Observational constraints on early time non-phantom behaviour of dynamical dark energy

TL;DR

This paper investigates early time non-phantom dynamical dark energy models using current cosmological data, finding tight constraints on early dark energy density and no significant impact on late-time expansion, with early scaling dynamics strongly disfavored.

Contribution

It provides the first comprehensive observational constraints on early time non-phantom dark energy models, including scaling and thawing scenarios, and assesses their impact on cosmological tensions.

Findings

Standard background parameters are stable across models.

Early dark energy density is constrained to below 1% around matter-radiation equality.

Early scaling dynamics are strongly disfavored by data.

Abstract

We study dynamical dark energy models that allow for general late time behaviour while admitting non-phantom dynamics at early times, including thawing, scaling$+$thawing, and effective fluid extensions. Using current cosmological data, we find that the standard background parameters remain tightly constrained and stable across all models, indicating no significant impact of these dynamics on the late time expansion history. At low redshift, parametrizations with a time-dependent equation of state, particularly CPL, show a $\sim2\sigma$ preference for evolution away from $w=-1$, driven by phantom-like behaviour. We then examine the observational consequences of allowing non-phantom dark energy at early times through scaling type dynamics. The data impose strong lower bounds on the steepness of the exponential potential $\gtrsim 20-30$, forcing the dark energy density to remain below the percent level around matter-radiation equality. Consequently, early dark energy is tightly constrained, and neither scaling nor tracker-type dynamics can produce a sufficiently large early dark energy component to alleviate the Hubble tension. While late time dynamical dark energy can improve the goodness of fit, the inclusion of early time scaling does not provide additional improvement and is strongly penalised by model selection criteria, leading to a disfavouring of such scenarios.