Reconstruction of primordial tensor power spectra from B-mode polarization of the cosmic microwave background

TL;DR

This paper presents a model-independent method to reconstruct primordial tensor power spectra from CMB B-mode polarization data, enabling the differentiation of various early Universe tensor mode sources with future experiments.

Contribution

It introduces a Fisher matrix-based approach for reconstructing tensor spectra without assuming a specific functional form, and demonstrates its utility in distinguishing inflationary models.

Findings

Optimal reconstruction near reionization and recombination bumps.

Error bars affected by cosmic variance, lensing, and noise.

Future experiments can differentiate tensor mode sources.

Abstract

Given observations of B-mode polarization power spectrum of the cosmic microwave background (CMB), we can reconstruct power spectra of primordial tensor modes from the early Universe without assuming their functional form such as a power-law spectrum. Shape of the reconstructed spectra can then be used to probe the origin of tensor modes in a model-independent manner. We use the Fisher matrix to calculate the covariance matrix of tensor power spectra reconstructed in bins. We find that the power spectra are best reconstructed at wavenumbers in the vicinity of $k\approx 6\times 10^{-4}$ and $5\times 10^{-3}~{\rm Mpc}^{-1}$, which correspond to the "reionization bump" at $\ell\lesssim 6$ and "recombination bump" at $\ell\approx 80$ of the CMB B-mode power spectrum, respectively. The error bar between these two wavenumbers is larger because of lack of the signal between the reionization and recombination bumps. The error bars increase sharply towards smaller (larger) wavenumbers because of the cosmic variance (CMB lensing and instrumental noise). To demonstrate utility of the reconstructed power spectra we investigate whether we can distinguish between various sources of tensor modes including those from the vacuum metric fluctuation and SU(2) gauge fields during single-field slow-roll inflation, open inflation and massive gravity inflation. The results depend on the model parameters, but we find that future CMB experiments are sensitive to differences in these models. We make our calculation tool available on-line.