StarHorse: a Bayesian tool for determining stellar masses, ages, distances, and extinctions for field stars

Understanding the formation and evolution of our Galaxy requires accurate distances, ages, and chemistry for large populations of field stars. Here, we present several updates to our spectrophotometric distance code, which can now also be used to estimate ages, masses, and extinctions for individual stars. Given a set of measured spectrophotometric parameters, we calculate the posterior probability distribution over a given grid of stellar evolutionary models, using flexible Galactic stellar-population priors. The code (called StarHorse) can accommodate different observational data sets, prior options, partially missing data, and the inclusion of parallax information into the estimated probabilities. We validate the code using a variety of simulated stars as well as real stars with parameters determined from asteroseismology, eclipsing binaries, and isochrone fits to star clusters. Our main goal in this validation process is to test the applicability of the code to field stars with known Gaia-like parallaxes. The typical internal precisions (obtained from realistic simulations of an APOGEE+ Gaia-like sample) are similar or equal to 8 per cent in distance, similar or equal to 20 per cent in age, similar or equal to 6 per cent in mass, and similar or equal to 0.04 mag in A(V). The median external precision (derived from comparisons with earlier work for real stars) varies with the sample used, but lies in the range of similar or equal to[0, 2] per cent for distances, similar or equal to[12, 31] per cent for ages, similar or equal to[4, 12] per cent for masses, and similar or equal to 0.07 mag for A(V). We provide StarHorse distances and extinctions for the APOGEE DR14, RAVE DR5, GES DR3, and GALAH DR1 catalogues.