Semi-analytic modeling of AGNs: auto-correlation function and halo occupation

The spatial clustering of active galactic nuclei (AGNs) is considered to be one of the important diagnostics for the understanding of the underlying processes behind their activities complementary to their abundance. We analyse the AGN clustering from a recent semi-analytic model performed on a large cosmological $N$-body simulation covering a cubic gigaparsec comoving volume. The recipe for the model has previously been shown to reproduce the observed luminosity function of AGN quite well over a wide redshift range ($z\lesssim 6$), with a newly introduced timescale to account for the loss of the angular momentum on small scales in the accretion process on to the supermassive black holes. The large simulation box allows us accurate determination of the auto-correlation function of the AGNs. The model prediction indicates that this timescale plays a significant role in allowing massive halos to host relatively faint population of AGNs, leading to a higher bias factor for those AGNs. The model predictions are in agreement with observations of X-ray selected AGNs in the luminosity range $10^{41.5}~\mathrm{erg} \ \mathrm{s}^{-1} \leq L_{2-10\mathrm{keV}} \leq 10^{44.5}~\mathrm{erg} \ \mathrm{s}^{-1}$, with the typical host halo mass of $10^{12.5-13.5} h^{-1} M_{\odot}$ at $z \lesssim 1$. Moreover, we find that not only the effective halo mass corresponding to the overall bias factor, but the extended shape of the predicted AGN correlation function shows remarkable agreement with those from observations. Further observational efforts in the near future towards the low luminosity end at redshift $\sim 1$ would give us stronger constraints on the triggering mechanisms of AGN activities through the clustering properties.