The lack of power anomaly is an unexpected feature observed at large angular scales in the maps of Cosmic Microwave Background (CMB) produced by the COBE, WMAP and Planck satellites. This signature, which consists in a missing of power with respect to that predicted by the ΛCDM model, might hint at a new cosmological phase before the standard inflationary era.
The main point of this paper is taking into account the latest Planck polarisation data to investigate how the CMB polarisation improves the understanding of this feature. With this aim, we apply to the latest Planck data, both PR3 (2018) and PR4 (2020) releases, a new class of estimators capable of evaluating this anomaly by considering temperature and polarisation data both separately and in a jointly way. This is the first time that the PR4 dataset has been used to study this anomaly. To critically evaluate this feature, taking into account the residuals of known systematic effects present in the Planck datasets, we analyse the cleaned CMB maps using different combinations of sky masks, harmonic range and binning on the CMB multipoles.
Our analysis shows that the estimator based only on temperature data confirms the presence of a lack of power with a lower-tail-probability (LTP), depending on the component separation method, ≤ 0.33% and ≤ 1.76% for PR3 and PR4, respectively. To our knowledge, the LTP≤ 0.33% for the PR3 dataset is the lowest one present in the literature obtained from Planck 2018 data, considering the Planck confidence mask. We find significant differences between these two datasets when polarisation is taken into account most likely due to a different level of systematics. Especially, the analysis with PR3 data, unlike that with PR4, seems to point towards a lack of power at large scales also for polarisation.
Moreover, we also show that for the PR3 dataset the inclusion of the subdominant polarisation information provides estimates that are less likely accepted in a ΛCDM cosmological model than the only-temperature analysis over the entire harmonic-range considered. In particular, at ℓmax = 26, we found that no simulation has a value as low as the data for all the pipelines.