Foreground influence on primordial non-Gaussianity estimates: needlet analysis of WMAP 5-year data

TL;DR

This paper assesses the impact of foreground residuals on primordial non-Gaussianity estimates in CMB data, using needlet bispectrum analysis of WMAP 5-year data to improve accuracy.

Contribution

It introduces a generalized needlet bispectrum estimator that marginalizes over foreground residuals, providing more reliable fNL estimates in CMB analysis.

Findings

Estimated fNL=38±47 with marginalization, compared to 35±42 without.

Foreground residuals are positively correlated with dust and synchrotron, negatively with free-free.

Foreground marginalization slightly increases the uncertainty in fNL estimates.

Abstract

We constrain the amplitude of primordial non-Gaussianity in the CMB data taking into account the presence of foreground residuals in the maps. We generalise the needlet bispectrum estimator marginalizing over the amplitudes of thermal dust, free-free and synchrotron templates. We apply our procedure to WMAP 5 year data, finding fNL= 38\pm 47 (1 \sigma), while the analysis without marginalization provides fNL= 35\pm 42. Splitting the marginalization over each foreground separately, we found that the estimates of fNL are positively cross correlated of 17%, 12% with the dust and synchrotron respectively, while a negative cross correlation of about -10% is found for the free-free component.