Is Dark Energy Dynamical in the DESI Era? A Critical Review

TL;DR

This paper evaluates whether DESI DR2 data favor dynamical dark energy by analyzing CPL-based ω(z) models and extensions with free ∑mν and N_eff, using a broad MCMC framework across DESI DR2 BAO/Lyα, CMB (CamSpec), BBN priors, cosmic chronometers, and multiple Type Ia SN compilations. It contrasts ΛCDM with ω0ωaCDM and related DM extensions, employing Bayesian evidence and Δχ² to assess the statistical preference while accounting for neutrino physics. The key findings show that DESI DR2 tightens late-time geometry but cannot lower the sound horizon or fully resolve the H0 tension; hints of dynamical dark energy emerge in several dataset combinations, yet these signals are largely driven by low-z SNe and fade when those data are removed. The study highlights phantom crossing and Quintom-B–type behavior as dynamical DE features, but cautions that systematic effects and dataset choices limit conclusive departures from ΛCDM, underscoring the need for upcoming Stage IV surveys to clarify the nature of dark energy.

Abstract

We investigate whether the recent DESI DR2 measurements provide or not evidences for dynamical dark energy by exploring the $ω_0ω_a$CDM model and its extensions with free $\sum m_ν$ and $N_{\mathrm{eff}}$. Using a comprehensive MCMC analysis with a wide range of cosmological datasets including DESI~DR2 BAO and Ly$α$ data, CMB compressed likelihoods, BBN, cosmic chronometers, and multiple Type~Ia supernova compilations, we assess the statistical preference for departures from $Λ$CDM.

Figures (19)

• Figure 1: The figure shows contours of the cosmological parameters for the $\omega_0\omega_a$CDM model at 68% (1$\sigma$) and 95% (2$\sigma$) confidence levels. The constraints are derived using combinations of DESI DR2, DESI DR2 Ly$\alpha$, CMB, BBN, and CC data, together with different Type Ia supernova samples (Pantheon$^+$, DES-SN5Y, DES-SN5Y with $z > 0.1$, and Union3).
• Figure 2: This figure shows the comparative analysis of the predicted values of the Hubble parameter ($h$) and the sound horizon scale ($r_d$) between the standard $\Lambda$CDM and $\omega_0\omega_a$CDM models. Each row corresponds to a specific combination of datasets, including DESI DR2, DESI DR2 Ly$\alpha$, CMB, BBN, and CC data, combined with different Type Ia supernova samples (Pantheon$^+$, DES-SN5Y, DES-SN5Y with $z > 0.1$, and Union3). The color scale indicates the statistical significance (in units of $\sigma$) of the deviation between the two models, with warmer tones representing larger differences.
• Figure 3: This figure shows the whisker plots of the Hubble constant $h$ (left panel) and the sound horizon at the drag epoch $r_d$ (right panel). Each point represents the marginalized mean value corresponding to each dataset combination, as indicated in the legend at the lower left of the right panel, with horizontal error bars showing the 68% (1$\sigma$) uncertainties. The green and red shaded vertical bands indicate the $1\sigma$ ranges of the Planck ($h = 0.674 \pm 0.005$, $r_d = 147.0 \pm 0.5$ Mpc) and SH0ES ($h = 0.735 \pm 0.014$, $r_d = 138.0 \pm 1.0$ Mpc) measurements, respectively.
• Figure 4: This figure shows how different choices for the dark-energy evolution, characterized by the ratio $f(z_p)=\rho_{\mathrm{DE}}(z_p)/\rho_{\mathrm{DE}}(0)$, affect the expansion history $H(z)/(1+z)$. Models with ($f(z_p)>1$, $\omega(z)>-1$) are shown as solid lines and predict a lower value of $H_0$, whereas models with ($f(z_p)<1$, $\omega(z)<-1$) are shown as dotted lines and predict a higher value of $H_0$.
• Figure 5: This figure shows the evolution of $H(z)/(1+z)$ as a function of redshift $z$ for different cosmological dataset combinations, including DESI DR2, DESI DR2 Ly$\alpha$, CMB, BBN, and CC data, together with various Type Ia supernova samples (Pantheon$^+$, DES-SN5Y, DES-SN5Y with $z > 0.1$, and Union3) The black dashed line represents the predicted o $\Lambda$CDM model, shown with its corresponding 1$\sigma$ and 2$\sigma$ confidence regions. The colored lines and shaded bands represent the predictions of the $\omega_0\omega_a$CDM model, also propagated with their respective 1$\sigma$ and 2$\sigma$ uncertainties for each dataset combination.