Resolving the high redshift Lyman-alpha forest in smoothed particle hydrodynamics simulations

TL;DR

This paper investigates how simulation resolution and box size affect the accuracy of synthetic Lyman-alpha forest spectra at high redshifts, emphasizing the need for higher resolution at z=5 due to the dominance of underdense regions.

Contribution

It provides specific resolution and box size requirements for SPH simulations to accurately model the Lyman-alpha forest at high redshifts, highlighting differences from previous studies.

Findings

Higher mass resolution is needed at z=5 for convergence.

Underdense regions dominate high-z Lya transmission.

Gas density distribution differs from previous literature at high overdensities.

Abstract

We use a large set of cosmological smoothed particle hydrodynamics (SPH) simulations to examine the effect of mass resolution and box size on synthetic Lya forest spectra at 2 \leq z \leq 5. The mass resolution requirements for the convergence of the mean Lya flux and flux power spectrum at z=5 are significantly stricter than at lower redshift. This is because transmission in the high redshift Lya forest is primarily due to underdense regions in the intergalactic medium (IGM), and these are less well resolved compared to the moderately overdense regions which dominate the Lya forest opacity at z~2-3. We further find that the gas density distribution in our simulations differs significantly from previous results in the literature at large overdensities (\Delta>10). We conclude that studies of the Lya forest at z=5 using SPH simulations require a gas particle mass of M_gas \leq 2x10^5 M_sol/h, which is >8 times the value required at z=2. A box size of at least 40 Mpc/h is preferable at all redshifts.