Calibration-independent consistency test of DESI DR2 BAO and SNIa

TL;DR

This work develops a calibration-independent, model-agnostic consistency test between DESI DR2 BAO and SNIa datasets by recasting SNIa observables into nuisance-free quantities via $F_{ m AP}(z)$ and the pair of functions $A(z)$ and $B(z)$, enabling a direct comparison with BAO without relying on $M_B$ or $r_d$. Using Gaussian Process reconstructions to obtain smooth $F_{ m AP}(z)$ and its derivatives, the study finds Pantheon+ and Union3 are consistent with DESI DR2 BAO at the $\sim 1\sigma$ level for $z\lesssim 0.5$, while DES-Y5 shows a significant tension, up to $>3\sigma$ near $z\approx 1$. Robustness tests confirm the results against higher-order polynomial reconstructions, and curvature effects are shown to be negligible. The analysis further discusses potential implications for cosmic distance duality and standard candle evolution, illustrating that the observed mismatch could signal new physics or unidentified systematics if confirmed.

Abstract

We investigate the consistency between DESI DR2 BAO and three SNIa datasets, Pantheon+, Union3, and DES-Y5. Our consistency test is {calibration}-independent since it is independent of cosmological nuisance parameters such as the absolute peak magnitude $M_B$ and the comoving sound horizon at the baryon drag epoch $r_d$. Importantly, the test is also model-agnostic, independent of any model of dark energy or modified gravity. We define a tension parameter to quantify tension across different datasets compared to DESI DR2 BAO. The Pantheon+ and Union3 data have tension $\lesssim\! 1σ$ across their redshift ranges, whereas the DES-Y5 tension is $\gtrsim3σ$ near $z=1$. This hints that DES-Y5 data has significant offset values for redshifts close to 1, compared to the other SNIa datasets. Since this consistency test is independent of cosmological nuisance parameters, the tension is minimal: other consistency tests involving differences in nuisance parameters may show greater tension.

Figures (7)

• Figure 1: Gaussian Process predictions (blue) compared to observational data (black), where $A$ and $B$ are defined in \ref{['eq:main']}.
• Figure 2: Left:$F_{\rm AP}$ comparison of SNIa data to DESI data. Right: Tension between SNIa and DESI datasets, as given by \ref{['eq:tension']}.
• Figure 3: GP applied to $m_B$ compared to $A$, using Pantheon+ data.
• Figure 4: Robustness of the results.
• Figure 5: The curvature correction term $|\gamma| \widetilde{D}_M^2(z)$.