Title:Lomb-Scargle periodograms struggle with non-sinusoidal supermassive BH binary signatures in quasar lightcurves

Abstract:Supermassive black hole binary (SMBHB) systems are expected to form as a consequence of galaxy mergers. At sub-parsec separations, SMBHBs are difficult to resolve, but can be identified as quasars with periodic variability. Previous periodicity searches have identified statistically significant candidates, but focused primarily on sinusoidal signals. However, theoretical models and hydrodynamical simulations predict that binaries produce more complex non-sinusoidal pulse shapes. Here we examine the efficacy of the Lomb-Scargle periodogram (LSP; one of the most popular tools for periodicity searches in unevenly sampled lightcurves) to detect periodicities with a saw-tooth shape mimicking results of hydrodynamical simulations. We simulate quasar lightcurves with damped random walk (DRW) variability and inject periodic signals. Our mock sample of 12,400 quasars consists either of idealised well-sampled lightcurves, or mimics the data in the Palomar Transient Factory (PTF) analyzed in Charisi et al, 2016. We assess the statistical significance of recovering two types of periodic signals, i.e. with sinusoidal and sawtooth pulse shapes. We find that the LSP detects 39.1% and 28.1% of the sinusoidal signals, in the PTF-like and idealised lightcurves, respectively. The fraction is significantly reduced for sawtooth periodicity, with only 7.5% and 1.1% detected in PTF-like and idealised lightcurves, respectively. These low recovery rates imply that previous searches have missed the large majority of binaries. Therefore, significant improvements are required beyond simple LSPs to successfully uncover SMBHBs in upcoming time-domain surveys.