On possible proxies of AGN light-curves cadence selection in future time domain surveys

Abstract

Motivated by upcoming photometric and spectroscopic surveys (Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST), Manuakea Spectroscopic Explorer), we design the statistical proxies to measure the cadence effects on active galactic nuclei (AGN) variability-observables (time-lags, periodicity, and structure-function (SF)). We constructed a multiple-regression model to statistically identify the cadence-formal error pattern knowing AGN time-lags and periodicity from different surveys. We defined the simple metric for the SF’s properties, accounting for the ’observed’ SF’s deviation relative to those obtained from the homogenously-sampled light curves. We tested the regression models on different observing strategies: the optical dataset of long light-curves of eight AGN with peculiarities and the artificial datasets based on several idealized and LSST-like cadences. The SFs metric is assessed on synthetic datasets. The regression models (for both data types) predict similar cadences for time-lags and oscillation detection, whereas for light curves with low variability ($\sim 10\%$), cadences for oscillation detection differ. For higher variability ($\sim 20\%$), predicted cadences are larger than for $F_{var}\sim 10\%$. The predicted cadences are decreasing with redshift. SFs with dense and homogenous cadences are more likely to behave similarly. SFs with oscillatory signals are sensitive to the cadences, possibly impacting LSST-like operation strategy. The proposed proxies can help to select spectroscopic and photometric-surveys cadence strategies, and they will be tested further in larger samples of objects.

The broad emission line asymmetry in a low mass ratio of supermassive binary black holes on elliptical orbits

“…To simulate this effect, the so‐called red noise is superposed to the generated signal. The red noise is modeled using the approach discussed in Kovačević et al (2021) or the paper describing the PoSKI model. We added the red noise as 30% of the light curve maximum, although different values can be adopted.…”

Photoreverberation mapping of quasars in the context of Legacy Survey of Space and Time observing strategies

“…In order to test our implementation of the photoRM method, we perform several experiments with a set of 19 artificially generated pairs of light curves of the continuum and line emission with different time‐lags ranging from a few days to several months. The light curves were generated as described in Kovačević et al (2021b), and are further entangled using the prescription in Chelouche & Daniel (2012). The procedure involves the entanglement of the generated continuum and emission‐line light curves (eq.…”

“…The continuum light curves were generated using the damped random walk (DRW) model (Kelly et al 2009), with the characteristic amplitude and timescale inferred from the SMBH mass and/or AGN luminosity. The DRW model parameters, and , are calculated from the first principles: assuming a priori theoretical distributions of AGN luminosity, SMBH mass, and BLR radius as given in Kovačević et al (2021b). The simulated light curves have idealized cadence of 1 day over the interval of 5,000 days (13.7 years).…”

“…We shortly describe how the artificial light curves are simulated, for more details see Kovačević et al (2021b). The continuum light curves were generated using the damped random walk (DRW) model (Kelly et al 2009), with the characteristic amplitude and timescale inferred from the SMBH mass and/or AGN luminosity.…”

Conditional Neural Process for nonparametric modeling of active galactic nuclei light curves

“…There are several approaches for modeling AGN light curves, such as damped random walk (DRW, Kelly et al 2009; Kozlowski 2017; Sánchez‐Sáez et al 2018), Gaussian Process (GP) with different kernels (Kovačević et al 2019; Shapovalova et al 2017; Shapovalova et al 2019) or Autoencoders (Tachibana et al 2020). Initial application of GPs (Kovačević et al 2021) shows that machine learning can be useful for time lag and periodicity detection for light curves with cadences above 100 observations per 10 years. The optical variability of AGNs has been most commonly represented by a stochastic model based on DRW process (Kelly et al 2009).…”