Cosmic Cartography of the Large-Scale Structure with Sloan Digital Sky Survey Data Release 6

TL;DR

This paper presents the largest high-resolution Wiener reconstruction of the cosmic density field from SDSS DR6 data, revealing complex large-scale structures and emphasizing the importance of Bayesian modeling for accurate density estimation.

Contribution

Introduces a novel Wiener reconstruction method using supersampling and Krylov inversion, improving resolution and accuracy in mapping the cosmic web from SDSS data.

Findings

Reconstructed the cosmic density field over 500 Mpc with ~10 Mpc effective resolution.

Identified complex non-Gaussian large-scale structures, including voids and filaments.

Demonstrated the superiority of Bayesian modeling over inverse weighting methods.

Abstract

We present the largest Wiener reconstruction of the cosmic density field made to date. The reconstruction is based on the Sloan Digital Sky Survey data release 6 covering the northern Galactic cap. We use a novel supersampling algorithm to suppress aliasing effects and a Krylov-space inversion method to enable high performance with high resolution. These techniques are implemented in the ARGO computer code. We reconstruct the field over a 500 Mpc cube with Mpc grid-resolution while accounting both for the angular and radial selection functions of the SDSS, and the shot noise giving an effective resolution of the order of ~10 Mpc. In addition, we correct for the redshift distortions in the linear and nonlinear regimes in an approximate way. We show that the commonly used method of inverse weighting the galaxies by the corresponding selection function heads to excess noise in regions where the density of the observed galaxies is small. It is more accurate and conservative to adopt a Bayesian framework in which we model the galaxy selection/detection process to be Poisson-binomial. This results in heavier smoothing in regions of reduced sampling density. Our results show a complex cosmic web structure with huge void regions indicating that the recovered matter distribution is highly non-Gaussian. Filamentary structures are clearly visible on scales up to ~20 Mpc. We also calculate the statistical distribution of density after smoothing the reconstruction with Gaussian kernels of different radii r_S and find good agreement with a log-normal distribution for ~10 Mpc < r_S < ~30 Mpc.