Dark Energy Survey Year 3 Results: Cosmology from Cosmic Shear and Robustness to Modeling Uncertainty

TL;DR

DES Year 3 cosmic shear analyzes over 100 million galaxies with four-bin tomography to constrain ΛCDM parameters, delivering S8 ≈ 0.759 with 3% precision and 0.772 with ΛCDM-optimized scales. The study emphasizes robustness to intrinsic alignments, nonlinear power spectrum modeling, and baryonic effects, validating fiducial choices on synthetic data and mocks. DES Y3 results are statistically consistent with Planck 2018 in ΛCDM but favor a lower S8 by ≈2.3σ, while remaining compatible with low-z probes; no significant constraint on w is found from cosmic shear alone. A comprehensive IA modeling (TATT) is tested against simpler models (NLA), with evidence favoring simpler IA in the real data, and the analysis demonstrates the critical role of scale cuts and SR information in tightening constraints. Overall, the work establishes percent-level precision in cosmic shear and highlights both methodological progress and remaining modeling challenges for future surveys.

Abstract

This work and its companion paper, Amon et al. (2021), present cosmic shear measurements and cosmological constraints from over 100 million source galaxies in the Dark Energy Survey (DES) Year 3 data. We constrain the lensing amplitude parameter $S_8\equivσ_8\sqrt{Ω_\textrm{m}/0.3}$ at the 3% level in $Λ$CDM: $S_8=0.759^{+0.025}_{-0.023}$ (68% CL). Our constraint is at the 2% level when using angular scale cuts that are optimized for the $Λ$CDM analysis: $S_8=0.772^{+0.018}_{-0.017}$ (68% CL). With cosmic shear alone, we find no statistically significant constraint on the dark energy equation-of-state parameter at our present statistical power. We carry out our analysis blind, and compare our measurement with constraints from two other contemporary weak-lensing experiments: the Kilo-Degree Survey (KiDS) and Hyper-Suprime Camera Subaru Strategic Program (HSC). We additionally quantify the agreement between our data and external constraints from the Cosmic Microwave Background (CMB). Our DES Y3 result under the assumption of $Λ$CDM is found to be in statistical agreement with Planck 2018, although favors a lower $S_8$ than the CMB-inferred value by $2.3σ$ (a $p$-value of 0.02). This paper explores the robustness of these cosmic shear results to modeling of intrinsic alignments, the matter power spectrum and baryonic physics. We additionally explore the statistical preference of our data for intrinsic alignment models of different complexity. The fiducial cosmic shear model is tested using synthetic data, and we report no biases greater than 0.3$σ$ in the plane of $S_8\timesΩ_\textrm{m}$ caused by uncertainties in the theoretical models.

Figures (17)

• Figure 1: The approximate footprints of Stage-III dark energy experiments: Dark Energy Survey Year 3 (DES Y3; green), Kilo-Degree Survey (KiDS-1000; blue) and first-year Hyper Suprime-Cam Subaru Strategic Program (HSC; red). The left and right panels show orthographic projections of the northern and southern sky respectively. The parallels and meridians show declination and right ascension. The different survey areas not only affect the final analysis choices, but also reflect the individual science strategies and the complementarity of Stage-III surveys.
• Figure 2: Cosmic shear two-point correlation measurements from DES Y3. We show here $\xi_+$ and $\xi_-$ (black data points, upper left and lower right halves respectively), with the different panels showing different combinations of redshift bins; in all cases the error bars come from our fiducial analytic covariance matrix. The lighter grey bands represent scales removed from our fiducial analysis, while the darker are the equivalent for the $\Lambda$CDM Optimized analysis. Also shown are the best-fit theory curve in $\Lambda$CDM (solid green) and the intrinsic alignment contributions to the signal: GI (dashed yellow), II (dot-dashed red) and GI+II (solid blue). For clarity, we multiply the IA contributions by a factor of 10, and in most bins the total IA signal is $\sim 1 \%$ of GG+GI+II. The detection significance of the cosmic shear signal after fiducial scale cuts is 27. The $\chi^2$ per effective d.o.f of the $\Lambda$CDM model is $237.7/222 = 1.07$ (a $p$-value of 0.22).
• Figure 3: The estimated redshift distributions and lensing kernels for the fiducial source galaxy sample used in this work. Most of the sensitivity of the DES Y3 cosmic shear signal to large scale structure is in the range between $z=0.1$ and $z=0.5$, where individual kernels peak. Each distribution is independently normalized over the redshift range $z=0-3$. The total effective number density heymans12 of sources is $n_\textrm{eff}=5.59$ galaxies per square arcminute and is divided almost equally into the 4 redshift bins.
• Figure 4: Window functions of $\xi_{+}$ (solid curves) and $\xi_{-}$ (dashed curves) over $k$-wavenumbers of the matter power spectrum at representative angular separations. Notice that our smallest angular scales (after cuts) in $\xi_{+}$ and $\xi_{-}$ are around 2.5 arcminutes and 30 arcminutes respectively, which means that only a relatively small contribution to the full signal comes from wavenumbers above $k\sim1\,h/$Mpc.
• Figure 5: The theoretical contribution of different modeling systematics to $\xi_{\pm}$; for each systematic we show the fractional impact, relative to the fiducial case $\Delta \xi_\pm \equiv (\xi_\pm-\xi_\pm^{\textrm{syst.}})/\xi_\pm$. Fiducial scale cuts are shown as light shaded bands and are derived jointly for cosmic shear, $2\times2$pt and $3\times2$pt. Darker bands correspond to (less stringent) scale cuts that are optimized for cosmic shear and $3\times2$pt in $\Lambda$CDM only. Error bars on $\xi_{+(-)}$ are, at their smallest, around 15% (20%) of the signal, so the maximum contamination shown here $\mathcal{O}(1\sim5\%)$ is still significantly below the sensitivity of our data. We also find that none of these forms of contamination project translate into a bias in cosmological parameters at Y3 precision, despite appearing coherent in some redshift bins. Eagle (black dashed), Horizon-AGN (solid blue), MassiveBlack-II (solid yellow) and OWLS-AGN (solid green) represent scenarios for baryonic physics, and obtained from the power spectra of hydrodynamic simulations, while Euclid Emulator (dot-dashed red) modifies the non-linear gravity-only power spectrum. Higher Order Corrections (dashed purple) is the theoretical impact of reduced shear and source magnification.