A joint analysis of 3D clustering and galaxy x CMB-lensing cross-correlations with DESI DR1 galaxies

TL;DR

This work develops a joint analysis framework that combines DESI DR1 3D galaxy clustering (full-shape $P_\ell(k)$ and post-reconstruction $\xi_\ell^{\rm post}(s)$) with angular cross-correlations to CMB lensing from Planck PR4 and ACT DR6, plus angular auto/cross spectra from DESI Legacy photometric galaxies. Using a hybrid theory–emulator approach (LPT for 3D, HEFT/Aemulus for non-linear real-space spectra, and Limber projections for 2D), the authors constrain the amplitude of structure with high precision, reporting $\sigma_8 = 0.803\pm0.017$, $\Omega_m = 0.3037\pm0.0069$, and $S_8 = 0.808\pm0.017$, while showing that including $C_\ell^{\kappa g}$ improves $\sigma_8$ uncertainties by about 30% over RSD+BAO alone. They test evolving dark energy via $w_0$-$w_a$ and find a ~3.5$\sigma$ tension with $\Lambda$CDM when combined with Union3 supernovae, and they explore gravitational slip with $\gamma = 1.17 \pm 0.11$, indicating mild tension with general relativity. The analysis demonstrates the power of joint DESI 3D clustering and lensing cross-correlations to tighten cosmological constraints and informs plans for DESI Y3.**

Abstract

The spectroscopic data from DESI Data Release 1 (DR1) galaxies enables the analysis of 3D clustering by fitting galaxy power spectra and reconstructed correlation functions in redshift space. Given low measurements of the amplitude of structure from cosmic shear at $z\sim1$, redshift space distortions (RSD) + Baryon Acoustic Oscillation (BAO) signals from DESI galaxies combined with weak lensing can break degeneracies and provide a tight alternative constraint on the $z\sim1$ amplitude of structure. In this paper we perform joint analyses that combine full-shape + post-reconstruction information from the DESI DR1 BGS and LRG samples along with angular cross-correlations with Planck PR4 and ACT DR6 CMB lensing maps. We show that adding galaxy-lensing cross-correlations tightens clustering amplitude constraints, improving $σ_8$ uncertainties by $\sim 40\%$ over RSD+BAO alone. We also include angular galaxy-galaxy and galaxy-lensing spectra using photometric samples from the DESI Legacy Survey to further improve constraints. Our headline results are $σ_8 = 0.803\pm 0.017$, $Ω_{\rm m} = 0.3037\pm 0.0069$, and $S_8 = 0.808\pm 0.017$. Given DESI's preference for higher $σ_8$ compared to lower values from BOSS, we perform a catalog-level comparison of LRG samples from both surveys. We test sensitivity to dark energy assumptions by relaxing our $Λ$CDM prior and allowing for evolving dark energy via the $w_0-w_a$ parameterization. We find our $S_8$ constraints to be relatively unchanged despite a $~3.5σ$ tension with the cosmological constant model when combining with the Union3 supernova likelihood. Finally we test general relativity (GR) by allowing the gravitational slip parameter ($γ$) to vary, and find $γ= 1.17\pm0.11$ in mild ($\sim1.5σ$) tension with the GR value of $1.0$.

Figures (14)

• Figure 1: Overlapping maps of the data samples used in this paper. DESI Y1 (BGS+LRG) and ACT DR6 footprints are shown in green and blue, respectively, while the red regions show their overlap. The footprint of the full photometric LRG sample is shown in grey. The Planck PR4 map overlaps almost entirely with each of these footprints and is therefore not shown here.
• Figure 2: Normalized redshift distributions for the Photometric and Spectroscopic samples. Note that the ELG and QSO samples are only used for 3D clustering.
• Figure 3: We show in the top row the pseudo-$C_{\ell}^{\kappa g}$ spectra computed from cross correlations between the DESI Y1 spectroscopic BGS and LRG galaxies and Planck PR4 and ACT DR6 CMB lensing maps. The middle and bottom rows show the 3D power spectrum multipoles and post-reconstruction correlation function multipoles, respectively, from the same BGS and LRG samples. For the post-reconstruction correlation function multipoles we show the data used for fitting as points with errorbars in the fit range $80\leq s\leq 120$$h$Mpc$^{-1}$ while the dashed lines show the continuation of these curves outside of the fitting range. We use only the monopole and quadrupole moments, consistent with the fiducial choice of the DESI Y1 analysisDESI2024.V.KP5 that was based on findings in Refs. KP5s2-MausKP5s3-NoriegaKP5s4-Lai that the hexadecapole contributed negligibly to $\Lambda$CDM constraints.
• Figure 4: Posterior Predictive Distributions for $C_{\ell}^{\kappa g}$ and $C_{\ell}^{g g}$ in the spectroscopic bins, using samples from the 3D full-shape + post-recon BAO fit. 500 samples are randomly drawn from the 1$\sigma$ region of the 3D analysis chains and used to predict corresponding $C_{\ell}^{\kappa g}$ and $C_{\ell}^{g g}$ curves. The data points in the top row show the cross-correlations of DESI galaxies with Planck PR4 (blue) and ACT DR6 (orange) lensing, while the data in the bottom row are the angular auto-spectra measured from the DESI galaxies. In the first two columns we also show a vertical line to denote the $\ell = k\times\chi(z)-1/2$ corresponding to the highest wavelength, $k_{\rm max}=0.2$$\,h{\rm Mpc}^{-1}$, used in the full-shape fits. We do not expect the 3D clustering to provide information for $C_{\ell}$ at higher bandpowers.
• Figure 5: Comparison of FS+BAO only, FS+BAO combined with cross-correlations with lensing (Planck PR4 and ACT DR6), and the whole combination of 3D clustering and angular spectra from both spectroscopic and photometric BGS and LRG galaxies from DESI along with PR4 and DR6 lensing data sets.