Screening vs. gevolution: in chase of a perfect cosmological simulation code

TL;DR

This paper compares two relativistic cosmological simulation approaches, gevolution and screening, showing they produce highly consistent results while the screening method offers computational efficiency improvements.

Contribution

The study demonstrates the high agreement between gevolution and screening simulation codes and highlights the efficiency advantage of the screening approach.

Findings

Power spectra agree within 0.04% for scalar perturbations.

Screening approach reduces computational time by nearly 40%.

Both methods reliably model large-scale structure.

Abstract

We compare two competing relativistic approaches to the N-body simulation of the Universe large-scale structure. To this end, employing the corresponding alternative computer codes ("gevolution" and "screening"), we conduct a series of cosmological simulations in boxes of different sizes and calculate the power spectra of the scalar perturbation $\Phi$, the frame-dragging vector potential ${\bf B}$ and the difference between scalar modes $\chi=\Phi-\Psi$. We demonstrate that the corresponding power spectra are in very good agreement between the compared schemes. For example, the relative difference of the power spectra for $\Phi$ is 0.04% maximum. Since the perturbed Einstein equations have much simpler form in the screening approach, the simulation with this code consumes less computational time, saving almost 40% of CPU hours.