SHAM beyond clustering: new tests of galaxy-halo abundance matching with galaxy groups

We construct mock catalogues of galaxy groups using subhalo abundance matching (SHAM) and undertake several new tests of the SHAM prescription for the galaxy-dark matter connection. All SHAM models we studied exhibit significant tension with galaxy groups observed in the Sloan Digital Sky Survey. The SHAM prediction for the field galaxy luminosity function is systematically too dim, and the group galaxy luminosity function systematically too bright, regardless of the details of the SHAM prescription. SHAM models connecting r-band luminosity, M-r, to V-max(acc), the maximum circular velocity of a subhalo at the time of accretion on to the host, faithfully reproduce the abundance of galaxy groups as a function of richness, g(N). However, SHAM models connecting M-r with V-max(peak), the peak value of V-max over the entire merger history of the halo, overpredict the abundance of galaxy groups. Our results suggest that SHAM models for the galaxy-dark matter connection may be unable to simultaneously reproduce the observed group multiplicity function and two-point projected galaxy clustering. Nevertheless, we also report a new success of the abundance matching prescription: an accurate prediction for Phi(m(12)), the abundance of galaxy groups as a function of magnitude gap m(12), defined as the difference between the r-band absolute magnitude of the two brightest group members. We demonstrate that it may be possible to use joint measurements of g(N) and Phi(m(12)) to provide tight constraints on the details of the SHAM implementation. Additionally, we show that the hypothesis that the luminosity gap is constructed via random draws from a universal luminosity function provides a poor description of the data, contradicting recent claims in the literature. Finally, we test a common assumption of the conditional luminosity function formalism that the satellite luminosity function Phi(sat)(L) need only be conditioned by the brightness of the central galaxy L-cen. We find this assumption to be well supported by the observed magnitude gap distribution.