CMB signal in WMAP 3yr data with FastICA

Abstract

We present an application of the fast Independent Component Analysis (FastICA) to the WMAP 3yr data with the goal of extracting the CMB signal. We evaluate the confidence of our results by means of Monte Carlo simulations including CMB, foreground contaminations and instrumental noise specific of each WMAP frequency band. We perform a complete analysis involving all or a subset of the WMAP channels in order to select the optimal combination for CMB extraction, using the frequency scaling of the reconstructed component as a figure of merit. We found that the combination KQVW provides the best CMB frequency scaling, indicating that the low frequency foreground contamination in Q, V and W bands is better traced by the emission in the K band. The CMB angular power spectrum is recovered up to the degree scale, it is consistent within errors for all WMAP channel combination considered, and in close agreement with the WMAP 3yr results. We perform a statistical analysis of the recovered CMB pattern, and confirm the sky asymmetry reported in several previous works with independent techniques.