Uncorrelated estimates of the primordial power spectrum

TL;DR

This paper employs localized principal component analysis to detect deviations from scale invariance in the primordial power spectrum, using CMB data and forecasting future constraints from the Planck satellite.

Contribution

It introduces a method for uncorrelated estimation of the primordial power spectrum and applies it to current data, providing new constraints and forecasts for future experiments.

Findings

More than 95% of models are incompatible with scale invariance.

Current data is consistent with a power-law spectrum.

Planck can improve constraints by a factor of 4 at small scales.

Abstract

We use the localized principle component analysis to detect deviations from scale invariance of the primordial power spectrum of curvature perturbations. With the technique we make uncorrelated estimates of the primordial power spectrum with five wavenumber bins. In the framework of a minimal LCDM model, using the latest cosmic microwave background data from the WMAP and ACT experiments we find that more than 95% of the preferred models are incompatible with the assumption of scale-invariance, but still compatible with a power-law primordial spectrum. We also forecast the sensitivity and constraints achievable by the Planck experiment by performing Monte Carlo studies on simulated data. Planck could significantly improve the constraints on the primordial power spectrum, especially at small scales by roughly a factor of 4.