The host halo masses of AGNs and quasars at z ∼ 3 − 7 with TNG-Cluster, FLAMINGO and other cosmological galaxy simulations

Most observations and clustering analyses suggest that quasars inhabit a narrow range of dark-matter halo masses (1012 − 13 M⊙) across cosmic time (z ≲ 7). Recent hydrodynamical simulations spanning gigaparsec-scale volumes allow direct comparison with self-consistent galaxy-formation models. We quantify the relation between AGN bolometric luminosity and host halo mass before Cosmic Noon in TNG300, TNG-Cluster, FLAMINGO L1_m8 and L2p8_m9, alongside smaller-volume simulations (Illustris, EAGLE, TNG300, and Simba). We consider AGNs with bolometric luminosities LAGNbol≥ 10⁴²\, erg\, s⁻¹. Across all models, more massive haloes host more luminous AGNs on average, but only up to a characteristic mass. The median relation is highly non-linear and not strictly monotonic, with substantial scatter; it flattens in FLAMINGO or turns over in TNG300 +TNG-Cluster above M200, crit ≳ 1012 M⊙, at least at z < 5 − 6. This high-mass AGN quenching also produces a characteristic quasar host halo mass. In TNG300 +TNG-Cluster, quasars (LAGNbol~ 10⁴⁵⁻⁴⁷\, erg\, s⁻¹) typically reside in 1012 − 12.5 M⊙ haloes at z = 3 − 6, whereas FLAMINGO quasars extend to median masses ∼1012.8 M⊙ at z ∼ 3 − 4. All simulations predict much greater scatter in AGN luminosity at fixed halo mass than in halo mass at fixed luminosity (up to 3 dex versus ≲ 1 dex for the 5th-95th percentiles), suggesting weak coupling between halo growth and instantaneous SMBH accretion. This large scatter and resulting diversity of host haloes help the simulated quasar populations broadly match observational estimates. The most luminous AGNs occupy increasingly rare haloes at earlier epochs but typically do not inhabit the most massive haloes at any redshift up to z ≈ 7.