Bayesian field-level inference of primordial non-Gaussianity using next-generation galaxy surveys

TL;DR

This paper introduces a Bayesian field-level inference method for primordial non-Gaussianity using galaxy survey data, leveraging the entire large-scale structure to improve constraints on the non-Gaussianity parameter nl and explore primordial physics.

Contribution

The paper presents a novel Bayesian forward-modeling approach that fully exploits the large-scale structure, including higher-order statistics and velocity fields, to infer primordial non-Gaussianity from galaxy surveys.

Findings

Achieves .78 constraint on nl for SDSS-like data, improving current limits by 2.5 times.

Demonstrates unbiased nl inference even with survey complexities and unknown biases.

Shows potential for further improvements with higher resolution data.

Abstract

Detecting and measuring a non-Gaussian signature of primordial origin in the density field is a major science goal of next-generation galaxy surveys. The signal will permit us to determine primordial physics processes and constrain models of cosmic inflation. While traditional approaches utilise a limited set of statistical summaries of the galaxy distribution to constrain primordial non-Gaussianity, we present a field-level approach by Bayesian forward-modelling the entire three-dimensional galaxy survey. Our method naturally and fully self-consistently exploits the entirety of the large-scale structure, e.g., higher-order statistics, peculiar velocity fields, and scale-dependent galaxy bias, to extract information on the local non-Gaussianity parameter, $\fnl$. We demonstrate the performance of our approach through various tests with mock galaxy data emulating relevant features of the \sdssiii{}-like survey, and additional tests with a \textit{Stage IV} mock data set. These tests reveal that the method infers unbiased values of $\fnl$ by accurately handling survey geometries, noise, and unknown galaxy biases. We demonstrate that our method can achieve constraints of $\sigma_{\fnl} \approx 8.78$ for \sdssiii{}-like data, an improvement of a factor $\sim 2.5$ over currently published constraints. Tests with next-generation mock data show that significant further improvements are feasible with sufficiently high resolution. Furthermore, the results demonstrate that our method can consistently marginalise all nuisance parameters of the data model. The method further provides an inference of the three-dimensional primordial density field, providing opportunities to explore additional signatures of primordial physics.