Reconstructing cosmological initial conditions from galaxy peculiar velocities. II. The effect of observational errors

TL;DR

This study evaluates how observational errors affect the Reverse Zeldovich Approximation's ability to reconstruct initial cosmological conditions from galaxy peculiar velocities, highlighting the importance of data sparseness and distribution.

Contribution

It systematically analyzes the impact of various observational errors on RZA reconstruction quality using mock catalogues, providing insights into optimal data collection strategies.

Findings

Sparseness of data significantly reduces reconstruction quality.

Wiener Filter helps mitigate the impact of distance measurement errors.

Homogeneous sky distribution of data improves reconstruction accuracy.

Abstract

The Reverse Zeldovich Approximation (RZA) is a reconstruction method which allows to estimate the cosmic displacement field from galaxy peculiar velocity data and to constrain initial conditions for cosmological simulations of the Local Universe. In this paper, we investigate the effect of different observational errors on the reconstruction quality of this method. For this, we build a set of mock catalogues from a cosmological simulation, varying different error sources like the galaxy distance measurement error (0 - 20%), the sparseness of the data points, and the maximum catalogue radius (3000 - 6000 km/s). We perform the RZA reconstruction of the initial conditions on these mock catalogues and compare with the actual initial conditions of the simulation. We also investigate the impact of the fact that only the radial part of the peculiar velocity is observationally accessible. We find that the sparseness of a dataset has the highest detrimental effect on RZA reconstruction quality. Observational distance errors also have a significant influence, but it is possible to compensate this relatively well with Wiener Filter reconstruction. We also investigate the effect of different object selection criteria and find that distance catalogues distributed randomly and homogeneously across the sky (such as spiral galaxies selected for the Tully-Fisher method) allow for a higher reconstruction quality than if when data is preferentially drawn from massive objects or dense environments (such as elliptical galaxies). We find that the error of estimating the initial conditions with RZA is always dominated by the inherent non-linearity of data observed at z=0 rather than by the combined effect of the observational errors. Even an extremely sparse dataset with high observational errors still leads to a good reconstruction of the initial conditions on a scale of about 5 Mpc/h.