Identifying strongly lensed gravitational waves through their phase consistency

Abstract: Strongly lensed gravitational waves (GWs) from binary coalescence manifest as repeated chirps from the original merger. At the detectors, the phase of the lensed GWs and its arrival time differences will be consistent modulo a fixed constant phase shift. We develop a fast and reliable method to efficiently reject event pairs that are not-lensed copies and appropriately rank the most interesting candidates. Our method exploits that detector phases are the best measured GW parameter, with errors only of a fraction of a radian and differences across the frequency band that are better measured than the chirp mass. The arrival time phase differences also avoid the shortcomings of looking for overlaps in highly non-Gaussian sky maps. Our basic statistic determining the consistency with lensing is the distance between the phase posteriors of two events and it directly provides information about the lens-source geometry which helps inform electromagnetic followups. We demonstrate that for simulated signals of not-lensed binaries specifically chosen with many coincident properties so as to trigger false lensing alarms none of the pairs have phases closer than $3\sigma$, and most cases reject the lensing hypothesis by $5\sigma$. Looking at the latest catalog, GWTC3, we find that only $6\%$ of the pairs are consistent with lensing at 99\% confidence level. Moreover, we reject about half of the pairs that would otherwise favor lensing by their parameter overlaps and demonstrate good correlation with detailed joint parameter estimation results. This reduction of the false alarm rate will be of paramount importance in the upcoming observing runs and the eventual discovery of lensed GWs. Our code is publicly available and could be applied beyond lensing to test possible deviations in the phase evolution from modified theories of gravity and constrain GW birefringence.