Landscape tomography through primordial non-Gaussianity

TL;DR

This paper explores how primordial non-Gaussianities in density perturbations, caused by isocurvature fields crossing potential barriers during inflation, can reveal the landscape structure of the early Universe.

Contribution

It introduces a novel method to probe the landscape potential via non-Gaussian features in primordial perturbations, using the full probability distribution function instead of polyspectra.

Findings

Non-Gaussian features in curvature perturbations reflect landscape potential structure.

Probability distribution function shows oscillations due to axionlike isocurvature fields.

Method can detect small-scale features in the primordial landscape.

Abstract

In this paper, we show how the structure of the landscape potential of the primordial Universe may be probed through the properties of the primordial density perturbations responsible for the origin of the cosmic microwave background anisotropies and the large-scale structure of our Universe. Isocurvature fields -fields orthogonal to the inflationary trajectory- may have fluctuated across the barriers separating local minima of the landscape potential during inflation. We analyze how this process could have impacted the evolution of the primordial curvature perturbations. If the typical distance separating consecutive minima of the landscape potential and the height of the potential barriers are smaller than the Hubble expansion rate parametrizing inflation, the probability distribution function of isocurvature fields becomes non-Gaussian due to the appearance of bumps and dips associated to the structure of the potential. We show that this non-Gaussianity can be transferred to the statistics of primordial curvature perturbations if the isocurvature fields are coupled to the curvature perturbations. The type of non-Gaussian structure that emerges in the distribution of curvature perturbations cannot be fully probed with the standard methods of polyspectra; instead, the probability distribution function is needed. The latter is obtained by summing all the $n$-point correlation functions. To substantiate our claims, we offer a concrete model consisting of an axionlike isocurvature perturbation with a sinusoidal potential and a linear derivative coupling between the isocurvature and curvature fields. In this model, the probability distribution function of the curvature perturbations consists of a Gaussian function with small superimposed oscillations reflecting the isocurvature axion potential.