Did DESI DR2 truly reveal dynamical dark energy?

TL;DR

This paper critically assesses the claim that DESI DR2 reveals dynamical dark energy by examining CPL-based DDE constraints from individual datasets and their combinations. It shows significant inter-dataset tensions among CMB, BAO, and SN measurements, which undermine the robustness of DDE when data are combined, even though independent probes sometimes favor DDE over $\Lambda$CDM. Bayesian evidence and BIC suggest that DDE likely exists, but the allowed parameter space remains ill-defined due to weak constraints and dataset inconsistencies; removing or marginalizing key factors (e.g., $A_L$) can further dilute significance. The work highlights that none of the single probes conclusively detects cosmic acceleration under DDE, and it argues that the ultimate fate of the universe may be matter-dominated in the far future, challenging the conventional view of a permanently accelerating cosmos.

Abstract

A fundamental question in cosmology is whether dark energy evolves over time, a topic that has gained prominence since the discovery of cosmic acceleration. Recently, the DESI collaboration has reported increasing evidence for evolving dark energy using combinations of cosmic microwave background (CMB), type Ia supernova (SN), and their new measurements of baryon acoustic oscillations (BAO). However, our analysis reveals that these combinations are problematic due to clear tensions among the CMB, BAO and SN datasets. Consequently, DESI's claim of dynamical dark energy (DDE) is not robust. A more reliable approach involves constraining the evolution of dark energy using each dataset independently. Through a statistical comparison for each dataset, on average, we find that DDE is strongly preferred over the $Λ$CDM model. This suggests that DDE likely exists, although its real parameter space remains elusive due to weak constraints on the dark energy equation of state and inconsistencies among the datasets. Interestingly, when considering DDE, none of the individual datasets -- including CMB, DESI DR2, Pantheon+, Union3, and DESY5 -- can independently detect cosmic acceleration at a significant level. Our findings not only clarify the current understanding of the nature of dark energy but also challenge the established discovery of cosmic acceleration and the long-held notion that dark energy exerts negative pressure. Both individual and combined datasets suggest that the ultimate fate of the universe is likely to be dominated by matter rather than dark energy.

Figures (6)

• Figure 1: One-dimensional posterior distributions of the parameters $\Omega_m$ and $H_0$ from CMB, DESI DR2 and SN observations in the $\Lambda$CDM ( upper) and CPL ( lower) models, respectively.
• Figure 1: Two-dimensional posterior distributions of the parameter pairs ($\omega_0$, $\omega_a$) from CMB, DESI DR2 and SN observations in the CPL model with and without the lensing amplitude $A_L$. The cross point of blue dashed lines corresponds to $\Lambda$CDM.
• Figure 2: Two-dimensional posterior distributions of the parameter pairs ($\Omega_m$, $H_0$) and ($\omega_0$, $\omega_a$) from different datasets in the CPL model. The vertical shaded grey regions are the constrained $1\,\sigma$ and $2\,\sigma$$\Omega_m$ values from DESI DR2. The cross point of blue dashed lines represents $\Lambda$CDM. The red dashed line denotes $\omega_0=0$.
• Figure 2: One-dimensional and two-dimensional posterior distributions of model parameters in the CPL model from the DESI DR2 data when considering different values of $H_0$ and $hr_d$, respectively. For the case of $hr_d$, we use a free $hr_d$ as a comparison.
• Figure 3: Two-dimensional posterior distributions of the parameter pairs ($\omega_0$, $\omega_a$) from calibrated and uncalibrated SN datsets in the CPL model. The cross point of blue dashed lines corresponds to $\Lambda$CDM.