Robust evidence for dynamical dark energy in light of DESI DR2 and joint ACT, SPT, and Planck data

TL;DR

This study tests six dynamical dark energy parametrizations against the most precise current data, including ACT, SPT, Planck CMB, DESI DR2 BAO, and SN compilations. Across CPL, JBP, BA, EXP, LOG, and SIN models, the analysis finds robust indications of dynamical behavior in the dark energy equation of state, with a transition from phantom-like in the past to quintessence-like today (Quintom-B regime) and a crossing redshift near $z_c\sim0.3$–$0.5$. Bayesian evidence favors several DDE models (notably CPL, BA, EXP) over $\Lambda$CDM for DESI+DESY5 data, though results can vary with SN dataset choice, highlighting the sensitivity to data combinations. The work emphasizes reconstructing $w(z)$, $f_{\mathrm{DE}}(z)$, and $q(z)$ to reveal departures from $\Lambda$CDM and discusses prospects for upcoming surveys (DESI full data, LSST, Euclid, CMB-S4) to decisively test DDE and control systematics.

Abstract

Recent baryon acoustic oscillation (BAO) measurements released by DESI, when combined with cosmic microwave background (CMB) data and type Ia supernova (SN) data, suggest a significant preference for dynamical dark energy (DDE) that exhibits the phantom-like behavior in the past and has transitioned into quintessence-like behavior today. In this work, we conduct a comprehensive analysis of six representative DDE parametrization models by utilizing the latest and most precise CMB data jointly from ACT, SPT, and Planck, in conjunction with BAO data from DESI DR2 and SN data from DESY5, PantheonPlus, and Union3. Our overall analysis indicates that the preference for DDE in the Quintom-B regime remains robust, regardless of the DDE parameterization model and the data combination employed. The trend of this preference is significantly strengthened with the support of DESY5 SN data. Specifically, when using the CMB+DESI+DESY5 data, for the Barboza-Alcaniz (BA) model, we obtain $w_0 = -0.785 \pm 0.047$ and $w_a = -0.43^{+0.10}_{-0.09}$, which significantly deviate from the $Λ$CDM values and provide evidence for DDE at the $4.2σ$ level. By the reconstruction of the dark energy equation of state $w(z)$, normalized dark energy density $f_{\mathrm{DE}}(z)$, and the deceleration parameter $q(z)$, we also observe clear departures from $Λ$CDM, further reinforcing the case for DDE. Furthermore, the Bayesian evidence analysis indicates that the Chevallier-Polarski-Linder, BA and Exponential models are moderately favored relative to $Λ$CDM based on the CMB+DESI+DESY5 data.

Figures (4)

• Figure 1: Two-dimensional marginalized contours ($1\sigma$ and $2\sigma$ confidence levels) in the $w_0$--$w_a$ plane by using the CMB+DESI, CMB+DESI+DESY5, CMB+DESI+PantheonPlus, and CMB+DESI+Union3 data in the CPL, JBP, BA, EXP, LOG, and SIN models. We have included boundary lines to delineate the regions of quintessence, phantom, and quintom. The red pentagram marks the $\Lambda$CDM model.
• Figure 2: The evidence for DDE in the CPL, JBP, BA, EXP, LOG, and SIN models using CMB+DESI, CMB+DESI+DESY5, CMB+DESI+PantheonPlus, and CMB+DESI+Union3 data.
• Figure 3: Reconstruct redshift evolution of cosmological quantities: DE EoS $w(z)$, normalized DE density $f_{\mathrm{DE}}(z)$, and deceleration parameter $q(z)$ at $1\sigma$ and $2\sigma$ confidence levels in the CPL, JBP, BA, EXP, LOG, and SIN models, using CMB+DESI+DESY5 data. Here, the black dashed lines represent predictions from the $\Lambda$CDM model, and $z_{\rm c}$ represent the redshift of the $w=-1$ crossing.
• Figure 4: Comparison of the Bayesian evidence for the DDE models and the $\Lambda$CDM model. The Bayes factor $\ln \mathcal{B}_{ij}$ (where $i$ = DDE, $j$ = $\Lambda$CDM) and its strength according to the Jeffreys scale are used to assess the preference between models, where a positive value indicates a preference for the DDE model.