Skip to main content
SearchLogin or Signup

Default priors cause biases in orbital parameter estimates of directly-imaged exoplanets

Presentation #516.02 in the session “Extrasolar Planets: Direct Imaging 2”.

Published onJan 11, 2021
Default priors cause biases in orbital parameter estimates of directly-imaged exoplanets

The eccentricity of a planet’s orbit and the inclination of its orbital plane carry important information about its formation and history. However, exoplanets detected via direct-imaging are often only observed over a very small fraction of their period, making it challenging to perform reliable physical inferences given wide, unconstrained posteriors. The aim of this project is to investigate biases (deviation of the median and mode of the posterior from the true values of orbital parameters, and the width of their credible intervals) in the estimation of orbital parameters of directly-imaged exoplanets, particularly their eccentricities, and to define general guidelines to perform better estimations. For this, we constructed various orbits and simulated data for each spanning around 0.5% of the orbital period. We used the Orbits For The Impatient (OFTI) algorithm to get orbit posteriors, and compared those to the true values of the orbital parameters. We found that the inclination of the orbital plane is the parameter that most affects our estimations of eccentricity, with orbits that appear near edge-on producing eccentricity distributions skewed away from the true values, and often bi-modal. we also found that when the median or the mode of the inclination posterior suggest a near-face-on inclination of the orbital plane, the mode of the eccentricity posterior is a slightly better estimator of eccentricity than the median. Finally, we found a degeneracy between eccentricity and inclination that makes it difficult to distinguish posteriors of face-on, eccentric orbits and edge-on, circular orbits. For the exoplanet-imaging community, we propose practical recommendations, guidelines and warnings relevant to orbit-fitting.


No comments here