Probing Cosmic Curvature with Fast Radio Bursts and DESI DR2

TL;DR

This paper presents a model-independent test of cosmic curvature by combining localized FRBs with DESI DR2 BAO data. It reconstructs the Hubble parameter $H(z)$ from 120 FRBs using an artificial neural network, derives $D_C(z)$ and $D_A(z)$ via FRB-only and FRB+BAO routes, and then constrains $oxed{Ω_k}$. The analysis shows $oxed{Ω_k=-0.31 \\pm 0.57}$ (covariance) and $oxed{Ω_k=-0.13 \\pm 0.46}$ (Gaussian), both compatible with a flat Universe at the $1σ$ level, with the covariance treatment yielding more conservative uncertainties. The results illustrate FRBs as a growing, cosmology-independent probe of large-scale geometry and demonstrate their synergy with BAO in constraining the Universe’s curvature. Looking forward, expanding the FRB sample and refining DM modeling will tighten these curvature bounds and bolster late-time geometry tests.

Abstract

The spatial curvature of the Universe remains a central question in modern cosmology. In this work, we explore the potential of localized Fast Radio Bursts (FRBs) as a novel tool to constrain the cosmic curvature parameter $Ω_k$ in a cosmological model-independent way. Using a sample of 120 FRBs with known redshifts and dispersion measures, we reconstruct the Hubble parameter $H(z)$ via artificial neural networks, and use it to obtain angular-diameter distances $D_A(z)$ through two complementary approaches. First, we derive the comoving distance $D_C(z)$ and $D_A(z)$ directly from FRBs without assuming a fiducial cosmology. Then, we combine the FRB-based $H(z)$ with Baryon Acoustic Oscillation (BAO) DESI DR2 measurements to infer $D_A(z)$. By comparing the FRB-derived and BAO+FRB-derived $D_A(z)$, we constrain spatial curvature. Our covariance-based likelihood (accounting for correlated uncertainties) yields $Ω_k = -0.31\pm0.57$, while a diagonal (Gaussian) treatment gives $Ω_k = -0.13\pm0.46$. Both estimations are consistent with spatial flatness at the $1σ$ level, albeit with a mild preference for negative curvature. Explicitly accounting for the full covariance broadens the intervals and avoids underestimation of uncertainties. These results highlight the growing relevance of FRBs in precision cosmology and their synergy with BAO as a powerful, cosmological model-independent probe of the large-scale geometry of the Universe.

Figures (4)

• Figure 1: Reconstructed FRBs observables. Left: redshift evolution of the average IGM dispersion measure $\langle \mathrm{DM}_{\mathrm{IGM}}(z) \rangle$. Right: Hubble parameter $H(z)$ inferred from FRBs.
• Figure 2: Comoving distance $D_C(z)$ reconstructed from FRBs.
• Figure 3: Left: One–dimensional posterior for the spatial curvature parameter $\Omega_k$ obtained with our two likelihoods. The red filled curve (COV) uses the full covariance between the FRB–derived $D_C$ and $D_A$ together with the BAO error propagation; the blue filled curve (Gaussian) adopts a diagonal treatment. The vertical dashed line marks $\Omega_k=0$. Right: Black points with $3\sigma$ bars are $D_A^{\rm BAO+FRB}(z)$ from Eq. \ref{['BAODA']}, using DESI DR2 data with the FRB–based $H(z)$ and propagated BAO uncertainties. Colored curves show the FRB–only estimator $D_A^{\rm FRB}(z;\Omega_k)$ evaluated at the mean $\Omega_k$ of each analysis; the shaded bands denote credible regions.
• Figure 4: Constraints on the spatial curvature $\Omega_k$. Horizontal bars show $1\sigma$ confidence intervals for our FRB + BAO (COV) and FRB + BAO (GAUSS) analyses, alongside a set of literature determinations (Planck 2018 + BAO; Planck 2018 (no lensing); EFTofLSS/BAO; BAO + Pantheon+; CC + SGL; Planck 2018 + CC; etc.). The vertical dashed line marks $\Omega_k=0$. Incorporating the full covariance (COV) yields broader, more conservative constraints than the Gaussian approximation (GAUSS), yet both remain consistent with spatial flatness at $1\sigma$.