Evidence for Evolving Dark Energy from LRG1-2 and Low-$z$ SNe Ia Data

TL;DR

This study tests whether dark energy evolves by integrating DESI DR2 BAO measurements with multiple Type Ia SN calibrations and a compressed CMB likelihood across numerous phenomenological $\omega(z)$ parameterizations. Using Bayesian inference, it finds that DESI DR2 LRG1/2 measurements drive departures from ΛCDM toward dynamical dark energy, typically with $\omega_0>-1$ and $\omega_a<0$ in a Quintom-B pattern; these signals strengthen when low-redshift SNe are included. However, removing low-$z$ SN data substantially reduces or eliminates the detected evolution, suggesting a strong influence from nearby SN systematics. The reconstructed evolution of $\omega(z)$ frequently shows phantom crossing around $z\sim0.5$, and growth- function trends $f_{DE}(z)$ support dynamical dark energy, emphasizing the need for careful handling of low-$z$ SN data and further low-redshift measurements to validate these hints of new physics beyond ΛCDM.

Abstract

In this paper, we present evidence for evolving dark energy using baryon acoustic oscillation (BAO) measurements from the recent Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2), combined with different Type Ia supernova calibrations (Pantheon$^+$, DES-SN5YR, and Union3) and the CMB compressed likelihood. In our analysis, we examined several dark energy models, including the Logarithmic, Exponential, CPL, BA, JBP, Thawing, Mirage, and GEDE parameterizations. We found that the evidence for evolving dark energy in the DESI measurements is primarily driven by the LRG1 and LRG2 tracers, as their combination yields a preferred value of $ω_0 > -1$. Further extending our analysis by combining other measurements, we find that all evolving dark energy models favor evolving dark energy scenario characterized by $ω_0 > -1$, $ω_a < 0$, and $ω_0 + ω_a < -1$ (Quintom-B). Indeed, our results show a departure from the $Λ$CDM model at a significance level of up to $3$-$3.6σ$, with the DES-SN5YR sample showing the most significant deviation. When we exclude low-$z$ SNe Ia measurements from the DES-SN5YR sample, we find that the preference for evolving dark energy is significantly biased by the low-redshift ($z < 0.01$) supernova sample. The evolution of $ω(z)$ shows a phantom crossing around $z \sim 0.5$ in most of the dynamical dark energy models, and the evolution of $f_{DE}(z)$ further supports the evolving nature of dark energy. Bayesian analysis of model evidence shows that the inclusion of low-$z$ supernovae data significantly strengthens the support for evolving dark energy models such as JBP and Mirage, while models revert to inconclusive or weak support when low-$z$ data are excluded.

Figures (15)

• Figure 1: The figure shows the posterior distributions at the 1$\sigma$ and 2$\sigma$ confidence levels in the $\Omega_m - h r_d$ contour plane for different tracers corresponding to various $z_{\text{eff}}$ values from the DESI DR2 dataset within the $\Lambda$CDM model. The horizontal lines correspond to the mean values of $\Omega_m$ for each tracer, and the gray band represents the Planck $\Lambda$CDM prediction for $\Omega_m = 0.315 \pm 0.007$.
• Figure 2: The figure shows the posterior distributions in the $\Omega_m - h r_d$ plane using different redshift bins from the DESI DR2 compilation. The blue band represents the Planck $\Lambda$CDM prediction for $\Omega_m$.
• Figure 3: The figure shows the posterior distributions $\omega$CDM, Logarithmic, Exponential, CPL, BA, JBP, Calib. Thawing, Calib. Mirage, and GEDE models at 68% (1$\sigma$) and 95% (2$\sigma$) confidence, using no LRG1, no LRG2, no LRG1 & LRG2, and the full DESI DR2 sample.
• Figure 4: The figure shows posterior distributions in the $\omega_0$–$\omega_a$ plane for the Logarithmic, Exponential, CPL, BA, and JBP models. The left panel excludes the LRG1 BAO point, while the right panel includes it. Contours indicate the 68% (1$\sigma$) and 95% (2$\sigma$) confidence intervals.
• Figure 5: The figure shows the $\sigma$ deviation of each model from the $\Lambda$CDM model using only the DESI DR2 BAO data alone