Inhomogeneous Initial Data and Small-Field Inflation

TL;DR

This paper investigates the robustness of small-field inflation against initial scalar field inhomogeneities, demonstrating that inflation can withstand such inhomogeneities across various models and deriving conditions for stability.

Contribution

It introduces a stability condition for small-field inflation considering inhomogeneities from different pre-inflationary scenarios, expanding understanding of initial data robustness.

Findings

Inflation remains stable despite initial inhomogeneities in broad models.

Derived a stability condition depending on Hubble scales and pre-inflationary properties.

Established a lower limit on the tensor-to-scalar ratio for Planck-scale initial data.

Abstract

We consider the robustness of small-field inflation in the presence of scalar field inhomogeneities. Previous numerical work has shown that if the scalar potential is flat only over a narrow interval, such as in commonly considered inflection-point models, even small-amplitude inhomogeneities present at the would-be onset of inflation at $\tau = \tau_i$ can disrupt the accelerated expansion. In this paper, we parametrise and evolve the inhomogeneities from an earlier time $\tau_{IC}$ at which the initial data were imprinted, and show that for a broad range of inflationary and pre-inflationary models, inflection-point inflation withstands initial inhomogeneities. We consider three classes of perturbative pre-inflationary solutions (corresponding to energetic domination by the scalar field kinetic term, a relativistic fluid, and isotropic negative curvature), and two classes of exact solutions to Einstein's equations with large inhomogeneities (corresponding to a stiff fluid with cylindrical symmetry, and anisotropic negative curvature). We derive a stability condition that depends on the Hubble scales $H(\tau_ i)$ and $H(\tau_{IC})$, and a few properties of the pre-inflationary cosmology. For initial data imprinted at the Planck scale, the absence of an inhomogeneous initial data problem for inflection-point inflation leads to a novel, lower limit on the tensor-to-scalar ratio.