1
MiniBooNE Oscillation Update
Mike Shaevitz
Columbia University
XII International Conference on “Neutrino Telescopes”
March 7, 2007
2
Outline
• Introduction: LSND, sterile neutrinos
• MiniBooNE Experiment:
Setup, reconstruction, calibration, particle ID, NuMI offaxis beam
• Systematic Uncertainties for oscillation analysis
• Main backgrounds and Constraints
• Oscillation analysis and sensitivity
3
The LSND Experiment
LSND observed a (~3.8s) excess ofe events: 87.9 ± 22.4 ± 6.0 events
Oscillation Probability: P(    e )  (0.264  0.067  0.045)%
4
Why Sterile Neutrinos?
Remember: mij2  mi2  m2j
 Three distinct neutrino oscillation signals
2
2
2
with msolar
 matm
 mLSND
 For three neutrinos, expect
2
2
2
m21
 m32
 m31
m23
m12
m13
3+2 models
m5
• One of the experimental
measurements is wrong
Need better
measurement in
LSND region
 MiniBooNE
• Additional “sterile” neutrinos
involved in oscillations
(M.Sorel, J.Conrad, M.Shaevitz, PRD 70(2004)073004 (hep-ph/0305255)
G. Karagiorgi et al., PRD75(2007)013011 (hep-ph/0609177)
hep- ph/ 0305255
5
The MiniBooNE Experiment
LMC
8GeV
Booster
•
•
•
•
•
•
•
K+
p+
magnetic horn
and target
+

decay pipe
25 or 50 m
?
e
450 m dirt
detector
Proposed in summer 1997，operating since 2002
Goal to confirm or exclude the LSND result
Similar L/E as LSND
– Baseline: L = 451 meters, ~ x15 LSND
– Neutrino Beam Energy: E ~ x(10-20) LSND
Different systematics: event signatures and backgrounds different from LSND
High statistics: ~ x5 LSND
5.579E20 POT for neutrino mode since 2002.
Switch horn polarity to run anti-neutrino mode since January 2006.
6
The MiniBooNE Collaboration
Y.Liu, D.Perevalov, I.Stancu
University of Alabama
S.Koutsoliotas
Bucknell University
R.A.Johnson, J.L.Raaf
University of Cincinnati
T.Hart, R.H.Nelson, M.Tzanov
M.Wilking, E.D.Zimmerman
University of Colorado
A.A.Aguilar-Arevalo, L.Bugel L.Coney,
J.M.Conrad, Z. Djurcic,
K.B.M.Mahn, J.Monroe, D.Schmitz
M.H.Shaevitz, M.Sorel, G.P.Zeller
Columbia University
D.Smith
Embry Riddle Aeronautical University
L.Bartoszek, C.Bhat, S.J.Brice
B.C.Brown, D. A. Finley, R.Ford,
F.G.Garcia, P.Kasper, T.Kobilarcik,
I.Kourbanis, A.Malensek, W.Marsh,
P.Martin, F.Mills, C.Moore, E.Prebys,
A.D.Russell , P.Spentzouris,
R.J.Stefanski, T.Williams
Fermi National Accelerator Laboratory
D.C.Cox, T.Katori, H.Meyer, C.C.Polly
R.Tayloe
Indiana University
G.T.Garvey, A.Green, C.Green, W.C.Louis, G.McGregor, S.McKenney
G.B.Mills, H.Ray, V.Sandberg, B.Sapp, R.Schirato, R.Van de Water
N.L.Walbridge, D.H.White
Los Alamos National Laboratory
R.Imlay, W.Metcalf, S.Ouedraogo, M.O.Wascko
Louisiana State University
J.Cao, Y.Liu, B.P.Roe, H.J.Yang
University of Michigan
A.O.Bazarko, P.D.Meyers, R.B.Patterson, F.C.Shoemaker, H.A.Tanaka
Princeton University
P.Nienaber Saint Mary's University of Minnesota
J. M. Link Virginia Polytechnic Institute
E.Hawker Western Illinois University
A.Curioni, B.T.Fleming Yale University
  e Oscillation Search
• Beam
 e /   0.5%
• Detector
•
•
•
•
12m diameter tank
Filled with 900 tons of pure mineral oil
Optically isolated inner region with 1280 PMTs
Outer veto region with 240 PMTs.
•
Detector Requirements:
– Detect and Measure Events: Vertex, E …
– Separate  events from e events
7
Oscillation Signal
 An Excess of “e” Events over Expectation
• Understanding the expected events is therefore the key
– Need to know the neutrino fluxes
• Electron neutrinos from , K+, and K0 decay
• Muon neutrinos can make background or give the signal
– Need to know the /e neutrino cross section vs. energy
• Events = flux × cross section
– Need to know the e reconstruction efficiency vs energy
• Observed events = efficiency × events
– Need to know the probability for  events to be mis-identified as
e events  Events with single EM showers look like e events
• Neutral current (NC) p0 events are the main mis-id background
• NC  production followed by radiative decay, N
• Photons entering from outside detector (“Dirt” background)
8
9
Event Reconstruction
• Use energy deposition and timing of hits
in the phototubes
– Prompt Cherenkov light
• Highly directional with respect to
particle direction
• Used to give particle track
direction and length
– Delayed scintillation light
• Amount depends on particle type
Delayed Scintillation
Spectrum of Michel electrons
from stopping
muons(MeV)
Michel
electron energy
Preliminary
Calibrations
15%
E resolution
at 53 MeV
PRELIMINARY
Energy vs. Range for
events stopping in
scintillator cubes
p0 Mass Distribution
Preliminary
10
•
•
Muon id from delayed decay electron
signature (92% non-capture probability)
11
Particle ID Algorithms
Identify events using
– hit topology
•
PID Vars
– Reconstructed physical observables
•
Track length, particle production angle
relative to beam direction
e
candidate
– Auxiliary quantities
•
Timing, charge related : early/prompt/late
hit fractions, charge likelihood
– Geometric quantities
•
•
Distance to wall

candidate
Two PID algorithms
1. Likelihood based analysis: e/ and e/p0
2. A “boosted decision tree” algorithm to
separate e, , p0
(See B. Roe et al. NIM A543 (2005))
p0
candidate
NuMI Offaxis  MiniBooNE’s Calibration Beam
12
Need to verify our PID with e in the signal energy range,but can’t due to blind analysis.
 Solution: use someone else’s beam!
MiniBooNE
Sitting off axis, we see a beam which is enhanced
in e flux and is in a useful energy range.
Results for Offaxis NuMI Beam in MiniBooNE
•
•
MiniBooNE sees events
from the offaxis NuMI beam
– They show up as
events during the 8 s
NuMI beam window
These events have a
significant e component
that can be used to test our
PID system
Likelihood Algorithm
Boosted
Decision
Tree Algorithm
Preliminary
PRELIMINARY
13
14
MiniBooNE Oscillation Search Method
• Do a combined oscillation fit to the observed  and e energy
distribution for data vs prediction
• Systematic (and statistical) uncertainties in (Mij)-1 matrix
– Uncertainties come from analyses of external and internal data
– Covariance matrix includes correlations between e and  events
• Predictions for the various backgrounds are directly constrained by
actual MiniBooNE measurements
– Constraints significantly reduce systematic uncertainties
– Combined fit also reduces e uncertainties using high stat  events
15
Expected Event Numbers
Events with  Selection Requirements
No Cuts
Events = 193,730 (mainly  CCQE)
(Final data sample for 5.58 × 1020 pot)
Simple Cuts:
E > 60 MeV
Veto Hits > 6
Events with e Selection Requirements
Events Time (ns)
Events Time (ns)
Total Expected Background = 915 events
Example Osc Signal = 315 events
(m2 = 0.4 eV2 , sin22q = 0.017)
16
Systematic Uncertainties
• Sources of uncertainty for the e candidate events
– Uncertainties come from modeling the beam, neutrino
interactions, and the detector
(These uncertainties will be reduced by using MiniBooNE data)
• Uncertainties from external and non-osc internal constraints
–
–
–
–
–
–
–
–
–
Neutrino flux from p+ decay
Neutrino flux from K+ decay
Neutrino flux from K0 decay
Neutrino flux from + decay
Xsec uncertainties
External interactions (“Dirt”)
NC p0 Mid-ID
Radiative N
Optical Model
Next: Few examples of these systematic uncertainty estimates.
Pion and Kaon Production
• pBe Pion production
– pBe production s measured by the
HARP collaboration at pproton = 8.9 GeV
– MiniBooNE uses a parameterization
with uncertainties set to cover
measurements.
17
• pBe K+ and K0 production
– Use external pBe cross section
measurements for beam momenta from
9.5 – 24 GeV
– MiniBooNE uses a parameterization with
uncertainties set to cover measurements.
18
 Cross Section Uncertainties
• Differential cross section for quasielastic scattering determined from
MiniBooNE data
• Shape fits are performed to observed
data Q2 distribution using a
relativistic-Fermi-gas model
• Two parameters (and their
uncertainties) are determined:
– Axial mass parameter, MA
– A Pauli blocking parameter
• Fit also agrees well with neutrino
energy distributions
• Other cross sections (i.e. CC1p) are
determined from MiniBooNE data
combined with previous external
measurements
Quasi-Elastic Scattering:    n     p
19
Optical Model Uncertainties
Uncertainties on the parameters associated
with optical model are used to determine the
uncertainties for the oscillation search
• Light Creation
– Cerenkov – well known
– Scintillation
• yield
• spectrum
• decay times
• Light Propagation
●
– Fluoresence
• rate
• spectrum
• decay times
– Scattering
• Rayleigh
• Particulate (Mie)
– Absorption
In Situ
–
●
Cosmics muons, Michel electrons, Laser
External
–
Scintillation from p beam (IUCF)
–
Scintillation from cosmic  (Cincinnati)
–
Fluorescence Spectroscopy (FNAL)
–
Time resolved spectroscopy (Princeton, JHU)
–
Attenuation (Cincinnati)
Backgrounds Constraints from MiniBooNE Data
• All of the major backgrounds for the oscillation search can be
constrained directly from measurements using MiniBooNE data
• NC p0 production
– Largest Mis-ID background is from NC p0 production where one of the
decay photons is missed. This background is constrained from the
large fraction of NC p0 events that are observed and measured in
MiniBooNE
• External events
– Backgrounds from the events in material outside of the MiniBooNE
detector are constrained by the isolation and measurement of such
events.
• Intrinsic kaon decay e’s
– Intrinsic e background from kaon decay can be constrained by
observed e events at high energy where there are no oscillation
events
• Intrinsic muon decay e’s
– Largest intrinsic e background is from muon decay and is highly
constrained by the observed  events. The constraint is applied by
using the combined e/  oscillation fit.
20
NC p0 Background Constraint
• Using PID variables isolate a
very pure sample p0 events from
  N    N + p0
(mainly from   N + p0 )
• Purity ~90% or greater
• Measure p0 production rate as a
function of p0 momentum and
compare to MC prediction to
calculate a correction factor.
• Correct NC p0 mis-ID rate using
this measured correction factor
(Also can be used to correct the
  N +  radiative background)
21
M Mass Distribution for Various pp0 Momentum Bins
Constraining External Event (“Dirt”) Background
• Neutrino beam interacts with material outside of MiniBooNE detector
creating photons (100 ~ 300 MeV) that come into the tank and produce
electron-like events.
• Dirt events contribute ~10% of background for oscillation nue search.
• N_dirt_measured / N_dirt_expected = 0.99 ± 0.15
Enhanced
Background
Cuts
Event Type of Dirt after PID cuts
22
Kaon Intrinsic e Background Constrained by High Energy Data
• At high energy, candidate e events are mainly from kaon decay
– Small contributions from -decay and p0 mis-id
• Normalization of the kaon intrinsic background can be partially
constrained within uncertainties by the level of this high energy
data
Preliminary
23
24
Constraining the Intrinsic e Background from Muon Decay
• Muon decay is the largest source of e background but is highly constrained
by the observed  events.
– MiniBooNE subtends a very small forward solid angle for neutrinos from pion
decay  observed E  0.6 Ep
– So, the measured  energy spectrum gives both the number and energy
spectrum of the decaying pions
– These decaying pions are the source of the e mu-decay background
• The combined / e oscillation fit:
– Automatically takes this correlation into account
– Effectively constrains the e background with an error that depends
primarily on the  event statistics.
Summary MiniBooNE Oscillation Analysis Strategy
• Develop a very detailed
simulation of the neutrino beam
and detector using both internal
and external information
90%CL
• Constrain the uncertainties
associated with the simulation
using actual MiniBooNE data and
measurements
• Accomplish the oscillation search
by doing a combined e/  fit to
the observed event distribution
vs. energy.
Monte Carlo
Sensitivity
Estimate
25
26
Summary
• From the beginning, MiniBooNE decided to do a “blind analysis”
– Candidate e events in the oscillation energy region were
sequester (~5000 events) (“Closed Box”)
– The other several 100,000 events were open for examination
(“Open Box”)
• Collaboration is in the final stages:
– Checking the “Open Box” data including new “Side Band” regions
– Assessing final cuts for enhancing the oscillations sensitivity
• Opening the “Closed Box” is close
– After opening, the result will be presented after about two weeks.