PoS(FPCP2017)025
results
neutrino It uses theIt uses latest -
asurements of http://pos.sissa.it/ and muon
h presentedh the River, Minnesota. The Minnesota. beam River,
the experimentthe include me
neutrino appearance phase. This a is proceedingwork phase. to a talk
-
based neutrino oscillation experiment.
- violating
ND ND 4.0). - POT.
CP NC of its measurements and whic -
20
10
x
05
.
baseline accelerator - 23 octant and the Fermilab and measures electron
Collaboration equivalent to 6 NoDerivatives 4.0 International License (CC BY 4.0(CC License NoDerivatives International the experimentthe and the status -
*
ý exposure
zech Republic 2017
A latest results latest A , C [email protected] , NonCommercial described - Speaker 13, mass hierarchy, theta 13, hierarchy, mass * June
mail: 9 Copyright owned by the author(s) under the terms of the Creative Commons of the under the by terms owned the Creative Copyright author(s) -
- Attribution The 15thonViolation Flavor PhysicsCP International Conference & 5 Prague
disappearance its between onsiteAsh near detector detectorandfar the in reachedrecently power 700kW design the benchmark. Goals of theta which based on the E experiment is a long The NOvA upgraded NuMI beam from Filip Jedin on behalf of the NOvA University in Prague Czech Technical Prague,7, Czech RepublicBrehova NOv
PoS(FPCP2017)025
- ý y in be by by
and and
these these Main Main
) with called
recent recent mode.
] - facility facility mixing mixing protons protons
2
, [ per pulse pulse per
the NuMI 23 secondar angles and and angles 20
Filip Jedin 2
so it can , the previous previous the
10 recycler recycler ring;
∆m assive detector assive ×
m protons
decay kinematics, kinematics, decay
13 of of NOvA between the neutrino the neutrino between pion pion decays result in Therefore, the NOvA
10 current (NC) neutrino neutrino (NC) current
- band beam peaked very very peaked beam band ×
with with the detectors placed on stacking stacking in the neutral narrow beam energy spectrum energy beam narrow
-
xperiment xperiment looking mostly for axis, most accelerator accelerator neutrinos, a so -
s leading neutrino experiments neutrino leading s physics physics goals ’ off
beam of neutrinos and a and neutrinos of beam
measure a narrow a measure
mrad oscillated acting acting oscillation parameters: 1.6 GeV). The GeV). 1.6
. The beam was recently upgraded upgraded recently was beam The . ~ hange between neutrino and antineutrino and neutrino between hange µ
v 1 ). To ). meet To the intensive µ v down of higher energy of higher down - very very
and
+ µ cient number of
ffi and the neutrino mass hierarchy. mass and the neutrino detector neutrino oscillation e u
- CP
und for neutrino oscillation searches δ the NuMI beam axis will axis beam NuMI the protons on target per year of running). From pion From of running). year per target on protons ned NuMI beam of beam NuMI ned
beam spill every 1.33 s in an intensity of 4.9 4.9 of intensity an in s 1.33 every spill beam
detect in detectors NOvA is the NuMI beam Neutrinos at the
laboratory laboratory frame. For 14 fi ff s 20 µ
de 10 - Appearance Appearance experiment (NOvA) was designed to answer some of
e 10 ×
v pulse cycle time of the Main Injector via slip -
baseline, baseline, two
s - .
in well ] µ
Axis Axis e oscillation maximum at 810 km (E km 810 at maximum oscillation
- v e 3,4 ff v . Fully operating since 2015, its one of the world the of one its 2015, since operating Fully . [
particles particles can decay into ]
to 1 experiment
+ [ µ π , CP violating phase phase violating , CP ased proton intensity. proton ased v 3
ship ship of the Intensity Frontier particle physics program of Fermi National Laboratory 2 θ latest results
generation experiment has to have a have to has experiment generation NuMI Beam
target in the "full" 14kton detector. This exposure is more than double that used in used that double than more is exposure This detector. 14kton "full" the in target
NOvA Introduction - pearance of pearance
OvA OvA nd on publications detectors placed 14 mrad o mrad 14 placed detectors the near feed from background the reduces strongly events, the dominant backgro axis. The oscillation results presented here from come an exposure equivalent to 6.05 As a result, we have a have we result, a As to 6.0 (corresponding angle will decay on depend we the the know characteristic energy neutrino and the parent pion in the neutrinos with E = some with2 energy GeV, smearing around that value. beam beam was upgraded from 300 kW to 700 kW of reducing the 10 nominal beam power. This is achieved increasing the intensity per cycle with 12 Booster batches instead of 11 by installing new stations and a RF new kicker injection magnet; and upgrading the target and horns to accommodate incre the 2.1 The source of neutrinos to the from protons TheMain 120GeV Injector created by extracting are neutrinos Injector. at Fermilab, and colliding them with a graphite which target, results in production of mesons ( To carry To out a measurement of s 2 detects the several NOvA NuMI hundreds beam of (same kilometers asaway. MINOS the target. from 1 km (ND) Detector and Near from the target km (FD) 810 Far Detector delivering 700 kW of beam power, with optional c optional with power, beam of kW 700 delivering From is these capable NOvA measurements, of extr angle 2. questions questions the flag (USA). IL Batavia, NOvA is a long ap 1. Neutrino physics, with many interesting topics and questions reaching beyond Standard Model, in made progress huge the Despite physicists. particle for appealing more and more being is mixing the ordering, mass the i.e. regarding questions, open many still are there years, sector. neutrino in violation the CP The NuMI O N PoS(FPCP2017)025
µ ý d al al v
the the ady ady
PVC PVC
liquid liquid signal signal of the
further further resents resents
and was was and network network
complex, complex, . When hits hits in the individu ]
using a set a set using ppearance
scintillator scintillator Filip Jedin
4 a ,
isappearance e 3 v [
like the readout readout the like of of the cell by a
events, events, charged
features from features the rst rst case of a HEP
gies gies and analyses.
in a visual input to photodiode (APD). (APD). photodiode
. The The . results results have alre layers layers of the 14kton has has 206 layers, each ]
cation and cation rep over over other likelihoo
7 nolo fi [ computation. computation. interleaving interleaving steel plates t
fi
the neural net is equivalent equivalent is net neural the cial images cial
sters, sters, without any cation fi LID and LEM LID
er was developed to select a sectionanalyses. ublication ublication is the fi - fi of classi event
channel avalanche - lled lled with mineral oil mixed with disappearance
analysis analysis used a very novel convolutional ll ll extends the full width or height
eld eld of computer vision. Recorded and and
future cross future he statistical power of power statistical he 2 t wide wide range of detector tech channels.
POT, two papers were recently published for the for published recently were papers two POT, joint analysis p
the
e 20 appearance v
e - lters lters are used to extract automatically 10 v
current (CC) classi appearance 4 cm. Each ce µ fi - er output. Training of the CVN was conducted was of the CVN Training output. er ×
v aid in aid
fi × y y and purity in particle identifi ciency. The CVN is a convolutional neural network neural convolutional a is CVN The ciency. si 6.05 6.05 will er takes the hits from these clu cienc C C and readout Each continuously. of the 928 ° , and the views, views, where the signal hits are treated as pixels ]
5 and the joint and the nely nely segmented, 65% active tracking calorimeters. The segmentation y
[ 15 ] - 6 344 344 000 total channels of readout. The 300ton ND [ and ~
he he FD, 4.1 m in the ND) and is fi x ed by the scintillator is collected and transported to the end
p learning p learning techniques from the fi
al results shifting shifting ber on terminates that a ofpixel a 32 - Detectors interactions, interactions, dark matter searches or other exotic physics. Initial
τ v latest results xperiment NOvA detectors are NOvA fi CVN NOvA
E mple with improved purity and effi and purity improved with mple
OvA OvA 3. of dataset analysis second the For analysis disappearance Other possible expansions for CVN in NOvA would include neutral current current shown improvements in the effi and NOvA by employed methods output of the CVN is used to classify the event, event, the classify to used is CVN the of output techniques, PID conventional previous than exposure more to 30% tois the a problem The algorithm approach CVN powerful arti with well extremely work can CNNs that concept of proof novel a of a expectably, or, of a sampling calorimeter, abstract abstract classifying features from the image. A multilayer perceptron at the end of the clas the to create features these uses of FD simulated events. Convolutional raw hit map in both the network, which takes advantage of the recent development in GPU sa designed using dee detectors are formed into clusters by grouping interactions. hits in The time CVN and space classifi to reconstruction, isolate as input and applies a series of trained linear operations to extract result obtained through the deep learning network, called Convolutional Visual Network (CVN). Network Visual Convolutional called network, learning deep the through obtained result With minimal event reconstruction, it is possible to build and train achieved a excellent separation of single signal and algorithm background for which both appearance and d channels. oscillation For the analysis, NOvA a new charged with standard detector layers, totalling 18 000 totalling layers, detector with standard 2.3 We omit the description of a very sophisticated simulation and reconstruction chain in NOvA software here. the However, PID of the neural network technique and the light produc wavelength The APDs are cooled to FD has 384 cells, for with 96 cells plus a muon range stack at the downstream end, made by The and the overall mechanical structure of the detectors are provided by a lattice of extruded cells with cross sectional size 6 cm detector (15.6 m in t the excite interactions neutrino from particles Secondary (pseudocumene). scintillator N 2.2 PoS(FPCP2017)025
- ý τ
v the the and and
The
3 were = 0.5 =
2
(more (more occur, occur, green, green, at two events events 3 NOvA NOvA nts nts are
θ
2 3 2 0.27 over the the over 1 1 shows 2
θ
C.L.
2 θ measured measured eve 2 sin
hypothesis hypothesis
Filip Jedin
publication publication %
led 93% C.L. CC, sin t t to energies
t results. t - sin 30 fi islands e candidates, candidates, of 68
GeV region is in the absence the fi energy energy bins of ±
v
, xed. e 3 2 at v 1 - most information information most and and
θ 2 2 archy (NH). For the the For (NH). archy with with studies of their are fi are CC CC candidate
eV sin - 3
µ 1 3 hier separated - v θ
expectations. level. Fixing Fixing level.
10
σ and ×
1 2 red at greater than at greater red
2 0.398 or 0.618 or 0.398 u
11) cantly change the change cantly . . Combined with the the . latest with Combined = fi
0 where two two where ∆m 3
elihood surface is pro surface elihood 9 inding, inding, 78 range provide the provide range 2 3 ± Cousins approach to account for 2 in Figure 2 for comparison. 2 for in Figure 2017. θ
- 2 θ 67 2 . sin t for oscillations using 19 sin oscillations. oscillations. Restricting the fi
r nd = (+2 and and u
a
2 23 2 23
red by the data at 2.6 at data the by red 2 avo 3 ed ed Feldman eV u
fl 3 fi ∆m er values, and the number of FD beam events beam of FD and the number values, er mainly from bins in the tail of the energy spectrum spectrum the bins tail energy in of the from mainly - -
∆m are also shown also are
] 10
8 (NH) (NH) with a prediction of 77.7 events. Figure
[ × dence levels. The likThe levels. dence
three
2
3.2/7 and does not signi not does and 3.2/7 t prediction, with t the with prediction, ratio to the prediction
nd physical nd boundaries physical eV 11)
. in the lower octant is disfavo octant lower the in 3 0 -
both at the C.L. of 68 % in the normal the in % 68 of C.L. the at both ers, ers, 82.4 events are expected. Figure 1 shows the
nalized, nalized, was performed on the second analysis dataset 3 ± 2 10 θ 5.0 GeV. induced events. In the case of no oscillations, 473 and T2K and = 0.5, is disfavo is 0.5, =
rent confi rent 72 ×
- . ected to be published later in in later published to be ected - 624 3 ] . /d.o.f. to/d.o.f. 2 2 2 2 - ffe θ [ was was fi χ ray 2
hile the so called solar parameters parameters solar called so the hile
- = ( w
= = 2.48 t paramet sin and 0 and 3
23 2 fi n the lower part. The data are fi 2 23 -
θ 2 404 . ∆m ∆m channel was obtained using the CVN technique described in Section 2.3. After 2.3. in Section described technique the CVN using obtained was channel and
octant. octant. These new results are consistent with those in our previous e
23 v g, where g, 3 /d.o.f. = 41.6/17, arises which = 41.6/17, /d.o.f. 2 2 t to our data, in red, for which the mixing is nonmaximal. The 1 2 t of
θ χ ects ofects low event a count ysis, where energy, classifi muon energy, where ysis, 0.8 (syst.) events are predicted background. Figure 3 shows the regions of regions the shows 3 Figure background. predicted are events (syst.) 0.8
∆m ff cances cances are calculated using the uni t to the second analysis data gives ± fi fi erence erence between the energy spectra for the maximal mixing prediction, in dashed disappearance t a t has latest results ff y spectrum along withspectrum y the fibest µ Joint Fit v isappearance data and external constraints from reactor experiments on on experiments reactor from constraints external and data isappearance space allowed at three di three space at allowed
. Contours from MINOS from . Contours d ]
OvA OvA bscured bscured until the analysis CP µ 3 v with hierarchy mass of inverted which 8.2 8.2 which δ parameters The signi statistical e [ 3.2 of the PID The the event selection criteria and analysis procedures, the FD data revealed 33 where the oscillation maximum occurs and the events in that occurs events and the maximum the oscillation where about the mixing angle. Visual scanning of the events in this region along with consistent results gave variables kinematic and location geometric of regions allowed C.L. 90% shows 2 Figure one for each below 2.5 GeV reduces the the reduces GeV below 2.5 mixin Maximal gives a best fi the di and the best fi The best 0 values degenerate hierarchy, inverted best fi that contain little information about the CC CC events, and 2.7 cosmic predicted. At the best energ of oscillations shown i 0.25 between width 0.25 GeV 3.1 blindA anal o than double of data from previous publications). After unbl were observed in the Far Detector with an expected background of 3.4 NC, 0.23 N exp are results analysis second PoS(FPCP2017)025
t ý fi for for -
3 2 θ 2
sin Filip Jedin
nd
a
23 2 ∆m background subtraction. FD data (black dots) and best 90% contours. C.L.
] 8 [ and T2Kand
] 2 [
4 oscillations for data and Monte Carlo after - . The dashed .curves show MINOS
he he systematic uncertainty band is shaded red. Combined beam and cosmic backgrounds t (black dots) and allowed 90% C.L. regions (solid black curves) of of curves) regionsblack(solid C.L. 90%allowed and dots) (black t The ratio to no
Hierarchy Top: Comparison Top: of the reconstructed energy spectrum of the Best fi Best
: : 1 latest results shown shown by the dashed blue histogram. The prediction assuming maximal mixing is shown in dashed Normal
OvA OvA
Figure 2 Figure the
Figure Figure prediction T (red). are Bottom:green. N PoS(FPCP2017)025
ý the the
at at the
National National ; and
ex]. - Filip Jedin
. [hep appearance dataappearance under Contract Contract under
e v
.
er (2016) 016) fi
rch Alliance, LLC Alliance, rch : (left) results from from results (left) : σ disappearance data.
µ v or or the contributions of the staff 01 (2007). ett. 116, ett.no. 15, 151806 116, (2016) - at 1, 2, and 3 and 2, 1, at 2007
- 5 CP
δ
appearance and
e DESIGN phase
- v POT POT per the year, expected production rate is estimated to CP
20
vs.
10 3 2 ×
θ 2 sin ex]. -
beam direction. beam
07CH11359 with the US DOE. the US with 07CH11359
-
results the from combination of
Allowed regions of regions Allowed
AC02 - ) LDM particle candidates per year, leaving the target area boosted forward into few latest results P. Adamson et al. (NOvA Collaboration), Phys. Rev. L Collaboration), Phys. Rev. et Adamson al. (NOvA P. Convolutional Neural Network Neutrino Classi Event A Aurisano et al., A. arXiv:1604.01444[hep D. S. Ayres et al. NOvA, FERMILAB Ayres D. S. 107, Lett. arXiv:1108.0015v1 181802 (2011), et Adamson al., MINOS Phys. Rev. P. (2 D 93, no. 051104 5, Collaboration), Phys. Rev. et Adamson al. (NOvA P. 14
Summary
(10 OvA OvA [4] [5] [1] [2] [3] University University of Argonne Minnesota module assembly facility Ash and River Laboratory, Resea Fermi by operated is Fermilab Fermilab. and Laboratory, De No. References Acknowledgments Foundation; Science National US the Energy; of Department US the by supported was work This the Department of Science India; and the Technology, European Research Council; the MSMT CR, Czech Republic; the RAS, RMES, and RFBR, Russia; CNPq and Brazil FAPEG, State and University of Minnesota. We are grateful f 4. At NuMI beam intensity of ~6 O the around degrees Figure 3: Figure and (right) N PoS(FPCP2017)025
ý appearance Filip Jedin
e nu
.
parameters from
ex]. -
Rev. Lett. 118, 151802 Lett.(2017) 118, Rev.
6
disappearance in NOvA (2017) arXiv:1703.03328disappearance [hep NOvA in dman and R. D. Cousins, Phys. Rev. D 57, 3873(1998). and R. D.dman Cousins, Phys. Rev. numu numu latest results P. Adamson et al. (NOvA Collaboration), on oscillationConstraints et Adamson al. (NOvA P. and D 91, 072010(2015). et Abe al., Phys. Rev. K. G. J. Fel P. Adamson et al. (NOvA Collaboration),Phys. et Adamson al. (NOvA P.
OvA OvA [7] [8] [9] [6] N