2014A&A...561A..58H

Our goal is to find a minimalistic model that describes the luminosity function and large-scale clustering bias of X-ray-selected active galactic nuclei in the general framework of the concordance ΛCDM model. We assume that a simple population-averaged scaling relation between the AGN X-ray luminosity L_X and the host dark matter halo mass M_h exists. With such a relation, the AGN X-ray luminosity function can be computed from the halo mass function. Using the concordance ΛCDM halo mass function for the latter, we obtain the M_h-L_X relation required to match the redshift-dependent AGN X-ray luminosity function known from X-ray observations. We find that with a simple power-law-scaling M_h∝L_X^Γ(z), our model can successfully reproduce the observed X-ray luminosity function. Furthermore, we automatically obtain predictions for the large-scale AGN clustering amplitudes and their dependence on the luminosity and redshift, which seem to be compatible with AGN clustering measurements. Our model also includes the redshift-dependent AGN duty cycle, which peaks at the redshift z≃1, and its peak value is consistent with unity, suggesting that on average there is no more than one AGN per dark matter halo. For a typical X-ray-selected AGN at z∼1, our best-fit M_h-L_X scaling implies low Eddington ratio L_X/L_Edd∼10^–4-10^–3 (2-10 keV band, no bolometric correction applied) and correspondingly high mass-growth e-folding times, suggesting that typical X-ray AGN are dominantly fueled via relatively inefficient ``hot-halo'' accretion mode.