Abstract: Context. At GeV energies, the sky is dominated by the interstellar emission from the Galaxy. With limited statistics and spatial resolution, accurately separating point sources is therefore challenging. Aims. Here we present the first application of deep learning based algorithms to automatically detect and classify point sources from gamma-ray data. For concreteness we refer to this approach as AutoSourceID. Methods. To detect point sources, we utilized U-shaped convolutional networks for image segmentation and k-means for source clustering and localization. We also explored the Centroid-Net algorithm, which is designed to find and count objects. Using two algorithms allows for a cross check of the results, while a combination of their results can be used to improve performance. The training data are based on 9.5 years of exposure from The Fermi Large Area Telescope (Fermi-LAT) and we used source properties of active galactic nuclei (AGNs) and pulsars (PSRs) from the fourth Fermi-LAT source catalog in addition to several models of background interstellar emission. The results of the localization algorithm are fed into a classification neural network that is trained to separate the three general source classes (AGNs, PSRs, and FAKE sources). Results. We compared our localization algorithms qualitatively with traditional methods and find them to have similar detection thresholds. We also demonstrate the robustness of our source localization algorithms to modifications in the interstellar emission models, which presents a clear advantage over traditional methods. The classification network is able to discriminate between the three classes with typical accuracy of similar to 70%, as long as balanced data sets are used in classification training. We published online our training data sets and analysis scripts and invite the community to join the data challenge aimed to improve the localization and classification of gamma-ray point sources.

Abstract: Research in many areas of modern physics such as, e. g., indirect searches for dark matter and particle acceleration in supernova remnant shocks rely heavily on studies of cosmic rays (CRs) and associated diffuse emissions (radio, microwave, X-rays, gamma-rays). While very detailed numerical models of CR propagation exist, a quantitative statistical analysis of such models has been so far hampered by the large computational effort that those models require. Although statistical analyses have been carried out before using semi-analytical models (where the computation is much faster), the evaluation of the results obtained from such models is difficult, as they necessarily suffer from many simplifying assumptions. The main objective of this paper is to present a working method for a full Bayesian parameter estimation for a numerical CR propagation model. For this study, we use the GALPROP code, the most advanced of its kind, which uses astrophysical information, and nuclear and particle data as inputs to self-consistently predict CRs, gamma-rays, synchrotron, and other observables. We demonstrate that a full Bayesian analysis is possible using nested sampling and Markov Chain Monte Carlo methods (implemented in the SuperBayeS code) despite the heavy computational demands of a numerical propagation code. The best-fit values of parameters found in this analysis are in agreement with previous, significantly simpler, studies also based on GALPROP.

Abstract: A multimessenger analysis optimized for a correlation of arrival directions of ultra-high energy cosmic rays (UHECRs) and neutrinos is presented and applied to 2190 neutrino candidate events detected in 2007-2008 by the ANTARES telescope and 69 UHECRs observed by the Pierre Auger Observatory between 2004 January 1 and 2009 December 31. No significant correlation is observed. Assuming an equal neutrino flux (E-2 energy spectrum) from all UHECR directions, a 90% CL upper limit on the neutrino flux of 5.0 x 10(-8) GeV cm(-2) s(-1) per source is derived.

Abstract: We present the results of searches for point-like sources of neutrinos based on the first combined analysis of data from both the ANTARES and IceCube neutrino telescopes. The combination of both detectors, which differ in size and location, forms a window in the southern sky where the sensitivity to point sources improves by up to a factor of 2 compared with individual analyses. Using data recorded by ANTARES from 2007 to 2012, and by IceCube from 2008 to 2011, we search for sources of neutrino emission both across the southern sky and from a preselected list of candidate objects. No significant excess over background has been found in these searches, and flux upper limits for the candidate sources are presented for E-2.5 and E-2 power-law spectra with different energy cut-offs.

Abstract: In past years the IceCube Collaboration has reported the observation of astrophysical high-energy neutrino events in several analyses. Despite compelling evidence for the first identification of a neutrino source, TXS 0506+056, the origin of the majority of these events is still unknown. In this paper, we search for a possible transient origin of the IceCube astrophysical events using neutrino events detected by the ANTARES telescope. The arrival time and direction of 6894 track-like and 160 shower-like events detected over 2346 days of livetime are examined to search for coincidences with 54 IceCube high-energy track-like neutrino events, by means of a maximum likelihood method. No significant correlation is observed and upper limits on the one-flavor neutrino fluence from the direction of the IceCube candidates are derived. The nonobservation of time and space correlation within the time window of 0.1 days with the two most energetic IceCube events constrains the spectral index of a possible point-like transient neutrino source to be harder than -2.3 and -2.4 for each event, respectively.