[Paper Review] Searching for quasar candidates with periodic variations from the Zwicky Transient Facility: results and implications

Yongjie Chen, Shuo Zhai+8 more|arXiv (Cornell University)|Jun 23, 2022

Bayesian Methods and Mixture Models4 citations

TL;DR

This study conducts a systematic search for quasars with periodic variability using photometric data from the Zwicky Transient Facility (ZTF), cross-matching with SDSS and Véron-Cetty & Véron quasar catalogs. Applying generalized Lomb-Scargle and auto-correlation methods, it identifies 86 high-significance periodic candidates using a single power-law red-noise model, with only three candidates showing consistent quasi-periodic behavior over 6–8 cycles, suggesting that quasar variability is likely quasi-periodic or stochastic rather than strictly periodic, challenging current detection methods.

ABSTRACT

We conduct a systematic search for quasars with periodic variations from the archival photometric data of the Zwicky Transient Facility by cross-matching with the quasar catalogs of the Sloan Digital Sky Survey and V{é}ron-Cetty \& V{é}ron. We first select out 184 primitive periodic candidates using the generalized Lomb-Scargle periodogram and auto-correlation function and then estimate their statistical significance of periodicity based on two red-noise models, i.e., damped random walk (DRW) and single power-law (SPL) models. As such, we finally identify 106 (DRW) and 86 (SPL) candidates with the most significant periodic variations out of 143,700 quasars. We further compare DRW and SPL models using Bayes factors, which indicate a relative preference of the SPL model for our primitive sample. We thus adopt the candidates identified with SPL as the final sample and summarize its basic properties. We extend the light curves of the selected candidates by supplying other archival survey data to verify their periodicity. However, only three candidates (with 6-8 cycles of periods) meet the selection criteria. This result clearly implies that, instead of being strictly periodic, the variability must be quasi-periodic or caused by stochastic red-noise. This exerts a challenge to the existing search approaches and calls for developing new effective methods.

Motivation & Objective

• To identify quasar candidates with periodic photometric variability in the ZTF survey data.
• To assess the statistical significance of periodicity in the presence of red-noise variability using robust models.
• To compare the performance of damped random walk (DRW) and single power-law (SPL) red-noise models in identifying true periodic signals.
• To verify periodicity by extending light curves with archival data from other surveys.
• To evaluate the implications for supermassive black hole binary detection and the limitations of current periodicity search methods.

Proposed method

• Cross-matching 223,061 quasars from SDSS DR14 and Véron-Cetty & Véron (2010) catalogs with ZTF photometric data using a 5 arcsec radius.
• Applying the generalized Lomb-Scargle periodogram and auto-correlation function to detect periodic signals in light curves.
• Estimating statistical significance of periodicity using two red-noise models: damped random walk (DRW) and single power-law (SPL).
• Using Bayes factors to compare model preference between DRW and SPL, favoring SPL for the primitive sample.
• Extending light curves with archival data from CRTS, PS1, CSS, and ATLAS surveys to verify periodicity over longer baselines.
• Applying selection criteria requiring at least 6–8 cycles of consistent periodicity to confirm true periodic behavior.

Experimental results

Research questions

• RQ1 What is the prevalence of periodic variability among quasars in the ZTF survey, and how many candidates can be robustly identified?
• RQ2 How do red-noise models (DRW vs. SPL) affect the statistical significance estimation of periodic signals in quasar light curves?
• RQ3 Which red-noise model provides a better fit to the observed quasar variability, and what does this imply for periodicity detection?
• RQ4 Can the periodicity of candidate quasars be confirmed over extended baselines using multi-survey archival data?
• RQ5 What are the implications of non-strictly periodic behavior for the detection of supermassive black hole binaries via periodic variability?

Key findings

• Out of 143,700 quasars, 106 candidates were identified as having significant periodicity using the damped random walk (DRW) red-noise model.
• 86 candidates were identified as having significant periodicity using the single power-law (SPL) red-noise model, which was statistically preferred over DRW based on Bayes factors.
• Only three candidates met the strict periodicity verification criteria, showing 6–8 cycles of consistent variation when light curves were extended with archival data.
• The low number of confirmed periodic candidates suggests that quasar variability is more likely quasi-periodic or driven by stochastic red noise rather than true periodicity.
• The results challenge existing periodicity search methods, highlighting the need for new techniques that account for quasi-periodic or red-noise-dominated variability.
• The SPL model is found to be a better fit for the primitive sample than the DRW model, indicating that the variability may not follow the canonical DRW process.