CMB component separation in the pixel domain

TL;DR

This paper introduces a new pixel domain method for CMB component separation that improves stability over traditional ILC, providing more accurate estimates of the power spectrum and addressing anomalies in WMAP data.

Contribution

The authors propose a novel approach for CMB component separation in the pixel domain that enhances stability and accuracy, reducing WMAP anomalies and improving power spectrum estimates.

Findings

More stable CMB signal restoration compared to ILC

Consistent quadrupole estimate with theoretical predictions

Suppression of the 'axis of evil' effect and increased hemispherical symmetry

Abstract

We show that the popular ILC approach is unstable in respect to the division of the sample of map pixels to the set of ``homogeneous'' subsamples. For suitable choice of such subsamples we can obtain the restored CMB signal with amplitudes ranged from zero to the amplitudes of the observed signal. We propose approach which allows us to obtain reasonable estimates of $C_\ell$ at $\ell\leq 30$ and similar to WMAP $C_\ell$ for larger $\ell$. With this approach we reduce some anomalies of the WMAP results. In particular, our estimate of the quadrupole is well consistent to theoretical one, the effect of the ``axis of evil'' is suppressed and the symmetry of the north and south galactic hemispheres increases. This results can change estimates of quadrupole polarization and the redshift of reionization of the Universe. We propose also new simple approach which can improve WMAP estimates of high $\ell$ power spectrum.