ALBERT

Description: A very large dynamic range with simultaneous capture of both large and small scales in the simulations of cosmic structures is required for correct modelling of many cosmological phenomena, particularly at high redshift. This is not always available, or when it is, it makes such simulations very expensive. We present a novel subgrid method for modelling low-mass ($10^5, { m M}_odot le M_{ m halo}le 10^9, { m M}_odot$) haloes, which are otherwise unresolved in large-volume cosmological simulations limited in numerical resolution. In addition to the deterministic halo bias that captures the average property, we model its stochasticity that is correlated in time. We find that the instantaneous binned distribution of the number of haloes is well approximated by a lognormal distribution, with overall amplitude modulated by this ‘temporal correlation bias’. The robustness of our new scheme is tested against various statistical measures, and we find that temporally correlated stochasticity generates mock halo data that is significantly more reliable than that from temporally uncorrelated stochasticity. Our method can be applied for simulating processes that depend on both the small- and large-scale structures, especially for those that are sensitive to the evolution history of structure formation such as the process of cosmic reionization. As a sample application, we generate a mock distribution of medium-mass (108 ≤ M/M⊙ ≤ 109) haloes inside a 500 Mpc $, h^{-1}$, 3003 grid simulation box. This mock halo catalogue bears a reasonable statistical agreement with a halo catalogue from numerically resolved haloes in a smaller box, and therefore will allow a very self-consistent sets of cosmic reionization simulations in a box large enough to generate statistically reliable data.