On the significance of power asymmetries in Planck CMB data at all scales

TL;DR

This study thoroughly analyzes Planck CMB data for hemispherical and dipolar asymmetries, accounting for velocity effects, and finds no significant deviations from isotropy across all scales.

Contribution

It introduces a new bias-mitigation method using antipodal masks and properly accounts for Doppler and aberration effects in CMB analysis.

Findings

No significant hemispherical anomalies detected (less than 1.5 sigma).

Dipolar modulation shows a 3.3 sigma discrepancy when aligned with the most asymmetric direction.

Proper removal of Doppler and aberration effects reduces false anomalies at high ell.

Abstract

We perform an analysis of the CMB temperature data taken by the Planck satellite investigating if there is any significant deviation from cosmological isotropy. We look for differences in the spectrum between two opposite hemispheres and also for dipolar modulations. We propose a new way to avoid biases due to partial-sky coverage by producing a mask symmetrized in antipodal directions, in addition to the standard smoothing procedure. We also properly take into account both Doppler and aberration effects due to our peculiar velocity and the anisotropy of the noise, since these effects induce a significant hemispherical asymmetry. We are thus able to probe scales all the way to ell = 2000. After such treatment we find no evidence for significant hemispherical anomalies (i.e. deviations are less than 1.5 sigma when summing over all scales). Although among the larger scales there are sometimes higher discrepancies, these are always less than 3 sigma. We also find results on a dipolar modulation of the power spectrum. Along the hemispheres aligned with the most asymmetric direction for 2 <= ell <= 2000 we find a 3.3 sigma discrepancy when comparing to simulations. However, if we do not restrict ourselves to Planck's maximal asymmetry axis, which can only be known a posteriori, and compare Planck data with the modulation of simulations along their respective maximal asymmetry directions, the discrepancy goes down to less than 1 sigma (with, again, almost 3 sigma discrepancies in some low-ell modes). We thus conclude that no significant power asymmetries seem to be present in the full data set. Interestingly, without proper removal of Doppler and aberration effects one would find spurious anomalies at high ell, between 3 sigma and 5 sigma. Even when considering only ell < 600 we find that the boost is non-negligible and alleviates the discrepancy by roughly half-sigma.