Improved search for Muon-neutrino to electron-neutrino oscillations in MINOS.

Abstract

We report the results of a search for ν(e) appearance in a ν(μ) beam in the MINOS long-baseline neutrino experiment. With an improved analysis and an increased exposure of 8.2 × 10(20) protons on the NuMI target at Fermilab, we find that 2 sin(2) (θ(23))sin(2)(2θ(13))<0.12(0.20) at 90% confidence level for δ = 0 and the normal (inverted) neutrino mass hierarchy, with a best-fit of 2sin(2) (θ(23))sin(2)(2θ(13)) = 0.041(-0.031)(+0.047) (0.079(-0.053) (+0.071)). The θ(13) = 0 hypothesis is disfavored by the MINOS data at the 89% confidence level.

Full text

Page 1

FERMILAB-PUB-11-351-PPD, BNL-96120-2011-JA
Improved search for muon-neutrino to electron-neutrino oscillations in MINOS
P. Adamson,7D. J. Auty,24D. S. Ayres,1C. Backhouse,18G. Barr,18M. Betancourt,15M. Bishai,3A. Blake,5
G. J. Bock,7D. J. Boehnlein,7D. Bogert,7S. V. Cao,26S. Cavanaugh,9D. Cherdack,27S. Childress,7
J. A. B. Coelho,6L. Corwin,12D. Cronin-Hennessy,15I. Z. Danko,19J. K. de Jong,18N. E. Devenish,24
M. V. Diwan,3M. Dorman,14C. O. Escobar,6J. J. Evans,14E. Falk,24G. J. Feldman,9M. V. Frohne,10
H. R. Gallagher,27R. A. Gomes,8M. C. Goodman,1P. Gouffon,21N. Graf,11R. Gran,16K. Grzelak,28A. Habig,16
J. Hartnell,24R. Hatcher,7A. Himmel,4A. Holin,14X. Huang,1J. Hylen,7G. M. Irwin,23Z. Isvan,19D. E. Jaffe,3
C. James,7D. Jensen,7T. Kafka,27S. M. S. Kasahara,15G. Koizumi,7S. Kopp,26M. Kordosky,29A. Kreymer,7
K. Lang,26G. Lefeuvre,24J. Ling,3,22P. J. Litchfield,15,20L. Loiacono,26P. Lucas,7W. A. Mann,27
M. L. Marshak,15M. Mathis,29N. Mayer,12A. M. McGowan,1R. Mehdiyev,26J. R. Meier,15M. D. Messier,12
D. G. Michael,4, ∗W. H. Miller,15S. R. Mishra,22J. Mitchell,5C. D. Moore,7L. Mualem,4S. Mufson,12J. Musser,12
D. Naples,19J. K. Nelson,29H. B. Newman,4R. J. Nichol,14J. A. Nowak,15J. P. Ochoa-Ricoux,4W. P. Oliver,27
M. Orchanian,4J. Paley,1,12R. B. Patterson,4G. Pawloski,23G. F. Pearce,20S. Phan-Budd,1R. K. Plunkett,7
X. Qiu,23J. Ratchford,26B. Rebel,7C. Rosenfeld,22H. A. Rubin,11M. C. Sanchez,13,1,9J. Schneps,27
A. Schreckenberger,15P. Schreiner,1P. Shanahan,7R. Sharma,7A. Sousa,9N. Tagg,17R. L. Talaga,1J. Thomas,14
M. A. Thomson,5R. Toner,5D. Torretta,7G. Tzanakos,2J. Urheim,12P. Vahle,29B. Viren,3J. J. Walding,29
A. Weber,18,20R. C. Webb,25C. White,11L. Whitehead,3S. G. Wojcicki,23T. Yang,23and R. Zwaska7
(The MINOS Collaboration)
1Argonne National Laboratory, Argonne, Illinois 60439, USA
2Department of Physics, University of Athens, GR-15771 Athens, Greece
3Brookhaven National Laboratory, Upton, New York 11973, USA
4Lauritsen Laboratory, California Institute of Technology, Pasadena, California 91125, USA
5Cavendish Laboratory, University of Cambridge, Madingley Road, Cambridge CB3 0HE, United Kingdom
6Universidade Estadual de Campinas, IFGW-UNICAMP, CP 6165, 13083-970, Campinas, SP, Brazil
7Fermi National Accelerator Laboratory, Batavia, Illinois 60510, USA
8Instituto de F´ ısica, Universidade Federal de Goi´ as, CP 131, 74001-970, Goiˆ ania, GO, Brazil
9Department of Physics, Harvard University, Cambridge, Massachusetts 02138, USA
10Holy Cross College, Notre Dame, Indiana 46556, USA
11Department of Physics, Illinois Institute of Technology, Chicago, Illinois 60616, USA
12Indiana University, Bloomington, Indiana 47405, USA
13Department of Physics and Astronomy, Iowa State University, Ames, Iowa 50011 USA
14Department of Physics and Astronomy, University College London, Gower Street, London WC1E 6BT, United Kingdom
15University of Minnesota, Minneapolis, Minnesota 55455, USA
16Department of Physics, University of Minnesota – Duluth, Duluth, Minnesota 55812, USA
17Otterbein College, Westerville, Ohio 43081, USA
18Subdepartment of Particle Physics, University of Oxford, Oxford OX1 3RH, United Kingdom
19Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, USA
20Rutherford Appleton Laboratory, Science and Technologies Facilities Council, OX11 0QX, United Kingdom
21Instituto de F´ ısica, Universidade de S˜ ao Paulo, CP 66318, 05315-970, S˜ ao Paulo, SP, Brazil
22Department of Physics and Astronomy, University of South Carolina, Columbia, South Carolina 29208, USA
23Department of Physics, Stanford University, Stanford, California 94305, USA
24Department of Physics and Astronomy, University of Sussex, Falmer, Brighton BN1 9QH, United Kingdom
25Physics Department, Texas A&M University, College Station, Texas 77843, USA
26Department of Physics, University of Texas at Austin, 1 University Station C1600, Austin, Texas 78712, USA
27Physics Department, Tufts University, Medford, Massachusetts 02155, USA
28Department of Physics, University of Warsaw, Ho˙ za 69, PL-00-681 Warsaw, Poland
29Department of Physics, College of William & Mary, Williamsburg, Virginia 23187, USA
(Dated: August 2, 2011)
We report the results of a search for νe appearance in a νµ beam in the MINOS long-baseline
neutrino experiment. With an improved analysis and an increased exposure of 8.2 × 1020pro-
tons on the NuMI target at Fermilab, we find that 2sin2(θ23)sin2(2θ13) < 0.12 (0.20) at 90%
confidence level for δ=0 and the normal (inverted) neutrino mass hierarchy, with a best fit of
2sin2(θ23)sin2(2θ13)=0.041+0.047
data at the 89% confidence level.
−0.031(0.079+0.071
−0.053). The θ13=0 hypothesis is disfavored by the MINOS
PACS numbers: 14.60.Pq, 14.60.Lm, 29.27.-a
arXiv:1108.0015v1 [hep-ex] 29 Jul 2011

Page 2

2
It has been experimentally established that neutrinos
undergo flavor change as they propagate [1–7]. This phe-
nomenon is well-described by three-flavor neutrino oscil-
lations, characterized by the spectrum of neutrino masses
together with the elements of the PMNS mixing ma-
trix [8]. This matrix is often parametrized by three Euler
angles θijand a CP-violating phase δ. While θ12and θ23
are known to be large [1, 4, 6], θ13appears to be relatively
small [9–13], with the tightest limits so far coming from
the CHOOZ [10] and MINOS [12] experiments. The T2K
collaboration has recently reported indications of a non-
zero value for θ13at the 2.5σ confidence level (C.L.) [14].
This letter reports new θ13constraints from the MINOS
experiment, using an increased data set and significant
improvements to the analysis.
MINOS is a two-detector long-baseline neutrino os-
cillation experiment situated along the NuMI neutrino
beamline [15]. The 0.98-kton Near Detector (ND) is lo-
cated on-site at Fermilab, 1.04 km downstream of the
NuMI target. The 5.4-kton Far Detector (FD) is located
735 km downstream in the Soudan Underground Labo-
ratory. The two detectors have nearly identical designs,
each consisting of alternating layers of steel (2.54 cm
thick) and plastic scintillator (1 cm). The scintillator lay-
ers are constructed from optically isolated, 4.1 cm wide
strips that serve as the active elements of the detectors.
The strips are read out via optical fibers and multi-anode
photomultiplier tubes. Details can be found in Ref. [16].
The data used in this analysis come from an expo-
sure of 8.2×1020protons on the NuMI target. The cor-
responding neutrino events in the ND have an energy
spectrum that peaks at 3 GeV and a flavor composition
of 91.7% νµ, 7.0% νµ, and 1.3% νe+νe, as estimated
by beamline and detector Monte Carlo (MC) simula-
tions, with additional constraints from MINOS ND data
and external measurements [6, 17]. The two-detector ar-
rangement and the relatively small intrinsic νe compo-
nent make this analysis rather insensitive to beam uncer-
tainties. Neutrino-nucleus and final-state interactions are
simulated using NEUGEN3 [18], and particle propagation
and detector response are simulated with GEANT3 [19].
MINOS is sensitive to θ13 through νµ → νe oscilla-
tions. To leading order, the probability for this oscilla-
tion mode is given by
P(νµ → νe) ≈ sin2(θ23)sin2(2θ13)sin2(1.27∆m2
32L/E) ,
where ∆m2
nant atmospheric oscillation parameters, L (in km) is the
distance between the neutrino production and detection
points, and E (in GeV) is the neutrino energy. We set
constraints on θ13by searching for an excess of νeevents
at the FD. Matter effects and possible leptonic CP vi-
olation modify the above probability significantly [20],
hence our results are presented as a function of δ and the
neutrino mass hierarchy.
32(in units of eV2) and θ23 are the domi-
Events in the MINOS detectors can be characterized
by the spatial patterns of energy deposition in the scin-
tillator strips. Charged-current (CC) νµinteractions are
identified by a muon track extending beyond the more lo-
calized hadronic recoil system. A single detector plane is
1.4 radiation lengths thick, so the electron from a νeCC
interaction penetrates only a few planes (typically 6−12),
leaving a transversely compact pattern of activity inter-
mingled with the associated hadronic shower. Neutral-
current (NC) interactions can mimic this pattern, partic-
ularly when neutral pions are present.
To obtain a νe-enriched sample, we apply a series of
selection criteria to the recorded neutrino events. We
require that the neutrino interaction occur within a fidu-
cial volume. We eliminate most νµ CC interactions by
rejecting events with a track longer than 24 planes and
events with a track extending more than 15 planes be-
yond the hadronic shower. We require that an event have
at least five contiguous planes with energy greater than
half that deposited on average by a minimum ionizing
particle. The calorimetrically determined event energy
must lie between 1 and 8 GeV, as events below 1 GeV are
overwhelmingly from NC interactions and events above
8 GeV have negligible νµ → νe oscillation probability.
The time and reconstructed direction of each event must
be consistent with the low-duty-cycle NuMI neutrino
source. These “pre-selection” criteria preserve 77% of
oscillation-induced νeCC events originating in the fidu-
cial volume while passing 8.5% of νµ CC, 39% of NC,
54% of ντ CC, and 35% of intrinsic νe CC events, as
estimated by the simulation.
Further background suppression requires a more so-
phisticated examination of the energy deposition pat-
terns. Earlier MINOS νe appearance searches used an
artificial neural network event classifier with eleven input
variables characterizing the transverse and longitudinal
profiles of an event’s activity in the detector [11, 12, 21].
The present analysis uses a nearest-neighbors algorithm
dubbed “library event matching” (LEM) [22]. In LEM,
each candidate event is compared to 5×107simulated sig-
nal and background events [23], one by one, to find the 50
that look most similar to the candidate event. A library
event is rejected if its reconstructed energy, number of
active strips, or number of active planes differs from that
of the candidate event by more than 20%. The similarity
of the candidate event to each remaining library event is
quantified by the following likelihood:
logL =
Nstrips
?
i=1
log
??∞
0
P(ni
cand;λ)P(ni
lib;λ)dλ
?
,
where the sum is taken over all strips with a signal above
3 photoelectrons in either of the two events, ni
charge (in photoelectrons) observed on strip i in event
x (with x either “candidate” or “library”), and P(n;λ)
is the Poisson probability for observing n given mean λ.
xis the

Page 3

3
Since events occur throughout the detector volume, each
event is translated to a fixed reference location before
L is evaluated. Strips far away from the event’s central
axis are combined before comparison. Additionally, li-
brary events are shifted by ±1 plane in search of a better
likelihood.
The final classifier is formed using a neural network
that takes as its inputs the reconstructed event energy
along with three variables derived from the best-match
ensemble: (1) the fraction of the 50 best-matched events
that are true νeCC events, (2) the average inelasticity
?y? of those νeCC events, and (3) the average fraction
of charge that overlaps between the input event and each
νeCC event. The resulting LEM discriminant is shown
in Fig. 1.
POT
19
Events / 10
00
20002000
40004000
6000 6000
80008000
POT
19
Events / 10
Data
Monte Carlo
Near Detector
LEM DiscriminantLEM Discriminant
00 0.10.10.2 0.2 0.30.30.4 0.4 0.50.50.6 0.60.7 0.70.80.8 0.9 0.911
POT
20
10
×
Events / 8.2
00
2020
40 40
6060
8080
POT
20
10
×
Events / 8.2
Background
10
×
Signal
Far Detector
) = 0.1
13
θ
(2
> 0,
32
m
2
sin
∆
4
π
=
23
θ
= 0,
CP
δ
2
FIG. 1: [Top] Distribution of the LEM discriminant for events
in the Near Detector that pass the pre-selection requirements.
Data (points) and Monte Carlo simulation (histogram) are
shown, with the magnitude of the systematic uncertainty in-
dicated by the band. This uncertainty is highly correlated
between the ND and FD and thus cancels out to a large de-
gree when we form our FD predictions. [Bottom] Expected
background and signal distributions in the Far Detector for
sin2(2θ13)=0.1. The signal distribution has been multiplied
by 10 for visibility.
We form a prediction for the FD event rate, in each of
Uncertainty source
Uncertainty on
background events
4.0%
2.1%
1.9%
1.1%
2.0%
5.4%
Event energy scale
ντ background
Relative FD/ND rate
Hadronic shower model
All others
Total
TABLE I: Systematic uncertainties on the number of pre-
dicted background events in the FD in the signal region, de-
fined by LEM>0.7. The final θ13 measurement uses multiple
LEM and reconstructed energy bins and thus uses a full sys-
tematics covariance matrix. These uncertainties, which are
small compared to the statistical errors, lead to a 7.0% loss in
sensitivity to sin2(2θ13). The “All others” category includes
uncertainties relating to the neutrino flux, cross sections, de-
tector modeling, and background decomposition.
15 bins (specified below) of LEM discriminant and recon-
structed energy, using the corresponding rate observed in
the ND. The ND rates are first broken down into indi-
vidual background contributions, as different background
types translate differently from the ND to the FD due to
oscillations and beamline geometry. To determine the rel-
ative background contributions in the ND rates, we apply
the νe selection to ND data collected in multiple beam
configurations with differing neutrino energy spectra and
thus differing background compositions. This allows the
construction of a system of linear equations that can be
solved for the relative contributions of NC, νµCC, and
intrinsic νe CC backgrounds in the primary low-energy
beam configuration [12]. The measured composition of
ND events, averaged over the range LEM>0.7 for recon-
structed energy between 1 and 8 GeV, is (61 ± 1)% NC,
(24 ± 1)% νµCC, and (15 ± 1)% νeCC.
We convert the resulting decomposed ND rates directly
into predictions for the FD rates using a Monte Carlo
simulation. More specifically, we use the simulated ra-
tio of FD and ND rates, for each background type and
for each LEM and energy bin, as the conversion factor
for translating the measured ND rate into the FD pre-
diction. We evaluate uncertainties on these ratios using
systematically modified samples of simulated ND and FD
data. The dominant systematic effects are summarized
in Table I.
Since the ND collects negligibly few events arising from
νµ → νe or νµ → ντ oscillations, the FD rates for
these events are estimated using the simulation plus the
observed νµ CC rates in the ND. For νe CC events,
we further apply an energy- and LEM-dependent cor-
rection to the FD predictions that is derived from hy-
brid events composed of electrons from simulation and
hadronic showers from data. The hadronic showers are
obtained by removing the muon hits from cleanly iden-
tified νµCC events [12, 24, 25], and the electromagnetic
shower simulation is verified using a pure sample of elec-

Page 4

4
Event class
sin2(2θ13)
0
34.1
6.7
6.4
2.2
0.2
49.6
0.1
34.1
6.7
6.2
2.1
19.1
68.2
NC
νµ CC
νe CC
ντ CC
νµ → νe CC
Total
TABLE II: Expected FD event counts for LEM>0.7, assum-
ing θ23=π
line refers to the intrinsic νe component in the beam. In the
θ13=0 case, a small amount of νµ → νeoscillation occurs due
to non-zero ∆m2
4, ∆m2
32=2.32×10−3eV2, and δ =0. The first νe
21.
trons recorded by the MINOS Calibration Detector [26].
The breakdown of expected FD events is given in Ta-
ble II. An analysis of beam-off detector activity yielded
no νe candidate events, resulting in a 90% C.L. upper
limit on cosmogenic backgrounds in the primary analysis
region of 0.3 events. We find that (40.4±2.8)% of νeCC
signal events end up in the signal region, LEM>0.7.
Most of the analysis procedures can be tested directly
on two signal-free or near-signal-free sideband samples.
First, the “muon-removed” hadronic showers described
above, before they are merged with simulated electrons,
represent a sample of NC-like events.
and observed LEM distributions in the FD agree for
this sample, with χ2/Nd.o.f.=9.7/8 using statistical errors
only. Second, FD events satisfying 0≤LEM<0.5 make
up a background-dominated sample for which we predict
370 ± 19 background events (statistical error only). We
observe 377 events, in agreement with prediction. Form-
ing the prediction for the latter sideband exercises all
aspects of the analysis up to the final signal extraction,
including the full ND decomposition procedure and the
ND-to-FD ratios derived from simulation.
In previous MINOS analyses [11, 12], the νe appear-
ance search was conducted by comparing the total num-
ber of νe candidate events in the FD to the expected
background. A similar approach applied to the present
data yields 62 events in the signal region of LEM>0.7,
with an expectation of 49.6±7.0(stat.)±2.7(syst.) if
θ13=0.However, we gain 12% in sensitivity by fit-
ting the FD sample’s LEM and reconstructed energy
(Ereco) distribution in 3×5 bins spanning LEM>0.6 and
1 GeV<Ereco<8 GeV.The energy resolutions for
hadronic and electromagnetic showers at 3 GeV are 32%
and 12%, respectively [16]. Figure 2 shows the FD data
and predictions used in the fit, along with the extracted
best-fit signal.
Figure 3 shows the regions of oscillation parame-
ter space allowed by these data.
use a three-flavor oscillation framework [20] includ-
ing matter effects [27], and we use the Feldman-
The predicted
For the fit, we
Events
0
5
10
15
20
Data
Background
Signal
0.6 < LEM < 0.7
MINOS Far Detector Data
Events
0
5
10
15
20
0.7 < LEM < 0.8
4
π
=
23
θ
= 0,
CP
δ
> 0,
32
m
2
∆
) = 0.041
13
θ
(2
2
Best Fit: sin
Reconstructed Energy (GeV)
2468
Events
0
5
10
15
20
Bins Merged for Fit
LEM > 0.8
FIG. 2: Reconstructed energy spectra for νe CC candidate
events in the Far Detector. The black points indicate the
data with statistical error bars shown. The histogram indi-
cates the expected background (unfilled area) together with
the contribution of νµ → νe signal (hatched area) for the
best-fit value of sin2(2θ13)=0.041.
Cousins procedure [28] to calculate the allowed re-
gions. We assume
32
∆m2
and θ12=0.60±0.02 [1]. The influence of these oscilla-
tion parameter uncertainties is included when construct-
ing the contours.
??∆m2
??=(2.32+0.12
−0.08)×10−3eV2[6],
21=(7.59+0.19
−0.21)×10−5eV2[1], θ23=0.785±0.100 [4],
Prior to unblinding the FD data, we planned to fit only
the LEM distribution integrated over energy. However,
the excess over background in the upper energy range
prompted the inclusion of energy information so that the
fit could weigh events appropriately when extracting θ13
constraints. If we had performed the signal extraction
over LEM bins only, the best fit and 90% C.L. upper
limit for sin2(2θ13) would each change by +0.006. A thor-
ough study of high-energy events in the signal and side-

Page 5

5
) π
(
δ
(
δ
0.0 0.0
2.0 2.0
0.5 0.5
1.0 1.0
1.51.5
2.02.0
> 0
2
m
∆

MINOS Best Fit
68% C.L.
90% C.L.
CHOOZ 90% C.L.
= 1 for CHOOZ
23
2sin
θ
2
) π
2323
θθ
22
)sin
13
θ
(2
13
θ
(2
22
2sin 2sin
00 0.10.1 0.20.2 0.30.30.4 0.4
) π
(
δ
(
δ
0.00.0
0.5 0.5
1.01.0
1.51.5
< 0
2
m
∆
MINOS
×
8.2 POT
20
10
)sin
) π
FIG. 3: Allowed ranges and best fits for 2sin2(θ23)sin2(2θ13)
as a function of δ. The upper (lower) panel assumes the nor-
mal (inverted) neutrino mass hierarchy. The vertical dashed
line indicates the CHOOZ 90% C.L. upper limit assuming
θ23=π
4and ∆m2
32=2.32×10−3eV2[10].
band samples, including events between 8 and 12 GeV,
indicates that the high-energy predictions are robust and
that the selected events are free of irregularities.
In conclusion, using a fit to νe discriminant and
reconstructed energy 2D distribution of FD νe can-
didate events, we find that 2sin2(θ23)sin2(2θ13)
0.041+0.047
−0.031
mass hierarchy and δ=0.
2sin2(θ23)sin2(2θ13)<0.12 (0.20) at 90% C.L. Using the
less sensitive techniques of the 2010 analysis [12] on the
current data set yields a consistent measurement [29].
The θ13=0 hypothesis is disfavored by the MINOS data
at the 89% C.L. This result significantly constrains the
θ13range allowed by the T2K data [14] and is the most
sensitive measurement of θ13to date.
This work was supported by the U.S. DOE; the U.K.
STFC; the U.S. NSF; the State and University of Min-
nesota; the University of Athens, Greece; and Brazil’s
FAPESP, CNPq, and CAPES. We are grateful to the
Minnesota Department of Natural Resources, the crew
of the Soudan Underground Laboratory, and the staff of
Fermilab for their contributions to this effort.
=
(0.079+0.071
−0.053) for the normal (inverted)
We further find that
∗Deceased.
[1] B. Aharmim et al. (SNO), Phys. Rev. Lett. 101, 111301
(2008).
[2] T. Araki et al. (KamLAND), Phys. Rev. Lett. 94, 081801
(2005).
[3] Y. Ashie et al. (Super-Kamiokande), Phys. Rev. Lett. 93,
101801 (2004).
[4] Y. Ashie et al. (Super-Kamiokande), Phys. Rev. D 71,
112005 (2005).
[5] K. Abe et al. (Super-Kamiokande), Phys. Rev. Lett. 97,
171801 (2006).
[6] P. Adamson et al. (MINOS), Phys. Rev. Lett. 106,
181801 (2011).
[7] P. Adamson et al. (MINOS), Phys. Rev. Lett. 107,
021801 (2011).
[8] B. Pontecorvo, JETP 34, 172 (1958); V. N. Gribov and
B. Pontecorvo, Phys. Lett. B 28, 493 (1969); Z. Maki,
M. Nakagawa, and S. Sakata, Prog. Theor. Phys. 28,
870 (1962).
[9] F. Boehm et al. (Palo Verde), Phys. Rev. Lett. 84, 3764
(2000).
[10] M. Apollonio et al. (CHOOZ), Eur. Phys. J. C 27, 331
(2003).
[11] P. Adamson et al. (MINOS), Phys. Rev. Lett. 103,
261802 (2009).
[12] P. Adamson et al. (MINOS), Phys. Rev. D 82, 051102
(2010).
[13] G. L. Fogli et al., Phys. Rev. Lett. 101, 141801 (2008);
M. C. Gonzalez-Garcia, M. Maltoni and J. Salvado,
JHEP 1004, 056 (2010); T. Schwetz, M. Tortola and
J. W. F. Valle, New J. Phys. 13, 063004 (2011).
[14] K. Abe et al. (T2K), Phys. Rev. Lett. 107, 041801
(2011).
[15] K.Anderson et al., FERMILAB-DESIGN-1998-01
(1998).
[16] D. G. Michael et al., Nucl. Inst. & Meth. A 596, 190
(2008).
[17] C. Alt et al. (NA49), Eur. Phys. J. C 49, 897 (2007).
[18] S.Dytman, H.Gallagher,
arXiv:0806.2119 (2008).
[19] R. Brun et al., CERN Program Library W5013 (1984).
[20] E. K. Akhmedov et al., J. High En. Phys. 04, 078 (2004).
[21] T. Yang, Ph.D. Thesis, Stanford Univ. (2009).
[22] A description of an earlier version of LEM can be found
in J. P. Ochoa, Ph.D. Thesis, Caltech (2009).
[23] The library composition is 40% νe CC (signal) and
60% NC (background). Only NC background events are
needed to obtain good performance, as potentially mis-
identified νµ CC and ντ CC events look quite similar
to NC events. We note that the performance of LEM is
measured on all event classes.
[24] J. Boehm, Ph.D. Thesis, Harvard Univ. (2009).
[25] A. Holin, Ph.D. Thesis, Univ. College London (2010).
[26] P. Adamson et al., Nucl. Inst. & Meth. A 556, 119 (2006);
A. Cabrera et al., Nucl. Inst. & Meth. A 609, 106 (2009);
P. Vahle, Ph.D. Thesis, Univ. of Texas at Austin (2004).
[27] Weassumeanelectron
8.28×1023cm−3.
[28] G. J. Feldman and R. D. Cousins, Phys. Rev. D 57, 3873
(1998).
[29] The older techniques applied to this data set yield a
nearly identical best fit value (shifted by +0.0005) but
a less constraining 90% C.L. upper limit (+0.015).
andM. Kordosky,
number densityof