AGILE detection of a candidate gamma-ray precursor to the ICECUBE-160731 neutrino event
F. Lucarelli, C. Pittori, F. Verrecchia, I. Donnarumma, M. Tavani, A. Bulgarelli, A. Giuliani, L. A. Antonelli, P. Caraveo, P. W. Cattaneo, S. Colafrancesco, F. Longo, S. Mereghetti, A. Morselli, L. Pacciani, G. Piano, A. Pellizzoni, M. Pilia, A. Rappoldi, A. Trois, S. Vercellone
DD raft version J uly
28, 2017Typeset using L A TEX preprint style in AASTeX61
AGILE
DETECTION OF A CANDIDATE GAMMA-RAY PRECURSOR TO THE ICECUBE-160731NEUTRINO EVENT
F. L ucarelli ,
1, 2
C. P ittori ,
1, 2
F. V errecchia ,
1, 2
I. D onnarumma , M. T avani ,
4, 5, 6
A. B ulgarelli , A. G iuliani , L. A. A ntonelli ,
1, 2
P. C araveo , P. W. C attaneo , S. C olafrancesco ,
10, 2
F. L ongo , S. M ereghetti , A. M orselli , L. P acciani , G. P iano , A. P ellizzoni , M. P ilia , A. R appoldi , A. T rois , and S. V ercellone ASI Science Data Center (ASDC), Via del Politecnico snc, I-00133 Roma, Italy INAF–OAR, via Frascati 33, I-00078 Monte Porzio Catone (Roma), Italy Agenzia Spaziale Italiana (ASI), Via del Politecnico snc, I-00133 Roma, Italy INAF / IAPS–Roma, Via del Fosso del Cavaliere 100, I-00133 Roma, Italy Univ. “Tor Vergata”, Via della Ricerca Scientifica 1, I-00133 Roma, Italy Gran Sasso Science Institute, viale Francesco Crispi 7, I-67100 L‘Aquila, Italy INAF / IASF–Bologna, Via Gobetti 101, I-40129 Bologna, Italy INAF / IASF–Milano, via E.Bassini 15, I-20133 Milano, Italy INFN–Pavia, Via Bassi 6, I-27100 Pavia, Italy University of Witwatersrand, Johannesburg, South Africa Dip. di Fisica, Universita di Trieste and INFN, Via Valerio 2, I-34127 Trieste, Italy INFN–Roma Tor Vergata, via della Ricerca Scientifica 1, 00133 Roma, Italy INAF – Osservatorio Astronomico di Cagliari, via della Scienza 5, I-09047 Selargius (CA), Italy INAF Oss. Astron. di Brera, Via E. Bianchi 46, I-23807 Merate (LC), Italy
ABSTRACTOn July 31st, 2016, the ICECUBE collaboration reported the detection of a high-energy starting eventinduced by an astrophysical neutrino. We report here about the search for a gamma-ray counterpart ofthe ICECUBE-160731 event made with the
AGILE satellite. No detection was found spanning the timeinterval of ± T using the AGILE “burst search” system. Looking fora possible gamma-ray precursor in the results of the
AGILE -GRID automatic
Quick Look procedure overpredefined 48-hours time-bins, we found an excess above 100 MeV between one and two days before T ,positionally consistent with the ICECUBE error circle, having a post-trial significance of about 4 σ . Arefined data analysis of this excess confirms a-posteriori the automatic detection. The new AGILE transientsource, named AGL J1418 + AGILE gamma-ray transient. Based on an extensive search for cataloged sources within the error regionsof ICECUBE-160731 and AGL J1418 + Corresponding author: Fabrizio [email protected] a r X i v : . [ a s t r o - ph . H E ] J u l L ucarelli et al . Keywords: neutrinos, BL Lacertae objects: general, gamma rays: galaxies, astronomicaldatabases: miscellaneousGILE observation of
ICECUBE-160731 3 INTRODUCTIONNeutrino astronomy by under-water and under-ice Cherenkov detectors has entered a new era since thecompletion of the ICECUBE and ANTARES telescopes (Halzen & Klein 2010; Ageron et al. 2011) andthe subsequent first clear detection of a di ff use background of Very High Energy (VHE) extra-terrestrialneutrinos (IceCube Collaboration 2013; Aartsen et al. 2015). No significant clustering of neutrinos abovebackground expectation has been observed yet (Aartsen et al. 2017), although the ICECUBE apparatusmight reach the sensitivity or accumulate enough statistics to unambiguously detect anisotropy or clusteringof events within a few more years of observations.Emission of TeV-PeV neutrinos might be due to exceptionally energetic transient phenomena like flaringactivities from Active Galactic Nuclei (AGNs), Gamma-Ray Bursts (GRBs) or Supernovae explosions (An-chordoqui et al. 2014). A direct correlation between gamma-rays and neutrinos from astrophysical sourcesis expected whenever hadronic emission mechanisms are at work. In particular, several theoretical worksassume that neutrinos production occurs in astrophysical beam dumps, where cosmic rays accelerated inregions of high magnetic fields near black holes or neutron stars interact via proton-proton ( pp ) or proton-photon ( p γ ) collisions with the matter or the radiation field surrounding the central engine or in a jet ofplasma ejected from it, giving raise also to gamma-rays emission (see (Halzen 2017) for a review).Supernovae remnants (SNRs) expanding in dense molecular clouds and microquasars in our Galaxy aswell as AGNs of the blazars category are the main neutrino source candidates up to PeV energies (Mannheim& Biermann 1989; Mannheim 1995; Halzen & Zas 1997; Protheroe et al. 1998; Bednarek 2005; Vissani2006; Sahakyan et al. 2014). Besides the identification of the pion excess in gamma-ray observations ofSNRs interacting with molecular clouds (Giuliani et al. 2011; Ackermann et al. 2013), detection and identi-fication of a clear neutrino point-like source would represent the evidence of proton and hadron accelerationprocesses, resolving as well the long-lasting problem of the cosmic ray origin (at least up to multi-PeVenergies).Since April 2016, the ICECUBE experiment alerts almost in real time the astronomical community when-ever an extremely high-energy single-track neutrino event (with energy in the sub-PeV to PeV range) isrecorded. The communication is sent through the ICECUBE HESE (a single high-energy starting ICE-CUBE neutrino) and the ICECUBE EHE (extremely high-energy ICECUBE neutrino) GCN / AMON no-tices system (Keivani et al. 2016) a few seconds after the event trigger. The instant notice provides a firstdetermination of the statistical relevance of the event and the reconstructed neutrino arrival direction, pro-jected onto the sky, with its 90% and 50% containment radius (c.r) .On July 31st, 2016, the ICECUBE Collaboration reported a HESE GCN / AMON notice announc-ing the detection of a high-energy neutrino-induced track-like event at time T = = and a signalness of ∼ AGILE satellite. The paper is organized as follows: in Section 2, wedescribe the main
AGILE instrumental characteristics and its unique capabilities for the search of gamma- For ICECUBE EHE notices, only source errors at 50% c.r. are given. http: // gcn.gsfc.nasa.gov / notices amon / As quoted in the ICECUBE EHE event information web page https: // gcn.gsfc.nasa.gov / amon ehe events.html Probability that the neutrino event is of astrophysical origin. L ucarelli et al .ray counterparts to such triggered events of very short duration. In Section 3, we present the results of the AGILE observations, both near the prompt neutrino event time T and in archival data. In Sections 4, wereport about the multi-wavelength (MWL) follow-up and in Section 5 we search for a possible e.m. coun-terpart candidate using the cross-catalog search tools available from the ASI Science Data Center (ASDC) . AGILE
AS DETECTOR OF TRANSIENT GAMMA-RAY SOURCESThe gamma-ray satellite
AGILE (Tavani et al. 2009), launched on 2007, has just completed its tenth yearof operations in orbit. The main on-board instrument is the gamma-ray imaging detector (GRID) sensitiveto gamma-rays in the energy range 30 MeV–50 GeV, composed by the gamma-ray Silicon Tracker, theMini-Calorimeter (MCAL) and the anti-coincidence (AC) system for the particle background rejection.The co-axial X-ray (20-60 KeV) detector Super-
AGILE completes the satellite scientific payload.Since Nov. 2009,
AGILE is operated in the so called spinning observation mode, in which the satelliterotates around the Sun-satellite versor. In this operation mode, the
AGILE gamma-ray imager approximatelyobserves the whole sky every day, with a sensitivity (at 5 σ detection level) to gamma-ray fluxes above100 MeV of the order of (3 ÷ × − ph cm − s − .As already demonstrated in the recent follow-up of the gravitational-wave event GW150914 (Tavani et al.2016) and in dozens of Astronomer’s Telegrams (ATel) and GCN circulars, AGILE is a very suitable in-strument to perform searches for short transient gamma-ray sources and gamma-ray counterparts to multi-messenger transient events like the neutrino event observed on July 31st, 2016.The main characteristics that make
AGILE in spinning mode an important instrument for follow-up ob-servations of multi-messenger counterparts are: • a very large field of view (FoV) of 2.5 sr for the AGILE -GRID; • best sensitivity to gamma-ray fluxes above 30 MeV of the order of (2 ÷ × − ph cm − s − fortypical single-pass integrations of 100 s; • a coverage of 80% of the whole sky every 7 minutes; • a gamma-ray exposure of ∼ • between 150-200 passes every day for any region in the accessible sky. • sub-milliseconds trigger for very fast events.Despite the small size (approximately a cube of side ∼
60 cm), the
AGILE -GRID achieves an e ff ectivearea of the order of 500 cm between 200 MeV and 10 GeV for on-axis gamma-rays, and an angularresolution (FWHM) of the order of 4 ◦ at 100 MeV, decreasing below 1 ◦ above 1 GeV (Cattaneo et al. 2011;Chen et al. 2013; Sabatini et al. 2015).A very fast ground segment alert system allows the AGILE Team to perform the full AGILE -GRID datareduction and the preliminary
Quick Look (QL) scientific analysis only 25 /
30 minutes after the telemetrydownload from the spacecraft (Pittori 2013; Bulgarelli et al. 2014).The
AGILE QL on-ground system implements two di ff erent kinds of automatic analysis: • A “burst search” system, involving both GRID and MCAL instruments, is used to look for transientsand GRB-like phenomena on timescales ranging from a few seconds to tens of seconds . The burst http: // A special sub-millisecond search for transient events detected by MCAL is operational on board (Tavani et al. 2009).
GILE observation of
ICECUBE-160731 5
Figure 1.
Hammer-Aito ff projection, in Galactic coordinates, of the AGILE gamma-ray exposure in [cm s sr] (binsize of 0.5 ◦ ) after one complete rotation in spinning mode, time-centered at the ICECUBE-160731 event time T . Theneutrino event error circle is shown in black. The magenta and yellow contours show, respectively, the Sun / anti-Sunexclusion regions and the average Earth occultation during the considered integration time: ( T -210; T + search system runs on predefined time windows of 100 seconds, and it may be also triggered byexternal GCN notices (Zoli et al. 2016). • A “standard”
AGILE -GRID QL analysis, based on a Maximum Likelihood (ML) algorithm (Mat-tox et al. 1996; Bulgarelli et al. 2012), is used to detect gamma-ray transients above 100 MeV ontimescales of 1-2 days (Bulgarelli et al. 2014). This automatic procedure routinely runs over prede-fined 48-hours time-bins.Given the AGILE e ff ective area and sensitivity, these collecting time intervals are the most appropriate toaccumulate enough statistics and to maximize the signal-to-noise ratio in both cases. AGILE
INVESTIGATIONS OF ICECUBE-160731The ICECUBE-160731 best-fit reconstructed neutrino arrival direction in equatorial coordinates is (fromRev. = (214.5440, -0.3347) +/ - 0.75 [deg](90% statistical plus systematic c.r.), corresponding to Galactic coordinates: l,b = (343.68, 55.52) [deg]. Inthe next sections, details of the automatic and refined AGILE data analysis of the ICECUBE-160731 eventare reported. 3.1.
Prompt event
The search for a GRB-like prompt event on short time-scales ranging from a few to tens of secondsconnected to the ICECUBE neutrino emission was performed with the
AGILE burst search system. Thesystem was triggered by the first ICECUBE GCN / AMON notice reported a few tens of seconds after T . L ucarelli et al .The automatic procedure searches for prompt gamma-ray emission on predefined 100 s time-interval binsranging from T -1000 to T + T , the reconstructed neutrino-source position was in good visibility for the AGILE -GRID FoV,neither occulted by the Earth nor by the exclusion regions around the Sun and anti-Sun positions (seeFig. 1). No significant detection was found in the GRID data from the event position in any of the 100 stime-bins scanned. The 3 σ Upper Limit (UL) for the emission in the range 30 MeV–50 GeV estimated inthe 100 s time-bin with the highest exposure on the event position is: 5 . × − ph cm − s − .Moreover, using the data of the AGILE -MCAL and the AC scientific ratemeters, we have searched forburst-like events in the energy range of 0.4 – 100 MeV and 70 keV – tens of MeV, respectively. Nosignificant event has been detected in neither of the two detectors.3.2.
Search for gamma-ray precursor and delayed emission
Since the astrophysics and the time scales of the phenomena related to the emission of these extremelyhigh-energy neutrinos are still uncertain, besides the investigations near T we also explored the AGILE -GRID data taken few days before and after T , searching for a possible gamma-ray precursor or delayedemission on longer (daily) time-scales possibly connected to the neutrino event.Interestingly, a gamma-ray excess above 100 MeV with a pre-trial ML significance of 4.1 σ compatiblewith the ICECUBE error circle appeared in the results of the AGILE -GRID automatic QL procedure betweenone and two days before T . This detection was reported in the ATel AGILE
QL procedure runs on predefined 2-days integration time since Nov. 2009, thestarting of the spinning observation mode. The AGILE source ML detection method derives, for eachcandidate source, the best parameter estimates of source significance, gamma-ray flux, and source location.The ML statistical technique, used since the analysis of EGRET gamma-ray data (Mattox et al. 1996)and adapted to the
AGILE data analysis (Bulgarelli et al. 2012; Chen et al. 2011), compares measuredcounts in each pixel with the predicted counts derived from the di ff use gamma-ray model to find statisticallysignificant excesses consistent with the instrument point spread function.An AGILE QL detection is in general defined by the condition √ T S ≥
4, where TS is the Test Statisticof the ML method defined as − log ( L / L ), where L / L is the ratio between the maximum likelihood ofthe null hypothesis over the point-like source hypothesis, given the di ff use AGILE gamma-ray backgroundmodel (Giuliani et al. 2004). This threshold has been calibrated over various timescales and di ff erent back-ground conditions (e.g., on or outside the Galactic plane) (Bulgarelli et al. 2012).To evaluate the post-trial significance of the automatic QL detection mentioned above, we used the prob-ability distribution of the ML Test Statistic (TS) computed in Bulgarelli et al. (2012). The probability ofhaving at least one detection due to a background fluctuation for any position within the predefined Re-gion of Interest (ROI) of 10 ◦ radius used in the ML fitting procedure with a significance √ T S ≥ h , in N independent trials, is given by P ( N ) = − (1 − p ) N , where p is the p -value (that is, the probability offinding a false positive detection in a single observation) corresponding to h . The p -value for a detectionwith √ T S ≥ . is 3 . × − . By considering all the generated maps havingenough exposure in spatial coincidence with the neutrino error circle (amounting to 226 since the begin- As expected by the Wilks’ theorem (Wilks 1938), the TS values follow in this case the χ distribution with one degree offreedom. GILE observation of
ICECUBE-160731 7
Figure 2.
A-posteriori refined analysis:
AGILE -GRID 1-day time-bin lightcurve starting at T -4 days(MJD = AGILE
ML analysis performed at the ICECUBE-160731 position over eachintegration bin. ning of the spinning observation mode), the probability of having one detection by chance in N =
226 trialsis P (226) = . × − . The chance probability of the AGILE detection becomes at least two orders ofmagnitudes lower if we consider the probability P of spatial coincidence of the AGILE -GRID excess withthe ICECUBE error region within the 10 ◦ radius ROI. The combined post-trial probability becomes then P × P ∼ . × − , which corresponds to a 3.9 σ post-trial significance.A refined analysis has been performed both to confirm the automatic QL result (applying more stringentcuts to further reduce the background contamination from albedo events) and to find a better temporalcharacterization of the gamma-ray transient positionally consistent with the ICECUBE-160731 position.In the refined GRID data analysis, we created a lightcurve symmetric with respect to T0, using a time-binof 24 hours, which is the minimum integration time needed by the GRID to detect a medium / high flaringgamma-ray source above 100 MeV with enough statistics .A search for gamma-ray emission above 100 MeV using the AGILE
ML around the ICECUBE positionhas thus been performed over the time interval ( T -4; T +
4) days. Exposure, counts and di ff use emissionmaps of each time-bin were generated using the o ffi cial AGILE scientific analysis software (release: BUILD21; response matrices: I0023) (Chen et al. 2011), applying a cut of 90 ◦ on the albedo events rejectionparameter and taking an AGILE -GRID FoV radius of 50 ◦ . In comparison, the predefined QL maps aregenerated with a looser albedo cut of 80 ◦ and a larger acceptance FoV radius of 60 ◦ . GRID data acquisitionduring the passage over the South Atlantic Anomaly (SAA) is suspended. Each time-bin of the lightcurvehas been analyzed by means of the ML algorithm assuming a gamma-ray source at the ICECUBE position. Only in some exceptional bright flares the integration time-bin may be reduced below 24 hours (see, e.g., Striani et al. (2011);Vercellone et al. (2011)). http://agile.asdc.asi.it/public/AGILE SW 5.0 SourceCode/ L ucarelli et al . Figure 3.
AGILE -GRID intensity map in [ph cm − s − sr − ] and Galactic coordinates, centered at the ICECUBE-160731 position, from T − . T − . AGILE -GRID ML detection, AGL J1418 + AGILE error circles.
Figure 2 shows the resulting gamma-ray light-curve, where for each bin, the ML gamma-ray flux estimateabove 100 MeV or the 95% C.L. UL at the input ICECUBE-160731 position is shown.A gamma-ray excess above 100 MeV with a ML significance of 4.1 σ is detected in the bin centered oneday and a half before the T (from MJD = = QL detection (Lucarelli et al. 2016). The candidate gamma-ray precursor has an estimated flux ofF(E >
100 MeV) = (3 . ± . × − ph cm − s − with centroid Galactic coordinatesl,b = (344.01, 56.03) ± ± AGILE a-posteriori refined analysis on a 24-hours basis shows that the excess is particularly short intime, mostly concentrated between July 29th and 30th, 2016. By examining the arrival times of the gammaevent file, we found a a clusterization of five counts in less than 7 hours around ( T -1) day within 1.5 degreesfrom the ICECUBE centroid. In particular, on the 24-hours integration from MJD 57598.25 to 57599.25(( T − . T − .
8) days), which fully contains the event clusterization, we obtained a ML significance ofthe peak gamma-ray emission of 4.9 σ at the Galactic centroid coordinates: l,b = (344.26, 55.86) ± ± >
100 MeV) = (3 . ± . × − ph cm − s − .The new AGILE transient, named AGL J1418 + observation of ICECUBE-160731 9Figure 3 shows the
AGILE -GRID intensity map centered at the ICECUBE-160731 position, in the 24-hours time interval correspondent to the peak significance. The white region defines the 95% C.L. ellipsecontour of the
AGILE -GRID detection AGL J1418 + AGILE or ICECUBE error circles. A further search in theSecond and Third FERMI-LAT high-energy sources Catalogs (2FHL and 3FHL, Ackermann et al. (2016);The Fermi-LAT Collaboration (2017)) does not show again any possible association with known gamma-ray counterparts. The closest 3FHL source is 3FHL J1418.4-0233 (associated to the BL Lac blazar 5BZBJ1418-0233 (Massaro et al. 2015)), which is more than 2 ◦ away from the neutrino position.3.3. Search for gamma-ray emission in AGILE archival data
The whole public
AGILE -GRID archival data from Dic. 2007 up to Nov. 2016 have been investigated inorder to search for other possible previous and later gamma-ray transient episodes around the ICECUBE-160731 position. This long time-scale search has been performed by using the
AGILE -LV3 online tool (Pit-tori et al. 2014) accessible from the ASDC Multi-Mission Archive (MMIA) web pages . This tool allowsfast online interactive analysis based on the Level-3 (LV3) AGILE -GRID archive of pre-computed counts,exposure and di ff use background emission maps.The search for transient emission above 100 MeV on 2-day integration times did not show any othersignificant detection but the one compatible with the AGILE QL result between one and two days before T (over a total of 271 analyzed maps).We finally performed a ML analysis centered on the ICECUBE position using the LV3 pre-computed mapsfor the whole AGILE observing time (9 years). We obtained an UL of 3 . × − ph cm − s − (E >
100 MeV,for a 95% C.L.). MULTI-WAVELENGTH FOLLOW-UP OF ICECUBE-160731The ICECUBE-160731 detection triggered a thorough campaign of MWL follow-up observations. Theseobservations covered a large part of the entire e.m. spectrum, from the optical band (Global MASTER net,iPTF P48, LCOGT) to the VHE gamma-rays (HAWC, MAGIC, HESS, ...).Very few observatories and space missions were observing the neutrino event position to T . Apart from AGILE and facilities like HAWC, ANTARES and FERMI-LAT, which have access to a large part of the skyalmost the whole day, all the others had to re-point to the ICECUBE position a few minutes or even hoursafter T . In this section, we will summarize the most interesting results of the MWL follow-up, remindingthe reader to the Appendix B for a summary of all other observations published in ATel and GCNs in thehours or the days after the event.In the X-ray band, SWIFT observed the ICECUBE-160731 error circle region starting approximatelyfrom ( T +
1) hrs till ( T +
12) hrs (Evans et al. 2016a,b). The XRT instrument on-board of the SWIFTsatellite detected six sources in the 0.3-10 keV band. Figure 4 shows a zoom of the
AGILE -GRID intensitymap over the integration of the
AGILE peak detection, with the location of the six SWIFT-XRT sources,numbered 1 to 6 (blue crosses in Fig. 4). After the revision of the best-fit neutrino arrival direction and itserror radius, three of the detected XRT sources eventually lay outside the revised ICECUBE-160731 error URL: http: // / mmia / index.php?mission = agilelv3mmia ucarelli et al . Figure 4.
AGILE -GRID intensity map in [ph cm − s − sr − ] zoomed around the ICECUBE-160731 position, in thetime interval ( T − . T − .
8) days. The black and white circles again show, respectively, the 90% c.r. of theICECUBE event and the 95% C.L. contour of the
AGILE -GRID detection AGL J1418 + circle. Only sources AGILE ellipsecontour), while source T + T +
21) hrs (Lipunov et al. 2016a,b). They only detected a point-like event,classified as MASTER OT J142038.73-002500.1, that might have been induced by particle crossing theCCD, and the bright NGC 5584 galaxy (which, anyhow, is already outside the revised error circle) (yellowboxes in Fig. 4). Rapid follow-up observations in the Optical / IR band, started only 3.5 hours after T ,were performed by the Palomar 48-inch telescope (iPTF P48) (Singer et al. 2016). They detected twooptical transient candidates at 1.1 and 2.0 ◦ from the initial neutrino candidate position (magenta diamondsin Fig. 4).In the gamma-ray band, FERMI-GBM could not observe the region at T since the position was occultedby the Earth (Burns & Jenke 2016) while FERMI-LAT reported only flux ULs (95% C.L.) above 100MeV of 10 − ph cm − s − in 2.25 days of exposure starting from a 2016-07-31 00:00 UTC, and of 0 . × − ph cm − s − in 8.25 days of exposure starting from 2016-07-25 at 00:00 UTC (Cheung et al. 2016). Asshown in Appendix A, the non-detection of any gamma-ray precursor by Fermi-LAT might be due to a lowexposure of the ICECUBE region during the AGILE gamma-ray transient.GILE observation of
ICECUBE-160731 11
Table 1. Optical and X-ray sources detected within the revised ICECUBE-160731 error circle during the MWLfollow-up
Mission / Observatory Source ID / name a R.A. (J2000) Dec (J2000) Association Class[deg] [deg]SWIFT-XRT (ATel + b a See Fig. 4. b The astrophysical origin of this transient is not confirmed.
At the time of the neutrino event T , the INTEGRAL satellite, which also has the capability to coveralmost the whole sky (Savchenko et al. 2016), was not observing because it was close to perigee inside theEarth radiation belts.The ICECUBE region was also observed in the VHE band by several experiments (see Appendix B fordetails). Apart from HAWC, that has a 24-hours duty cycle, all the others could re-point to the ICECUBEposition hours later than T , reporting only flux ULs above di ff erent energy thresholds. On a search forsteady source using archival data, the HAWC Collaboration reported about a location with a pre-trial signif-icance of 3.57 σ at R.A.,Dec (J2000) = (216.43, 0.15) [deg] (Taboada 2016) (shown as cyan cross in Fig. 4),although more than 2 ◦ away from the neutrino error circle. Considering the number of trials quoted in theHAWC GCN, this is not a significant detection. POSSIBLE NEUTRINO-EMITTER E.M. SOURCES IN THE ICECUBE-160731 AND
AGILE
AGLJ1418 + / transient sources found duringthe MWL follow-up are good candidates as the ICECUBE-160731 emitter. In particular, we decided toreview only the e.m. sources still within the revised ICECUBE error region plus the closest optical transientdetected by iPTF48 (named iPTF16elf, Singer et al. (2016)) (see Fig. 4). Table 1 shows the main char-acteristics of the five e.m. sources satisfying the chosen selection criteria. The table also shows the mostlikely known association as reported from each of the ATel announcing the detection obtained during thefollow-up.To find some of the key features of one of the most promising neutrino-emitter candidates, the High-energy peaked BL Lac (HBL) types of AGNs (Padovani et al. 2016; Resconi et al. 2017), we reviewed theinitial counterpart association and, moreover, we investigated the broad-band spectral properties of eachobject.The first two SWIFT-XRT sources detected during the follow-up, (2QZCat, Croom et al. (2001)) while source . By looking to theirspectral energy distributions (SEDs), built using both the XRT detections and MWL archival data, neither Also known as [VV2010] J141936.0-010840 (VV2010 Cat., V´eron-Cetty & V´eron (2010)) and SDSS J141935.99-010840.2(SDSS Cat. – Release Also known as [VV2010] J141949.9-000644 (V´eron-Cetty & V´eron 2010), 2MASS J14194982-0006432 (2MASS Cat.,Cutri et al. (2003)), and SDSS J141949.83-000643.7 (Abazajian et al. 2009). ucarelli et al . Figure 5.
R.A.-Dec sky map (J2000) obtained with the ASDC
SkyExplorer tool showing known radio, optical andX-ray sources within 50 arcmin from the ICECUBE-160731 position. The map also covers most of the 95% C.L.error circle of the
AGILE detection described in Sect. 3.2. Black circles show sources from the SDSSWHLGC and theZWCLUSTER catalogs (Wen et al. 2009; Zwicky et al. 1961); blue circle sources from the ROSAT All Sky Survey(RASS) catalogs (Voges et al. 1999, 2000); red circles are radio sources from the FIRST survey at 1.4 GHz (Whiteet al. 1997). The dashed circle indicates the position of the RASS 1RXS J141658.0-001449 source and the nearbyFIRST 1.4 GHz radio-source (blue circle with smallest red circle inside), a possible HBL AGN candidate (see text fordetails). of the two quasars shows hints of high-peaked synchrotron emission, which is one of the key feature usedto identify a HBL type of AGN. Moreover, they completely lack radio emission, which leads us to concludethat they might be radio-quiet quasars and we can discard them as possible emitter of the ICECUBE-160731neutrino.XRT source ∼ .
5” from 2MASS J14182661 + = = SkyExplorer tool . In particular, we focused oursearch to known radio and X-ray sources which might show the typical characteristics of HBL / HSP AGNblazars (Chang et al. 2017): low radio fluxes and low IR-radio spectrum slopes; high X-ray-to-radio fluxratios; ν synchrotron peaks above 10 Hz. https: // tools.asdc.asi.it GILE observation of
ICECUBE-160731 13
Figure 6.
Spectral energy distribution (SED) of the possible HBL candidate, the faint SDSS J141658.90-001442.5galaxy, found within the ICECUBE-160731 error circle. The galaxy appears within the 25” error circle of the RASSsource 1RXS J141658.0-001449 ( ν F ν value shown as black point in the SED), along with a FIRST 2 mJy radiosource (red point). Optical and IR data of the SDSS J141658.90 galaxy are from: Sloan Digital Sky Survey (SDSS)– Release A query of 50 arcmin around the ICECUBE-160731 centroid Galactic coordinates l,b = (343.68, 55.52deg) selecting, among others, radio and X-ray sources from the FIRST (White et al. 1997) and the RASSCatalogs (Voges et al. 1999, 2000), returns several objects (see Figure 5). Following the search criteriadefined above, one of the most interesting object resulting from the query is a RASS source appearingat ∼
19 arcmin from the center, 1RXS J141658.0-001449, with position and related uncertainty R.A.,Dec(J2000) = (14 h m s . , − ◦ (cid:48) ± = = (14 h m s . , − ◦ (cid:48) . ∼ / optical / X-ray emission comes from the same galaxy, we have produced the SEDshown in Figure 6. The high value of the ratio between the 1RXS J141658.0-001449 flux density in the0.1-2.4 keV band and the FIRST radio source ν F ν value at 1.4 GHz (respectively, black and red points inFig. 6) might hint to a non-thermal synchrotron emission peaking above 10 Hz, typical of a HBL AGNblazar. Considering these types of e.m. sources as the most likely neutrino-emitters, the X-ray source1RXS J141658.0-001449 (and the plausible host galaxy SDSS J141658.90-001442.5) appears as one of thecandidate as origin of the ICECUBE-160731 event.This source was not in the field covered by the July 31st, 2016, SWIFT series of ToO observations (Evanset al. 2016a). Interestingly, the source lies also within the 95% error ellipse contour of the
AGILE detectionoccurred before the neutrino event time T (see Fig. 4).4 L ucarelli et al .
400 600 800
SWIFT XRT NONE 2016 Dec 11 Exposure: 4922 sX Pixels Y P i x e l s −
12 345 h m s h m s h m s h m s h m s − o ’ " − o ’ " − o ’ " − o ’ " − o ’ " Figure 7.
Smoothed SWIFT-XRT count map (0.3–10 keV) centered on the ROSAT / RASS-FSC 1RXS J141658.0-001449 source, obtained from the SWIFT ToO executed on Dic. 2016, almost six months later than the ICECUBE-160731 neutrino detection. Total exposure: ∼ . detect algorithm. No significant X-ray excess is found at the 1RXS J141658.0 position. SWIFT ToO data on the 1RXS J141658.0-001449 field
To better estimate the position and the spectrum of the RASS 1RXS J141658.0-001449 source (which wasnot in the field covered by the first SWIFT series of ToO observations (Evans et al. 2016a)) and determine astronger spatial correlation with the radio and optical sources described above, a new SWIFT ToO has beensubmitted and executed in December 2016, almost six months later than the ICECUBE-160731 neutrinodetection.The data were collected in five distinct ∼ .Figure 7 shows the (smoothed) cumulative XRT count map in the 0.3-10 keV energy range, with anoverall exposure of 4.9 ks. The position of the 1RXS J141658.0 source (with its quoted error circle) issuperimposed to the map (white circle near the map center). No apparent X-ray excess is visible at the1RXS J141658.0 position.Using the XIMAGE sosta algorithm, we derive a 3 σ UL of 3 . × − cts s − in the XRT energy bandon the 1RXS J141658.0 position. Assuming a source with a power-law photon index of 1.7, we evaluatedan upper limit of 4 . × − cts s − in the ROSAT PSPC band. This value is well below the count rate of(2 . ± . × − quoted for 1RXS J141658.0-001449 in the RASS-FSC Catalog. This might indicate Correspondent SWIFT OBSERVATION IDs: from 00034815001 to 00034815005.
GILE observation of
ICECUBE-160731 15
Table 2. SWIFT-XRT detections in the 0.3–10 keV band from the ToO centered on the 1RXS J141658.0-001449source
ID Count rate R.A. (J2000) Dec (J2000) prob. SNR[cts s − ] [hh mm ss] [dd mm ss]1 3.16E-03 ± ± ± ± ± an intrinsic variability of the source, which was significant only during the RASS observation. It should benoted that this source does not appear anymore in the second ROSAT all-sky survey (2RXS) Catalog (Bolleret al. 2016), an extended and revised version of the 1RXS Catalog that contains a significant reduced numberof low reliability sources.Applying the XIMAGE detect algorithm on the overall 5 ks XRT count map, weighted by the correspon-dent sum of each single XRT exposure, five (uncataloged) X-ray field sources are detected within the FoV(see Fig. 7). Table 2 reports count rates, source coordinates, SNR ratio and probability to be a backgroundfluctuation for all the five detections. Studies of the characteristics of the five field sources is ongoing. DISCUSSION AND CONCLUSIONSWe reported the results of
AGILE gamma-ray observations of the ICECUBE-160731 neutrino event errorregion. These observations covered the event sky location at the event time T and also allowed us to searchfor e.m. gamma-ray counterparts before and after the event.The analysis of the AGILE -GRID data in the time window T ± AGILE burst search systemhas not shown any significant gamma-ray excess above 30 MeV from the neutrino position. Moreover, noburst-like events using the
AGILE -MCAL and the AC ratemeters around T have been detected. Instead,an automatic detection above 100 MeV, compatible with the ICECUBE position, appeared from the AGILEQL procedure on a predefined 48-hours interval centered around one day and a half before T . Consideringall the number of trials performed by the AGILE QL system and the chance probability to have a gamma-ray excess in coincidence with the neutrino position, the automatic detection reaches a combined post-trialsignificance of about 4 σ . A refined data analysis confirms the QL detection already reported in the ATel AGILE -GRID gamma-ray transient, named AGL J1418 + T -1) days, and reaching a peakML significance of 4.9 σ on the 24-hours integration covering the interval ( T -1.8; T -0.8) days. AGLJ1418 + T . This non-detection of an e.m. counterpart in any of thewavelengths covered by the ICECUBE-160731 follow-up does not exclude the possibility of a bright rapidgamma-ray flare precursor just before the neutrino detection. Most of the instruments involved in the e.m.follow-up, in fact, could re-point their instruments only hours or even a day after T , and might have missedthe flaring episode seen by AGILE at E >
100 MeV.6 L ucarelli et al .As said in the MLW follow-up summary, FERMI-LAT did not report any evidence of a precursor above100 MeV. As we show in Appendix A, this might be due to a very high FERMI-LAT observing angle and avery low exposure of the ICECUBE region with respect to the
AGILE observations.Given the high Galactic latitude of the ICECUBE neutrino arrival direction (b = ff use neutrino background seen by ICECUBE, the extra-galactic cosmic-ray component and the isotropic di ff use gamma-ray background observed by FERMI (Ackermann et al.2015). Kadler et al. (2016) found for the first time a significant probability that one of the ICECUBE PeVevent was spatially and temporally coincident with a major gamma-ray outburst of the Flat Spectrum RadioQuasar (FSRQ) PKS B1424-418. Considering that there is a substantial fraction of the blazar populationnot resolved yet, Kadler et al. estimate that around 30% of the detected multi-TeV / PeV neutrinos will not beassociated with any known gamma-ray blazar, like appears to be the case of the ICECUBE-160731 event.Recently, Resconi et al. (2017) found that a significant correlation between known HBL blazars, ICE-CUBE neutrinos and UHECRs detected by Auger and the Telescope Array (TA) exists. We thus searchedfor a HBL candidate counterpart inside the common ICECUBE and
AGILE
AGL J1418 + σ UL of 3 . × − cts s − in the 0.3–10 keV band. We then cannot confirmat the moment our hypothesis about the HBL nature of this source that, anyhow, might have been detectedduring the ROSAT survey because in an intrinsic X-ray high-state.Other possible PeV neutrino-emitters have been proposed, like Starburst galaxies, giant radio galaxieswith misaligned jets, gamma-ray bursts (GRBs) (see Ahlers & Halzen (2014) for a review). Lipunov et al.(2016c), for example, correlate another recent ICECUBE HESE neutrino event (ICECUBE-160814) withan optical transient occurred almost ten days after the event time. They postulate the possibility that theneutrino emitter might be an ejecting white dwarf in a binary system. This is an intriguing possibility,although the power budget available in these systems (optical companion plus compact object) could not besu ffi cient to accelerate protons up to multi-PeV energies in order to produce sub-PeV / PeV neutrinos from pp collisions.Eventually, none of the other e.m. sources proposed up to now as neutrino-emitter candidates are able toexplain the bulk of multi-wavelength / multi-messenger (neutrinos plus cosmic rays) observational data likethe HBL / HSP class of blazars (Resconi et al. 2017). Indeed, the probability to find a blazar of this class ina 1 ◦ radius sky-area like the ICECUBE-160731 error circle is quite low. Assuming, in fact, an HSP densityof the order of 5 × − deg − from the 2WHSP catalog (Chang et al. 2017), there are approximately 5HSP / HBL AGNs every 100 squared degrees of sky. Thus, the probability to find one of these objects withinthe roughly 3 squared degrees covered by the ICECUBE error circle is of about 0.15%. In the specific caseof the ICECUBE-160731 neutrino, for example, we have not found yet any other potential HBL candidatebut the one not confirmed with the dedicated SWIFT ToO observations. Moreover, the
AGILE transient, notconfirmed by FERMI (although caused by a poor FERMI-LAT visibility just before T ) might indicate aGILE observation of ICECUBE-160731 17possible soft gamma-ray source, in disagreement with the hard-spectrum gamma-ray features expected forthe HBLs.Nevertheless, the HBL scenario can still hold if we assume a lepto-hadronic process occurring withinthe blazar jet (Righi et al. 2017), where the bulk of broad-band e.m. emission is due to synchrotron andInverse Compton leptonic processes, while protons would be mainly responsible for the neutrino flux (fromthe decay of charged pions produced by photo-meson production on the soft photons field within the jet).In this case, Righi et al. (2017) foresee that a soft gamma-ray component, peaking at MeV / GeV energies,would be expected from re-processing of VHE photons from the decay of π ’s originated in the p γ collisionswithin the jet. The AGILE observation of the gamma-ray transient AGL J1418 + T , if associated with the ICECUBE event, could bethen explained by such hadronic mechanism.To conclude, there is also the possibility that the source of the ICECUBE-160731 neutrino event might beeither a di ff erent AGN type or a di ff erent class of source, even though we cannot exclude at the moment amoderately bright HBL not yet identified.We would like to thank Paolo Giommi and Matteo Perri, for many fruitful discussions and the valuablehelp with the analysis of the SWIFT ToO, and Paolo Lipari for the very useful comments about the paper. Wealso thank the Swift Team for making the SWIFT ToO observations possible, in particular M. H. Siegel asthe Swift Observatory Duty Scientist. Lastly, we thank the anonymous referees for their valuable commentsthat helped to improve our paper. AGILE is an ASI space mission with programmatic support from INAFand INFN. We acknowledge partial support through the ASI grant no. I / / /
0. Part of this work is basedon archival data, software or online services provided by the ASI SCIENCE DATA CENTER (ASDC). Itis also based on data and / or software provided by the High Energy Astrophysics Science Archive ResearchCenter (HEASARC), which is a service of the Astrophysics Science Division at NASA / GSFC and the HighEnergy Astrophysics Division of the Smithsonian Astrophysical Observatory. This research has also madeuse of the SIMBAD database and the VizieR catalog access tool, operated at CDS, Strasbourg, France.
Software:
AGILE scientific analysis software (BUILD 21; Chen et al. (2011)), XIMAGE8 L ucarelli et al . o ff - a x i s a n g l e [ ◦ ] AGILEFermi-LAT
Figure 8.
Time-evolution of the ICECUBE-160731 region o ff -axis angles as observed by AGILE and Fermi-LATduring the 48-hrs time interval ( T − T ) days (MJD 57598.07991 ÷ APPENDIX A. COMPARISON BETWEEN
AGILE
AND FERMI-LAT DATA DURING THE ICECUBE-160731EVENTIn this Appendix, we verify that the FERMI-LAT non-detection of the
AGILE possible gamma-ray pre-cursor of the neutrino 160731 event might be due to a poor exposure and non optimal viewing angle of theICECUBE error circle.We have compared the FERMI-LAT attitude data with the
AGILE ones during the time interval ( T − T ) days (MJD 57598.07991 ÷ ff -axis angle lower than 50 ◦ only for a 3.9% of its total exposure time, while for AGILE theexposure time below the same o ff -axis angle amounted to 27.4% of the total (see Figure 8) .Further investigations of the FERMI spacecraft data show also several periods of not data-taking duringthe same time-interval (amounting to ∼
15% of the total observation time), particularly near ( T -1) days (asit is possible to see from Fig. 8), where AGILE found a clusterization of gamma-like events compatible withthe ICECUBE error circle.To prove that during this period the
AGILE and FERMI-LAT exposures on the ICECUBE region were at least comparable, we have evaluated the exposures for both instruments on time intervals of 24, 12and 6 hours centered at ( T −
1) days (MJD = AGILE detection reached its peaksignificance. At high values of the o ff -axis angle ( > ◦ ), the Fermi / LAT sensitivity is up to 50% lower than the nominal on-axis value.
GILE observation of
ICECUBE-160731 19
Table 3.
AGILE and FERMI-LAT exposures on the ICECUBE-160731 error circle during the period of thedetection of the possible gamma-ray precursor AGL J1418 + ff -axisangle of 50 ◦ between source and FoV center has been assumed. Interval duration
AGILE mean exp FERMI-LAT mean exp[MJD] [hrs] (cm s) (cm s)57598.25 ÷ +
06 3.8E + ÷ +
06 1.2E + ÷ +
05 4.7E + We downloaded Pass8 data around the position of ICECUBE-160731 and, using version v10r0p5 ofthe Fermi Science Tools provided by the Fermi satellite team and the instrument response functionP8R2 SOURCE V6, we calculated the mean exposure values on the neutrino error circle on those di ff erentintegration times. We selected Pass8 FRONT and BACK source class events and, in order to be comparablewith the AGILE spectral sensitivity (optimized to the observation of soft gamma-ray sources with typicalspectral indexes of 2 ÷ AGILE exposures on the di ff erent time intervals chosenand for a maximum o ff -axis angle between source and FoV center of 50 ◦ .The LAT exposure on the 24-hours interval MJD 57598.25 ÷ AGILE exposure of 3 . × cm s obtained under the same maximum viewing angle and the same integra-tion time. On the shorter intervals of 12 and 6 hours around ( T -1) days, the AGILE exposure becomes evenlarger than the FERMI one. Assuming, thus, a very short gamma-ray flare, as the
AGILE detection indicates,it might imply the possibility that FERMI, given the very low exposure and the large viewing angle of theICECUBE-160731 position during this period, lost most of the gamma-ray transient episode. Di ff erences inthe event classification algorithms between the two instruments can also bring to a detection / non-detectionin such cases of short gamma-ray transients at the level of 4 σ above the background. B. SUMMARY OF THE ICECUBE-160731 MWL FOLLOW-UP From the FERMI data ASDC mirror (https: // tools.asdc.asi.it). http: // fermi.gsfc.nasa.gov ucarelli et al . T a b l e . S u mm a r yo f t h e M W L f o ll o w - upo f t h e I C E C U B E - e v e n t M i ss i on / O b s e r v a t o r y A T e l G C N C i r c u l a r O b s e r v a ti on / i n t e g r a ti on ti m e C o mm e n t s ( E n e r gyb a nd ) [ U T C ] HA W C ( T e V g a mm a -r a y s )- - - : : - - : : N od e t ec ti on a r oundn e u t r i no e v e n tti m e T ( m o s t s i gn i fi ca n tl o ca ti on ( . σ ) a t R . A ., D ec ( J ) = . , . e g ) . F r o m a r c h i v a l d a t a , a p r e - t r i a l . σ d e t ec ti on fr o m R . A ., D ec ( J ) = . , . e g i s r e po r t e d . S W I F T ( X -r a y , O p ti ca l / UV ) - - : : - - : : S i xkno w no r ca t a l og e d X -r a y s ou r ce s d e - t ec t e d ( . - e V ) bu t no t r a n s i e n t e v e n t s . N o t r a n s i e n t s ou r ce s d e t ec t e d i n t h e s i m u lt a - n e ou s UVO T d a t a . AG I LE ( G a mm a -r a y s ) - - - : - - :
00 2016 - - : - - : > σ p r e - t r i a l s d e t ec ti onon t h e i n t e r v a l - - / - - ( : ) U T . G l ob a l M A S TE R n e t ( O p ti ca l ) F r o m - - : : N oop ti ca lt r a n s i e n t s d e t ec t e d i n s i d e s qu a r e d e g r ee s a r ound ce n t e r o fI C E C U B E - R e v . e rr o r c i r c l e . D e t ec t e don e li k e l yp a r- ti c l e CC D e v e n t ( O T J . - . ) a nd t h e NG C a l a xy . F A C T ( T e V g a mm a -r a y s )- - - : - - : N od e t ec ti on . H E SS ( T e V g a mm a -r a y s ) - - - / - ( r) - - / ( r) N od e t ec ti on . F E R M I- L A T ( G a mm a -r a y s ) - . a y s fr o m - - :
00 8 . a y s fr o m - - : N od e t ec ti on a bov e M e V . F E R M I- G B M ( X -r a y / G a mm a -r a y s )- N e u t r i no e v e n tt r i gg e r ti m e ( T ) . P o s iti ono cc u lt e dby E a r t h a t T . F l ux U . L . a t σ l e v e l ( - e V ) on t h e i n t e r v a l J u l y 30 t h - A ug . s t . i P T FP ( O p ti ca l / I R )- F r o m - - : . N oop ti ca lt r a n s i e n t s d e t ec t e d c l o s e t o t h e I C E C U B E upd a t e d e rr o r c i r c l e . T w oop - ti ca lt r a n s i e n t s ca nd i d a t e ( i P T F e l f a nd i P T F e l g ) d e t ec t e d a t . a nd2 . e g fr o m t h e n e u t r i no ca nd i d a t e po s iti on , bo t h c on s i s - t e n t w it hkno w ng a l a x i e s . M AX I / G S C ( X -r a y ) - A t - - : . F r o m - - t o2016 - - . N od e t ec ti onon t h e - e V b a nd w it h i n t h e I C E C U B E e rr o r c i r c l e , n e it h e r n ea r T no r i n t h e p e r i od J u l y20 t h– A ug . r d . σ U . L . a r e p r ov i d e d . T ab l e c on ti nu e donn ex t pag e GILE observation of
ICECUBE-160731 21 T a b l e ( c on ti nu e d ) M i ss i on / O b s e r v a t o r y A T e l G C N C i r c u l a r O b s e r v a ti on / i n t e g r a ti on ti m e C o mm e n t s ( E n e r gyb a nd ) [ U T C ] M AG I C ( T e V g a mm a -r a y s ) - - - : - - : N od e t ec ti on a bov e G e V . AN T A R E S ( T e V / P e V n e u t r i no s ) T ± r T ± a y N oup - go i ng m uonn e u t r i no ca nd i d a t ee v e n t s r ec o r d e d w it h i n t h r ee d e g r ee s o f t h e I C E - C U B E e v e n t c oo r d i n a t e s . % U . L . on t h e fl u e n ce fr o m a po i n t - li k e s ou r cea r e r e po r t e d . K onu s - W i nd ( X -r a y / G a mm a -r a y s )- T ± s F r o m a y s b e f o r e t o1d a y a f t e r T . N o t r i gg e r e d e v e n t s d e t ec t e d . % C . L . upp e r li m it s a r e r e po r t e don t h e - e V fl u - e n ce f o r t yp i ca l s ho r t a nd l ong G RB s p ec t r a . L C OG T ( O p ti ca l ) - F r o m - - : : till - - : : . N od e t ec ti ono f n e w op ti ca l s ou r ce s do w n t o 3 σ li m iti ng m a gn it ud e s > . ucarelli et al .REFERENCES Aartsen, M. G., Abraham, K., Ackermann, M., et al.2015, Physical Review Letters, 115, 081102—. 2017, ApJ, 835, 151Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ueros,M. A., et al. 2009, Ap. J. Supp., 182, 543Acero, F., Ackermann, M., Ajello, M., et al. 2015, Ap.J. Supp., 218, 23Ackermann, M., Ajello, M., Albert, A., et al. 2015,ApJ, 799, 86Ackermann, M., Ajello, M., Allafort, A., et al. 2013,Science, 339, 807Ackermann, M., Ajello, M., Atwood, W. B., Baldini,L., et al. 2016, Ap. J. Supp., 222, 5Ageron, M., Aguilar, J. A., Al Samarai, I., et al. 2011,Nucl. Instrum. Methods A, 656, 11Ahlers, M., & Halzen, F. 2014, Physical Review D, 90,043005Anchordoqui, L. A., Barger, V., Cholis, I., et al. 2014,Journal of High Energy Astrophysics, 1, 1Bednarek, W. 2005, ApJ, 631, 466Boller, T., Freyberg, M. J., Tr¨umper, J., et al. 2016,A&A, 588, A103Bulgarelli, A., Chen, A. W., Tavani, M., et al. 2012,A&A, 540, A79Bulgarelli, A., Trifoglio, M., Gianotti, F., et al. 2014,ApJ, 781, 19Burns, E., & Jenke, P. 2016, GRB CoordinatesNetwork, 19758Cattaneo, P. W., Argan, A., Bo ff elli, F., et al. 2011,Nucl. Instrum. Methods A, 630, 251Chang, Y.-L., Arsioli, B., Giommi, P., & Padovani, P.2017, A&A, 598, A17Chen, A. W., Argan, A., Bulgarelli, A., et al. 2013,A&A, 558, A37Chen, A. W., Bulgarelli, A., Contessi, T., et al. 2011,GRID Scientific Analysis – USER MANUAL, http://agile.asdc.asi.it Cheung, C. C., Toomey, M. W., Kocevski, D., &Buson, S. 2016, The Astronomer’s Telegram, 9303Croom, S. M., Smith, R. J., Boyle, B. J., et al. 2001,Mon. Not. R. Astron. Soc., 322, L29Cutri, R. M., & et al. 2014, VizieR Online DataCatalog, 2328Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003,2MASS All Sky Catalog of point sources. (Cutri)Drake, A. J., Djorgovski, S. G., Mahabal, A., et al.2009, ApJ, 696, 870 Edge, A., Sutherland, W., Kuijken, K., et al. 2013, TheMessenger, 154, 32Evans, P. A., Kennea, J. A., Keivani, A., et al. 2016a,GRB Coordinates Network, 19747—. 2016b, The Astronomer’s Telegram, 9294Giuliani, A., Chen, A., Mereghetti, S., et al. 2004,Mem. SAIt Suppl., 5, 135Giuliani, A., Cardillo, M., Tavani, M., et al. 2011,ApJL, 742, L30Halzen, F. 2017, Nature Physics, 13, 232Halzen, F., & Klein, S. R. 2010, Review of ScientificInstruments, 81, 081101Halzen, F., & Zas, E. 1997, ApJ, 488, 669IceCube Collaboration. 2013, Science, 342, 1242856Kadler, M., Krauß, F., Mannheim, K., et al. 2016,Nature Physics, 12, 807Keivani, A., Tesic, G., & Cowen, D. 2016, ”IceCubeRealtime HESE”, AMON - GCN Public DocumentLi, T.-P., & Ma, Y.-Q. 1983, ApJ, 272, 317Lipunov, V., Gorbovskoy, E. S., Tyurina, N., et al.2016a, The Astronomer’s Telegram, 9298—. 2016b, GRB Coordinates Network, 19748Lipunov, V. M., Tyurina, N. V., Gorbovskoy, E. S., &Buckley, D. 2016c, GRB Coordinates Network,19888Lucarelli, F., Pittori, C., Verrecchia, F., et al. 2016, TheAstronomer’s Telegram, 9295Mannheim, K. 1995, Astroparticle Physics, 3, 295Mannheim, K., & Biermann, P. L. 1989, AstroparticlePhysics, 221, 211Massaro, E., Maselli, A., Leto, C., et al. 2015,Astrophysics and Space Science, 357, 75Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996,ApJ, 461, 396Padovani, P., Resconi, E., Giommi, P., Arsioli, B., &Chang, Y. L. 2016, Mon. Not. R. Astron. Soc., 457,3582Pittori, C. 2013, Nuclear Physics B ProceedingsSupplements, 239, 104Pittori, C., Lucarelli, F., & Verrecchia, F. 2014,”Tutorial for the AGILE-LV3 online analysis”,
Protheroe, R. J., Bednarek, W., & Luo, Q. 1998,Astroparticle Physics, 9, 1Resconi, E., Coenders, S., Padovani, P., Giommi, P., &Caccianiga, L. 2017, Mon. Not. R. Astron. Soc.,468, 597
GILE observation of