ARIADNE: Measuring accurate and precise stellar parameters through SED fitting

TL;DR

ARIADNE is a Bayesian model averaging tool that uses spectral energy distribution fitting and Gaia data to accurately determine stellar parameters, addressing biases in existing methods and improving consistency across models.

Contribution

The paper introduces ARIADNE, a novel Bayesian model averaging code that combines multiple stellar models with SED fitting and Gaia data for precise stellar parameter estimation.

Findings

ARIADNE achieves excellent agreement with interferometric radii, with a mean fractional difference of 0.001 ± 0.070.

Significant offsets up to 550 K in temperature and 0.6 R_sun in radius are found between different models.

Discrepancies are more pronounced for stars smaller than 0.4-0.5 R_sun, indicating areas for further model improvement.

Abstract

Accurately measuring stellar parameters is a key goal to increase our understanding of the observable universe. However, current methods are limited by many factors, in particular, the biases and physical assumptions that are the basis for the underlying evolutionary or atmospheric models, those that these methods rely upon. Here we introduce our code spectrAl eneRgy dIstribution bAyesian moDel averagiNg fittEr (ARIADNE), which tackles this problem by using Bayesian Model Averaging to incorporate the information from all stellar models to arrive at accurate and precise values. This code uses spectral energy distribution fitting methods, combined with precise Gaia distances, to measure the temperature, log g, [Fe/H], A$_{\text V}$, and radius of a star. When compared with interferometrically measured radii ARIADNE produces values in excellent agreement across a wide range of stellar parameters, with a mean fractional difference of only 0.001 $\pm$ 0.070. We currently incorporate six different models, and in some cases we find significant offsets between them, reaching differences of up to 550 K and 0.6 R$_\odot$ in temperature and radius, respectively. For example, such offsets in stellar radius would give rise to a difference in planetary radius of 60%, negating homogeneity when combining results from different models. We also find a trend for stars smaller than 0.4-0.5 R$_\odot$, which shows more work needs to be done to better model these stars, even though the overall extent is within the uncertainties of the interferometric measurements. We advocate for the use of ARIADNE to provide improved bulk parameters of nearby A to M dwarfs for future studies.