Comparing LambdaCDM, wCDM, and w0waCDM models with DESI DR2 BAO: Redshift-Resolved Diagnostics and the Role of rd

TL;DR

This paper reexamines DESI DR2 BAO measurements to compare ΛCDM, ωCDM, and CPL-based w0waCDM using redshift-resolved BAO diagnostics in the ratios $D_M/r_d$ and $D_H/r_d$, and explores how the sound-horizon anchor $r_d$ influences late-time inferences. It employs MCMC over the parameter space with BAO-only constraints on $h r_d$ and dark-energy parameters, then adds a Planck DR3-based $r_d$ prior to isolate the BAO ruler calibration from the full CMB likelihood. The main finding is that all three models provide good fits to DESI DR2 BAO data, with only a mild preference for ΛCDM; dynamical dark energy is not statistically warranted, especially once the $r_d$ prior is applied. Pivot diagnostics show ω_p ≈ -0.899 ± 0.087 at z_p ≈ 0.34, consistent with -1, and degeneracies between $h r_d$ and DE parameters dominate, highlighting the importance of independent $r_d$ calibration. The results underscore that apparent evidence for dynamical dark energy in some multi-probe analyses can hinge on how early-universe information is incorporated, and they advocate robustness tests comparing fixed $r_d$ to fixed θ*, along with precise rd calibration for decisive conclusions.

Abstract

We reanalyze DESI DR2 baryon acoustic oscillation (BAO) measurements to compare LCDM, wCDM, and w0waCDM. Using DM/rd, DH/rd in seven redshift bins, we reconstruct the covariance and run Markov Chain Monte Carlo in Omegam, h rd, w0, wa. In the BAO-only case, all models fit well (tilde chi2 simeq 0.8 - 1.05). Model-selection metrics show at most weak preference for LCDM; the slightly lower chi2 of w0waCDM is offset by complexity, and the pivoted equation of state is consistent with -1 (omegap = -0.899 pm 0.087 at zp simeq 0.34). These results agree with the DESI DR2 analysis. To assess the role of early-universe information, we add a Gaussian prior on rd from Planck DR3 rather than using the full CMB likelihood. Fixing rd isolates the BAO-ruler calibration and yields no significant evidence for dynamical dark energy. The key discriminator is which early-time anchor is held fixed; anchoring theta-ast can raise Omegam in w0waCDM, increasing r-ast and DA(z-ast) to keep theta-ast constant, thereby mimicking late-time evolution, whereas anchoring rd does not. We therefore advocate a robustness test comparing fixed-rd and fixed theta-ast analyses; under the former, DESI DR2 BAO remain fully consistent with LCDM.