Including Gravitational Waveform Uncertainty in Bayesian Inference of Source Properties

Gravitational-wave models for the inspiral and merger of black holes have intrinsic uncertainties, for example arising from approximations of general relativity used to generate waveforms. Differences between models with the same astrophysical parameters can be quantified as corrections to the frequency domain amplitude and phase. In this work, we present updated results with spline inspired phenomenological models to add variability to a baseline waveform model during parameter estimation runs. The positions of the nodes in frequency space are variable and are informed by expected detector sensitivities. The prior distributions of the correction parameters change with the frequency nodes and are motivated by the faithfulness of different waveform models with each other in the signal's posterior parameter space. We demonstrate that injected waveform modifications can be recovered, and that adding waveform variability during parameter estimation runs can reduce systematic bias in recovered astrophysical parameters.