Reconstruction of a direction-dependent primordial power spectrum from Planck CMB data

TL;DR

This paper investigates potential direction-dependent features in the primordial power spectrum using Planck CMB data, detecting a quadrupolar modulation with moderate significance and suggesting possible links to known CMB anomalies.

Contribution

It introduces a non-parametric reconstruction method for the scale-dependent primordial power spectrum allowing for direction dependence, extending previous analyses.

Findings

Detection of quadrupolar modulation at 2.2 sigma significance

Spectral features suggest anomalies at k ~ 0.006/Mpc

Preferred directions relate to CMB hemispherical asymmetry

Abstract

We consider the possibility that the primordial curvature perturbation is direction-dependent. To first order this is parameterised by a quadrupolar modulation of the power spectrum and results in statistical anisotropy of the CMB, which can be quantified using `bipolar spherical harmonics'. We compute these for the Planck DR2-2015 SMICA map and estimate the noise covariance from Planck Full Focal Plane 9 simulations. A constant quadrupolar modulation is detected with 2.2 sigma significance, dropping to 2 sigma when the primordial power is assumed to scale with wave number k as a power law. Going beyond previous work we now allow the spectrum to have arbitrary scale-dependence. Our non-parametric reconstruction then suggests several spectral features, the most prominent at k ~ 0.006/Mpc. When a constant quadrupolar modulation is fitted to data in the range 0.005 < k Mpc < 0.008, its preferred directions are found to be related to the cosmic hemispherical asymmetry and the CMB dipole. To determine the significance we apply two test statistics to our reconstructions of the quadrupolar modulation from data, against reconstructions of realisations of noise only. With a test statistic sensitive only to the amplitude of the modulation, the reconstructions from the multipole range 30 < l < 1200 are unusual with 2.1 sigma significance. With the second test statistic, sensitive also to the direction, the significance rises to 6.9 sigma. Our approach is easily generalised to include other data sets such as polarisation, large-scale structure and forthcoming 21-cm line observations which will enable these anomalies to be investigated further.