Paper detail

Fundamental stellar parameters and metallicities from Bayesian spectroscopy. Application to low- and high-resolution spectra

We present a unified framework to derive fundamental stellar parameters by combining all available observational and theoretical information for a star. The algorithm relies on the method of Bayesian inference, which for the first time directly integrates the spectroscopic analysis pipeline based on the global spectrum synthesis and allows for comprehensive and objective error calculations given the priors. Arbitrary input datasets can be included into our analysis and other stellar quantities, in addition to stellar age, effective temperature, surface gravity, and metallicity, can be computed on demand. We lay out the mathematical framework of the method and apply it to several observational datasets, including high- and low-resolution spectra (UVES, NARVAL, HARPS, SDSS/SEGUE). We find that simpler approximations for the spectroscopic PDF, which are inherent to past Bayesian approaches, lead to deviations of several standard deviations and unreliable errors on the same data. By its flexibility and the simultaneous analysis of multiple independent measurements for a star, it will be ideal to analyse and cross-calibrate the large ongoing and forthcoming surveys, like Gaia-ESO, SDSS, Gaia and LSST.