Fast and effective localization of gravitational wave (GW) events could play a crucial role in identifying possible electromagnetic counterparts, and thereby help usher in an era of GW multi-messenger astronomy. We discuss an algorithm for accurate and very low latency ($<$ 1 second) localization of GW sources using only the relative times of arrival, relative phases, and relative signal-to-noise ratios for pairs of detectors. The algorithm is independent of distances and masses to leading order, and can be generalized to all discrete sources detected by ground-based detector networks. Our approach, while developed independently, is similar to that of BAYESTAR with a few modifications in the algorithm which result in increased computational efficiency. For the LIGO two detector configuration (Hanford+Livingston) expected in late 2015 we find a median 50\% (90\%) localization of 143 deg$^2$ (558 deg$^2$) for binary neutron stars (for network SNR threshold of 12, corresponding to a horizon distance of $\sim 130$ Mpc), consistent with previous findings. We explore the improvement in localization resulting from high SNR events, finding that the loudest out of the first 4 (or 10) events reduces the median sky localization area by a factor of 1.9 (3.0) for the case of 2 GW detectors, and 2.2 (4.0) for 3 detectors. We consider the case of multi-messenger joint detections in both the GW and the electromagnetic (EM) spectra. We specifically explore the case of independent, and possibly highly uncertain, localizations, showing that the joint localization area is significantly reduced. We also show that a prior on the binary inclination, potentially arising from GRB observations, has a negligible effect on GW localization. Our algorithm is simple, fast, and accurate, and may be of particular utility in the development of multi-messenger astronomy.
↧