The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Baryon Acoustic Oscillations in the Data Release 9 Spectroscopic Galaxy Sample

TL;DR

This paper presents precise BAO distance measurements from the SDSS-III BOSS CMASS DR9 galaxy sample, using both the angle-averaged correlation function and power spectrum with density-field reconstruction. By employing extensive mock catalogs to estimate covariances and robust weighting schemes to mitigate systematics, the authors obtain a consensus distance scale of D_V(z=0.57)/r_s = 13.67 ± 0.22, corresponding to D_V(z=0.57) ≈ 2094 ± 34 Mpc for a fiducial sound horizon. The results provide a highly precise, model-consistent view of the expansion history, aligning with supernova and CMB constraints within a flat ΛCDM framework and strengthening the BAO distance ladder. The analysis demonstrates BAO’s power to constrain cosmological parameters and curvature, and it showcases the reliability of reconstruction and cross-validation between real- and Fourier-space BAO analyses. This work significantly tightens low-to-intermediate redshift distance measurements and informs future cosmological inferences from BAO data.

Abstract

We present measurements of galaxy clustering from the Baryon Oscillation Spectroscopic Survey (BOSS), which is part of the Sloan Digital Sky Survey III (SDSS-III). These use the Data Release 9 (DR9) CMASS sample, which contains 264,283 massive galaxies covering 3275 square degrees with an effective redshift z=0.57 and redshift range 0.43 < z < 0.7. Assuming a concordance Lambda-CDM cosmological model, this sample covers an effective volume of 2.2 Gpc^3, and represents the largest sample of the Universe ever surveyed at this density, n = 3 x 10^-4 h^-3 Mpc^3. We measure the angle-averaged galaxy correlation function and power spectrum, including density-field reconstruction of the baryon acoustic oscillation (BAO) feature. The acoustic features are detected at a significance of 5σin both the correlation function and power spectrum. Combining with the SDSS-II Luminous Red Galaxy Sample, the detection significance increases to 6.7σ. Fitting for the position of the acoustic features measures the distance to z=0.57 relative to the sound horizon DV /rs = 13.67 +/- 0.22 at z=0.57. Assuming a fiducial sound horizon of 153.19 Mpc, which matches cosmic microwave background constraints, this corresponds to a distance DV(z=0.57) = 2094 +/- 34 Mpc. At 1.7 per cent, this is the most precise distance constraint ever obtained from a galaxy survey. We place this result alongside previous BAO measurements in a cosmological distance ladder and find excellent agreement with the current supernova measurements. We use these distance measurements to constrain various cosmological models, finding continuing support for a flat Universe with a cosmological constant.

Figures (35)

• Figure 1: The sky coverage of the galaxies used in this analysis. The light grey shaded region shows the expected total footprint of the survey, totalling $10\,269$ deg$^2$. The coloured and dark grey regions indicate the DR9 spectroscopic coverage of the survey, totalling $3792$ deg$^2$. Colours indicate the completeness within each sector used to build the random catalog as defined in Eq. \ref{['eq:comp']}. Sectors coloured dark grey are removed from the analysis by the cuts described in Section \ref{['sec:targ_summary']}. The total effective area (accounting for all applied cuts and the completeness in every sector included) used in our analysis is $3275$ deg$^2$. The low completeness at many edges is due to unobserved tiles that will overlap the current geometry in future data releases.
• Figure 2: The galaxy number density as a function of redshift for the BOSS DR9 CMASS sample (thick blue line) used in this analysis, which ranges in redshift between $0.43<z<0.7$. For comparison, we also plot the density for a SDSS-II DR7 LRG sample (thin red line) covering $0.16<z<0.47$, which was used in Pad12. Note that both selections include a small fraction of objects that fall outside the redshift cuts shown here.
• Figure 3: The CMASS correlation function before (left) and after (right) reconstruction (crosses) with the best-fit models overplotted (solid lines). Error bars show the square root of the diagonal covariance matrix elements, and data on similar scales are also correlated. The BAO feature is clearly evident, and well matched to the best-fit model. The best-fit dilation scale is given in each plot, with the $\chi^2$ statistic giving goodness of fit.
• Figure 4: Average of the mock correlation functions before and after reconstruction showing that the average acoustic peak sharpens up significantly after reconstruction. This indicates that, on average, our reconstruction technique effectively removes some of the smearing caused by non-linear structure growth, affording us the ability to more precisely centroid the acoustic peak.
• Figure 5: Comparisons of $\sigma_\alpha$ errors in mock catalogs before and after reconstruction as measured from $\xi(r)$. Reconstruction tends to improve our ability to measure $\alpha$; on a mock-by-mock basis, the average amount of improvement in $\sigma_\alpha$ is a factor of 1.54. However, the amount of improvement varies, and 26 (out of 600) of the mocks actually see $\sigma_\alpha$ increase from pre-reconstruction to post-reconstruction. The CMASS DR9 point is overplotted as the black star and falls within the locus of the mock points. 44 (out of 600) of the mocks have a ratio of $\sigma_\alpha$ after reconstruction compared to before reconstruction that is greater than the CMASS DR9 value. Hence, the fact that the error on $\alpha$ measured from CMASS DR9 does not decrease significantly after reconstruction is not unexpected in the context of the mocks. One can also see that most of the extreme outliers in $\sigma_\alpha$ before reconstruction have significantly smaller errors after reconstruction.