Speaker
Description
A gravitational wave (GW) background (or foreground) emerges from the sum of individual GWs emitted by a population of sources. The discreteness of the population induces non-Gaussian signatures in the statistics of that signal. Therefore, detecting deviations from a Gaussian statistics helps supporting the astrophysical origin of the observed signal. In this talk, we will review Bayesian methods to detect non-Gaussianities using variance mixture distributions. We show that for some families of distributions, the excess kurtosis can be used directly as a physical parameter of the model and the likelihood reduces to a Gaussian (Whittle) likelihood when excess kurtosis goes to zero. The model is tested with both Pulsar Timing Array simulated data for the supermassive black hole GW background, and with LISA simulated data, for the galactic binary GW foreground.