BASILISK: Bayesian Hierarchical Inference of the Galaxy-Halo Connection using Satellite Kinematics--I. Method and Validation

Abstract

We present a Bayesian hierarchical inference formalism (Basilisk) to constrain the galaxy-halo connection using satellite kinematics. Unlike traditional methods, Basilisk does not resort to stacking the kinematics of satellite galaxies in bins of central luminosity, and does not make use of summary statistics, such as satellite velocity dispersion. Rather, Basilisk leaves the data in its raw form and computes the corresponding likelihood. In addition, Basilisk can be applied to flux-limited, rather than volume-limited samples, greatly enhancing the quantity and dynamic range of the data. And finally, Basilisk is the only available method that simultaneously solves for halo mass and orbital anisotropy of the satellite galaxies, while properly accounting for scatter in the galaxy-halo connection. Basilisk uses the conditional luminosity function to model halo occupation statistics, and assumes that satellite galaxies are a relaxed tracer population of the host halo's potential with kinematics that obey the spherical Jeans equation. We test and validate Basilisk using mocks of varying complexity, and demonstrate that it yields unbiased constraints on the galaxy-halo connection and at a precision that rivals galaxy-galaxy lensing. In particular, Basilisk accurately recovers the full PDF of the relation between halo mass and central galaxy luminosity, and simultaneously constrains the orbital anisotropy of the satellite galaxies. Basilisk's inference is not affected by potential velocity bias of the central galaxies, or by slight errors in the inferred, radial profile of satellite galaxies that arise as a consequence of interlopers and sample impurity.