Impact of Secondary non-Gaussianities on the Search for Primordial Non-Gaussianity with CMB Maps

Abstract

When constraining the primordial non-Gaussianity parameter f_NL with cosmic microwave background anisotropy maps, the bias resulting from the covariance between primordial non-Gaussianity and secondary non-Gaussianities to the estimator of f_NL is generally assumed to be negligible. We show that this assumption may not hold when attempting to measure the primordial non-Gaussianity out to angular scales below a few tens arcminutes with an experiment like Planck, especially if the primordial non-Gaussianity parameter is around the minimum detectability level with f_NL between 5 and 10. In future, it will be necessary to jointly estimate the combined primordial and secondary contributions to the CMB bispectrum and establish f_NL by properly accounting for the confusion from secondary non-Gaussianities.

Figures (2)

• Figure 1: Absolute values of signal-to-noise ratio squared for the detection of primordial bispectrum (black lines) assuming $f_{\rm NL}=1$ as a function of $l_3$. The signal-to-noise ratio squared for WMAP and Planck are shown with dashed and dot-dashed line, respectively. The red, blue, and green lines show the confusion resulting from the covariance between primary and point source, primary and SZ-lensing, and primary and ISW-lensing bispectra, respectively.
• Figure 2: The maximum non-Gaussianity measured with an optimal estimator for the primordial bispectrum $f_{\rm NL}^{\rm tot}$, which includes the true underlying primordial non-Gaussianity with $f_{\rm NL}$ as labeled on the figure and the bias correction coming from the unaccounted secondary anisotropies. The bias is generally small and non existing if primordial non-Gaussianity measurements are limited out to $l < 500$, but depending on the value of $f_{\rm NL}$ and the residual point source contamination, the correction is generally a factor of 1.5 to 2. If $f_{\rm NL} \lesssim 10$, for Planck, it is necessary to account for secondary non-Gaussianities properly.