Constraining Dynamical Dark Energy from Galaxy Clustering with Simulation-Based Priors

TL;DR

This paper introduces a novel EFT-SBP method combining perturbation theory and simulation-based priors to analyze galaxy clustering data, testing the preference for dynamical dark energy and finding it less significant when combined with other datasets.

Contribution

The paper presents a new modeling approach using Gaussian mixture models for EFT priors, improving the analysis of galaxy clustering data and constraining dark energy models.

Findings

Dynamical dark energy preference disappears when combining datasets.

Simulation-based priors weaken the evidence for dynamical dark energy.

The method improves the figure of merit for dark energy constraints by ~20%.

Abstract

The effective-field theory based full-shape analysis with simulation-based priors (EFT-SBP) is the novel analysis of galaxy clustering data that allows one to combine merits of perturbation theory and simulation-based modeling in a unified framework. In this paper we use EFT-SBP with the galaxy clustering power spectrum and bispectrum data from BOSS in order to test the recent preference for dynamical dark energy reported by the DESI collaboration. While dynamical dark energy is preferred by the combination of DESI baryon acoustic oscillation, \textit{Planck} Cosmic Microwave Background, and Pantheon+ supernovae data, we show that this preference disappears once these data sets are combined with the usual BOSS EFT galaxy power spectrum and bispectrum likelihood. The use of the simulation-based priors in this analysis further weakens the case for dynamical dark energy by additionally shrinking the parameter posterior around the cosmological constant region. Specifically, the figure of merit of the dynamical dark energy constraints from the combined data set improves by $\approx 20\%$ over the usual EFT-full-shape analysis with the conservative priors. These results are made possible with a novel modeling approach to the EFT prior distribution with the Gaussian mixture models, which allows us to both accurately capture the EFT priors and retain the ability to analytically marginalize the likelihood over most of the EFT nuisance parameters. Our results challenge the dynamical dark energy interpretation of the DESI data and enable future EFT-SBP analyses of BOSS and DESI in the context of non-minimal cosmological models.