Graviton non-Gaussianities and Parity Violation in the EFT of Inflation

TL;DR

This work analyzes tensor non-Gaussianities in the EFT of Inflation, showing how foliation-coupled higher-derivative operators modify the graviton sector beyond Einstein gravity. At leading order, graviton correlators reproduce Maldacena's results, while at next-to-leading order parity-even and parity-odd operators introduce new structures in the tensor bispectrum and, for certain NNLO operators, sizable enhancements in scalar-tensor-tensor correlators. The in-in calculations reveal that parity-odd operators at NLO affect the γγγ bispectrum and helicity splitting, but do not appreciably enhance mixed correlators at leading slow-roll; NNLO analyses identify two parity-even and two parity-odd operators with observable consequences, including potential large ⟨ζγγ⟩ when H^2/Λ^2 is sizeable relative to ε. The results have implications for primordial gravity experiments and contribute to the broader boostless cosmological bootstrap program by constraining allowed higher-derivative interactions during inflation.

Abstract

We study graviton non-Gaussianities in the EFT of Inflation. At leading (second) order in derivatives, the graviton bispectrum is fixed by Einstein gravity. There are only two contributions at third order. One of them breaks parity. They come from operators that directly involve the foliation: we then expect sizable non-Gaussianities in three-point functions involving both gravitons and scalars. However, we show that at leading order in slow roll the parity-odd operator does not modify these mixed correlators. We then identify the operators that can affect the graviton bispectrum at fourth order in derivatives. There are two operators that preserve parity. We show that one gives a scalar-tensor-tensor three-point function larger than the one computed in Maldacena, 2003 if $M^2_{\rm P}A_{\rm s}/Λ^2\gg 1$ (where $Λ$ is the scale suppressing this operator and $A_{\rm s}$ the amplitude of the scalar power spectrum). There are only two parity-odd operators at this order in derivatives.