Observational constraints on early dark energy

TL;DR

This paper reviews and updates constraints on Early Dark Energy using various cosmological data sets, finding a consistent preference for the standard Lambda Cold Dark Matter model over EDE, and explores the impact of priors and analysis methods.

Contribution

It provides the first analyses of EDE using CAMSPEC and Hillipop likelihoods for Planck data, and compares Bayesian and frequentist approaches to assess prior effects.

Findings

Bayesian analysis favors Lambda CDM over EDE

CAMSPEC and Hillipop yield tighter EDE bounds than Plik

Prior volume effects are not significant in EDE analyses

Abstract

We review and update constraints on the Early Dark Energy (EDE) model from cosmological data sets, in particular Planck PR3 and PR4 cosmic microwave background (CMB) data and large-scale structure (LSS) data sets including galaxy clustering and weak lensing data from the Dark Energy Survey, Subaru Hyper Suprime-Cam, and KiDS+VIKING-450, as well as BOSS/eBOSS galaxy clustering and Lyman-$\alpha$ forest data. We detail the fit to CMB data, and perform the first analyses of EDE using the CAMSPEC and Hillipop likelihoods for Planck CMB data, rather than Plik, both of which yield a tighter upper bound on the allowed EDE fraction than that found with Plik. We then supplement CMB data with large-scale structure data in a series of new analyses. All these analyses are concordant in their Bayesian preference for $\Lambda$CDM over EDE, as indicated by marginalized posterior distributions. We perform a series of tests of the impact of priors in these results, and compare with frequentist analyses based on the profile likelihood, finding qualitative agreement with the Bayesian results. All these tests suggest prior volume effects are not a determining factor in analyses of EDE. This work provides both a review of existing constraints and several new analyses.