Role of priors in CMB-based spatial curvature inference

Ascertain the quantitative impact of adopted prior distributions on both the marginalized parameters and the curvature parameter prior in Bayesian analyses of Cosmic Microwave Background anisotropy datasets (e.g., Planck), in order to determine whether the reported preference for positive spatial curvature is driven by prior choices rather than by the data likelihood.

Background

The paper discusses recent claims that analyses of CMB primary anisotropies can prefer closed geometries when spatial flatness is not imposed. This preference is complicated by the well-known geometric degeneracy of the CMB and by the influence of priors in Bayesian parameter estimation.

The authors note that the role of the adopted priors—both on parameters marginalized in the posterior and on the curvature parameter itself—has not been fully clarified. Understanding this is essential for interpreting whether any apparent preference for non-zero curvature is a genuine data-driven result or an artifact of prior choices.

While this work focuses on BAO analyses and their robustness to fiducial curvature assumptions, it highlights that the prior dependence in CMB-only curvature inference remains a key unresolved issue in the literature.