
A statistical property of primordial fluctuations indicating non-random, skewed distributions.Non-Gaussianity refers to any deviation from a perfect Gaussian (normal) distribution in the statistical properties of primordial fluctuations — the tiny density variations in the early universe that seeded the formation of galaxies and large-scale structure. In the simplest inflationary models, these fluctuations are predicted to be nearly perfectly Gaussian, with random phases and a symmetric bell-shaped distribution. Any measurable non-Gaussianity would therefore be a powerful fingerprint of the specific physics that operated during the universe's first moments, potentially revealing interactions or exotic fields that simple models overlook.
The concept was introduced in the early 1980s alongside inflationary theory, with key contributions from theorists like Alexei Starobinsky and later developed by others in the 1990s–2000s, but no single definitive originator is universally credited.
Primordial fluctuations are typically characterized by their power spectrum, which assumes Gaussian statistics. Non-Gaussianity is quantified through higher-order correlation functions, most commonly the bispectrum (three-point correlation). A nonzero bispectrum indicates that fluctuations are not independent but correlated in a way that skews the distribution. In inflationary theory, non-Gaussianity arises when the field driving inflation has self-interactions or when additional light fields contribute to the curvature perturbation, producing characteristic shapes (e.g., local, equilateral, orthogonal) that can be searched for in cosmic microwave background (CMB) maps and large-scale structure surveys.
Searching for non-Gaussianity in CMB data (e.g., from Planck satellite) and galaxy surveys is a major observational program to test and discriminate between inflationary models.
The Planck satellite's measurements have constrained local non-Gaussianity to f_NL = -0.9 ± 5.1, consistent with zero but still leaving room for subtle effects.













