부산대학교 도서관

Most tests of general relativity with gravitational-wave observations rely on inferring the degree to which a signal deviates from general relativity in conjunction with the astrophysical parameters of its source, such as the component masses and spins of a compact binary. Due to features of the signal, measurements of these deviations are often highly correlated with the properties of astrophysical sources. As a consequence, prior assumptions about astrophysical parameters will generally affect the inferred magnitude of the deviations. Incorporating information about the underlying astrophysical population is necessary to avoid biases in the inference of deviations from general relativity. Current tests assume that the astrophysical population follows an unrealistic fiducial prior chosen to ease sampling of the posterior -- for example, a prior flat in component masses -- which is is inconsistent with both astrophysical expectations and the distribution inferred from observations. We propose a framework for fortifying tests of general relativity by simultaneously inferring the astrophysical population using a catalog of detections. Although this method applies broadly, we demonstrate it concretely on massive graviton constraints and parameterized tests of deviations to the post-Newtonian phase coefficients. Using observations from LIGO-Virgo-KAGRA's third observing run, we show that concurrent inference of the astrophysical distribution strengthens constraints and improves overall consistency with general relativity. We provide updated constraints on deviations from the theory, finding that, upon modeling the astrophysical population, the 90\%-credible upper limit on the mass of the graviton improves by $25\%$ to $m_g \leq 9.6 \times 10^{-24}\, \mathrm{eV}/c^2$ and the inferred population-level post-Newtonian deviations move ${\sim} 0.4 \sigma$ closer to zero.
Comment: 20 pages, 11 figures