Bayesian sampling with BeAtlas, a grid of synthetic Be star spectra I. Recovering the fundamental parameters of \alpha Eri and \beta CMi

TL;DR

This paper introduces BeAtlas, a comprehensive grid of Be star models coupled with Bayesian methods, enabling efficient and accurate inference of stellar and disc parameters for Be stars from observational data.

Contribution

The paper presents BeAtlas, a large grid of NLTE radiative transfer models for Be stars, combined with Bayesian-MCMC techniques, to robustly estimate fundamental stellar and disc parameters.

Findings

Successfully recovered literature parameters for ri and mi

Confirmed disc truncation in mi and analyzed disc density slopes

Estimated lower rotational rates than previous studies

Abstract

Classical Be stars are fast rotating, near main sequence B-type stars. The rotation and the presence of circumstellar discs profoundly modify the observables of active Be stars. Our goal is to infer stellar and disc parameters, as well as distance and interstellar extinction, using the currently most favoured physical models for these objects. We present BeAtlas, a grid of 61.600 NLTE radiative transfer models for Be stars, calculated with the HDUST code. The grid was coupled with a Monte Carlo Markov chain code to sample the posterior distribution. We test our method on two well-studied Be stars, $\alpha$ Eri and $\beta$ CMi, using photometric, polarimetric and spectroscopic data as input to the code. We recover literature determinations for most of the parameters of the targets, in particular the mass and age of $\alpha$ Eri, the disc parameters of $\beta$ CMi, and their distances and inclinations. The main discrepancy is that we estimate lower rotational rates than previous works. We confirm previously detected signs of disc truncation in $\beta$ CMi and note that its inner disc seems to have a flatter density slope than its outer disc. The correlations between the parameters are complex, further indicating that exploring the entire parameter space simultaneously is a more robust approach, statistically. The combination of BeAtlas and Bayesian-MCMC techniques proves successful, and a powerful new tool for the field: the fundamental parameters of any Be star can now be estimated in a matter of hours or days.