Effective Field Theories and Inflation

TL;DR

The paper investigates whether high-energy physics beyond the inflationary scale can modify CMB predictions despite decoupling expectations. By studying a toy EFT and supersymmetric realizations (global SUSY and supergravity), it shows that heavy-field loops can generate slow-roll corrections that scale as $M_0^2/M^2$ with $M_0 \gg H$, and that in SUSY contexts these effects can be sizable and potentially observable while maintaining theoretical control. The results imply that heavy physics need not be completely invisible in the CMB, though detectability hinges on prior assumptions about the inflationary mechanism and the underlying high-energy theory. This work sharpens criteria for when trans-Planckian physics could leave an imprint in cosmological data and guides how to search for deviations from standard inflationary predictions.

Abstract

We investigate the possible influence of very-high-energy physics on inflationary predictions focussing on whether effective field theories can allow effects which are parametrically larger than order H^2/M^2, where M is the scale of heavy physics and H is the Hubble scale at horizon exit. By investigating supersymmetric hybrid inflation models, we show that decoupling does not preclude heavy-physics having effects for the CMB with observable size even if H^2/M^2 << O(1%), although their presence can only be inferred from observations given some a priori assumptions about the inflationary mechanism. Our analysis differs from the results of hep-th/0210233, in which other kinds of heavy-physics effects were found which could alter inflationary predictions for CMB fluctuations, inasmuch as the heavy-physics can be integrated out here to produce an effective field theory description of low-energy physics. We argue, as in hep-th/0210233, that the potential presence of heavy-physics effects in the CMB does not alter the predictions of inflation for generic models, but does make the search for deviations from standard predictions worthwhile.