Search for point sources of cosmic neutrinos with ANTARES J. P. Gómez-González IFIC (CSIC-Universitat de València) The ANTARES detector [1] is currently the largest deep-sea neutrino telescope in the Northern Hemisphere. It consists of a three-dimensional array of 885 photomultiplier tubes which detect the Cherenkov light induced by the charged particles produced in the interaction of cosmic neutrinos with the matter surrounding the detector. The trajectories of the resulting muons are reconstructed with high precision, revealing the direction of the incoming neutrinos. The main scientific goal of ANTARES is the detection and identification of astrophysical sources of high energy neutrinos. This contribution describes such a search using the data collected in two years of detector operation, and a dedicated clustering algorithm based on the analytical maximization of the likelihood of the events. INTRODUCTION DATA SELECTION AND TRACK RECONSTRUCTION The ANTARES neutrino telescope started data taking in 2007 and it is fully operational since May The detector consists of 12 detection lines anchored to the seabed at a depth of 2475 m and sustained vertically by means of buoys. Each line has 25 storeys composed by a triplet of photo-multiplier tubes housed in pressure resistant glass spheres called optical modules (OMs), The OMs are facing downward at 45 from the vertical for an increased detection- efficiency for up-going neutrinos. Two different searches were conducted: In the candidate list search we looked for correlation of events with (both galactic and extra- galactic) sources known to be TeV gamma-ray emitters. In this case, the lowest p-value corresponds to HESS J , with a (post-trial) probability of 18%. For this cluster, the maximum likelihood is Q = 4.32, and the number of events estimated by the algorithm is In the all sky survey the largest Q value found was 7.9. This value, or a larger one, occurs in about the 90% of the background-only experiments. No statistically significant excess of events has been found neither in the search using a candidate list of interesting sources, nor in the full sky search. Using a different event selection the results obtained agree with an independent analysis which used a search method based on the likelihood of the events. Upper limits for the 24 candidate sources using the latter method and the Feldman-Cousins prescription were submitted for publication. Cosmic accelerator reach the detector, not deflected  absorbed by matter and EBL p deflected by magnetic fields, GZK effect CMB Earth Data runs were selected requiring 75%-80% of the detector elements to be active when averaging over the full run. The total livetime of the analysis is 304 days; 144 days correspond to data taken with the 5 line configuration, while for the remaining 160 days the detector consisted of 9, 10 and 12 lines. The reconstruction method [3] used is an online algorithm that performs multiple fitting steps of increasing sophistication. The final step consists on the likelihood maximization of the observed hit times as a function of the muon direction and position. The goodness of the track reconstruction is described by the log- likelihood of the fitted muon. Distribution of the quality of the reconstruction parameter for upgoing events. The contribution from the different components of background is included. The simulation reproduces well the data. CLUSTERING METHOD AND RESULTS The final sample of neutrino candidates was chosen on the basis of a selection criteria that resulted to yield the optimal the sensitivity, or expected median upper limit on the neutrino flux. A total of 2190 upgoing events, reconstructed with  >-5.4 and an angular uncertainty <1  were selected. The problem of the search for point sources is the identification of clusters of signal events, which are originated independently and appear spread over a non-Gaussian distribution made up by the atmospheric neutrino and atmospheric muon events that constitute the physical background for a neutrino telescope. The search method used in this analysis is the Expectation-Maximization clustering algorithm [4], which is a general approach to maximum-likelihood estimation where the different groups of data involved are described by different density components. This algorithm works in two steps. In the first step the expected value of the complete data log-likelihood is computed for a given set of parameters. In the Maximization step a new set of parameters that maximizes the likelihood is found, and then, the test statistic is computed. The two main parameters describing the performance of a neutrino telescope are the angular resolution and the effective area. Both parameters are estimated from simulation. Left plot shows the cumulative angular resolution for an E^-2 neutrino spectrum and for events complying the final selection criteria. The median value of the reconstruction error being 0.5  0.1 . This uncertainty incorporates all the effects leading to a deterioration of the detector timing resolution. In addition, the uncertainty on the absolute orientation of the detector, which is estimated to be of the order of 0.1 , is also taken into account in the sensitivity computation. The plot on the right shows averaged effective area as a function of the neutrino energy for a cosmic neutrino spectrum and considering the different detector configurations. Based on the agreement between data and simulations a 15% systematic reconstruction error on the detection efficiency has been considered for the limits calculation. 32 st International Cosmic Ray Conference, Beijing (China) August July 2011 [1] M. Ageron et. al., ANTARES: the first undersea neutrino telescope, 2011, Nucl. Instr. Meth., 2011 (Accepted), arXiv: v1. [2] J. A. Aguilar et al., Time Calibration of the ANTARES neutrino Telescope., Astroparticle Physics, 2011, 9, 227. [3] A. Heijboer, Track reconstruction and point source searches with ANTARES, PhD thesis, Universitet van Amsterdam, 2004 [4] A. P. Dempster, N. M. Laird, and D. B. Rubin. Maximum likelihood from incomplete data via the EM algorithm, J. Royal Statistical Soc. Series B, 38:1-38, 1977 ANTARES is a three-dimensional photo-detector array which detects the Cherenkov light emitted by the charged leptons originated in the interaction of high energy neutrinos with the matter surrounding the detector. The tracks of the produced muons can be reconstructed using the position and timing information of the hits arriving to the PMTs. An accurate timing and position calibration [2] of the detector OMs is necessary in order to achieve the best attainable angular resolution. The main advantages of the neutrinos with respect to other particles are that they are not deflected by magnetic fields and are weakly interacting so that they can travel long distances. The drawback is that the weak interactions of neutrinos imply that a very massive detector is required to observe a measurable flux from astrophysical sources. Astrophysical sources of interest are expected to produce multi-messenger signatures as cosmic rays, gamma-rays, and neutrinos. The equatorial coordinates of the 24 sources used in the candidate list search. The corresponding Q value, the number of signal events fitted by the algorithm, as well as the (pre-trial) p-values are also shown. Skymap with the equatorial coordinates of the 2190 events selected. The position Of the 24 sources selected in the candidate search are also included.