An emulation-based model for the projected correlation function

TL;DR

This paper presents an emulator-based halo model to accurately predict the projected galaxy-galaxy correlation function from large-scale structure data, aiding cosmological parameter inference with upcoming surveys.

Contribution

It introduces a novel emulator approach using the AbacusSummit simulations to model the 2PCF across a wide scale range, facilitating efficient cosmological analysis.

Findings

The emulator predicts the projected correlation function with high accuracy over key scales.

The projected correlation function alone provides weak constraints on cosmological parameters.

Additional clustering statistics are needed for more precise cosmological inference.

Abstract

Data from the ongoing \textit{Euclid} survey will map out billions of galaxies in the Universe, covering more than a third of the sky. This data will provide a wealth of information about the large-scale structure (LSS) of the Universe and will have a significant impact on cosmology in the coming years. In this paper, we introduce an emulator-based halo model approach to forward model the relationship between cosmological parameters and the projected galaxy-galaxy two-point correlation function (2PCF). Utilizing the large \textsc{AbacusSummit} simulation suite, we emulate the 2PCF by generating mock-galaxy catalogues within the Halo Occupation Distribution (HOD) framework. Our emulator is designed to predict the 2PCF over scales $0.1 \leq r / (h^{-1}\text{Mpc}) \leq 105$, from which we derive the projected correlation function, independent of redshift space distortions. We demonstrate that the emulator accurately predicts the projected correlation function over scales $0.5 \leq r_\perp/(h^{-1}\text{Mpc}) \leq 40$, given a set of cosmological and HOD parameters. This model is then employed in a parameter inference analysis, showcasing its ability to constrain cosmological parameters. Our findings indicate that while the projected correlation function places weak constraints on several cosmological parameters due to its intrinsic lack of information, additional clustering statistics are necessary to better probe the underlying cosmology. Despite the simplified covariance matrix used in the likelihood model, the posterior distributions of several cosmological parameters remain broad, underscoring the need for a more comprehensive approach.