Improving BayesWave waveform reconstructions using informed priors

BayesWave is an algorithm that uses a Morlet-Gabor wavelet frame to model instrumental transients (glitches) and gravitational wave bursts. The BayesWave algorithm employs transdimensional Bayesian inference to find the optimal number of wavelets and wavelet parameters to fit the data. The current model makes no assumptions about the overall amplitude or phase evolution of the signals, while physically we expect the gravitational wave signals generated by astrophysical systems to exhibit a smooth phase and amplitude evolution. We show that imposing a prior that penalizes non-smooth phase evolution, improves waveform reconstructions, especially for low signal-to-noise sources.