The ISW effect and the lack of large-angle CMB temperature correlations

TL;DR

This paper investigates the contributions of the integrated Sachs-Wolfe effect to the large-angle temperature correlations in the CMB, revealing that cancellations between early and late contributions can explain the observed anomaly.

Contribution

It provides a detailed analysis of the ISW effect's role in the large-angle CMB correlation anomaly within the ΛCDM model, highlighting the importance of cross-terms.

Findings

Early contributions dominate $S_{1/2}$ in ΛCDM

Low $S_{1/2}$ realizations involve cancellations

Pure late ISW auto-correlations are statistically unremarkable

Abstract

It is by now well established that the magnitude of the two-point angular-correlation function of the cosmic microwave background temperature anisotropies is anomalously low for angular separations greater than about 60 degrees. Physics explanations of this anomaly typically focus on the properties of the Universe at the surface of last scattering, relying on the fact that large-angle temperature fluctuations are dominated by the Sachs-Wolfe effect (SW). However, these fluctuations also receive important contributions from the integrated Sachs-Wolfe effect (ISW) at both early (eISW) and late ($\ell$ISW) times. Here we study the correlations in those large-angle temperature fluctuations and their relative contributions to $S_{1/2}$ -- the standard measure of the correlations on large angular scales. We find that in the best-fitting $\Lambda$CDM cosmology, while the auto-correlation of the early contributions (SW plus eISW) dominates $S_{1/2}$, there are also significant contributions originating from cross-terms between the early and late contributions. In particular, realizations of $\Lambda$CDM with low $S_{1/2}$ are typically produced from a combination of somewhat low pure-early correlations and accidental cancellations among early-late correlations. We also find that if the pure $\ell$ISW auto-correlations were the only contribution to $S_{1/2}$ in $\Lambda$CDM, then the $p$-value of the observed cut-sky $S_{1/2}$ would be unremarkable. This suggests that physical mechanisms operating only at or near the last scattering surface could explain the observed lack of large-angle correlations, though this is not the typical resolution within $\Lambda$CDM.