Using Computational Models to Uncover the Parameters of Three Kepler Binaries: KIC 5957123, KIC 8314879, and KIC 10727668

TL;DR

This study uses computational modeling and MCMC analysis of Kepler binary star data to estimate system parameters, testing the approach's effectiveness with photometry and distance data, and discussing improvements for future research.

Contribution

It demonstrates a method to determine binary star parameters solely from photometry and distance estimates using MCMC, with application to Kepler data and evaluation of its feasibility.

Findings

Models for two binaries matched observed data well.

Model for one binary did not produce a good fit.

The approach shows promise but requires modifications for better success.

Abstract

Theories of stellar convective core overshoot can be examined through analysis of pulsating stars. Better accuracy can be achieved by obtaining external constraints such as those provided by observing pulsating stars in eclipsing binary systems, but this requires that the binary parameters be identified so photometric variations of the pulsating component may be isolated from the binary periodicity. This study aims to uncover the physical parameters of three binaries observed by the Kepler spacecraft. We also seek to evaluate the feasibility of accurately constraining binaries using only readily available time-series photometry and distance estimates. Binary models were constructed using the Physics of Eclipsing Binaries (PHOEBE) software package. Markov Chain Monte Carlo methods were used to sample the parameter space of these models and provide estimates of the posterior distributions for these systems. An initial run using binned light curve data was performed to identify general parameter trends and provide initializing distributions for a subsequent analysis incorporating the full data set. We present theoretical models for all three binaries, along with posterior distributions from our MCMC analyses. Models for KIC 8314879 and KIC 10727668 produced a good match to the observed data, while the model of KIC 5957123 failed to generate an appropriate synthetic light curve. For the two successful models, we interpret the posterior distributions and discuss confidence in our parameter estimates and uncertainties. We also evaluate the feasibility of this procedure in various contexts, and propose several modifications to improve the success of future studies.