Thawing quintessence and transient cosmic acceleration in light of DESI

TL;DR

The paper addresses whether DESI-era data indicating dynamical dark energy can be reconciled without phantom behavior. It tests a thawing quintessence model with a generalized exponential potential, parameterized by $w_T(z)=-1+\alpha(1+z)^{-\beta}$, against ΛCDM and CPL using Planck 2018 CMB, DESI DR2 BAO, and multiple SNe samples via MCMC in Cobaya with physically motivated priors. The results show that thawing quintessence can match the data as well as CPL and ΛCDM, with best-fit parameters producing a transient acceleration in the future and yielding neutrino mass bounds comparable to ΛCDM; for several SN samples, ΛCDM is only marginally disfavored, while the thawing model often receives weak-to-moderate Bayesian support. This work highlights the importance of physics-based dynamical dark energy descriptions and suggests a transient-acceleration scenario that alleviates theoretical issues associated with eternal acceleration and the trans-Planckian problem, pending further observational tests.

Abstract

Recent analysis of the DESI Collaboration challenges the $Λ$-Cold Dark Matter ($Λ$CDM) model, suggesting evidence for a dynamic dark energy. These results are obtained in the context of generic parameterizations of the dark energy equation of state (EoS), which better fit the data when they exhibit an unphysical phantom behavior in the past. In this paper, we briefly analyze how ambiguous this latter conclusion can be in light of the background degeneracy between EoS parameterizations and minimally coupled quintessence scenarios. We then investigate whether the current observational data can be accommodated with a non-phantom, thawing dark energy EoS, typical of a broad class of quintessence models. We show that the thawing behavior of this EoS performs comparabily to the Chevallier-Polarski-Linder parameterization and is statistically competitive with $Λ$CDM while predicting cosmic acceleration as a transient phenomenon. Such a dynamic behavior aligns with theoretical arguments from string theory and offers a way out of the trans-Planckian problem that challenges the ever-accelerated $Λ$CDM paradigm.

Figures (2)

• Figure 1: A summary of the arguments presented in Sec. \ref{['sec:1']}. The figure shows marginalized constraints at 68% and 95% C.L. on the $w_0-w_a$ plane from DESI BAO + CMB + Pantheon+ data. The blue contours reproduces the DESI results reported in DESI:2025zgx while the black/dark gray contours on the left upper corner represent the allowed parametric space (68% and 95% C.L.) when the non-phantom constraint $w(a) \geq -1$ (which translates to $w_0 + w_a \geq -1$ for CPL) is imposed. As discussed in Shlivko:2024llw, applying this restriction to CPL would exclude a priori families of well-motivated models of thawing quintessence, such as the exponential potential represented by the red line. The orange band represents the region in which the CPL parameterization is mapped into the $\alpha - \beta$ plane with $\alpha, \beta > 0$ -- see Eq. (\ref{['wt']}).
• Figure 2: The first and last columns show, respectively, the evolution of $w(z)$ and $q(z)$ as a function of redshift for the mean values of the thawing model discussed in the text, as well as their 1 and 2$~\sigma$ confidence intervals. The middle column shows the marginalized constraints at 68% and 95% C.L. on the $\alpha-\beta$ plane. The upper, middle and lower rows refer to each of the SNe samples, namely DESY5 (red), Union3 (magenta) and Pantheon+ (blue), all added to the DESI BAO + CMB data combination. In all cases, the green band delimits the region of transient acceleration.