Two statistical procedures for mapping large-angle non-Gaussianity

Research paper by W. Cardona, A Bernui, M. J. Reboucas

Indexed on: 09 Oct '13Published on: 09 Oct '13Published in: arXiv - Astrophysics - Cosmology and Nongalactic Astrophysics


A convincing detection of primordial non-Gaussianity in the cosmic background radiation (CMB) is essential to probe the physics of the early universe. Since a single statistical estimator can hardly be suitable to detect the various possible forms of non-Gaussianity, it is important to employ different statistical indicators to study non-Gaussianity of CMB. This has motivated the proposal of a number statistical tools, including two large-angle indicators based on skewness and kurtosis of spherical caps of CMB sky-sphere. Although suitable to detect fairly large non-Gaussianity they are unable to detect non-Gaussianity within the Planck bounds, and exhibit power spectra with undesirable oscillation pattern. Here we use several thousands simulated CMB maps to examine interrelated problems regarding advances of these spherical patches procedures. We examine whether a change in the choice of the patches could enhance the sensitivity of the procedures well enough to detect large-angle non-Gaussianity within the Planck bounds. To this end, a new statistical procedure with non-overlapping cells is proposed and its capability is established. We also study whether this new procedure is capable to smooth out the undesirable oscillation pattern in the skewness and kurtosis power spectra of the spherical caps procedure. We show that the new procedure solves this problem, making clear this unexpected power spectra pattern does not have a physical origin, but rather presumably arises from the overlapping obtained with the spherical caps approach. Finally, we make a comparative analysis of this new statistical procedure with the spherical caps routine, determine their lower bounds for non-Gaussianity detection, and make apparent their relative strength and sensitivity.