Power Suppression at Large Scales in String Inflation

TL;DR

This paper explores how a sharp transition in string inflation models can explain the observed large-scale power suppression in the cosmic microwave background, linking it to background dynamics and string coupling tuning.

Contribution

It introduces a string inflation model where a transition from fast-roll to slow-roll inflation causes power suppression, with the number of e-foldings tied to the string coupling.

Findings

Power suppression arises from a transition in inflation dynamics.

Small string coupling naturally yields sufficient e-foldings.

The transition causes a shift from blue to red spectrum, explaining large-scale anomalies.

Abstract

We study a possible origin of the anomalous suppression of the power spectrum at large angular scales in the cosmic microwave background within the framework of explicit string inflationary models where inflation is driven by a closed string modulus parameterizing the size of the extra dimensions. In this class of models the apparent power loss at large scales is caused by the background dynamics which involves a sharp transition from a fast-roll power law phase to a period of Starobinsky-like slow-roll inflation. An interesting feature of this class of string inflationary models is that the number of e-foldings of inflation is inversely proportional to the string coupling to a positive power. Therefore once the string coupling is tuned to small values in order to trust string perturbation theory, enough e-foldings of inflation are automatically obtained without the need of extra tuning. Moreover, in the less tuned cases the sharp transition responsible for the power loss takes place just before the last 50-60 e-foldings of inflation. We illustrate these general claims in the case of Fibre Inflation where we study the strength of this transition in terms of the attractor dynamics, finding that it induces a pivot from a blue to a redshifted power spectrum which can explain the apparent large scale power loss. We compute the effects of this pivot for example cases and demonstrate how magnitude and duration of this effect depend on model parameters.