Multifield positivity bounds for inflation

bstract Positivity bounds represent nontrivial limitations on effective field theories (EFTs) if those EFTs are to be completed into a Lorentz-invariant, causal, local, and unitary framework. While such positivity bounds have been applied in a wide array of physical contexts to obtain useful constraints, their application to inflationary EFTs is subtle since Lorentz invariance is spontaneously broken during cosmic inflation. One path forward is to employ a Breit parameterization to ensure a crossing-symmetric and analytic S-matrix in theories with broken boosts. We extend this approach to a theory with multiple fields, and uncover a fundamental obstruction that arises unless all fields obey a dispersion relation that is approximately lightlike. We then apply the formalism to various classes of inflationary EFTs, with and without isocurvature perturbations, and employ this parameterization to derive new positivity bounds on such EFTs. For multifield inflation, we also consider bounds originating from the generalized optical theorem and demonstrate how these can give rise to stronger constraints on EFTs compared to constraints from traditional elastic positivity bounds alone. We compute various shapes of non-Gaussianity (NG), involving both adiabatic and isocurvature perturbations, and show how the observational parameter space controlling the strength of NG can be constrained by our bounds.