Comparison between time-domain and frequency-domain Bayesian inferences to inspiral-merger-ringdown gravitational-wave signals

Abstract

Time-domain (TD) Bayesian inference is important in ringdown analysis for gravitational wave (GW) astronomy. The validity of this method has been well studied by Isi and Farr [1]. Using GW190521 as an example, we study the TD method in detail by comparing it with the frequency-domain (FD) method as a complement to previous study. We argue that the autocovariance function (ACF) should be calculated from the inverse fast Fourier transform of the power spectral density (PSD), which is usually estimated by the Welch method. In addition, the total duration of the GW data that are used to estimate the PSD and the slice duration of the truncated ACF should be long enough. Only when these conditions are fully satisfied can the TD method be considered sufficiently equivalent to the FD method.