Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Dark Energy Survey Year 3 results: Cosmology from combined galaxy clustering and lensing validation on cosmological simulations

J. DeRose et al. (DES Collaboration)
Phys. Rev. D 105, 123520 – Published 17 June 2022
PDFHTMLExport Citation
Abstract
Authors
See Also
Article Text
  • INTRODUCTION
  • buzzard2.0 SIMULATIONS
  • MEASUREMENT OF THE 3×2-POINT DATA VECTOR
  • 3×2-POINT LIKELIHOOD
  • VALIDATION OF THE DES Y3 3×2-POINT…
  • SUMMARY AND CONCLUSIONS
  • ACKNOWLEDGMENTS
  • APPENDICES
    • References

    Abstract

    We present a validation of the Dark Energy Survey Year 3 (DES Y3) 3×2-point analysis choices by testing them on Buzzard2.0, a new suite of cosmological simulations that is tailored for the testing and validation of combined galaxy clustering and weak-lensing analyses. We show that the buzzard2.0 simulations accurately reproduce many important aspects of the DES Y3 data, including photometric redshift and magnitude distributions, and the relevant set of two-point clustering and weak-lensing statistics. We then show that our model for the 3×2-point data vector is accurate enough to recover the true cosmology in simulated surveys assuming the true redshift distributions for our source and lens samples, demonstrating robustness to uncertainties in the modeling of the nonlinear matter power spectrum, nonlinear galaxy bias, and higher-order lensing corrections. Additionally, we demonstrate for the first time that our photometric redshift calibration methodology, including information from photometry, spectroscopy, clustering cross-correlations, and galaxy–galaxy lensing ratios, is accurate enough to recover the true cosmology in simulated surveys in the presence of realistic photometric redshift uncertainties.

    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    4 More
    • Received 3 June 2021
    • Accepted 27 April 2022

    DOI:https://doi.org/10.1103/PhysRevD.105.123520

    © 2022 American Physical Society

    Physics Subject Headings (PhySH)

    1. Research Areas
    Cosmological parametersCosmologyDark energyLarge scale structure of the Universe
    1. Physical Systems
    Galaxies
    1. Techniques
    Astrophysical & cosmological simulations
    Gravitation, Cosmology & Astrophysics

    Authors & Affiliations

    Click to Expand

    See Also

    Dark Energy Survey Year 3 results: Cosmological constraints from galaxy clustering and galaxy-galaxy lensing using the MagLim lens sample

    A. Porredon et al. (DES Collaboration)
    Phys. Rev. D 106, 103530 (2022)

    Article Text (Subscription Required)

    Click to Expand

    References (Subscription Required)

    Click to Expand
    Issue

    Vol. 105, Iss. 12 — 15 June 2022

    Reuse & Permissions
    Access Options
    Author publication services for translation and copyediting assistance advertisement

    Authorization Required

    Log In
    Other Options
    ×

    Images

    • Figure 1
      Figure 1

      Comparison of ugrizJHK color distributions as a function of redshift between buzzard (blue) and the DES Y3 redshift sample (black). Different rows depict color distributions for different band combinations (listed in the left column), while different columns show different redshift bins. Agreement is good, except for in the u-band and at redshifts z>1, where the SED templates used in buzzard are poorly constrained by data [41].

      Reuse & Permissions
    • Figure 2
      Figure 2

      Comparison of red-sequence colors between buzzard and DES Y3 as measured by redMaPPer. Top: mean red-sequence color as a function of redshift for the DES Y3 data (black) compared to the buzzard simulations (blue). The largest differences occur at high redshift in g−r, where the mean color is poorly constrained in the data. Bottom: scatter in red-sequence colors as a function of redshift for the DES Y3 data and buzzard.

      Reuse & Permissions
    • Figure 3
      Figure 3

      Comparison of γt(θ) and w(θ) between buzzard (lines) and the measurements from the Y3 data using the redMaGiC lens sample (points), with error bars given by the fiducial Y3 covariance matrix. Shaded regions indicate the angular scale cuts applied in the fiducial 3×2-point analysis. Rows alternate between showing the measured signals and the fractional difference between data and simulations. Numbers in each panel label the bin combinations shown, with the numbers in the γt(θ) panels representing the lens-source bin pair and numbers in w(θ) panels denoting the lens bins alone. The intent of this comparison is to gauge how well the galaxy clustering and lensing properties in buzzard match those observed in the data, with the caveat that disagreement may arise due to differences in cosmology, and source and lens galaxy redshift distributions. Agreement in w(θ) is generally better than for γt(θ) especially for the second through fifth lens bins, which may be a result of slightly different source redshift distributions between the simulations and data, especially where there is significant overlap between source and lens redshift distributions. We also observe a marked deficit of power on small scales in γt(θ) for the first lens bin at smaller scales than those used in the analysis, part of which is likely due to resolution effects in buzzard. The excellent match between γt(θ) on larger scales in the first lens bin is interesting in light of the large discrepancy in the amplitude of w(θ) between the DES Y3 data and buzzard for this lens bin. The DES Collaboration [44] demonstrates that under the assumption of LCDM the redMaGiC w(θ) and γt(θ) imply bias values that differ by 12%, so this may play some role in the discrepancies seen between the buzzard and DES Y3 w(θ) measurements in the first lens bin.

      Reuse & Permissions
    • Figure 4
      Figure 4

      Comparison of ξ±(θ) between buzzard and the DES Y3 data. As in Fig. 3, rows alternate between ξ±(θ) and the fractional difference between simulations and data, while the numbers in each panel denote which source bin pairs are plotted. The intent of this comparison is to show that the cosmic shear measurements from our simulations largely agree with those in the DES Y3 data, so analyses performed on the simulations can be trusted to have similar constraining power to those performed on the DES Y3 data. The residuals shown here are largely scale independent, which is likely a representation of the imperfect match in source redshift distributions between our simulations and data, along with a small difference in the best-fit cosmology from the ξ±(θ) analysis in the DES Y3 data and that used in the buzzard simulations.

      Reuse & Permissions
    • Figure 5
      Figure 5

      Top: comparison of source (left) and lens (right) redshift distributions in buzzard (dashed) to DES Y3 data (solid). The mean redshifts of the source tomographic bins are [0.326,0.511,0.744,0.871] in buzzard compared with [0.382,0.563,0.759,0.913] in the Y3 data. We have matched the effective source number density and shape noise in buzzard to that found in the DES Y3 matacalibration sample [10]. In combination with the close match in mean redshifts of each tomographic bin, this means that the total signal to noise of our lensing measurements in buzzard should be approximately the same as that found in the DES Y3 data. Bottom: comparison between true (solid) and photometric (dashed) redshift distributions in buzzard for sources (left) and lenses (right). The differences between true and photometric redshift distributions illustrated here are shown to be negligible for the simulated analyses presented in this work in Sec. 5d.

      Reuse & Permissions
    • Figure 6
      Figure 6

      Comparison of our fiducial ξ±(θ) model prediction (lines) at the true buzzard cosmology to the mean measurement from our simulations without shape noise (points). Rows alternate between the signals themselves and fractional deviations between models and simulations, and the gray regions are our fiducial scale cuts. We find χ2=0.78 for ξ+ and χ2=0.59 for ξ−. The differences on large scales are likely caused by a combination of source galaxy clustering and source galaxy magnification [28, 31]. On scales below the scale cuts used in this analysis, the observed differences are likely sourced by ray-tracing resolution effects [33].

      Reuse & Permissions
    • Figure 7
      Figure 7

      Same as Fig. 6, but for γt(θ) and w(θ). Best-fit configuration A1 (linear bias, blue) and A2 (nonlinear bias, yellow) models fixed to the true buzzard cosmology are compared to the mean buzzard data vector without shape noise. Rows alternate between the signals themselves and fractional deviations between models and simulations. For configuration A1, we find χ2=4.5 for γt(θ) and χ2=9.1 for w(θ) for our fiducial scale cuts, shown as light gray shaded regions. For A2, we find χ2=7.2 for γt(θ) and χ2=8.4 for w(θ) using rmin=4  h−1 Mpc scale cuts, depicted by the dark gray shaded regions.

      Reuse & Permissions
    • Figure 8
      Figure 8

      Constraints on S8 and Ωm from 1×2- (left), 2×2- (middle), and 3×2-point (right) analyses on the mean data vector from the full suite of simulations. Top: constraints marginalizing over only cosmology, linear bias, and IAs, assuming the true source and lens redshift distributions. Posteriors obtained using the mean buzzard data vector are shown as solid lines, while dashed contours use a data vector generated at the true cosmology of the simulations with the best-fit linear bias model from analysis configuration A1, i.e., the blue line in Fig. 7. The shaded buzzard contours are the 1/18 and 2/18 confidence regions, while the dashed contours represent the 0.3 and 1σ confidence regions for a single simulation realization. The crosshairs represent the true buzzard cosmology, and the difference between the dashed contours and these is a product of parameter projection effects. Middle: same as the top row, but posteriors are obtained using analysis configuration C (nonlinear bias), where the uncontaminated data vector is the best-fit nonlinear bias model from analysis configuration A2, i.e., the yellow line in Fig. 7. Bottom: same as the top and middle rows, but using analysis configuration D, i.e., using calibrated photometric redshift distributions to make our model predictions, and marginalizing over source and lens photometric redshift uncertainties. This isolates the effect of photometric redshift biases on our analysis. Dashed contours are the same as the solid contours in the top row, but scaled to represent the constraining power of a single Y3 simulation. In all cases, the probability to exceed a parameter bias of more than 0.3σ is less than 60%, as summarized in Tables 2 and 3.

      Reuse & Permissions
    • Figure 9
      Figure 9

      Same as Fig. 8, but for a wCDM parameter space, showing constraints on Ωm and w. In all cases, the probability to exceed a parameter bias of more than 0.3σ is less than 59%, as summarized in Table 2. The wCDM constraints in the presence of nonlinear bias as shown for configuration C (middle row) are impacted by parameter projection effects, as discussed further in Sec. 5c. The inclusion of shear ratios improves the constraining power of ξ±(θ) alone in configuration D (bottom left) with respect to configuration B, mostly by partially breaking degeneracies with intrinsic alignment parameters, even though configuration D also marginalizes over additional nuisance parameters.

      Reuse & Permissions
    • Figure 10
      Figure 10

      Comparison of magnitude distributions for sources (top) and redMaGiC lenses (bottom) between buzzard (dashed) and the DES Y3 data (solid). Different colors represent different photometric bands, where we compare riz for sources and griz for lenses, because these are the bands that are used for each respective selection. Overall agreement is good, with fractional differences between buzzard and the DES Y3 data not exceeding 20%.

      Reuse & Permissions
    • Figure 11
      Figure 11

      1 and 0.3σ posteriors of our analysis configuration C, wCDM model fit to uncontaminated 2×2-point and 3×2-point data vectors. These are the same posteriors as the dashed lines in the second and third columns of the middle row of fig. 9. There is a degeneracy between b21 (the b2 nonlinear bias parameter in the first lens bin) and w in both cases. In the less-constraining 2×2-point case, there is a tail to high values of b21, which correlate with low values of w. When including the cosmic shear data, this tail is truncated, symmetrizing the b21 posterior and thus reducing the projection effect on the w posterior.

      Reuse & Permissions
    ×

    Sign up to receive regular email alerts from Physical Review D

    Log In

    Cancel
    ×

    Search

    Article Lookup

    Paste a citation or DOI
    Enter a citation
    ×