Robust Evidence for Dynamical Dark Energy from DESI Galaxy-CMB Lensing Cross-Correlation and Geometric Probes

TL;DR

This work combines DESI LRG–CMB lensing cross-correlation with geometric probes (DESI-DR2 BAO and SN Ia samples) to constrain dynamical dark energy in the $w_0w_a$CDM and $w$CDM frameworks. Using the Limber-approximated angular power spectra, magnification bias, and Halofit nonlinearities within a CPL dark-energy model, the authors perform joint MCMC analyses and Bayesian model comparison. They find that including DESY5 SN Ia data yields strong to very strong evidence for dynamical dark energy, with constraints such as $w_0\approx-0.675$ and $w_a\approx-1.24$ (68% CL) in the CPL case, and $S_8$ values around $0.83$; in the $w$CDM case, $w\approx-0.899$ with milder evidence against $\Lambda$CDM. The results underscore the power of galaxy–CMB lensing cross-correlations as a robust, geometry-informed test of the dark sector, complementary to primary CMB analyses, and highlight prospects for future surveys to further sharpen these constraints.

Abstract

Recent analyses joining data from the Cosmic Microwave Background (CMB), Baryon Acoustic Oscillations (BAO), and Type Ia Supernovae (SNIa) have provided strong evidence in favor of dynamical dark energy (DDE) over a simple cosmological constant. Motivated by these findings, we present new observational constraints on DDE based on the cross-correlation between DESI Luminous Red Galaxies (LRG) samples and CMB lensing ($\mathrm{CMB}_κ \times \mathrm{LRG}$), which effectively probes the impact of cosmological parameters on the growth of structure at the perturbative level. We demonstrate that, when combined with geometric measurements such as BAO and SNIa, this cross-correlation yields compelling statistical evidence for DDE exceeding $4σ$, including within simpler parametrizations such as the $w$CDM model. Remarkably, this evidence is independent of constraints from primary Planck CMB anisotropies data. These results highlight the robustness and potential of Galaxy-CMB lensing cross-correlation as a powerful observational probe of the dark sector, particularly when used in conjunction with geometric observables.

Figures (5)

• Figure 1: Triangle plot comparing the posterior distributions of cosmological parameters obtained in this work using our independently implemented likelihood using Monetpython (blue) with those reported by Sailer:2024coh (gray). The parameters shown include the matter density parameter $\Omega_{\rm m}$, the amplitude of matter fluctuations $\sigma_8$, the derived parameters $S_8$ and $S_8^X = \sigma_8 (\Omega_m/0.3)^{0.4}$. The contours correspond to the 68% and 95% confidence levels. The agreement between the two results serves as a consistency check and confirms the accuracy of our likelihood implementation.
• Figure 2: Two-dimensional confidence contours at the 68% and 95% levels for the parameters $w_0$ and $w_{\rm a}$ are shown for various data combinations, as indicated in the legend, within the framework of the $w_0w_{\rm a}$CDM model.
• Figure 3: In the top row, we present the measurements $C_{\ell}^{\kappa g}$ for the DESI LRG sample with Planck PR4 (blue) and ACT DR6 (orange). The third row shows $C_{\ell}^{gg}$ for the same redshift bins. All measurements and their associated covariance matrices are taken from Sailer:2024coh. The solid black lines indicate the best-fit predictions obtained from our joint fit CMB$\kappa$$\times$LRG+DESI-DR2+DESY5 for $w_{0}w_{a}$CDM, using all three redshift bins within the linear theory framework described in Section \ref{['modeling']}. The second and fourth rows display the residuals concerning the best-fit model for the cross- and auto-correlations, respectively. For each measurement, we display its individual $\chi^2$ value indicated by colors. The grey shaded regions indicate the scales excluded from our fits, while the red shaded areas represent the scales additionally excluded for ACT DR6 data, as summarized in Table \ref{['bins']}. Lastly, the blue region indicates from which $\ell_{\rm max}$ we truncate the sum when convolving our theoretical predictions with the corresponding window functions, as discussed in §\ref{['cross-samples']}.
• Figure 4: Two-dimensional confidence contours at the 68% and 95% levels for the parameters $w_0$ and $H_0$ are shown for various data combinations, as indicated in the legend, within the framework of the $w$CDM model.
• Figure 5: Triangle plot comparing the posterior distributions of cosmological parameters obtained in this work using our baseline approach explained in §\ref{['cross-samples']} with some pipeline variations. The contours correspond to the 68% and 95% confidence levels. The agreement between all the results serves as a consistency check and confirms the accuracy of our likelihood implementation.