[Paper Review] Model-agnostic interpretation of 10 billion years of cosmic evolution traced by BOSS and eBOSS data

S. Brieden, Héctor Gil-Marín+1 more|arXiv (Cornell University)|Apr 25, 2022

Galaxies: Formation, Evolution, Phenomena116 references73 citations

TL;DR

This paper presents a model-agnostic analysis of BOSS and eBOSS data spanning 10 billion years of cosmic evolution, extracting baryon acoustic oscillations (BAO), redshift-space distortions (RSD), and power spectrum shape parameters without assuming a specific cosmological model. The method achieves the most precise model-independent constraints to date using spectroscopic large-scale structure data alone, yielding h = 0.6816 ± 0.0067, Ωm = 0.3001 ± 0.0057, and 10⁹×As = 2.43 ± 0.20 under flat-ΛCDM with BBN priors.

ABSTRACT

We present the first model-agnostic analysis of the complete set of Sloan Digital Sky Survey III (BOSS) and -IV (eBOSS) catalogues of luminous red galaxy and quasar clustering in the redshift range $0.2\leq z \leq 2.2$ (10 billion years of cosmic evolution), which consistently includes the baryon acoustic oscillations (BAO), redshift space distortions (RSD) and the shape of the transfer function signatures, from pre- and post-reconstructed catalogues in Fourier space. This approach complements the standard analyses techniques which only focus on the BAO and RSD signatures, and the full-modeling approaches which assume a specific underlying cosmology model to perform the analysis. These model-independent results can then easily be interpreted in the context of the cosmological model of choice. In particular, when combined with $z>2.1$ Ly-$\alpha$ BAO measurements, the clustering BAO, RSD and {\it Shape} parameters can be interpreted within a flat-$\Lambda$CDM model yielding $h=0.6816\pm0.0067$, $\Omega_{ m m}=0.3001\pm0.0057$ and $10^{9} imes A_s= 2.43\pm0.20$ (or $\sigma_8=0.858\pm0.036$) with a Big Bang Nucleosynthesis prior on the baryon density. Without any external dataset, the BOSS and eBOSS data alone imply $\Omega_{ m m}=0.2971\pm 0.0061$ and $10^{9} imes A_s=2.39^{+0.24}_{-0.43}$ (or $\sigma_8=0.857\pm0.040$). For models beyond $\Lambda$CDM, eBOSS data alone (in combination with Planck) constrain the sum of neutrino mass to be $\Sigma m_ u< 0.40$ eV with a BBN prior ($\Sigma m_ u <0.082$ eV) at 95\% CL, the curvature energy density to $\Omega_\mathrm{k} = -0.022_{-0.038}^{+0.032}$ ($\Omega_\mathrm{k} = 0.0015\pm 0.0016$) and the dark energy equation of state parameter to $w=-0.998_{-0.073}^{+0.085}$ ($w=-1.093_{-0.044}^{+0.048}$) at 68\% CL without a BBN prior.

Motivation & Objective

• To provide model-agnostic constraints on cosmic evolution using the full BOSS and eBOSS clustering data across 0.2 ≤ z ≤ 2.2.
• To extract BAO, RSD, and power spectrum shape parameters independently of any assumed cosmological model.
• To improve upon classic and full-modeling approaches by combining the robustness of compression with the information content of broadband shape fitting.
• To enable direct interpretation of results in any cosmological model of choice, including ΛCDM extensions.
• To deliver the tightest model-independent constraints from spectroscopic large-scale structure data alone.

Proposed method

• The analysis compresses pre- and post-reconstruction power spectrum multipoles into three physical observables: BAO scaling parameters (α∥, α⊥), RSD growth rate (fσs8), and a phenomenological shape parameter (m) from the broadband power spectrum.
• The ShapeFit compression technique is applied to extract shape information from the large-scale power spectrum, improving on fixed-template methods and approaching full-modeling performance.
• Reconstruction is applied to enhance BAO signal-to-noise, and the resulting post-recon power spectra are included with proper covariance modeling.
• A numerical BAO wiggles removal method is used to ensure consistency in shape parameter estimation across cosmologies.
• The method is validated on synthetic galaxy mocks to assess systematic errors and ensure robustness.
• Results are interpreted in both standard ΛCDM and extended models (e.g., massive neutrinos, curvature, varying dark energy), with priors applied only where necessary.

Experimental results

Research questions

• RQ1 How can BAO, RSD, and broadband power spectrum shape be extracted in a model-agnostic way from BOSS and eBOSS clustering data?
• RQ2 What are the most precise model-independent constraints on cosmological parameters using only spectroscopic large-scale structure data?
• RQ3 How do the BAO, RSD, and shape parameters evolve over 10 billion years of cosmic history?
• RQ4 To what extent do the results from this model-agnostic approach agree with official BOSS/eBOSS and full-modeling analyses?
• RQ5 What are the constraints on ΛCDM extensions (e.g., massive neutrinos, curvature, varying dark energy) when interpreted from this model-agnostic framework?

Key findings

• The model-agnostic analysis yields h = 0.6816 ± 0.0067, Ωm = 0.3001 ± 0.0057, and 10⁹×As = 2.43 ± 0.20 under flat-ΛCDM with a BBN prior on baryon density.
• Without external data, BOSS and eBOSS alone constrain Ωm = 0.2971 ± 0.0061 and 10⁹×As = 2.39+0.24−0.43 (or σ8 = 0.857 ± 0.040).
• With Planck and BBN priors, eBOSS data constrain the sum of neutrino masses to Σmν < 0.40 eV (95% CL), with a tighter bound of Σmν < 0.082 eV when using a BBN prior.
• The curvature energy density is constrained to Ωk = −0.022+0.032−0.038 (or 0.0015 ± 0.0016) at 68% CL, and the dark energy equation of state to w = −0.998+0.085−0.073 (or −1.093+0.048−0.044) at 68% CL.
• Systematic checks on synthetic mocks confirm the method’s robustness, with residual systematics well below statistical uncertainties.
• The method achieves the tightest model-independent constraints from spectroscopic LSS data alone to date, outperforming classic compression and approaching full-modeling precision.