Improving the modelling of redshift-space distortions: I. A bivariate Gaussian description for the galaxy pairwise velocity distributions

TL;DR

This paper introduces a bivariate Gaussian model for galaxy pairwise velocity distributions to improve the accuracy of redshift-space distortion modeling in galaxy surveys, capturing both linear and nonlinear dynamics.

Contribution

The paper develops a novel bivariate Gaussian framework for galaxy pairwise velocities, enhancing the modeling of redshift-space distortions across all scales.

Findings

Accurately models redshift-space distortions on all scales.

Replicates observed Gaussian, exponential, and skewed velocity distributions.

Includes the single-Gaussian model as a special case.

Abstract

As a step towards a more accurate modelling of redshift-space distortions in galaxy surveys, we develop a general description of the probability distribution function of galaxy pairwise velocities within the framework of the so-called streaming model. For a given galaxy separation $\vec{r}$, such function can be described as a superposition of virtually infinite local distributions. We characterize these in terms of their moments and then consider the specific case in which they are Gaussian functions, each with its own mean $\mu$ and dispersion $\sigma$. Based on physical considerations, we make the further crucial assumption that these two parameters are in turn distributed according to a bivariate Gaussian, with its own mean and covariance matrix. Tests using numerical simulations explicitly show that with this compact description one can correctly model redshift-space distorsions on all scales, fully capturing the overall linear and nonlinear dynamics of the galaxy flow at different separations. In particular, we naturally obtain Gaussian/exponential, skewed/unskewed distribution functions, depending on separation as observed in simulations and data. Also, the recently proposed single-Gaussian description of redshift-space distortions is included in this model as a limiting case, when the bivariate Gaussian is collapsed to a two-dimensional Dirac delta function. We also show how this description naturally allows for the Taylor expansion of $1+\xi_S(\vec{s})$ around $1+\xi_R(r)$, which leads to the Kaiser linear formula when truncated to second order, expliciting its connection with the moments of the velocity distribution functions. More work is needed, but these results indicate a very promising path to make definitive progress in our program to improve RSD estimators.