Bayesian Inference of Stellar Parameters and Interstellar Extinction with Heterogeneous Data

摘要

I outline a probabilistic method for estimating stellar parameters. It uses not only the spectral energy distribution but also the apparent magnitude, parallax (if available) and the strong prior information provided by the Hertzsprung-Russell Diagram. This (a) improves the accuracy and precision over the use of just the spectrum, and (b) ensures that the inferred parameters (e.g. effective temperature, interstellar extinction and absolute magnitude) are both physically realistic and are consistent with the distance and apparent magnitude. The method - which gives covariate probability distributions over the parameters - is applied to some 85,000 Hipparcos-2MASS stars.