The Milky Way's (MW) satellite population is a powerful probe of warm dark matter (WDM) models as the abundance of small substructures is very sensitive to the properties of the WDM particle. However, only a partial census of the MW's complement of satellite galaxies exists because surveys of the MW's close environs are incomplete both in depth and in sky coverage. We present a new Bayesian analysis that combines the sample of satellites recently discovered by the Dark Energy Survey (DES) with those found in the Sloan Digital Sky Survey (SDSS) to estimate the total satellite galaxy luminosity function down to $M_{\rm V}=0$. We find that there should be at least $124^{+40}_{-27}$ ($68\%$ CL, statistical error) satellites as bright or brighter than $M_{\rm V}=0$ within 300 kpc of the Sun, with only a weak dependence on MW halo mass. When it comes online the Large Synoptic Survey Telescope should detect approximately half of this population. We also show that WDM models infer the same number of satellites as in $\Lambda$CDM, which will allow us to rule out those models that produce insufficient substructure to be viable.

The total number and luminosity function of the population of dwarf galaxies of the Milky Way (MW) provide important constraints on the nature of the dark matter and on the astrophysics of galaxy formation at low masses. However, only a partial census of this population exists because of the flux limits and restricted sky coverage of existing Galactic surveys. We combine the sample of satellites recently discovered by the Dark Energy Survey (DES) with the satellites found in Sloan Digital Sky Survey (SDSS) Data Release 9 (together these surveys cover nearly half the sky) to estimate the total luminosity function of satellites down to $M_{\rm V}=0$. We apply a new Bayesian inference method in which we assume that the radial distribution of satellites independently of absolute magnitude follows that of subhaloes selected according to their peak maximum circular velocity. We find that there should be at least $124^{+40}_{-27}$ (68 per cent CL, statistical error) satellites brighter than $M_{\rm V}=0$ within $300$ kpc of the Sun. As a result of our use of new data and better simulations, and a more robust statistical method, we infer a much smaller population of satellites than reported in previous studies using earlier SDSS data only; we also address an underestimation of the uncertainties in earlier work by accounting for stochastic effects. We find that the inferred number of faint satellites depends only weakly on the assumed mass of the MW halo and we provide scaling relations to extend our results to different assumed halo masses and outer radii. We predict that half of our estimated total satellite population of the MW should be detected by the Large Synoptic Survey Telescope. The code implementing our estimation method is available online.