Quasi-periodic oscillations in optical color evolutions to support sub-pc binary black hole systems in broad line active galactic nuclei

TL;DR

The paper tackles the problem of distinguishing optical QPOs in BLAGN arising from sub-pc BBHs from spurious QPOs caused by intrinsic red noise. It proposes a color-evolution approach using two-band light curves and a CAR(1) model for intrinsic variability, with Lomb-Scargle analysis to assess periodicity and a Monte Carlo framework to estimate false-alarm probabilities. It finds that single-band QPOs due to red noise occur with probability about $P_1 \approx 3.1\times10^{-2}$, while color QPOs have $P_2 < 3.3\times10^{-7}$, implying >$5\sigma$ confidence when a color QPO is detected. The method is validated on SDSS J1609+1756 (no color QPO detected) and is proposed for application to ZTF and LSST to enable robust searches for BBH signatures in BLAGN.

Abstract

Optical quasi-periodic oscillations (QPOs) with periodicity around hundreds to thousands of days have been accepted as an efficient indicator for sub-pc binary black hole systems (BBHs) in broad line active galactic nuclei (BLAGN). However, considering intrinsic variability (red noises) of BLAGN, it is still an open question on physical origin of detected optical QPOs from AGN variability or truly from sub-pc BBHs. Here, a simple method is proposed to support optical QPOs related to sub-pc BBHs by detecting QPOs in time dependent optical color evolutions of BLAGN. Periodic variations of obscurations are expected on optical light curves related to sub-pc BBHs, but there should be non-periodic variable obscurations on optical light curves in normal BLAGN. Through simulated optical light curves for intrinsic AGN variability by Continuous AutoRegressive process with time durations around 2800days (similar as time durations of light curves in ZTF), the probability is definitely smaller than $3.3\times10^{-7}$ that QPOs can be detected in the corresponding optical color evolutions, but about $3.1\times10^{-2}$ that optical QPOs with periodicity smaller than 1400days (at least two cycles) can be detected in the simulated single-band light curves. Therefore, confidence level is definitely higher than 5$σ$ to support the QPOs in color evolutions not related to intrinsic AGN variability but truly related to sub-pc BBHs. In the near future, the proposed method can be applied for searching reliable optical QPOs in BLAGN through multi-band light curves from the ZTF and the upcoming LSST.

Figures (4)

• Figure 1: On the correlation between SDSS g- and r-band apparent magnitudes of 3530 low redshift quasars. Solid and dashed red lines show the best fitting results and corresponding 5$\sigma$ confidence bands.
• Figure 2: Left panels show the light curves of $LC_{1,0}(t)$ (in blue) and $LC_{2,0}(t)$ (in red) (top left panel), $LC_{1}(t)$ (in blue) and $LC_{2}(t)$ (in red) (middle left panel) and $Color(t)$ (bottom left panel). Right panels show the corresponding LS powers. In top right panel (middle right panel), solid line in blue and in red show the results through $LC_{1,0}(t)$ ($LC_{1}(t)$) and $LC_{2,0}(t)$ ($LC_{2}(t)$), respectively. In each right panel, horizontal dashed red line marks the 5$\sigma$ significance level (determined by the input false alarm probability to be $3\times10^{-7}$), vertical dashed red line marks the position of periodicity around 812days determined in $LC_{1}(t)$.
• Figure 3: Top panel shows the ZTF g-band (blue symbols) and r-band (red symbols) light curves of SDSS J1609+1756. Middle panel shows the $CL(t)$. Bottom panel shows the determined LS powers of $CL(t)$, with horizontal dashed red line as 5$\sigma$ significance level.
• Figure 4: Distributions of the $\tau$, $\sigma$, $\Delta t$ and $\sigma_t$ of the $N_1=94527$ artificial $LC_1(t)$ including fake QPOs.