Revisiting the extreme clustering of z ≈ 4 quasars with large volume cosmological simulations

Abstract

ABSTRACT Observations from wide-field quasar surveys indicate that the quasar autocorrelation length increases dramatically from z ≈ 2.5 to ≈ 4. This large clustering amplitude at z ≈ 4 has proven hard to interpret theoretically, as it implies that quasars are hosted by the most massive dark matter haloes residing in the most extreme environments at that redshift. In this work, we present a model that simultaneously reproduces both the observed quasar autocorrelation and quasar luminosity functions. The spatial distribution of haloes and their relative abundance are obtained via a novel method that computes the halo mass and halo cross-correlation functions by combining multiple large-volume dark-matter-only cosmological simulations with different box sizes and resolutions. Armed with these halo properties, our model exploits the conditional luminosity function framework to describe the stochastic relationship between quasar luminosity, L, and halo mass, M. Assuming a simple power-law relation L ∝ Mγ with lognormal scatter, σ, we are able to reproduce observations at z ∼ 4 and find that: (i) the quasar luminosity–halo mass relation is highly non-linear (γ ≳ 2), with very little scatter (σ ≲ 0.3 dex); (ii) luminous quasars ($\log _{10} L/{\rm erg}\, {\rm s}^{-1}\gtrsim 46.5-47$) are hosted by haloes with mass log10M/M⊙ ≳ 13–13.5; and (iii) the implied duty cycle for quasar activity approaches unity ($\varepsilon _{\rm DC}\approx 10\,\mathrm{ per}\,\mathrm{ cent}-60~{{\ \rm per\ cent}}$). We also consider observations at z ≈ 2.5 and find that the quasar luminosity–halo mass relation evolves significantly with cosmic time, implying a rapid change in quasar host halo masses and duty cycles, which in turn suggests concurrent evolution in black hole scaling relations and/or accretion efficiency.