Extraction of the angular power spectrum produced by inflation from observations of experiments such as Simons Observatory

We demonstrate an approach that allows separating two-point correlations created by a Gaussian random field from correlations created by cosmic foregrounds such as polarized dust emission, gravitational lensing and other non-Gaussian signals. The result of traditional approaches should typically be a 'foreground-cleaned' two-dimensional CMB map of the anisotropy or polarization. Our method does not create a clean map, but extracts the part of the two-point correlations, or equivalently the part of the power spectrum, which is due only to the Gaussian component of the observed signal produced by inflation.