1 Searches for High Energy Neutrino Emission in the Galaxy with

2 the Combined IceCube-AMANDA Detector

3 Received ; accepted – 2 –

4 ABSTRACT

5 We present searches for neutrino sources from the Galaxy with energies above ∼ 500 GeV, using the data collected by the South Pole neutrino telescopes Ice- Cube (in incomplete configurations) and AMANDA. Data collected during the season 2007 to 2008 with 22 IceCube strings and during 2008 to 2009 with the 40 string configuration have been used. In the two seasons, a total life time of 276 days and 375 days has been respectively obtained with IceCube. In order to improve the sensitivity of IceCube at low energy, we also used during the first season 143 days and during the second season 306 days of data of the precursor AMANDA which was still in operation. This analysis exploits a large energy range and focuses on neutrino searches from galactic sources at both small and large scales. For instance, the range 180.0 GeV - 20.5 TeV includes 90% of the signal from a E−3 spectrum and the range 2.4 - 750.0 TeV includes 90% of the signal from a E−2 spectrum. The accessible part of the galactic plane in the northern sky, which includes the Local Arm towards the Cygnus region and the Perseus Arm, has been scanned in search of excesses of neutrinos with respect to the background of atmospheric neutrinos. The search was conducted for a candidate list of interesting galactic sources for which we provide flux upper lim- its since no excess was observed. For the Cygnus star forming region we have developed a dedicated search based on the measurement of the spatial correla- tions between events. Finally, we describe a search for neutrino emission from Cygnus X-3 in coincidence with the ejection of relativistic particles. No evidence for a signal is found in any of the searches performed. Upper limits are set for neutrino fluxes from astrophysical sources over the northern sky. The obtained 90% confidence level muon neutrino flux upper limit lies in the range between E3dN/dE ∼ 5.4 − 19.5 × 10−11 TeV2 cm−2 s−1. This upper limit is the most – 3 –

stringent one calculated for the neutrino sample collected in IceCube in the 40 strings configuration.

6 Subject headings: acceleration of particles, cosmic rays, neutrinos – 4 –

7 1. Introduction

8 The IceCube neutrino telescope has been successfully completed at the South Pole

9 during December 2010. IceCube is the most sensitive telescope to date to search for high

10 energy neutrino sources. High energy neutrinos from astrophysical sources are intimately

11 related to the efficient acceleration of hadrons and their interaction in the environment of

12 the accelerator. The pp and pγ processes responsible for the production of neutrinos in

13 astrophysical sites will also lead to the generation of gamma-rays of similar energy (Kelner

14 et al. 2006; Kelner & Aharonian 2008). However, it is difficult to infer the contribution

15 of a possible hadronic component from the observed gamma-rays, since gamma-rays can

16 also be produced by relativistic electrons via Inverse Compton scattering, or even subject

17 to absorption if the radiation field at production site is sufficiently intense. Moreover,

18 gamma-rays can be absorbed on their way to us. Measurements of the neutrino flux from

19 astrophysical sources, even if challenging, reveals unique insights into the acceleration

20 mechanisms and into the origin of cosmic-rays.

21 The IceCube neutrino telescope has the potential to discover high energy neutrino sources

−11 −12 −2 −1 22 at the flux level of F (E > 1 T eV ) ∝ 10 − 10 erg cm s , for a source spectrum

−2 23 following E in the relevant energy range from 1 TeV up to a few PeV (Abbasi et al. 2011).

33 −1 24 This translates in source luminosities of L ∝ 10 ergs assuming a relatively nearby source

25 at a distance of 2 kpc.

26 Among the most promising candidate sources of cosmic-rays in the Galaxy we cite

27 here the remnants of supernovae (both shell-type and pulsar wind nebulae), the jets

28 of microquasars, and the collective winds of massive stars (Hillas 2005; Tavani et al.

29 2009; Corbel& Fermi LAT collaboration 2010; Aharonian et al. 2007; Ohm et al. 2010;

30 Marcowith et al. 2008). Due to the large amount of energy released in a supernova explosion

41 41.8 31 (∼10 -10 erg/s), supernova remnants are prime candidate sources of the galactic cosmic

32 rays. In microquasars, the kinetic energy carried by the jet is at least comparable to the – 5 –

36 33 observed non-thermal luminosities accounting for ∼ 10 erg/s or even higher (Gallo et al.

34 2005; Margon 1984). The total energy inject in the interstellar medium produced by the

39 35 winds of OB and Wolf-Rayet (WR) stars can be as high as ∼10 erg/s, like in the case

36 of the Cygnus OB2 association (Lozinskaya et al. 2002). What remains undetermined is

37 the fraction of total energy that goes into cosmic ray acceleration, as well as the amount of

38 cosmic rays interacting close to source. Measurements of the products of the cosmic ray

39 interactions, i.e., both gamma-rays and neutrinos, are expected to shed light to this problem

40 in the upcoming future. Recently, a large family of galactic accelerators have been observed

41 to have the bulk of their gamma-ray emission at energies below 50 TeV. This statement is

42 relative to the limited experimental techniques implied for the survey. Moreover, they show

−2 43 a characteristic gamma-ray spectrum softer than dN/dE ∝ E expected from first-order

44 Fermi shock acceleration (Fermi 1949, 1954). In this paper we study the capabilities of

−2 45 IceCube to observe sources which depart from the standard dN/dE ∝ E spectrum

46 unbroken up to the PeV scale. This study makes use of the 22- and 40-strings configurations

47 of IceCube as well as the Antarctic Muon And Neutrino Detector Array (AMANDA). With

48 the goal to enhance sensitivity for soft spectra sources, or sources with an energy cutoff

49 below 1 PeV, we have implied AMANDA as low energy extension of IceCube and develop

50 an analysis strategy optimized for a high retention of signal below 10 TeV. The impact

51 of the use of AMANDA as low energy extension of IceCube is shown in Figure 1 and in

52 Figure 1.

53 Moreover, we use our low-energy optimized data samples to search for neutrino

54 emission from the Galaxy at energies above ∼ 500 GeV. The part of the Galaxy accessible

55 to IceCube at the energies considered in this analysis covers the range in galactic longitude

◦ ◦ −2 56 of 210 > l > 40 . For IceCube recent results for an E neutrino spectrum we refer to

57 Abbasi et al. (2011). The search program here reported includes a scan of the part of the

58 galactic Plane accessible to IceCube, a dedicated analysis of the Cygnus region, the search – 6 –

Fig. 1.— Effective area for IC22+AMANDA of the galactic source analysis at final level: upper curve for 143 days of lifetime (AMANDA operational) and lower curve for 133 days. – 7 –

Fig. 2.— Sensitivity and Discovery Potential of the IC40+AMANDA analysis for E−3 neu- trino spectra. The sensitivity is calculated with the method from Feldman and Cousins (Feldman & Cousins 1998) with the additions proposed in (Conrad et al. 2003) and (Hill 2003) to include a systematic uncertainty of ±17% on the neutrino flux. – 8 –

59 for neutrino emission from a pre-defined list of interesting sources, and a time dependent

60 analysis which searches for neutrino emission from Cygnus X-3 in correlation with radio

61 flares. The paper is organized as follows: in Section 2 we report some technical aspects

62 of IceCube, of AMANDA and on the merging of its data in the IceCube data stream. In

63 Section 3, we report about the analysis methods developed and the respective astrophysical

64 targets considered. Following this, the extraction of the neutrino samples is described in

65 Section 4 and in Section 5 the obtained results are reported.

66 2. The Combined Detector: IceCube and AMANDA

67 2.1. IceCube

68 During the construction phase from 2004 to 2010, the operational configuration of

69 IceCube increased year by year (see Figure 3) to finally cover a volume of approximately

70 one cubic kilometer. IceCube is composed of 86 strings each holding 60 digital optical

71 modules (DOMs). Each DOM is composed by a 10 inch photomultiplier tube (PMT) and

72 the relative read-out electronics all housed in a glass pressure vessel. In the array, 78 out

73 of 86 strings form a hexagonal grid with a typical distance of 125 m between neighboring

74 strings. The vertical distance between DOMs on the same string is 17 m. The remaining 8

75 strings are part of the low energy extension DeepCore (R. Abbasi 2011) and are deployed

76 in the center of the detector with a smaller vertical and horizontal spacing between the

77 DOMs. The DOMs detect Cherenkov radiation emitted by secondary charged particles

78 produced in interactions of high energy neutrinos with nuclei in the ice or the bedrock

79 below the ice. To enhance the detection of light from upwards going particles, the PMTs

80 point downwards. In order to avoid deterioration of the PMT analog signal, the waveforms

81 are digitized directly in the DOMs with a set of Analog Transient Waveform Digitizer

82 (ATWDs) and a Fast Analog to Digital Converters (FADCs) (Abbasi et al. 2010). At – 9 –

83 trigger level, the events that are used in the analyses presented in this paper are selected

84 by a simple multiplicity trigger. The trigger used in this analysis requires at least 8 hit

85 DOMs within a time window of 5 µsec. The DOMs send hit times and charge amplitude

86 to the surface once one of the trigger conditions is met and an event is formed. The events

87 contain all DOM readouts associated with the trigger as well as all further readouts within a

88 ±10 µsec around the trigger. If there are several triggers with overlapping readout windows

89 or the trigger condition remains fulfilled for a longer time, an event with a correspondingly

90 adjusted readout window is generated.

91 2.2. AMANDA

92 After a construction phase from 1993 until 2000, the completed AMANDA-II detector

93 took data as standalone neutrino telescope from February 2000 until December 2006. This

94 configuration consisted of 677 optical modules (OMs) on 19 strings. Most of the optical

95 modules were deployed at depths between 1500 and 2000 m whereas IceCube extends down

96 to 2500 m. For data analysis, a total of 526 OMs were used. The AMANDA string position,

97 forming a roughly cylindrical geometry, is shown in Figure 3.

98 The typical distance between adjacent strings is around 40 m. From February 2007

99 until April 2009, AMANDA was operated as an integrated part of IceCube. In many

100 aspects, IceCube is technologically more advanced than AMANDA, reflecting general

101 progress as well as experience collected during the operation of AMANDA (Ackermann

102 et al. 2006). In particular, the signal transfer from the optical modules to the surface is

103 different. As mentioned above, IceCube DOMs digitize the PMT signal directly in the

104 ice. They also generate the HV for the PMTs in the DOMs. In contrast, AMANDA OMs

105 produced analog signals that were sent to the AMANDA data acquisition system which was

106 located in the Martin A. Pomerantz Observatory (MAPO). The original data acquisition – 10 –

Fig. 3.— Left: top view of strings geometry of IC40 and AMANDA (NOTE: THIS FIGURE IS WRONG, THE POSITION OF AMANDA IS OFF, TO BE CHANGE WITH A COR- RECT ONE). Right: Artistic view of the IceCube detector. In this work, we have based the data analysis on the 22-string and the 40-string configurations. – 11 –

107 system (DAQ) could register the leading and trailing edge time of up to 8 pulses per event

108 and the total charge. The same cables were used to transfer the analog data collected by

109 the optical modules and to provide the HV to the PMTs. For AMANDA strings 11 to

110 19, an additional connection via optical fibers was installed to transmit the PMT signals

111 with a better time resolution. Moreover, AMANDA string 18 (Ackermann et al. 2006)

112 was equipped with prototypes for the IceCube DOMs, including the capability of on-board

113 waveform digitization. This option however was used only in testing mode, but it was not

114 included in the data acquisition schemes used for physics analysis.

115 The AMANDA DAQ has been upgraded starting from 2002. Flash ADC modules called

116 Transient Waveform Recorders (TWRs) have been installed in the new DAQ in order

117 to digitize the analog waveforms from the AMANDA OMs at the surface. The upgrade

118 data acquisition operated in parallel to the analog one until 2006. From 2007 on, only

119 the TWR-DAQ was operational. This surface waveform digitization allowed to store

120 more information such as the number and the arrival times of individual pulses. As the

121 TWR-DAQ was also faster than the previous DAQ, the trigger threshold could be reduced.

122 Trigger thresholds from 8 to 13 hit optical modules were used during different years.

123 This upgrade significantly improved AMANDA performances allowing an enhancement of

124 IceCube at low energy.

125 2.2.1. AMANDA integrated into IceCube

126 The operational integration of AMANDA into IceCube required the establishment of

127 connections between the two detectors for the exchange of trigger information to be able to

128 merge events as well as for a fine synchronization in time. MAPO is about 300 m far from

129 the IceCube Control Lab (ICL) that houses the IceCube surface data acquisition. In order

130 to connect the two buildings, optical fibers have been used. Moreover, a TCP/IP connection – 12 –

131 has been established for the communication between the buildings. A GPS module was

132 installed to synchronize the TWRs responsible for the digitization of the AMANDA

133 waveforms and to synchronize the detector with IceCube. By default the IceCube clock

134 was used. For fine synchronization, the optical fibre connection was used to transmit the

135 AMANDA trigger signal to IceCube. This allowed to extract the precise time of AMANDA

136 pulses relative to IceCube pulses even in the case of disturbances of the GPS signal. In

137 the integrated mode, AMANDA and IceCube were still triggered separately. As AMANDA

138 triggered on lower energetic events with respect to those from IceCube, when an AMANDA

139 trigger occurred also IceCube strings were jointly read out even if the IceCube hits were not

140 sufficient for an IceCube trigger. On the other hand, in correspondence to IceCube triggers

141 only AMANDA was not read-out. In fact, the contribution of AMANDA hits in events

142 below the threshold to trigger IceCube would have been negligible. Events from AMANDA

143 and IceCube were merged by a dedicated software (Joint Event Builder, JEB) on the base

144 of a time coincidence. As the duration of AMANDA events was fixed to 10.24 µsec, while

145 the length of IceCube events is increased in case several overlapping triggers are present, it

146 was possible that several AMANDA events were associated with one IceCube event. The

147 JEB software merged all these into one combined IceCube-AMANDA event.

148 With the prospect of IceCube DeepCore ahead, AMANDA was decommissioned in 2009.

149 DeepCore provides a more densely instrumented region in the bottom and central part of

150 the full IceCube detector and makes use of the clearest ice at the largest depths.

151 3. Methods and Targets

152 The IceCube telescope monitors the entire sky with different sensitivities at different

153 energies (Abbasi et al. 2011) without the need of explicit pointing. Search strategies are

154 different depending of the physics goal and statistics of the method. Throughout this paper – 13 –

155 we consider different power-laws for the energy distribution of signal events with the goal

156 to represent possible different source scenarios. For the optimization of the analysis and

157 the evaluation of the detector performance at low energies, a very soft power-law spectrum

−3 158 following dN/dE ∝ E has been used. Moreover, we have also considered the spectrum

159 expected from the Crab nebula if the H.E.S.S. gamma-ray observations are interpreted in

160 terms of an hadronic model of pp interactions (Kelner et al. 2006) which corresponds to

−2.4 161 dN/dE ∝ E with an exponential cut-off at 7 TeV. This spectrum is representative of a

162 ”low-energy” source and very instructive in order to understand the impact of an energy

163 cutoff on IceCube sensitivity. We will refer to this spectrum throughout the paper as a

164 ”Crab-like” spectrum. We report in the following/below on the different searches that have

165 been performed.

166 3.1. Galactic Plane Scan and Source List

167 3.1.1. Unbiased Scan

168 The location of the Solar System in the Local spiral arm gives us a particular

169 perspective of the Galaxy. Due to projection effects and given the distribution of cold gas in

170 the Galaxy most of the galactic accelerators are observed in our local reference system in a

171 narrow band close to the plane of the sky. In the energy range of interest IceCube accesses

172 only the northern sky. Even if the galactic center region is not included, other important

173 parts of the galactic region are in the IceCube field of view considered in this analysis.

174 These include part of the first quadrant of the Galaxy, the whole second quadrant, and a

175 small portion of the third galactic quadrant.

176 The exploration of the galactic plane, seeking for sources of high energy neutrinos, is done

177 by superimposing a fine grid over the event distribution within the galactic coordinate

◦ ◦ ◦ ◦ 178 35.875 < l < 209.875 , −5.125 < b < 5.125 . The grid step is chosen to be smaller than – 14 –

179 the angular resolution of the telescope achieved in analysis [CHECK WHERE IS THE

◦ ◦ 180 ANG RES]. In the analysis of IceCube 22-strings, a step of 0.5 × 0.5 is used, while for

◦ ◦ 181 the IceCube 40-strings detector a grid spacing of 0.25 × 0.25 was chosen. At each grid

182 point, the likelihood of composite signal and background hypothesis is compared to the

183 background hypothesis with the unbinned likelihood method described in Braun et al.

184 (2008).

185 3.1.2. Specific Sources

186 A total of seven particularly interesting sources have been individually studied in this

187 analysis as representatives of different types of galactic accelerators. In the specific case of

188 Cygnus X-3, due to the high variability in the radio and X-rays, we tested the hypothesis of

189 variable neutrino emission and performed a time-dependent analysis. The other considered

190 sources here below listed are treated as steady point-source candidates.

191 Crab nebula: The Crab nebula is powered by a pulsar with a spin-down luminosity of

38 192 ∼ 5 × 10 erg/s. This energy is injected into relativistic particles and magnetic fields

193 (Kennel & Coroniti 1984), although the exact composition of the pulsar wind, as well as

194 the mechanism by which the total power of the pulsar is transported and dissipated is

195 not known. It is an efficient particle accelerator, where ∼60% of the total power of the

196 pulsar is injected into relativistic electrons which emit synchrotron radiation from radio to

197 X-rays (Hester 2008). Although it appears as the strongest gamma-ray source in the sky,

198 the ratio between the gamma-ray luminosity at E > 1 TeV and the spin down luminosity

−5 199 is of the order of 10 (Aharonian et al. 2004, 2006). The simplest implications of this are

200 that electrons lose rapidly their energy through synchrotron radiating at lower frequencies,

201 and that the majority of cosmic-rays, if present in a significant proportion, escape from

202 the source without interaction. Even if this argument implies a mainly leptonic origin of – 15 –

203 the gamma-rays emission, the high intensities observed in the Crab in gamma-rays can be

204 reached by IceCube after a few years of operation, providing information about the nature

205 of the high energy processes that take place in the Crab. Due to the unusual flaring state

206 of the Crab observed during 2010, the IceCube Collaboration has performed a dedicate fast

207 response analysis of the 79-string configuration data to search for neutrinos that might be

208 emitted along with the observed γ-rays (Abbasi 2011).

209 Cas A: This source is a classical shell-type SNR at a distance of 3.4 kpc. It was detected

210 by HEGRA in the energy region between 1 TeV to 10 TeV without any evidence of

211 an energy cutoff (Aharonian et al. 2000), and detected by MAGIC down to 250 GeV

−2.3 212 following a steep power-law spectrum ∝ E and a photon flux above 1 TeV of

−13 −2 −1 213 F (> 1 T eV ) ∼ 7.3 × 10 cm s (Albert et al. 2007).

214 IC 443: IC 443, at a distance of ∼1.5 kpc, is an asymmetric shell-type SNR, where part

215 of the shell is impacting a molecular cloud, accelerating particles to very high energies in

216 the process. TeV gamma-rays are observed in coincidence with the molecular line emission

217 region, giving support to an hadronic origin of the TeV gamma-rays. The differential

218 flux measured in the energy range from 100 GeV to 1.6 TeV is well fitted by a very

−3.1 219 steep power-law ∝ E (Albert et al. 2007). The photon flux above 1 TeV obtained by

−13 −2 −1 220 extrapolation is F (> 1 T eV ) ∼ 3.2 × 10 cm s .

221 LS I+61 303: This source is a high mass X-ray binary in an eccentric orbit around a Be

222 star. The nature of the compact object is not known, and both a pulsar wind model and

223 a microquasar model have been suggested for this source. MAGIC detected VHE emission

224 modulated with the orbital period (Jogler et al. 2008). The highest significant detection

225 is obtained around apastron, at orbital phases 0.6-0.7, with a spectrum following F (E

−12 −2.6 −1 −2 −1 226 > 300 GeV) = 2.6 × 10 E TeV cm s unbroken. No TeV emission is observed

227 at periastron, although significant gamma-ray absorption in the strong radiation field of

228 the Be star is expected in this case (Sierpowska-Bartosik et al. 2009). This scenario is – 16 –

229 supported by the detection of MeV-GeV gamma-rays from the Fermi satellite (Abdo et

± 230 al. 2009), which may be the result of the cascade process in γγ → e . The hypothesis

231 of particle injection along the whole orbit is then a plausible option. This together with

232 the considerable amount of both matter and radiation from the companion star available

233 for cosmic-ray interaction makes this source an interesting candidate for steady neutrino

234 emission. For the search of periodic neutrino emission from binary systems performed by

235 IceCube, we refer to (Abbasi 2011).

236 SS 433: SS 433 is a confirmed microquasar and a black hole candidate in orbit around a

237 massive star. The source exhibits two oppositely directed relativistic jets which are thought

−6 −1 238 to eject material at a rate larger then 10 M year (Begelman et al. 1980). It is the

239 only X-ray binary system in which hadrons have been found in the jet (Migliari et al.

240 2002). The entire source is embedded within a nebulous structure (W50) which is thought

241 to be the expanding supernova shells of the progenitor star of the black hole in SS 433.

242 The source has been observed by the HEGRA, MAGIC, and CANGAROO-II Cherenkov

243 telescopes (Aharonian et al. 2005; Saito et al. 2009; Hayashi et al. 2009), resulting in upper

244 limits to the gamma-ray emission from both the inner system and the different interaction

245 regions with the W50 nebula. Strong gamma-ray absorption is expected from this system,

246 due to the periodic companion eclipses as well as attenuation due to the precession of

247 the accretion disk envelope (Reynoso et al. 2008). As in the case of LS I+61 303, the

248 presence of significant amount of targets for cosmic-ray interactions as well as the possibility

249 for a higher energy emission that what is inferred from gamma-ray observations due to

250 absorption, makes this an interesting target for IceCube. Also this source has been tested

251 for a possible periodic neutrino emission in (Abbasi 2011).

252 W51C: This is a supernova remnant within a molecular cloud located at 5.5 kpc which

253 was detected by the Fermi-LAT telescope showing a very high luminosity in gamma-rays,

36 254 greater then 10 erg/s (Abdo et al. 2009). The strong gamma-ray luminosity of W51C and – 17 –

255 the hints of an hadronic origin for the gamma-ray spectrum makes this an interesting target

256 for IceCube.

257 3.2. Cygnus X-3

258 The high energy sky presents strong variability in time. If neutrino emission

259 from a particular object is expected to vary with time and if it is in coincidence with

260 electromagnetic emission, it is advantageous to include the time information in the data

261 analysis like described for example in the case of different objects in Resconi et al. (2009).

262 One such case is the microquasar Cygnus X-3 (Cyg X-3). In coincidence to the period

263 of data taking considered in this analysis, first observation of production of high energy

264 photons within the system of Cyg X-3 was published by the Fermi (Abdo et al. 2009) and

265 AGILE (Tavani et al. 2009) satellite missions independently. The reported gamma-ray

266 fluxes are in the energy band between 100 MeV and 100 GeV (Fermi), following a power

267 law of spectral index 2.70 ± 0.25, and between 100 MeV and 10 GeV (AGILE) with spectral

268 index 1.8 ± 0.2. However, γ-rays from Cyg X-3 are only detectable during certain periods of

269 time, probably correlated with strong radio outbursts and certain X-ray emission states of

270 the system (Abdo et al. 2009; Koljonen et al. 2010). Cyg X-3 has not been observed in TeV

271 energies by Cherenkov telescopes like MAGIC and VERITAS so far, a possible explanation

272 being strong absorption of high-energy photons or limited observation time. Assuming a

273 hadronic origin of at least part of the γ-rays, the high densities in the system of Cyg X-3

274 could provide an environment for copious neutrino production at TeV energies, detectable

275 by neutrino telescopes such as IceCube (Abdo et al. 2009; Bednarek 2005). The specific

276 scenario of a possible periodic neutrino emission has been studied in (Abbasi 2011).

277 In order to select the most promising period of neutrino emission, the phenomenology

278 of emission states of Cyg X-3 has been studied carefully using mostly radio and X-ray – 18 –

279 observations (Szostek et al. 2008; Tudose et al. 2007; Koljonen et al. 2010). With radio

280 and X-ray data, there are two ways to identify active periods of Cyg X-3 associated with

281 jet ejection: the observation of radio flux above 1 Jy, following (Szostek et al. 2008) and

282 the observation of hyper-soft X-ray state and subsequent hardening of the X-ray spectrum,

283 following (Koljonen et al. 2010). The identification of potential flaring periods of Cyg X-3

284 is thus split in a radio and an X-ray part that are defined as in Table 1. The resulting

285 time intervals from both selections can be overlapping and are combined with a logical

286 OR operation, resulting in the final time windows. The utilized radio data were taken

287 with the Arcminute Microkelvin Imager (AMI) radio telescope (Zwart et al. 2008; Pooley

288 & Fender 1997) at a frequency of 15 GHz between May 2008 and May 2009 in irregular

289 intervals. X-ray data were obtained from the Rossi X-ray Timing Explorer/All Sky Monitor

290 (RXTE/ASM) (Levine et al. 1996), using the B band between 3 and 5 keV, and from the

291 Burst Alert Telescope on board the Swift satellite (Swift/BAT) Barthelmy et al. (2005)

1 292 sensitive within 15 - 50 keV. The ratio of BAT (hard X-ray) to ASM B (soft X-ray)

293 counts provides a spectral hardness parameter. Gamma-ray data from Fermi or AGILE are

294 not explicitly taken into consideration in this analysis since they are not available for the

295 entire period of data-taking of the analyzed IceCube and AMANDA data. Applying the

296 selection criteria to the radio and X-ray data from the IceCube 40 strings running time

297 (between MJD 54560 and MJD 54989) results in the four time windows in Table 2 and

298 Figure 4. Even though there is no AMI data from the first 2.5 months of this period, there

299 was an ATel (Trushkin et al. 2008) issued for a strong radio flare around MJD 54574 that

1AMI data from Pooley,G., http://www.mrao.cam.ac.uk/ guy/cx3/data/ (2010-01- 14). RXTE/ASM data from Bradt, H., Chakrabarty, D., Cui, W. et al., http://xte.mit.edu/ASM lc.html (2010-03-11). Swift/BAT data from Krimm, H., http://swift.gsfc.nasa.gov/docs/swift/results/transients/CygX-3/ (2010-06-04). – 19 –

300 is consistent with the time windows selected from X-ray data.

301 The statistical method applied to this source is a maximum likelihood test using a

302 time-dependent version of the unbinned likelihood ratio method (Braun et al. 2010). The

303 search time windows are incorporated into the signal probability density function (p.d.f.) of

304 the likelihood function as normalized Gaussians at center of the time interval with FWHM

305 equal to the window duration. During maximization of the likelihood, the windows are

306 allowed to be shifted up to 20 days to earlier or later times. This allows to find neutrino

307 emission that comes before or after a radio flare. The value of 20 days is motivated by

308 the hypothesis of emission during radio quenched state, that can happen up to around 20

309 days before the onset of a major radio flare (Koljonen et al. 2010; Trushkin et al. 2007). In

310 the search for neutrinos from the microquasar Cygnus X-3, five searches are performed in

311 total, one with each of the four windows as hypothesis of neutrino signal light curve and

312 one search using all four windows simultaneously. Doing this analysis, only about 50% of

313 the flux from a time-integrated search (that uses no information about acitivity of Cyg

314 X-3) is needed for a 5σ discovery, assuming the neutrino emission happens only during

315 the windows or within ±20 days. The discovery flux with 50% detection probability is

−10 −1 −2 −1 −3 −11 −2 −1 316 ≈ 1.0 · 10 T eV cm s for an E spectrum and ≈ 1.2 · 10 T eV cm s for an

−2 317 E spectrum.

318 3.2.1. The Cygnus Region

◦ 319 The Cygnus region is approximately bounded by galactic longitudes l = 70 to l =

◦ 320 90 , where our line of sight is directed nearly along the Local spiral arm of the Galaxy. At

321 a distance of approximately 5 kpc our line of sight has left the Local arm and crosses the

322 Perseus arm, and even the outer arm further away (∼10 kpc). Here many different sources

323 that are located at different distances superpose in a relative small area in the sky, resulting – 20 –

Table 1: Criteria followed for the selection of Cyg X-3 flaring periods.

Wavelength Telescope START STOP

Radio AMI radio telescope (Zwart et al. 2008) S15GHz a> 1 Jy S15GHz a< 1 Jy

X-ray RXTE/ASM (Levine et al. 1996) & Sfit b> 1 Jy Sfit b <1 Jy Swift/BAT (Barthelmy et al. 2005)

aMeasured radio flux density at 15 GHz. bAverage of Gaussian fitted to 28 radio flares. One Gaussian is centered on each X-ray state with hardness > 0.001 following a state with hardness < 0.001 within 10 days. See text for definition of X-ray hardness.

Table 2: Selected search windows for the neutrino search from Cyg X-3 direction extracted from radio and X-ray data. Window Start Stop Duration (MJD) (MJD) (days) 1 54571.4 54582.5 11.1 2 54584.5 54607.4 22.9 3 54637.6 54649.5 11.9 4 54811.5 54824.6 13.1 – 21 –

Table 3: Upper limit on the flux of muon neutrinos from six γ-ray sources, assuming an E−3 neutrino spectrum.

90% Object R.A. Dec ns pre-trial p-value Φνµ Crab Nebula 83.63◦ 22.02◦ 0 1 7.3 LSI +61 303 40.13◦ 61.23◦ 1.6 0.247 8.3 W51 290.82◦ 14.15◦ 0.6 1 8.3 CasA 350.85◦ 58.82◦ 0 1 5.9 SS433 287.96◦ 4.98◦ 0 1 9.8 IC443 94.18◦ 22.53◦ 0 1 7.3

aThe flux limits are given in units of 10−11T eV −1cm−2s−1

Table 4: Feldman-Cousins upper limits on neutrino flux from Cyg X-3 averaged over the whole period of data-taking. Time Shift Ev. Exp. bg. E−3 Upper Limit E−2 Upper Limit Window (days) TeV2 cm−2 s−1 sr−1 TeV cm−2 s−1 sr−1 1 +4.46 5 2.6 ± 1.3 4.7 · 10−11 7.3 · 10−12 2 -15.05 5 4.6 ± 1.9 4.0 · 10−11 6.0 · 10−12 3 -0.58 3 3.0 ± 1.5 4.5 · 10−11 5.9 · 10−12 4 +20.00 0 3.0 ± 1.5 6.1 · 10−11 7.7 · 10−12 All +2.05 12 10.4 ± 3.0 5.0 · 10−11 7.0 · 10−12 – 22 –

Fig. 4.— The radio light curve and hardness of Cyg X-3 with the four time windows of the analysis (blue shading). – 23 –

324 in a complex region which harbors some of the closest and most massive regions of star

325 formation in the Galaxy. The vast majority of molecular gas detected in the Cygnus region

326 concentrate on the local arm (Schneider et al. 2006), and hence covering distances between

327 1-3 kpc. One of the most massive giant molecular cloud complexes in the Galaxy resides

328 within this region, at a distance of ∼1.7 kpc, and it is thought to be the birth place of one

329 of the most massive objects in the area, the Cyg OB2 association, and probably also Cyg

330 OB9 nd Cyg OB1, as well as a number of less massive star clusters with young or ongoing

331 star formation (Le Duigou & Kn¨odlseder 2002). The strong stellar winds and radiation

332 pressure of the massive stars in the Cygnus have strongly influenced the spatial distribution

333 of the molecular gas in the region, displacing and compressing the gas forming filametary

334 structures and dense clumps which surround the less dense enviroment of the cluster, in

335 which the gas has been evacuated. If high energy particles are generated at some point

336 within the stellar associations, they can interact with the UV radiation fields producing

337 TeV gamma-rays through the Inverse Compton and pγ processes. However, protons and

338 nuclei can travel longer distances than electrons, and they may leave the photon dominated

339 regions around the massive star clusters and interact with the nearby molecular clouds. The

340 resulting neutrino event pattern in the region would be then generated by a spatial process

341 which depends on the complicated distribution of the gas in the region. This picture is

342 even more complicated when we consider that the injection of cosmic-rays may take place

343 at different locations due to presence of several particle accelerators inside the Cygnus. The

344 existence of particle accelerators in this region is confirmed by the observation of strong

◦ ◦ 345 TeV gamma-ray emission throughout an area of approximately 10 × 10 (Abdo et al. 2007;

346 Aliu et al. 2011).

347 Due to the complexity of the possible spatial distribution of neutrinos produced in the

348 Cygnus, we have applied a correlation analysis for the discovery of a significant neutrino

349 event pattern in an extended region, as described in (Sestayo & Resconi 2011). The – 24 –

◦ ◦ 350 analysis is performed inside an area of 11 × 7 centered in the most active part of the

351 Cygnus complex in TeV gamma-rays.

352 4. Neutrino Samples

353 4.1. AMANDA data

354 Contrary to IceCube, the waveforms collected in the optical modules of AMANDA

355 are not digitalized in the optical modules but instead are transferred to the surface as

356 analog signals. This leads to the presence of two undesired effects in the data which are not

357 present in IceCube data. The first one concerns the crosstalk between cables. The cables

358 connect different optical modules on the same string to the surface and they can be very

359 close to each other. Limited shielding results in the possibility that a signal is caused by

360 electromagnetic induction from neighboring cables. In particular, large pulses in one optical

361 module can cause a detectable signal at the surface end on cables connected to other optical

362 modules. The second issue concerns pickup of electromagnetic noise. The PMT signals

363 have to be transferred over a distance of over 1.5 km to the surface. The cables needed for

364 this task are vulnerable to pick up electromagnetic noise from the surface as they act as

365 electromagnetic antennae.

366 AMANDA, as a detector that explored several techniques for later neutrino telescopes, does

367 not use a uniform cabling. It turned out that the optical modules on AMANDA string 5-10,

368 which are connected by twisted pair cables are most vulnerable to these effects. Methods

369 dedicated to the identification of non-particle induced signals based on the waveforms have

370 been developed and result in an efficient separation from particle induced signal. The

371 integral over the entire collected waveform pulse is used in order to remove crosstalk pulses.

372 In fact, crosstalk pulses do not originate from a charge deposit in the PMT. Hence they

373 consist of (positive and negative) fluctuations around the baseline with the total integral – 25 –

374 close to zero. Waveforms from a particle induced signal in the PMT have a characteristic

375 width, which is wider at the surface due to dispersion in the cables. In case of AMANDA

376 strings 5-10, this is typically 250-300 ns for single photon electrons (SPE). Waveforms from

377 multiple photon electrons (MPE) result from the (linear) overlay of many SPE waveforms

378 and typically are wider than those. In contrast, non-particle induced waveforms are often

379 very spiky, i.e. they have many peaks within the width corresponding to a typical SPE

380 pulse. This feature of non-particle-induced waveforms has been used to remove noisy events

381 from the data: if the median peak rate in the waveforms recorded in AMANDA string

382 5-10 is incompatible with a PMT signal and if there is a high number (more than 20) of

383 waveforms in these strings, the event is considered non-particle induced and it is removed

384 from the data set. Figure 5 illustrates this cut. Both analyses presented in this paper

385 apply cross-talk cleaning and the illustrated rejection of non-particle induced events on the

386 AMANDA data before further event selections are made for the final hypothesis tests.

387 4.2. IceCube 22-strings and AMANDA

388 Data has been collected from May 2007 until April 2008 when IceCube was operating

389 in a 22-string configuration. The lifetime of the IceCube 22-string run is 276 days in

390 which 143 days AMANDA was operating in stable mode. The unusual long downtime of

391 AMANDA during the IceCube 22 operation has been due to various hardware failures

392 during May 2007 (trigger system) and during August 2007 (high voltage supply system).

393 In this section, the event selection from the trigger level up to the final analysis level is

394 described and the characteristics of the combined neutrino sample are highlighted. – 26 –

Fig. 5.— Distributions of the parameters used to identify non-particle induced events in AMANDA. Shown are the number of waveform fragments versus the median peak rate per waveform (see text ) on strings 5-10. The bottom plot represents a run with stronger pickup of electromagnetic noise (109325) while the upper plot is obtained from a normal run (110594). Events with more than 20 waveform fragments on strings 5-10 and a median peak rate above 0.005 GHz are rejected as non-particle induced events. – 27 –

395 4.2.1. Trigger Logic

396 IceCube and AMANDA are triggered separately in the combined detector mode.

397 The trigger rate of IceCube 22-strings is 550 Hz, while AMANDA triggers at 200 Hz.

398 Seasonal variations affect the trigger rate for a factor 10%. Once AMANDA triggers, the

399 full IceCube detector is read out, while once IceCube trigger AMANDA is not read out

400 since the event will be dominated by IceCube information anyhow. The overall trigger rate

401 of the combined IceCube 22-string and AMANDA detector after correction for overlaps

402 between the two triggers is 640 Hz. At trigger level, the data are strongly dominated

403 by down-going atmospheric muons induced from cosmic rays air showers outnumbering

6 404 atmospheric neutrinos by a factor of about 10 .

405 4.2.2. Track Reconstruction and On-line Filtering

406 For muon track reconstructions and data filtering, the data events collected are divided

407 into two streams. The first stream concerns events which trigger AMANDA. As explained

408 above, these events are then complemented with IceCube collected information and for

409 this reasons are considered as combined events (C-events). The second stream involves

410 events that trigger only IceCube, then the information from AMANDA is disregarded

411 and these events are called IceCube only events (IO-events). First event selections are

412 applied online at the South Pole based on track reconstruction algorithms which do not

413 require sophisticated computation (first guess reconstruction). For IO-events, a plane wave

414 reconstruction, is chosen as first guess reconstruction (linefit) (Ahrens et al. 2004). For

415 C-events, a pattern recognition algorithm (called JAMS REF) was chosen. Events are

416 removed if the first guess reconstruction is down-going. The passing rate for reconstructed

417 up-going or horizontal events corresponds to 22 Hz for IO-events and 8 Hz for C-events,

418 producing a total event rate of 30 Hz. At this stage, data are transferred out from South – 28 –

419 Pole to the north in order to allow more sophisticated reconstruction.

420 4.2.3. Neutrino Sample

421 After the rejection of down-wards reconstructed events based on simple first guess

422 reconstruction, the data are still dominated by mis-reconstructed atmospheric muons.

423 In particular coincidences between multiple muons from different air showers can mimic

424 up-going event topologies. Further event selections are needed in order to arrive to a

425 sample of events dominated by atmospheric neutrinos. More CPU intensive track and

426 energy reconstructions are performed to improve the angular reconstruction and to provide

427 quality parameters for the rejection of the down-going atmospheric muon background.

428 These reconstructions are maximum likelihood fits which are based on the probability

429 density function for the arrival time of a photon given the track hypothesis. Two likelihood

430 reconstruction have been studied: a simple one based only on the first photon in each

431 optical module (SPE) and a more complete one which includes the possible presence of

432 multiple photo electrons (MPE). For more details about SPE and MPE we refer to (Ahrens

433 et al. 2004). A list of quality parameters available from these reconstructions is given in

434 Abbasi et al. (2011). A typical quality parameter used in this analysis is the number of

435 unscattered photons detected which are characterized by a small time residual with respect

436 to the Cherenkov cone theoretical hypothesis.

437 In order to optimize the retention of lower energetic events, a multivariate approach

438 was chosen. The final signal likelihood is defined as the product of the signal likelihood from

439 the considered variables and is compared to the likelihood of the background hypothesis.

440 According to a lemma by Neyman and Pearson (Neyman et al. 1933), this signal likelihood

441 leads to the best possible discrimination power if all variables are uncorrelated. For

442 correlated variables, as in our case, this signal likelihood turns out to be powerful as – 29 –

443 well. We use unbiased data to model the background and atmospheric neutrinos for the

444 signal. C-events and IO-events were treated independently here. For IO-events, further

445 cuts were used to reject coincident air shower muons. These cuts are based on the smooth

446 distribution of hits along the track and on reconstructions only on the early resp. late hits

447 in the event. For C-events, the rate of coincident muons is significantly lower due to the

448 smaller size of AMANDA. While a tighter time-window cleaning helped to further reject

449 the coincident muons in the combined events stream, no dedicated cuts/reconstructions

450 were used to remove these. The final cuts were optimized for best discovery potential, and

◦ 451 finally tracks are selected if the angular resolution estimator return a value lower then 4 .

452 The resulting neutrino sample contains a significantly larger number of events compared to

453 the search presented in (Abbasi et al. 2009). In total, 8727 events are selected, of which

−4 454 3430 are C-events. These event numbers correspond to a data rate of 4.7 · 10 Hz while

−4 455 AMANDA was operational (143 days) and to 2.4 · 10 Hz in IceCube 22-string only mode

456 (133 days). The final effective area of the IceCube 22-strings plus AMANDA analysis is

457 shown in Figure 1. Below energies of a few TeV, there is a strong improvement by using

458 the combined detector including AMANDA with respect to the possible performance of a

459 low energy optimized analysis on IceCube only. The energy distribution of atmospheric

460 neutrinos for this selection according to simulations is shown in Figure 6.

◦ 461 The estimated angular resolution is given by a median of 1.9 for atmospheric neutrinos

◦ 462 and 1.7 for a Crab-like spectrum. About 10% of the final events are expected to be

463 mis-reconstructed muon background. The resulting discovery potential (minimum flux

464 which produces a 5σ detection in 50%) for a Crab-like spectrum is shown in Figure 7 as a

465 function of declination.

466 The minimum detectable flux is more than an order of magnitude above the neutrino

467 emission expected from the Crab assuming that the H.E.S.S. observations are consistent – 30 –

Fig. 6.— Event energy distribution for atmospheric neutrinos at the final selection level of the galactic point source analysis normalized to the lifetime of the IceCube 22-strings run according to simulations. – 31 –

Fig. 7.— Discovery potential of the galactic point source analysis in terms of the ”Crab-like” spectrum as a function of declination. – 32 –

468 with a model of pp interactions. While with this expectation, a positive detection is

469 unlikely but not excluded as the photon flux may be absorbed, the analysis is useful as a

470 starting point for future improvements with the full IceCube detector and the DeepCore

471 subdetector. It demonstrates how a low energy optimized cut strategy and the use of

472 additional information from a more densely instrumented sub-detector can largely improve

473 the discovery potential. In comparison to the analysis presented in (Abbasi et al. 2009), the

474 sensitivity obtained for the Crab-like spectrum is almost a factor of two lower.

475 4.3. IceCube 40-strings and AMANDA

476 Following the successful analysis strategy developed for the IceCube 22-strings and

477 AMANDA detector, a similar analysis has been conducted on the more competitive data

478 sample collected with the combined IceCube 40-strings and AMANDA detector from April

479 5, 2008 to May 20, 2009. Both parts of the combined IceCube-AMANDA detector operated

480 very stably during this time. For IceCube about 375 days of data were collected and

481 used in this analysis and for AMANDA about 306 days. This corresponds to 92% of the

482 entire IceCube data taking during the season. The main causes for AMANDA downtime

483 were scheduled operations in the course of the integration of new strings into the detector.

484 Moreover, the decommission of AMANDA began few weeks before the completion of the

485 IceCube 40-strings run. The event selection is in many aspects similar to the one applied to

486 the IceCube 22-strings and AMANDA data as the targeted energy range is the same as well

487 as the physics driving the analysis. Again, different cut criteria are developed for combined

488 IceCube-AMANDA events (C-events) and for IceCube only events (IO-events). – 33 –

489 4.3.1. IceCube Only Events

490 Similar as in the previous combined analysis, IO-events are selected by a series of

491 one-dimensional cuts on event quality parameters and combined with a multivariate

492 classification based on the Neyman-Pearson lemma (Neyman et al. 1933). The probability

493 density functions for five quality parameters (paraboloid sigma, MPE plogl, number and

494 length of direct pulses, smoothness of direct pulses, see Abbasi et al. (2011)) are generated

495 from atmospheric muon-dominated data as background and from atmospheric neutrino

496 simulation and combined in the cut. The distribution of the resulting cut variable is shown

497 in Figure 8 for data and for atmospheric neutrino simulation as well as for two example

−3 498 signal neutrino spectra. An optimization of the discovery potential for a soft E spectrum

499 results in an optimal cut value of 1.0.

500 While the analysis is optimized for soft spectra such as this one, it is desirable to retain

501 the best possible efficiency for high energy neutrino events as well. With a multivariate

502 cut optimized on a very soft spectrum however, the retention of high energy events is not

503 necessarily optimal as these deposit much more light in the detector and may thus have

504 event topologies that are not caught in the low energy event selection. To remedy this,

505 additional events are included in the event selection if their signatures are likely to be

506 induced by very high energy neutrinos. These additional events are selected with a series of

507 one-dimensional cuts based among others on their reconstructed energy.

508 4.3.2. Combined Events

509 C-events are first cleaned as described in section 4.1. The performance of the cuts

510 developed there has been validated again on the IceCube 40-strings and AMANDA data

511 before they were applied. Subsequently, a series of one-dimensional cuts is applied. This – 34 –

Fig. 8.— Distribution of the main cut variable for IceCube events in data (black), simulated atmospheric neutrinos (red), and simulated neutrino signal with two different spectra of arbitrary flux scale. A cut at llh value of 1 is applied to the data to select a neutrino sample. The agreement of data and simulated atmospheric neutrinos is very good in the signal region. – 35 –

512 series of cuts has been shown to result in a similar performance as the likelihood ratio based

513 on the Neyman-Pearson lemma for this particular data set.

514 4.3.3. Neutrino Sample

515 Maximum likelihood reconstructions are available both using a single photo electron

516 (SPE) pdf and using a multiple photo electron (MPE) pdf. The latter has a better angular

517 resolution for high energy events; both have a similar performance for lower energies in

518 IceCube. The IceCube 40-strings and AMANDA analysis uses the MPE pdf for all events

519 with sufficient hits in IceCube to obtain a good resolution with these alone. For all low

520 energetic combined IceCube-AMANDA events, the SPE pdf is used.

521 The total number of selected neutrino candidates is 19,797. The purity of the

522 atmospheric neutrino sample is between 97-98% based on estimated with background

523 simulation. Of the selected neutrino candidates, 81.3% are IO-events selected with the

524 multivariate Neyman-Pearson likelihood ratio cut. 2.4% are additional IO-events with high

525 energy estimates. The remaining 16.3% events are C-events. Despite the larger lifetime of

526 AMANDA in 2008/2009 with respect to the previous year, the fraction of C-events in this

527 analysis is smaller than in the previous one. This is partially due to the larger size of the

528 IceCube detector but also to the higher purity of the IceCube-40 strings and AMANDA

529 sample. The resulting energy distribution as derived from atmospheric neutrino simulation

530 is shown in Figure 9. The declination and azimuth distributions of the final neutrino sample

531 are reported in Figure 10 and in Figure 11. The selected C-events peak at lower energies

532 than the IO-events. Also the effective area (see Figure 1) shows the effect of AMANDA at

533 the lower energies. Through the longer lifetime and larger size of the detector, the effective

534 area is improved significantly with respect to the previous analysis. – 36 –

Fig. 9.— Energy distribution of the final sample. – 37 –

Fig. 10.— Declination distribution of the final neutrino sample. – 38 –

Fig. 11.— Azimuth distribution of the final neutrino sample. – 39 –

535 5. Results

536 No statistically significant excess has been found in any of the tests performed. In the

537 absence of detection, upper limits on the muon neutrino flux from the considered regions

538 of the Galaxy have been calculated. The upper limits are derived for soft neutrino spectra,

539 with and without energy cutoff, using the method by Feldman and Cousins (Feldman

540 & Cousins 1998) at 90% CL. Systematic uncertainties have been included in the limit

541 calculation using the method defined in (Conrad et al. 2003) with the modification in

542 (Hill 2003). The upper limits are calculated under the hypothesis of a muon neutrino flux

◦ 543 only. Neutrino oscillations with the large mixing angle Θ23 ∼ 45 result however in the

544 expectation of equal fluxes of νµ and ντ at the detector independent from their ratios at

545 the source. This analysis is sensitive to ντ as well, mainly due to the decay of τ created

546 in charged current interactions into muons with a branching ratio of 17%. At higher

547 energies (above the energy scale targeted in this analysis), the track length of τ is sufficient

548 to generate a track-like signature in the detector. Taking into account these effects and

549 the details of energy losses, the contribution of ντ is estimated to 10% − 17% of the νµ

550 contribution for IceCube analyses optimized for muon neutrinos.

551 5.0.4. Galactic plane scan and source list

552 In Figure 12 and in Figure 13, we report the result of the scan of the galactic plane with

553 IceCube 22-strings and IceCube 40-strings. In the analysis of IceCube 22-strings data, the

554 lowest background probability at point-source angular scales is found at galactic coordinates

◦ ◦ 555 l = 75.87 , b = 2.67 , with a pre-trial p value of 0.37%. Considering the intrinsic trials of

556 the scan, an equal or higher significance in at least one of the sampled locations is found

557 in 95% of pure background skymaps. The most significant point-like spot in the analysis

558 of IceCube 40-strings, with a pre-trial p-value of 0.093%, is found at galactic coordinates – 40 –

◦ ◦ 559 l = 85.5 , b = −2, 0 . After accounting for the trials, an equal or higher significance is found

560 in 88% of pure background samples.

561 Constraints to the neutrino emission have been calculated for the six selected candidate

562 sources: the Crab Nebula, LSI +61 303, W51, CasA, SS433 and IC443. In Table 5, we

−3.0 −1 −2 −1 563 summarize the results for a flux proportional to (E/TeV) TeV cm s without energy

564 cut-off.

565 In the IceCube 22-strings data sample, the highest excess was observed at the position

566 of the Crab Nebula, with a pre-trial p value of 13% (37% post-trial). In the IceCube

567 40-string data sample, the most significant clustering of events has been observed around

568 LSI +61 303. The best fit number of events from this location is 1.6 and the observation

569 corresponds to a pre-trial p value of 24.7%. Accounting for the trials from testing six

570 different positions, the post-trial p value of this search is 42.3%, i.e. 42.3% of pure

571 background samples show a similar or stronger accumulation of events around one of the

572 six objects.

573 5.0.5. The Cygnus region

574 The clustering function, Φ(Θ) of the neutrino events within the galactic coordinates

◦ ◦ ◦ ◦ 575 72 < l < 83 , −3 < b < 4 has been computed for both IceCube 22-strings and IceCube

576 40-strings samples, producing a non-significant result. Figure 14 shows the clustering

577 function of the observed event pattern. No significant concentration of events is seen at any

578 of the angular scales of observation.

579 The result obtained on the IceCube 22-strings data sample shows a positive fluctuation

580 at the level of 2.3σ. This result contains already the correction of the trials associated to the

581 observation at different angular scales. The region considered showed a positive fluctuation – 41 –

Fig. 12.— Pre-trial significances (p value) of the galactic plane scan in IceCube 22-strings.

Table 5: Upper limit on the flux of muon neutrinos from six γ-ray sources, assuming an E−3 neutrino spectrum.

90% Object R.A. Dec ns pre-trial p-value Φνµ Crab Nebula 83.63◦ 22.02◦ 0 1 7.3 LSI +61 303 40.13◦ 61.23◦ 1.6 0.247 8.3 W51 290.82◦ 14.15◦ 0.6 1 8.3 CasA 350.85◦ 58.82◦ 0 1 5.9 SS433 287.96◦ 4.98◦ 0 1 9.8 IC443 94.18◦ 22.53◦ 0 1 7.3

aThe flux limits are given in units of 10−11T eV −1cm−2s−1 – 42 –

Fig. 13.— Result of the scan of the galactic Plane with data collected by IceCube 40-strings combined with AMANDA. The pre-trial p-value at each point of the grid is presented along with the distribution of the detected events shown as black dots. The most significant excess of events is located at (85.5◦, −2, 0◦) and has a pre-trial p-value of 0.093% (− log(pvalue) = 3.03). Accounting for the trials induced by the scanning of many points inside the galactic Plane, the p-value of this test is 88.02%. – 43 –

Fig. 14.— Clustering function of the IceCube 22-string events (dots) and IceCube 40-string events (stars) within the region defined by the galactic coordinates 72◦ < l < 83◦, −3◦ < b < 4◦ – 44 –

582 of the background, which translates in excess values of Φ(Θ) at all angular scales but no

583 significant structure is observed. On the contrary, the IceCube 40-string analysis reveals

584 an under-fluctuation within the analyzed area, showing a rather dispersed distribution of

585 events with respect to the average background case.

586 The conclusions from both the galactic plane scans and the correlation analysis are

◦ ◦ 587 therefore that the variations in the event density from the 11 × 7 region analyzed are

588 consistent with background fluctuations.

589 The under-fluctuation observed in IceCube 40-strings provides restrictive upper limits

590 to the neutrino emission above 500 GeV from the Cygnus region. Upper limits have

−2.6 591 been computed for a representative E power-law model under a point-like source

592 case. That is, assuming that the correlation of signal events in the region is given only

593 by the PSF of the analysis. The upper limits from the Cygnus region are at the level of

−11 −1 −2 −1 −2.6 594 3 × 10 T eV cm s for an E spectrum without energy cutoff below 10 TeV.

595 5.0.6. Cygnus X-3

596 The neutrino events coming from the direction of Cyg X-3 are found to be completely

597 compatible with pure background. No hint for a signal is found in the neutrino sample.

598 The smallest pre-trial p-value is 22.3% (coming from the search with window 1). After

599 correction for trials, we arrive at a probability of ∼ 57% that this observation occurs in a

600 background only sample (final p-value). Upper limits on neutrino emission from Cyg X-3

601 during and up to 20 days before or after the time windows have been calculated using the

602 method proposed by Feldman & Cousins (Feldman & Cousins 1998) and can be looked up

603 in Table 4. Figure 15 shows the neutrino events close to Cyg X-3 as a function of time. – 45 –

Fig. 15.— Arrival times of neutrino events reconstructed within 5◦ from Cyg X-3 and the position of the shifted search windows, each window being shifted individually. The height of the line depicts the log10 of the spatial event weight (higher line means closer to the source). The dashed lines show the unshifted positions of the windows. The red arrows indicate the shift. – 46 –

604 6. Conclusions

605 In this paper we present a dedicated search for high energy neutrino events from

606 the Galaxy. The tests have been realized on the data collected by the IceCube and the

607 AMANDA neutrino telescope. IceCube was not jet completed and functioned in 22 and

608 40-strings configuration. We have performed a scan of the Galactic Plane with the aim to

609 study the spatial properties of neutrino events observed in the part of the Milky accessible

610 from the northern hemisphere. The Galactic Plane was sampled and the data fitted to a

611 model of point-like emission. Although no significant local clustering of events has been

612 observed, IceCube’s ability to observe the whole northern hemisphere without the need of

613 pointing provides an unprecedented description of the large-scale neutrino emission from

614 the Galaxy. Departures from the single point-like source have been tested in the Cygnus

615 region, yielding very restrictive upper limits which are already at the level of ∼1 Crab with

616 IceCube 40-strings. This is only a factor 2 above what it is expected in neutrinos if all the

617 TeV gamma-rays observed in the region have an hadronic origin, implying that in the next

618 years IceCube results will have the potential to constraint the nature of the gamma-ray

619 emission in one of the most active parts of the Galaxy, or to see something significant.

620 Finally, a dedicated time optimized search from the direction of the binary system Cygnus

621 X3 has been realized. The analysis realized on this first combined sample is in various

622 aspects a pioneering analysis. For the first time, a nested detector has been operated inside

623 a more spaced one. We have investigated first the capability to improve the performance

624 towards low energy and tested new search strategies. We stress here that on the base of the

625 demonstration of the fact that a nested array permits a significant improvement at lower

626 energies with respect to the spaced array alone, the design study of DeepCore started. – 47 –

627 7. Acknowledgments

628 We acknowledge the support from the following agencies: U.S. National Science

629 Foundation Office of Polar Programs, U.S. National Science Foundation Physics Division,

630 University of Wisconsin Alumni Research Foundation, the Grid Laboratory Of Wisconsin

631 (GLOW) grid infrastructure at the University of Wisconsin - Madison, the Open

632 Science Grid (OSG) grid infrastructure; U.S. Department of Energy, and National

633 Energy Research Scientific Computing Center, the Louisiana Optical Network Initiative

634 (LONI) grid computing resources; National Science and Engineering Research Council of

635 Canada; Swedish Research Council, Swedish Polar Research Secretariat, Swedish National

636 Infrastructure for Computing (SNIC), and Knut and Alice Wallenberg Foundation, Sweden;

637 German Ministry for Education and Research (BMBF), Deutsche Forschungsgemeinschaft

638 (DFG), Research Department of Plasmas with Complex Interactions (Bochum), Germany;

639 Fund for Scientific Research (FNRS-FWO), FWO Odysseus programme, Flanders Institute

640 to encourage scientific and technological research in industry (IWT), Belgian Federal

641 Science Policy Office (Belspo); University of Oxford, United Kingdom; Marsden Fund,

642 New Zealand; Japan Society for Promotion of Science (JSPS); the Swiss National Science

643 Foundation (SNSF), Switzerland; A. Groß acknowledges support by the EU Marie Curie

644 OIF Program; J. P. Rodrigues acknowledges support by the Capes Foundation, Ministry of

645 Education of Brazil. – 48 –

646 REFERENCES

647 Neyman,J. , Pearson, E. (1933). On the Problem of the Most Efficient Tests of Statistical

648 Hypotheses. Philosophical Transactions of the Royal Society of London. Series A,

649 231: 289337

650 Resconi, E. et al. Astron.Astrophys. 502 (2009) 499-504

651 Kappes, A., Hinton, J., Stegmann, C., & Aharonian, F. A. 2007, ApJ, 656, 870

652 Hester, J. J. 2008, ARA&A, 46, 127

653 IceCube Collaboration, Abbasi, R., Abdou, Y., et al. 2011, arXiv:1106.3484

654 Aharonian, F. A., Krawczynski, H., Puehlhofer, G., Rowell, G. P., HEGRA Collaboration

655 2000, Bulletin of the American Astronomical Society, 32, 1239

656 Abdo, A. A., Ackermann, M., Ajello, M., et al. 2009, ApJ, 701, L123

657 Begelman, M. C., Hatchett, S. P., McKee, C. F., Sarazin, C. L., Arons, J. 1980, ApJ, 238,

658 722

659 Schneider, N., Bontemps, S., Simon, R., et al. 2006, A&A, 458, 855

660 Le Duigou, J.-M., & Kn¨odlseder,J. 2002, A&A, 392, 869

661 Sestayo, Y., & Resconi, E. submitted

662 Jogler, T., et al. 2008, Journal of Physics Conference Series, 120, 062019

663 Sierpowska-Bartosik, A., & Torres, D. F. 2009, ApJ, 693, 1462

664 Kelner, S. R., Aharonian, F. A., & Bugayov, V. V. 2006, Phys. Rev. D, 74, 034018

665 Kelner, S. R., & Aharonian, F. A. 2008, Phys. Rev. D, 78, 034013 – 49 –

666 Abbasi, R., et al. 2011, ApJ, 732, 18

667 Kelner, S. R., & Aharonian, F. A. 2008,Phys. Rev. D, 78, 034013

668 Kelner, S. R., Aharonian, F. A., & Bugayov, V. V. 2006, Phys. Rev. D, 74, 034018

669 Hillas, A. M. 2005, Journal of Physics G Nuclear Physics, 31, 95

670 Tavani, M., et al. 2009, Nature, 462, 620

671 Corbel, S., & Fermi LAT collaboration 2010, Bulletin of the American Astronomical Society,

672 42, 692

673 Aharonian, F., et al. 2007, A&A, 467, 1075

674 Ohm, S., et al. 2010, 25th Texas Symposium on Relativistic Astrophysics,

675 Marcowith, A., Komin, N., Gallant, Y. A.,& et al. 2008, International Cosmic Ray

676 Conference, 2, 787

677 Gallo, E., Fender, R., Kaiser, C., Russell, D., Morganti, R., Oosterloo, T., & Heinz, S. 2005,

678 Nature, 436, 819

679 Margon, B. 1984, ARA&A, 22, 507

680 Lozinskaya, T. A., Pravdikova, V. V., & Finoguenov, A. V. 2002, Astronomy Letters, 28,

681 223

682 Klepach, E. G., Ptuskin, V. S., & Zirakashvili, V. N. 2000, Astroparticle Physics, 13, 161

683 Heinz, S., & Sunyaev, R. 2002, A&A, 390, 751

684 Bosch-Ramon, V. 2007, Ap&SS, 309, 321

685 Aharonian, F., et al. 2007, A&A, 464, 235 – 50 –

686 Aharonian, F., et al. 2006, A&A, 457, 899

687 Aharonian, F., et al. 2004, ApJ, 614, 897

688 Albert, J., et al. 2007, A&A, 474, 937

689 Mart´ı,J., Paredes, J. M., & Peracaula, M. 2001, A&A, 375, 476

690 Mirabel, I. F., & Rodr´ıguez,L. F. 1998, Nature, 392, 673

691 Brinkmann, W., Aschenbach, B., & Kawai, N. 1996, A&A, 312, 306

692 Dubner, G. M., Holdaway, M., Goss, W. M., & Mirabel, I. F. 1998, AJ, 116, 1842

693 Abbasi, R., et al. 2010, Nuclear Instruments and Methods in Physics Research A, 618, 139

694 Ackermann, M., et al. 2006, Nuclear Instruments and Methods in Physics Research A, 556,

695 169

696 Braun, J., Dumm, J., de Palma, F., Finley, C., Karle, A., & Montaruli, T. 2008,

697 Astroparticle Physics, 29, 299

698 Braun, J., Baker, M., Dumm, J. et al., 2010, Astropart. Phys., 33, 175, arXiv:0912.1572

699 Aharonian, F., et al. 2005, A&A, 439, 635

700 Saito, T. Y., et al. 2009, arXiv:0907.1017

701 Hayashi, S., et al. 2009, Astroparticle Physics, 32, 112

702 Reynoso, M. M., Romero, G. E., & Christiansen, H. R. 2008, MNRAS, 387, 1745

703 Abdo, A. A., et al. 2009, ApJ, 706, L1

704 Kennel, C. F., & Coroniti, F. V. 1984, ApJ, 283, 710 – 51 –

705 Fermi, E. 1949, Physical Review , 75, 1169

706 Fermi, E. 1954, ApJ, 119, 1

707 Pooley, G., private communication

708 Abbasi, R (The IceCube Collaboration), submitted in 2011 (astro-ph/1109.6096v1)

709 Abdo, A. A., Allen, B., Berley, D., et al. 2007, ApJ, 658, L33

710 Aliu, E. for the VERITAS coll. 32nd ICRC, Beijing 2011

711 Albert, J., Aliu, E., Anderhub, H., et al. 2007, ApJ, 664, L87

712 Migliari, S., Fender,R., & M´endez,M. 2002, Science, 297, 1673

713 Ahrens, J. (The IceCube Collaboration) 2004, Nuclear Instruments and Methods in Physics

714 Research A, 524, 169

715 Abbasi, R. (The IceCube Collaboration) 2009, ApJ, 701, L47

716 Abdo, A., Ackermann, M., Ajello, M. et al. (Fermi LAT coll.), 2009, Science, 326, 1512

717 Barthelmy, S. D., Barbier, L. M., Cummings, J. R. et al., 2005, Space Sci. Rev., 120, 143

718 Bednarek, W. 2005, ApJ, 631, 466

719 Bednarek, W. 2010, MNRAS, 406, 689

720 Bonnet-Bidaud, J., &Chardin, G. 1988, Phys. Rep., 170, 325

721 Feldman, G. & Cousins, R.1998, Phys. Rev. D, 57, 3873

722 Koljonen, K., Hannikainen, D.,McCollough, M. et al., 2010, MNRAS, 406, 307,

723 arXiv:1003.4351 – 52 –

724 Levine, A. M., Bradt, H., Cui, W. et al., 1996, ApJ, 469, L33

725 Pooley, G. G. & Fender, R. P. 1997, MNRAS, 292, 925

726 Szostek, A., Zdziarski, A. A. &McCollough, M. L. 2008, MNRAS, 388, 1001

727 Tavani, M., Bulgarelli, A.,Piano, G. et al. (AGILE coll.), 2009, Nature, 462, 620

728 Trushkin, S. 1998, Astron. Lett., 24,16

729 Trushkin, S. A., Bursov, N. N., Kotani, T. et al., 2007, in 6th Integral Workshop, ESA

730 Special Publication, 622, 357, arXiv:astro-ph/0702393

731 Trushkin, S. A., Nizhelskij, N. N. & Sotnikova, J. V. 2008, Astronomer’s Telegram, ATel

732 #1483, http://www.astronomerstelegram.org/?read=1483

733 Tudose, V. et al., 2007, MNRAS, 375, L11

734 Zwart, J., Barker, R., Biddulph, P. et al., 2008, MNRAS, 391, 1545

735 Conrad, J., Botner, O., Hallgren, A., & P´erezde Los Heros, C. 2003, Phys. Rev. D, 67,

736 012002

737 Hill, G. C. 2003, Phys. Rev. D, 67, 118101

738 Abbasi, R. (The IceCube Collaboration) 2011 submitted to ApJ,

739 http://arxiv.org/abs/1108.3023

740 Abbasi, R. (The IceCube Collaboration) 2011 submitted to ApJ,

741 http://arxiv.org/abs/1106.3484

This manuscript was prepared with the AAS LATEX macros v5.2.