Empirical Determination of the Dark Matter Velocity Distribution

Using the hydrodynamic simulation Eris, as well as various realizations of the Milky Way from the FIRE simulation, we found that the kinematics of dark matter follows closely the kinematics of old metal poor stars. We use this correspondence to obtain the first empirical measurement of the local velocity distribution of dark matter, by first analyzing the Gaia data release coupled with RAVE as well as the ninth release from the Sloan Digital Sky Survey and computing the velocity distribution of metal poor stars. We find that this velocity distribution is peaked at lower velocities than the generally assumed Maxwell Boltzmann distribution, leading to a weakening of direct detection limits at dark matter masses less than 10 GeV by a factor of a few. We also found a few kinematic outliers in the stellar data that might be hints of dark matter substructure.