Density Perturbations and Primordial Non-Gaussianities in a Closed Universe

TL;DR

This paper develops an analytical framework to compute density perturbations and non-Gaussianities in a closed universe, highlighting potential observable signatures at large scales even with minimal inflation.

Contribution

It introduces a Hamiltonian formalism for density perturbations in a closed universe, deriving explicit expressions for correlators and exploring non-Gaussianities with potential observational implications.

Findings

Curvature effects are negligible at large multipoles, matching flat universe results.

Non-Gaussianities may be observable at large scales due to curvature effects.

New terms in $f_{NL}$ arise from curvature, absent in flat space, affecting large-scale observations.

Abstract

The spatial curvature of the universe is not yet known. Even though at present the Universe is very close to being essentially flat and most signatures of curvature appear to have been diluted by inflation, if the number of e-foldings during inflation is close to the minimum necessary to explain the horizon problem, the curvature of the universe may have left imprints in the cosmic microwave background (CMB) that may be observable, especially at large angles. Motivated by general results on quantum cosmology and using effective field theory techniques, we develop a general approach for analytically computing the power spectrum of density perturbations for a closed universe. Following a Hamiltonian formalism we determine the corresponding Bunch-Davis vacuum, find analytic expressions for two-point functions and higher correlators, expanding in terms of $S^3$ harmonics. In particular we concentrate on potential implications for observable non-Gaussianities. We consider cubic interactions as well as higher derivative ones to explore the consequence of a speed of sound $c_s\neq 1$. For large multipoles curvature effects are negligible and reproduce the known results for the flat case. However, they depart from the flat space result for relatively small multipoles. In this limit non-Gaussianities may lead to potentially observable values of $f_{\rm NL}$. In particular we find terms in $f_{\rm NL}$ that are absent in the flat space case, which are important at large scales and may be observable even if a long period of inflation dilutes the curvature. We compare our results with previous discussions in the literature.