BASILISK II. Improved Constraints on the Galaxy-Halo Connection from Satellite Kinematics in SDSS

Basilisk is a novel Bayesian hierarchical method for inferring the galaxy-halo connection, including its scatter, using the kinematics of satellite galaxies extracted from a redshift survey. In this paper, we introduce crucial improvements, such as updated central and satellite selection, advanced modelling of impurities and interlopers, extending the kinematic modelling to fourth order by including the kurtosis of the line-of-sight velocity distribution, and utilizing satellite abundance as additional constraint. This drastically enhances Basilisk's performance, resulting in an unbiased recovery of the full conditional luminosity function (central and satellite) and with unprecedented precision. After validating Basilisk's performance using realistic mock data, we apply it to the SDSS-DR7 data. The resulting inferences on the galaxy-halo connection are consistent with, but significantly tighter than, previous constraints from galaxy group catalogues, galaxy clustering and galaxy-galaxy lensing. Using full projected phase-space information, Basilisk breaks the mass-anisotropy degeneracy, thus providing precise global constraint on the average orbital velocity anisotropy of satellite galaxies across a wide range of halo masses. Satellite orbits are found to be mildly radially anisotropic, in good agreement with the mean anisotropy for subhaloes in dark matter-only simulations. Thus, we establish Basilisk as a powerful tool that is not only more constraining than other methods on similar volumes of data, but crucially, is also insensitive to halo assembly bias which plagues the commonly used techniques like galaxy clustering and galaxy-galaxy lensing.