Search for a neutrino emission from the Fermi Bubbles with the ANTARES telescope
aa r X i v : . [ a s t r o - ph . H E ] M a r th Fermi Symposium : Monterey, CA : 28 Oct-2 Nov 2012 Search for a neutrino emission from the Fermi Bubbleswith the ANTARES telescope
S. Biagi ∗ on behalf of the ANTARES Collaboration Dipartimento di Fisica dell’Universit `a and INFN Sezione di Bologna,Viale Berti Pichat 6/2, 40127 Bologna, Italy
ANTARES is the largest neutrino telescope in the Northern hemisphere. The main scientific goal is the search forcosmic neutrinos coming from galactic and extragalactic sources. Neutrinos are detected through the Cherenkovlight emitted along the path of charged particles produced in neutrino interactions inside or in the vicinity ofthe detector. ANTARES is sensitive to all flavors though it is optimized for muon neutrinos. The detector hasbeen taking data in its complete configuration since May 2008.Using data collected in the period 2007-2010, the first analysis devoted to the search for neutrinos from theFermi Bubbles is presented. The Fermi Bubbles are characterized by gamma emission with a E − spectrumand a relatively constant intensity all over the space. According to a proposed hadronic mechanism for thisgamma-ray emission, the Fermi Bubbles can be a source of high-energy neutrinos. No evidence of a neutrinosignal is found in the ANTARES data. Therefore upper limits are calculated for neutrino fluxes with differentenergy cutoffs.
1. INTRODUCTION
The Fermi Bubbles (FBs) are extended regionscharacterized by gamma emission with a spectrum ∝ E − [1]. They cover ∼ . · − GeV cm − s − sr − .According to a proposed hadronic mechanism forgamma ray emission [2], the FBs can be a sourceof high-energy neutrinos. From the measured gammaflux it is possible to derive the neutrino flux [3]:Φ ν ≈ Φ γ . E dΦ ν d E ≈ . · − GeV cm − s − sr − (2)According to [2], the neutrino energy spectrum shouldpresent an exponential cutoff Φ ∼ E − e − E/X . Fourdifferent values for the cutoff are assumed in the fol-lowing: no cutoff ( X = ∞ ), 500 TeV, 100 TeV, and 50TeV. In Fig. 2 the number of expected neutrinos fromthe FBs for different energy cutoffs compared with theconventional atmospheric neutrino flux is shown as afunction of the simulated neutrino energy. ∗ [email protected] Figure 1: Left: Gamma-ray flux of the FBs multiplied for E taken from [1]. Right: Green line indicates the shapeof the FBs in galactic coordinates from [1], black regionsare the approximation used in this analysis.
2. THE ANTARES NEUTRINOTELESCOPE
The ANTARES neutrino telescope is located in theMediterranean Sea close to the southern French coastof the city of Toulon [4]. 885 photomultipliers tubes(PMTs) mounted on 12 strings are installed at a depthof 2500 meters (see Fig. 3) and detect the Cherenkovlight emitted by ultra relativistic neutrino-inducedmuons along their path. The time and the chargecollected by the PMTs (the hits ) are digitized andsent on-shore for triggering and storing on disk. Thecollected hits are used to reconstruct the direction ofthe primary neutrino and to estimate its energy. Thetrack reconstruction algorithm is based on a likelihoodfit that uses a detailed parametrization of the prob-ability density function for the photon arrival timesand gives as outputs the position and direction of the eConf C121028 th Fermi Symposium : Monterey, CA : 28 Oct-2 Nov 2012
Figure 2: Expected number of neutrino events from theFBs region corresponding to 4 years of data as a functionof the neutrino energy. The gray area represents theconventional atmospheric neutrinos and points are theneutrinos from the FBs: without energy cutoff (fullcircles), 500 TeV cutoff (empty circles), 100 TeV cutoff(crosses) and 50 TeV cutoff (black lines). ©F. Montanet ~
70 m m
350 m a storey
Figure 3: The ANTARES neutrino telescope is athree-dimensional array of 885 photomultipliersdistributed over 12 lines anchored on the seabed atdistances of about 70 m from each other and tensionedby a buoy at the top of each line. muon track, the information on the number of hits( N hit ) used for the reconstruction, and a quality pa-rameter Λ. The neutrino-induced tracks are selectedas “upgoing” to reject the dominant background ofatmospheric muons; cutting on the Λ quality param-eter is possible to reduce the contamination of mis-reconstructed as upgoing atmospheric muons to thelevel of few percent. -90-80-70-60-50-40-30-20-1001020 30 40 50 60 70 80 90 -180-160-140-120-100-80-60-40-20020406080100120140160180 -90-80-70-60-50-40-30-20-1001020 30 40 50 60 70 80 90 -180-160-140-120-100-80-60-40-20020406080100120140160180 -90-80-70-60-50-40-30-20-1001020 30 40 50 60 70 80 90 -180-160-140-120-100-80-60-40-20020406080100120140160180 -90-80-70-60-50-40-30-20-1001020 30 40 50 60 70 80 90 -180-160-140-120-100-80-60-40-20020406080100120140160180 Figure 4: Position in galactic coordinates of the FBsregion (the ON zone) and of the three chosen regionswith the same visibility (the OFF zones) superimposedon the total ANTARES visibility.
3. ON/OFF ZONES APPROACH
As it can be seen in Fig. 2, the expected eventflux from the FBs is several order of magnitude lowerthan the atmospheric neutrino flux, that represents anirreducible background. A discrimination will be doneon the basis of the energy and a correct estimation ofthe atmospheric background is fundamental in thisanalysis.MonteCarlo (MC) ν µ events generated to reproducethe atmospheric neutrinos are weighted using the so-called conventional “Bartol” flux [5]. Atmosphericneutrinos are produced by the interaction of high-energy cosmic rays in the atmosphere. The uncer-tainties on the flux of atmospheric neutrinos are atthe level of 25 ÷
30% due to the lack of measurementsof the cosmic ray fluxes at high energies and to theuncertainties on the cross sections of cosmic rays withthe light atoms in the upper atmosphere.In addition systematic uncertainties about the de-tector simulation (absorption length of light in sea wa-ter, PMT efficiency) can make the data/MC compar-ison inefficient for revealing the possible signal. Thesolution adopted is to estimate the background in theFBs zone directly from data, defining different zonesin the sky with the same coverage and visibility. Themain idea of this method is to compare the measure-ment done in the FBs area (ON zone) with a back-ground estimated from data itself looking into an areaoutside the FBs (OFF zone).In local coordinates, the FBs move in the sky. Azone with the same position but shifted in time followsthe FBs zone. A proper choice of the time shift avoidsan overlapping of the zones, allowing an unambiguousdefinition of the OFF zones. The expected number ofbackground events is proportional to the efficiency ofthe detector, that is a function of the local coordinatesonly. In this analysis, three OFF zones (Fig. 4) areidentified corresponding to 6, 12 and 18 hours timeshifts from the FBs (ON zone); moreover this choicereduces the background uncertainty. eConf C121028 th Fermi Symposium : Monterey, CA : 28 Oct-2 Nov 2012 Figure 5: Left: Reconstruction quality parameter Λ distributions for data and MC. Events with the best N hit areselected. Several models for the neutrino spectrum are displayed. Right: Data and MC events as a function of the totalnumber of hits. The cut that defines the high energy region is obtained with the MRF procedure – see text.
4. SENSITIVITY OPTIMIZATION ANDBLINDING PROCEDURE
Background in this analysis is estimated directlyfrom data. To verify that the OFF zones have thesame visibility, the data event rate is compared withMC expectation; a conservative systematic error of3% is evaluated to account for different visibilities ofthe zones. The optimization of the analysis cuts isdone using MC simulations in order to maximize thesensitivity to a FBs neutrino signal and to reject at-mospheric muons and atmospheric neutrinos (blindedanalysis).Cuts are optimized according to the prescriptionsof the Model Rejection Factor (MRF) procedure [6].This method uses the Feldman and Cousins statisticsto calculate upper limits at 90% c.l. [7], for an en-semble of hypothetical experiments with no signal anda background with a Poissonian probability of occur-rence. The MRF represents the average upper limitthat can be obtained in case of no discovery, i.e. thesensitivity of ANTARES to the assumed signal flux.The minimization of the MRF produces the best cuton the chosen parameters, in our case the reconstruc-tion quality parameter Λ and the number of hits usedin reconstruction, N hit .The rejection of downward-going atmosphericmuons is achieved cutting on the Λ parameter; theenergy estimator, N hit , can provide a discriminationbetween atmospheric and signal neutrinos (Fig. 5).Neutrino tracks reconstructed with only one line andwith an angular error greater than 1 ◦ are rejected.For simplicity, a unique event selection is chosenfor the four tested spectrum models (Λ > − .
24 and N hit > Table I MRF and sensitivity to a neutrino flux ∝ E − e − E/X (in units of 10 − GeV cm − s − sr − ) fordifferent energy cutoffs. Λ and N hit cuts are optimized toget the best sensitivity. Cutoff [TeV] MRF Sensitivity no cutoff 2.75 3.30500 3.79 4.55100 5.74 6.8950 7.58 9.09
5. RESULTS AND UPPER LIMIT
Data are unblinded in the ON region searching foran excess of events in comparison to the measuredbackground in the three OFF regions. 75 events areobserved in the ON region with 90 ± ± +15% − for data/MC comparison is taken intoaccount in the limit calculation. Due to a negativefluctuation of background in the ON zone, the quotedupper limits are lower than the ANTARES sensitivityto neutrino fluxes. A comparison of these limits withthe assumed theoretical models is presented in Fig. 6;the limits for the most optimistic cutoffs are very closeto the expected fluxes.Very soon, data collected in 2011 will be addedto increase the statistical significance of the analy-sis. Furthermore, ANTARES is currently developinga method to combine various observables through alikelihood approach in order to increase the energy res-olution. An Artificial Neural Network (ANN) is usedfor the mapping of the likelihood between the chosen eConf C121028 th Fermi Symposium : Monterey, CA : 28 Oct-2 Nov 2012
Figure 6: ANTARES upper limits at a 90% c.l. for a E − neutrino flux from the Fermi Bubbles with different energycutoffs. Theoretical predictions normalized at the value of eq. 2 are plotted with dotted lines. observables and the energy: about 50 parameters areused as input of the ANN (hits from various triggers,parameters from tracking algorithms, etc) [8]. Afterthe selection of the input parameters, the ANN mustbe trained with a sample of neutrino events generatedwith MC simulations; the final achieved energy resolu-tion is about 0.3 of the logarithm of the reconstructedenergy in the region between 1 TeV and 300 TeV. TheANN will be used in the final version of this analysis,improving the sensitivity of ANTARES. References (2010).(2010).