Efficient hybrid technique for generating sub-grid haloes in reionization simulations

TL;DR

This paper introduces a hybrid simulation method that combines large-volume low-resolution data with small-volume high-resolution data to efficiently generate accurate models of sub-grid haloes during reionization, reducing computational costs significantly.

Contribution

The authors develop a hybrid approach that merges low- and high-resolution simulation data to produce high-dynamic range reionization models with minimal computational resources.

Findings

Achieves high accuracy in halo and HI field statistics within 10% at large scales.

Uses only 13% of the resources compared to full high-resolution simulations.

Provides a resource-efficient tool for high-redshift HI distribution modeling.

Abstract

Simulating the distribution of cosmological neutral hydrogen (HI) during the epoch of reionization requires a high dynamic range and is hence computationally expensive. The size of the simulation is dictated by the largest scales one aims to probe, while the resolution is determined by the smallest dark matter haloes capable of hosting the first stars. We present a hybrid approach where the density and tidal fields of a large-volume, low-resolution simulation are combined with small haloes from a small-volume, high-resolution box. By merging these two boxes of relatively lower dynamic range, we achieve an effective high-dynamic range simulation using only 13% of the computational resources required for a full high-dynamic range simulation. Our method accurately reproduces the one- and two-point statistics of the halo field, its cross-correlation with the dark matter density field, and the two-point statistics of the HI field computed using a semi-numerical code, all within 10% accuracy at large scales and across different redshifts. Our technique, combined with semi-numerical models of reionization, provides a resource-efficient tool for modeling the HI distribution at high redshifts.