One simulation to fit them all - changing the background parameters of a cosmological N-body simulation

TL;DR

This paper presents a novel scaling algorithm that accurately transforms the output of a cosmological N-body simulation to represent different cosmologies, enabling efficient exploration of parameter space.

Contribution

The authors introduce a three-step scaling method for N-body simulations that accurately reproduces large-scale structure across different cosmological parameters.

Findings

Reproduces mass power spectra within 0.5% on large scales

Accurately replaces BAO features for different cosmologies

Halo properties and assembly histories are reproduced with high fidelity

Abstract

We demonstrate that the output of a cosmological N-body simulation can, to remarkable accuracy, be scaled to represent the growth of large-scale structure in a cosmology with parameters similar to but different from those originally assumed. Our algorithm involves three steps: a reassignment of length, mass and velocity units, a relabelling of the time axis, and a rescaling of the amplitudes of individual large-scale fluctuation modes. We test it using two matched pairs of simulations. Within each pair, one simulation assumes parameters consistent with analyses of the first-year WMAP data. The other has lower matter and baryon densities and a 15% lower fluctuation amplitude, consistent with analyses of the three-year WMAP data. The pairs differ by a factor of a thousand in mass resolution, enabling performance tests on both linear and nonlinear scales. Our scaling reproduces the mass power spectra of the target cosmology to better than 0.5% on large scales (k < 0.1 h/Mpc) both in real and in redshift space. In particular, the BAO features of the original cosmology are removed and are correctly replaced by those of the target cosmology. Errors are still below 3% for k < 1 h/Mpc. Power spectra of the dark halo distribution are even more precisely reproduced, with errors below 1% on all scales tested. A halo-by-halo comparison shows that centre-of-mass positions and velocities are reproduced to better than 90 kpc/h and 5%, respectively. Halo masses, concentrations and spins are also reproduced at about the 10% level, although with small biases. Halo assembly histories are accurately reproduced, leading to central galaxy magnitudes with errors of about 0.25 magnitudes and a bias of about 0.13 magnitudes for a representative semi-analytic model.