Systematic and Statistical Uncertainties in the Characterization of Gravitational-Wave Sources

Abstract

With the increasing number of gravitational-wave (GW) detections made by LIGO-Virgo-Kagra during the first three observing runs, the field of GW astrophysics has a growing need for prompt and precise parameter estimation (PE) of GW sources. To better understand the limitations of the current PE practices and to develop a more robust approach to analyze future GW sources, this thesis explores the impact of various aspects of data analysis, including priors, waveform models, noise characterization, and instrumental calibration errors, on the final PE results. This thesis demonstrates that in the case of marginal signals, the choice of priors can greatly impact the PE results and the subsequent astrophysical interpretation, especially when population-informed priors are not yet available. As the detector sensitivity improves, two other sources of systematic errors become increasingly relevant: waveform approximants and instrumental calibration errors. In the second half of the thesis, we conclude that the current waveform approximants for neutron star-black hole mergers are unlikely to introduce systematic errors comparable to the statistical uncertainties for sources that can be detected with current and near-future detector sensitivity. In terms of the calibration errors, we show that they will not impede the standard siren measurement of the Hubble constant in the coming decades. This thesis examines the impact of various data analysis choices on the final results with extensive PE runs unparalleled in the previous literature and can continue to provide valuable guidance for PE analysis for future generations of GW detectors.