Bayesian Model Testing of Ellipsoidal Variations on Stars due to Hot Jupiters

TL;DR

This paper applies Bayesian model testing to identify the most accurate model of stellar ellipsoidal variations caused by hot Jupiters, improving planetary mass estimates using Kepler data.

Contribution

It introduces a Bayesian framework for selecting the best ellipsoidal variation model, comparing several models on synthetic and real Kepler data, notably confirming EVIL-MC as most probable.

Findings

EVIL-MC is the most preferred model for Kepler-13.

Modified Kane & Gelino model is a fast alternative.

Bayesian testing enhances model selection accuracy.

Abstract

A massive planet closely orbiting its host star creates tidal forces that distort the typically spherical stellar surface. These distortions, known as ellipsoidal variations, result in changes in the photometric flux emitted by the star, which can be detected within the data from the Kepler Space Telescope. Currently, there exist several models describing such variations and their effect on the photometric flux. By using Bayesian model testing in conjunction with the Bayesian-based exoplanet characterization software package EXONEST, the most probable representation for ellipsoidal variations was determined for synthetic data and the confirmed hot Jupiter exoplanet Kepler-13Ab.The most preferred model for ellipsoidal variations observed in the Kepler-13 light curve was determined to be EVIL-MC. Among the trigonometric models, the Modified Kane & Gelino model provided the best representation of ellipsoidal variations for the Kepler-13 system and may serve as a fast alternative to the more computationally intensive EVIL-MC. The computational feasibility of directly modeling the ellipsoidal variations of a star are examined and future work is outlined. Providing a more accurate model of ellipsoidal variations is expected to result in better planetary mass estimations.