Phase effects from strong gravitational lensing of gravitational waves

Assessing the probability that two or more gravitational waves (GWs) are lensed images of the same source requires an understanding of the image properties, including their relative phase shifts in strong lensing (SL). For non-precessing, circular binaries dominated by quadrupole radiation these phase shifts are degenerate with either a shift in the coalescence phase or a detector and inclination dependent shift in the orientation angle. This degeneracy is broken by the presence of higher harmonic modes with $|m| e 2$ in the former and $|m| e l$ in the latter. Precession or eccentricity will also break this degeneracy. This implies that lensed GWs will not necessarily be consistent with (unlensed) predictions from general relativity (GR). Therefore, unlike EM lensing, GW SL can lead to images with an observable modified phase evolution. However, for a wide parameter space, the lensed waveform is similar enough to an unlensed waveform that detection pipelines will still find it. For present detectors, templates with a shifted detector-dependent orientation angle have SNR differences of less than $1%$ for mass ratios up to 0.1, and less than $5%$ for precession parameters up to 0.5 and eccentricities up to 0.4 at 20Hz. The mismatch is lower than $10%$ with the alternative detector-independent coalescence phase shift. Nonetheless, for a loud enough source, even with one image it may be possible to directly identify it as a SL image from its non-GR waveform. In more extreme cases, lensing may lead to considerable distortions, and the lensed GWs may be undetected with current searches. Nevertheless, an exact template with a phase shift in Fourier space can always be constructed to fit any lensed GW. We conclude that an optimal search strategy would incorporate phase information in all stages, with an exact treatment in the final assessment of the probability of multiple lensed events.