Time-series analyses of Cepheid and RR Lyrae variables in the wide-field variability surveys

TL;DR

This study employs Fourier analysis and machine learning to examine the light curves of Cepheid and RR Lyrae variables, revealing insights into their physical parameters and model-observation discrepancies.

Contribution

It introduces a combined Fourier and machine learning approach to analyze variable stars and compares observational data with theoretical models.

Findings

Fourier parameters highlight model-observation offsets for short-period Cepheids.

RR Lyrae models align well with observations across most periods.

Machine learning constrains physical parameters like mass and luminosity effectively.

Abstract

We discuss time-series analyses of classical Cepheid and RR Lyrae variables in the Galaxy and the Magellanic Clouds at multiple wavelengths. We adopt the Fourier decomposition method to quantify the structural changes in the light curves of Cepheid and RR Lyrae variables. A quantitative comparison of Cepheid Fourier parameters suggests that the canonical mass-luminosity models that lie towards the red-edge of the instability strip show a greater offset with respect to observations for short-period Cepheids. RR Lyrae models are consistent with observations in most period bins. We use ensemble light curve analysis to predict the physical parameters of observed Cepheid and RR Lyrae variables using machine learning methods. Our preliminary results suggest that the posterior distributions of mass, luminosity, temperature and radius for Cepheids and RR Lyraes can be well-constrained for a given metal abundance, provided a smoother grid of models is adopted in various input physical parameters.