Unveiling Microlensing Biases in Testing General Relativity with Gravitational Waves

Gravitational waves (GW) from chirping binary black holes (BBHs) provide unique opportunities to test general relativity (GR) in the strong-field regime. However, testing GR can be challenging when incomplete physical modeling of the expected signal gives rise to systematic biases. In this study, we investigate the potential influence of wave effects in gravitational lensing (which we refer to as microlensing) on tests of GR using GWs for the first time. We utilize an isolated point-lens model for microlensing with the lens mass ranging from $10-10^5~$M$_\odot$ and base our conclusions on an astrophysically motivated population of BBHs in the LIGO-Virgo detector network. Our analysis centers on two theory-agnostic tests of gravity: the inspiral-merger-ringdown consistency test (IMRCT) and the parameterized tests. Our findings reveal two key insights: First, microlensing can significantly bias GR tests, with a confidence level exceeding $5\sigma$. Notably, substantial deviations from GR $(\sigma > 3)$ tend to align with a strong preference for microlensing over an unlensed signal, underscoring the need for microlensing analysis before claiming any erroneous GR deviations. Nonetheless, we do encounter scenarios where deviations from GR remain significant ($1 < \sigma < 3$), yet the Bayes factor lacks the strength to confidently assert microlensing. Second, deviations from GR correlate with pronounced interference effects, which appear when the GW frequency ($f_\mathrm{GW}$) aligns with the inverse time delay between microlens-induced images ($t_\mathrm{d}$). These false deviations peak in the wave-dominated region and fade where $f_\mathrm{GW}\cdot t_\mathrm{d}$ significantly deviates from unity. Our findings apply broadly to any microlensing scenario, extending beyond specific models and parameter spaces, as we relate the observed biases to the fundamental characteristics of lensing.