Evidence for Dynamical Dark Energy using DESI DR2 Ly$α$ Forest

TL;DR

This work tests a suite of dynamical dark energy models against DESI DR2 Ly-alpha forest BAO measurements, combined with DESI DR2 galaxy BAO, Type Ia SNe (Pantheon+, DES-SN5Y, Union3), and a compressed CMB likelihood. Using Metropolis–Hastings MCMC implemented in SimpleMC and Bayesian evidence computed with MCEvidence, the authors compare LambdaCDM to models with ω(z) parameterizations (CPL, Logarithmic, Exponential, JBP, BA, GEDE) as well as ωCDM and non-flat extensions. Across datasets, they find ω0 > -1, ωa < 0, and ω0 + ωa < -1 in redshift-dependent models, pointing to a Quintom-B type dynamical dark energy; curvature remains consistent with flatness, and oωCDM shows the strongest, though not decisive, preference over ΛCDM in several combinations (2σ–3σ in some cases). The inclusion of Ly-alpha data modestly enhances sensitivity to deviations from ΛCDM and supports a dynamical DE scenario, with Bayesian evidence indicating mild-to-moderate gains for several models but no decisive evidence. The results underscore the diagnostic value of high-redshift Ly-alpha measurements and anticipate that upcoming Stage IV surveys will further illuminate the nature of dark energy and its possible evolution.

Abstract

We present a comprehensive analysis of the cosmological implications of the Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2) Lyman-$α$ forest baryon acoustic oscillation (BAO) measurements, combined with complementary datasets DESI DR2 galaxy BAO, Type Ia supernova (Pantheon$^+$, DES-SN5Y, and Union3), and compressed CMB likelihood. We test several dynamical dark energy models CPL, logarithmic, exponential, JBP, BA, and GEDE as well as $ω$CDM and non-flat extensions of the standard $Λ$CDM and $ω$CDM models. Using the Metropolis Hastings MCMC algorithm implemented in \texttt{SimpleMC}, we constrain the cosmological parameters of each model and evaluate the Bayesian evidence using \texttt{MCEvidence} to assess each model's performance relative to $Λ$CDM. Our results show that the inclusion of DESI DR2 Ly$α$ data modestly enhances sensitivity to departures from $Λ$CDM. Our results show non-flat extensions remain consistent with spatial flatness ($Ω_k \approx 0$ within $1$-$2σ$). Further, all redshift dependent dark energy models predict $ω_0 > -1$, $ω_a < 0$, and $ω_0 + ω_a < -1$, favor a dynamical dark energy scenario characterized by Quintom-B type, and we found that dynamical dark energy favors over the $Λ$CDM model ranging from 0.24-2.60$σ$, depending on the choice of model and corresponding combination of datasets. Bayesian evidence comparison favors the $ω$CDM model with moderate evidence over $Λ$CDM, whereas other parameterizations remain statistically inconclusive.

Figures (1)

• Figure 1: 2D Posterior distributions of different planes of the o$\Lambda$CDM, CPL, $\omega_a \omega_0$CDM, Logarithmic, Exponential, JBP, BA, and GEDE models using DESI DR2 Lyman-$\alpha$ forest measurements in combination with the CMB, Galaxy BAO, and SNe Ia measurements at the 68% (1$\sigma$) and 95% (2$\sigma$) confidence intervals.