains this analysis to a
strict ?µ ? ?e appearance-only search
38
?µ Constraint on Intrinsic ?e from Beam µ Decay
p ? µ ?µ
Measure the ?µ flux Kinematics allows connection
to the p flux
E? (GeV)
En 0.43 Ep
µ ? e ?µ ?e
Ep (GeV)
Once the p flux is known, the µ flux is determined
39
K and K0 Decay Backgrounds
At high energies, well above the signal range, ?µ
and ?e-like events are largely from neutrinos
from kaon decay. The events in these high energy
bins are used to constrain the ?e background from
kaon decay.
signal range
40
p0 Production Constrained with MiniBooNE Data
This reduces the error on predicted mis-identified
p0s
Reweighting improves agreement in other
variables, e.g.?
Because this constrains the ? resonance rate, it
also constrains the rate of ??N?
41
External Sources of Background
External events (sometimes referred to a Dirt
events) are from ? interactions outside of the
detector Ndata/NMC 0.99 0.15
Event Type of External Events after PID cuts
Cosmic Rays
Measured from out-of-beam data 2.1 0.5 events
42
Summary of Predicted Background Rates
Beam Intrinsic ?e
?µ Misidentification
43
Handling Uncertainties in the Analyses
What we begin with...
... what we need
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 EnQE and information on
the correlations between bins
44
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
45
Multisims Used to Convert Uncertainties to Errors
Using Multisims to convert from parameter
uncertainties to errors in EnQE 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.
1000 multisims for K production
70 multisims for the Optical Model
standard MC
Number of events passing cuts in 500 lt EnQE lt600
MeV
46
Form Large Error Matrix for Uncertainties and
Correlations
Error Matrix Elements
Where N is number of events passing cuts MC is
standard monte carlo a represents a given
multisim M is the total number of multisims i,j
are EnQE bins
BDT
Total error matrix is sum from each source.
TB ?e-only total error matrix BDT ?µ-?e total
error matrix
47
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
48
The Initial Results
49
Box Opening Procedure
Progress cautiously, in a step-wise fashion

• After applying all analysis cuts
• Fit sequestered data to an oscillation
  hypothesis, returning no fit parameters.
• Return the c2 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 c2 for a fit to EnQE , without
  returning fit parameters.
• Compare EnQE in data and Monte Carlo, returning
  the fit parameters.
• At this point, the box is open.

50
Step 1
Return the c2 of the data/MC comparison for a
set of diagnostic variables
12 variables are tested for Track Based 46
variables are tested for Boosted Decision Tree
All analysis variables were returned with good
probability except... Track Based analysis c2
Probability for the Evisible fit was only 1
This probability was sufficiently low to merit
further consideration
51
Step 1

• 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

EnQEgt 475 MeV We agreed to report events
over the original full range EnQEgt 300 MeV,
52
Step 1 Redux
Return the c2 of the data/MC comparison for a
set of diagnostic variables
c2 probabilities returned
BDT
TB (EnQEgt475 MeV)
12 variables
46 variables
Parameters of the oscillation fit were not
returned.
53
Step 2
Open up the plots from step 1 for inspection
Examples of what we saw
Evisible
c2 Prob 59
c2 Prob 28
fitted energy (MeV)
TB (EnQEgt475 MeV)
BDT
MC contains fitted signal at an unknown level
54
Step 3
Report the c2 for a fit to EnQE across full
energy range
TB (EnQEgt475 MeV) c2 Probability of fit 99
BDT analysis c2 Probability of fit 52
Leading to...
Step 4
Open the box...
The Track-based ?µ??e Appearance-only Result
Signal significance 0.55 s
Counting experiment within 475ltEnQElt1250 MeV
380 events observed and 358 ?19 (stat) ? 35
(sys) events expeced
55
Track Based Energy Dependent Fit Result
Data are in good agreement with background
prediction.
Error bars are only the diagonal elements
of error matrix. Fit errors for E gt 475
MeV Normalization 9.6 Energy scale 2.3
Best Fit (dashed) (sin22q, Dm2) (0.001, 4 eV2)
56
Oscillation Limits from the Track Based Analysis
Under the hypothesis of ?µ? ?e appearance-only as
the oscillation mechanism
c2 probability, null hypothesis 93
Energy fit 475ltEnQElt3000 MeV
57
Boosted Decision Tree Result
Counting experiment within 300ltE?QElt1600 MeV
971 events observed and 1070 ?33 (stat) ? 225
(sys) events expected
58
Boosted Decision Tree E?QE data/MC Comparison
Error bars are statistical and
systematic (diagonals of matrix)
(sidebands used for constraint not shown)
59
Oscillation Limits from the BDT Analysis
Boosted Decision Tree analysis shows no evidence
for ?µ? ?e appearance-only oscillations.
solid Track based dashed Boosted
Decision Tree
The two independent analyses are in good
agreement.
60
What About the Low Energy Region (Elt475 MeV)?
96 17 20 events above expectation, in the
range of 300ltE?QElt475MeV Deviation 3.7s
(preliminary)
to Egt475 MeV
Background-subtracted
61
Are the Excess Events Consistent with
Oscillations?
Fit to the gt 300 MeV range
Best Fit (dashed) (sin22q, Dm2) (1.0, 0.03
eV2) c2 Probability 18
Examples in LSND allowed range
62

• 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 nm?ne searches

This will be addressed by further study
63
A 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

64
A MiniBooNE-LSND Compatibility Test
Maximum Joint Probability
Dm2 (eV2)
MiniBooNE is incompatible with a ?µ??e
appearance only interpretation of LSND at 98 CL
65
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 ?µ CCQE
production p0 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.
66
Conclusions
Today Ive shown an analysis of MiniBooNE data
within a ?µ??e appearance-only context
The observed reconstructed energy distribution is
inconsistent with oscillations in a ?µ??e
appearance-only model The limit set is
compatible with LSND at a level of 2 or less.
67
Conclusions
Within the energy range defined by this
oscillation analysis, the event rate is
consistent with background.
The observed low energy deviation is under
investigation.
68
Questions
69
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
70
Each event is characterized by 7 reconstructed
variables vertex (x,y,z), time, energy,
and direction (q,f)?(Ux, Uy, Uz). Resolutions
vertex 22 cm direction 2.8?
energy 11
nm CCQE events
2 subevents Veto Hitslt6 Tank Hitsgt200
71
Step 1 Convert the Fundamental
information into Analysis Variables
Fundamental information from
PMTs Analysis Hit Position Charge
Hit Timing variables Energy ? ? Time
sequence ? ? Event shape ? ? ? Physics
? ? ?
Physics p0 mass, EnQE, etc.
72
Examples of Analysis Variables
Resolutions vertex 24 cm direction 3.8º energy
14
Reconstructed quantities which are inputs to EnQE
nm CCQE
nm CCQE
Evisible
UZ cosqz
73
Step 2 Reduce Analysis Variables to a Single
PID Variable
Boosted Decision Trees
A procedure that combines many weak
classifiers to form a powerful committee

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

hit level (charge, time, position)
analysis variables
One single PID score
74
A Decision Tree
Variable 1
(sequential series of cuts based on MC study)
(Nsignal/Nbkgd)
bkgd-like
signal-like
Variable 2
9755/23695
bkgd-like
sig-like
Variable 3
30,245/16,305
1906/11828
7849/11867
bkgd-like
sig-like
9790/12888
20455/3417
etc.
This tree is one of many possibilities...
75
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
Background- like
negative
positive
signal-like
76
BDT cuts on PID score as a function of energy. We
can define a sideband just outside of the
signal region
77
BDT cuts on PID score as a function of energy. We
can define a sideband just outside of the
signal region
78
BDT Efficiency and backgrounds after cuts
Analysis cuts on PID score as a function of Energy
Efficiency after precuts
signal
background
79
Other Single Photon Sources
Neutral Current n N ? n N g
negligible
From Efrosinin, hep-ph/0609169, calculation
checked by Goldman, LANL
Charged Current
lt 6 events _at_ 95 CL
n N ? m N g
where the presence of the g leads to
mis-identification
Use events where the m is tagged by the michel
e-, study misidentification using BDT algorithm.
80
Checked or Constrained by MB data
Further reduced by tying ne to nm
Track Based /Boosted Decision Tree error in
Source of Uncertainty On ne background
Flux from p/m decay 6.2 / 4.3 v
v Flux from K decay 3.3 / 1.0 v
v Flux from K0 decay 1.5 / 0.4 v
v Target and beam models 2.8 / 1.3 v
n-cross section 12.3 / 10.5
v v NC p0 yield 1.8 / 1.5 v
External interactions (Dirt) 0.8 / 3.4
v Optical model 6.1 / 10.5
v v DAQ electronics model 7.5 / 10.8
v
81
Example Cross Section Uncertainties
(Many are common to nm and ne and cancel in the
fit)
MAQE, elosf 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 MAcoh, coh ?????25
? ? N??rate function of ? mom 7 BF EB, pF
9 MeV, 30 MeV ??s
10 MA1? 25 MAN?
40 DIS ? 25
determined from MiniBooNE ?? QE data
determined from MiniBooNE ?? NC ?0 data
determined from other experiments
82
How the Constraints Enter
Two Approaches
TB Reweight MC prediction to match measured ?µ
result (accounting for systematic error
correlations)

• BDT Include the correlations of ?µ to ?e in the
  error matrix

Systematic (and statistical) uncertainties are
included in (Mij)-1
(i,j are bins of EnQE)
83
Two points on interpreting our limit

• 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.

Church, et al., PRD 66, 013001