Wednesday, November 16, 2011

1111.3390 (A. Curto et al.)

Why wavelets provide an optimal estimator for the primordial non-Gaussianity?    [PDF]

A. Curto, E. Martinez-Gonzalez, R. B. Barreiro
We present a comprehensive study of the wavelet estimator to detect primordial non-Gaussianity in Cosmic Microwave Background data. We first summarise the main assumptions made to derive the estimator based on the direct bispectrum in harmonic space. We then analyse the main statistical properties of the wavelet estimator and show the conditions to reach optimality. In particular, we clarify some points regarding the spherical Mexican hat wavelet fnl estimator and the recent issue raised about the need to introduce a linear term to achieve optimality. We define the likelihood for the fnl parameter in terms of the cubic statistics constructed from the wavelet coefficient maps. We compute the fnl estimator from the previous likelihood. The analytical results obtained are checked with simulations containing different degrees of complexity, including anisotropic noise. The results of this analysis show that, in the particular case of this estimator, the optimality is achieved without the need of including a linear term. For illustration we analyse simulated CMB maps with resolution and noise level matching those of the WMAP V+W 7-year.
View original: http://arxiv.org/abs/1111.3390

No comments:

Post a Comment