Rapid and Robust Parameter Inference for Binary Mergers

The detection rate for compact binary mergers has grown as the sensitivity of the global network of ground based gravitational wave detectors has improved, now reaching the stage where robust automation of the analyses is essential. Automated low-latency algorithms have been developed that send out alerts when candidate signals are detected. The alerts include sky maps to facilitate electromagnetic follow up observations, along with probabilities that the system might contain a neutron star, and hence be more likely to generate an electromagnetic counterpart. Data quality issues, such as loud noise transients (glitches), can adversely affect the low-latency algorithms, causing false alarms and throwing off parameter estimation. Here a new analysis method is presented that is robust against glitches, and capable of producing fully Bayesian parameter inference, including sky maps and mass estimates, in a matter of minutes. Key elements of the method are wavelet-based de-noising, penalized maximization of the likelihood during the initial search, rapid sky localization using pre-computed inner products, and heterodyned likelihoods for full Bayesian inference.