Simulating Structure Formation of the Local Universe

TL;DR

This paper presents constrained cosmological N-body simulations of the Local Universe using a Bayesian approach and a novel structure formation model, achieving high correlation with observed galaxy distributions at small scales.

Contribution

Introduction of the KIGEN-code with ALPT for accurate constrained simulations of the Local Universe, combining large-scale 2LPT with spherical collapse and improved modeling of coherent flows and fingers-of-god.

Findings

High correlation (98.3%) between simulated and observed density fields at large scales.

Simulations reproduce matter statistics, power spectra, and mass functions consistent with reference models.

Effective modeling of small-scale structures with minimal artificial filament formation.

Abstract

In this work we present cosmological N-body simulations of the Local Universe with initial conditions constrained by the Two-Micron Redshift Survey (2MRS) within a cubic volume of 180 Mpc/h side-length centred at the Local Group. We use a self-consistent Bayesian based approach to explore the joint parameter space of primordial density fluctuations and peculiar velocity fields, which are compatible with the 2MRS galaxy distribution after cosmic evolution. This method (the KIGEN-code) includes the novel ALPT (Augmented Lagrangian Perturbation Theory) structure formation model which combines second order LPT (2LPT) on large scales with the spherical collapse model on small scales. Furthermore we describe coherent flows with 2LPT and include a dispersion term to model fingers-of-god (fogs). These implementations are crucial to avoid artificial filamentary structures, which appear when using a structure formation model with 2LPT and data with compressed fogs. We assume LCDM cosmology throughout our method. The recovered initial Gaussian fields are used to perform a set of 25 constrained simulations. Statistically this ensemble of simulations is in agreement with a reference set of 25 simulations based on randomly seeded Gaussian fluctuations in terms of matter statistics, power-spectra and mass functions. Considering the entire volume, we obtain correlation coefficients of about 98.3% for the comparison between the simulated density fields and the galaxy density field in log-space with Gaussian smoothing scales of r_S=3.5 Mpc/h (74% for r_S=1.4 Mpc/h). The cross power-spectra show correlations with the galaxy distribution which weaken towards smaller length scales until they vanish at scales of 2.2-3.0 Mpc/h. The simulations we present provide a fully nonlinear density and velocity field with a high level of correlation with the observed galaxy distribution at scales of a few Mpc.