Adiabatic versus Isocurvature Non--Gaussianity

TL;DR

This paper investigates how to distinguish between adiabatic and isocurvature primordial non-Gaussianities in the CMB, analyzing their correlations, scale dependence, and contamination effects to improve differentiation methods.

Contribution

It introduces a joint analysis framework for multiple NG models, utilizing the skew spectrum estimator to separate and compare adiabatic and isocurvature signals in CMB data.

Findings

Adiabatic and isocurvature modes are strongly correlated but distinguishable by phase differences.

Point source contamination can be effectively separated at high multipoles (l>100).

Including correlations increases NG parameter errors by 20-30% for WMAP and 5% for Planck.

Abstract

We study the extent to which one can distinguish primordial non--Gaussianity (NG) arising from adiabatic and isocurvature perturbations. We make a joint analysis of different NG models based on various inflationary scenarios: local-type and equilateral-type NG from adiabatic perturbations and local-type and quadratic-type NG from isocurvature perturbations together with a foreground contamination by point sources. We separate the Fisher information of the bispectrum of CMB temperature and polarization maps by l for the skew spectrum estimator introduced by Munshi & Heavens (2009) to study the scale dependence of the signal-to-noise ratio of different NG components and their correlations. We find that the adiabatic and the isocurvature modes are strongly correlated, though the phase difference of acoustic oscillations helps to distinguish them. The correlation between local-and equilateral-type is weak, but the two isocurvature modes are too strongly correlated to be discriminated. Point source contamination, to the extent to which it can be regarded as white noise, can be almost completely separated from the primordial components for l>100. Including correlations among the different components, we find that the errors of the NG parameters increase by 20-30% for the WMAP 5-year observation, but 5% for Planck observations.