Non-Gaussian structure of the lensed CMB power spectra covariance matrix

TL;DR

This paper develops an analytical model for the non-Gaussian covariance matrix of lensed CMB power spectra, highlighting the significance of second-order terms for accurate cosmological parameter estimation.

Contribution

It introduces a new analytical approach to model the non-Gaussian covariance matrix, including previously neglected second-order effects, improving accuracy over prior models.

Findings

Model captures parameter errors within a few percent.

Non-Gaussian terms significantly affect neutrino mass and dark energy constraints.

Comparison with simulations validates the analytical model.

Abstract

Gravitational lensing of the Cosmic Microwave Background (CMB) encodes cosmological information in the observed anisotropies of temperature and polarization. Accurate extraction of this additional information requires precise modeling of the covariance matrix of the power spectra of observed CMB fields. We introduce a new analytical model to describe the non-Gaussian structure of this covariance matrix and display the importance of second-order terms that were previously neglected. When compared with direct numerical simulations our model captures parameter errors to better than a few percent for cases where the non-Gaussianity causes an order unity degradation in errors. We also provide a detailed comparison between the information content of lensed CMB power spectra and ideal reconstruction of the lensing potential. We illustrate the impact of the non-Gaussian terms in the power spectrum covariance by providing Fisher errors on the sum of the masses of the neutrinos, the dark energy equation of state, and the curvature of the Universe.