Conditioning halos on the tidal environment for fast and accurate HI power spectra during reionization

Abstract

Predicting the statistical properties of the neutral hydrogen (HI) density field during reionization is an important step in using upcoming 21 cm observations to constrain models of reionization. Semi-numerical models of reionization are often coupled with the collapse fraction field $f_{\text{coll}}(\mathbf{x})$, which determines the fraction of dark matter within halos. In this work, we improve upon earlier prescriptions that compute $f_{\text{coll}}$ based on the dark matter overdensity $\delta(\mathbf{x})$ alone, to include more information about the environment in the form of eigenvalues of the tidal tensor. We compute the mean of the $f_{\text{coll}}$ conditioned on these eigenvalues from a set of high-resolution, small-volume simulations and use them to sample the $f_{\text{coll}}$ field of a low-resolution, large-volume simulation. We subsequently use a semi-numerical code for reionization to compute the HI density field and its power spectrum, and benchmark our results against a reference high-resolution, large-volume simulation. Across variations in redshift, ionized fraction, grid resolution, and minimum halo mass, our method recovers the large-scale HI power spectrum with errors at the $\lesssim 2\%-5\%$ level for $k \lesssim 0.5~ h~ \text{Mpc}^{-1}$, providing a substantial improvement over the $\sim 10\%$ results previously obtained using density-only conditioning. Overall, this makes our method a simple yet efficient tool for forward modeling HI maps during reionization.