Binary orbits from combined astrometric and spectroscopic data

Abstract

An efficient Bayesian technique for estimation problems in fundamental stellar astronomy is tested on simulated data for a binary observed both astrometrically and spectroscopically. Posterior distributions are computed for the components' masses and for the binary's parallax. One thousand independent repetitions of the simulation demonstrate that the 1- and 2-$\!\sigma$ credibility intervals for these fundamental quantities have close to the correct coverage fractions. In addition, the simulations allow the investigation of the statistical properties of a Bayesian goodness-of-fit criterion and of the corresponding p-value. The criterion has closely similar properties to the traditional chi^{2} test for minimum-chi^{2} solutions.