Multi-frequency variabilities: blazar classification and statistical properties

Abstract

Blazars are a class of active galactic nuclei whose jets are aligned with the observer’s line of sight. They are powerful multi-frequency emitters that exhibit rapid and violent variation. Classification of blazars requires multi-frequency observation which may be able to achieve through careful planning. However, in the age of automated surveys, we might be able to complete the same task via data mining and machine learning. In this work, we explore the possibility of using the data from robotic surveys for blazar classifications, particularly the variability of their multi-frequency light curves. The 5 th edition of the Roma-BZCAT is used as our reference blazar catalog. The optical light curves of blazar studied here are taken from the Zwicky Transient Facility (ZTF) public Data Release 4 (2018-2020). The distributions of variability and fractional variability amplitudes in g and r bands are presented and compared for BL Lacs and FSRQs. The Principle Component Analysis (PCA) is then applied to various features extracted from the discrete correlation function (DCF) between the two bands as well as the variability and fractional variability amplitudes of the two bands. Although, our machine learning application for the the BL Lac-FSRQ classification shows unpromising result, the PCA has shown that around 80% of the populations can be explained with 13 features which mainly are the DCF-based ones.