Variability, periodicity and contact binaries in WISE

TL;DR

This study analyzes WISE infrared light curves to identify and classify ~56,000 periodic variables, especially contact binaries, demonstrating efficient methods for variable classification that can be used in future surveys.

Contribution

It introduces computationally inexpensive methods for classifying periodic variables using infrared and Gaia data, improving efficiency over traditional periodogram analysis.

Findings

Identified ~56,000 periodic variables from WISE data.

Non-parametric methods are as effective as periodograms for classification.

Proposed methods reduce computational costs for future surveys.

Abstract

The time-series component of WISE is a valuable resource for the study of variable objects. We present an analysis of an all-sky sample of ~450,000 AllWISE+NEOWISE infrared light curves of likely variables identified in AllWISE. By computing periodograms of all these sources, we identify ~56,000 periodic variables. Of these, ~42,000 are short-period (P<1 day), near-contact or contact eclipsing binaries, many of which are on the main sequence. We use the periodic and aperiodic variables to test computationally inexpensive methods of periodic variable classification and identification, utilizing various measures of the probability distribution function of fluxes and of timescales of variability. The combination of variability measures from our periodogram and non-parametric analyses with infrared colors from WISE and absolute magnitudes, colors and variability amplitude from Gaia is useful for the identification and classification of periodic variables. Furthermore, we show that the effectiveness of non-parametric methods for the identification of periodic variables is comparable to that of the periodogram but at a much lower computational cost. Future surveys can utilize these methods to accelerate more traditional time-series analyses and to identify evolving sources missed by periodogram-based selections.