Non-Gaussianity from Symmetry

TL;DR

The paper proposes ungaussitons, light scalar fields near symmetry-enhanced points during inflation, as a generic source of large primordial non-Gaussianity. It leverages the $\delta N$ formalism to show that symmetry suppresses linear $\sigma$-terms so higher-order fluctuations dominate, producing measurable bispectrum and trispectrum signals with a logarithmic scale dependence. A key result is a universal consistency relation $\tau_{\rm NL} \approx 1\times10^3 C f_{\rm NL}^{4/3}$ (with $C$ of order unity) that ties the trispectrum to the bispectrum, plus model-dependent predictions for the scale dependence. The paper discusses concrete realizations (moduli, right-handed sneutrinos, and flat directions with $Q$-balls), the required cosmological timing (late decays after reheating but before BBN), and observational implications that could distinguish ungaussitons from curvaton- and other non-Gaussianity scenarios if the relation is confirmed by Planck-era data.

Abstract

We point out that a light scalar field fluctuating around a symmetry-enhaced point can generate large non-Gaussianity in density fluctuations. We name such a particle as an "ungaussiton", a scalar field dominantly produced by the quantum fluctuations,generating sizable non-Gaussianity in the density fluctuations. We derive a consistency relation between the bispectrum and the trispectrum, tau_NL = 10^3 f_NL^(4/3), which can be extended to arbitrary high order correlation functions. If such a relation is confirmed by future observations, it will strongly support this mechanism.

Figures (1)

• Figure 1: Contours of the non-linearity parameters $f_{\rm NL}$ and $\tau_{\rm NL}$; $(f_{\rm NL},\tau_{\rm NL}) = (100, 5\times 10^5)$ (solid line) and $(1,10^3)$ (dotted line). Shaded region is already excluded by the WMAP 5yr data. Here we set $g_{*{\rm osc}}=g_{*d}=g_{*R}=100$ and ${\bar{\sigma}}=H_I/2\pi$, and drop the logarithmic dependence.