Reproducing Cosmic Microwave Background anisotropies with mixed isocurvature perturbations

TL;DR

The paper investigates how CMB-based cosmological parameter inferences depend on the assumed initial conditions by allowing a generic mix of perturbation modes. It introduces a framework with four modes (AD, CI, NID, NIV) and a correlation matrix M to encode cross-correlations, computing the CMB spectra as a linear combination of mode spectra and fitting to COBE and BOOMERANG data. The results show that relaxing the purely adiabatic assumption greatly broadens the allowed region in parameter space, potentially accommodating higher baryon density and Hubble parameter values, with the isocurvature content quantified by alpha and often dominated by the NID mode. The study highlights that CMB analyses test model assumptions about the primordial conditions and that polarization and external (non-CMB) measurements are essential to breaking degeneracies and constraining the nature of the initial perturbations.

Abstract

Recently high quality data of the cosmic microwave background anisotropies have been published. In this work we study to which extent the cosmological parameters determined by using this data depend on assumptions about the initial conditions. We show that for generic initial conditions, not only the best fit values are very different but, and this is our main result, the allowed parameter range enlarges dramatically.

Figures (3)

• Figure 1: CMB anisotropy spectrum for different values of the cosmological parameters $\omega_{\rm b}$ and $h$. We have shown the best-fit corresponding to a purely adiabatic case (dashed line) and allowing the most general set of initial conditions (solid line). The calibration and the beam size of the BOOMERanG data have been optimized to fit the mixed model (red/dark error bars) or the adiabatic model (green/light error bars). The parameter choice on top ($\omega_{\rm b} = 0.02$, $h = 0.65$) can be fitted by both models while the values $\omega_{\rm b} = 0.042$, $h = 0.65$, can only be fitted by a mixed model.
• Figure 2: The likelihood contours of 50%, 68%, 95%, 99% are indicated in the $(\omega_{\rm b}, h)$ plane for mixed models (top) and for purely adiabatic models (bottom). The likelihoods are obtained by marginalization over the BOOMERanG calibration and beam size as well as over the initial conditions given by the matrix $M$ for mixed models and by the amplitude of the adiabatic mode for adiabatic models. For the mixed model, the lowest $\chi^2$ corresponds to even higher values of $\omega_{\rm b}$ and $h$ than those shown in the plot.
• Figure 3: The isocurvature content $\alpha$ of the best fit mixed model at fixed parameter values $(\omega_{\rm b}, h)$ is indicated. The contours $\alpha = 0.2$ to $0.9$ in steps of $0.1$ are shown.