Primordial Non-Gaussianity in the Effective Field Theory of Large Scale Structure

Abstract

In the search for understanding the initial conditions of the universe, it is crucial describe with high precision the evolution of the matter density in the universe on large scales as a function of time. The Effective Field Theory of Large Scale Structure (EFT of LSS) provides exactly this: it is a consistent perturbation theory of the evolution of the density on large scales, that parametrizes analytically unknown short scale behavior such as galaxy formation. In this thesis we discuss the inclusion of particular initial conditions – primordial non-Gaussianity (PNG) – in the EFT of LSS. It turns out that the evolution equations have to be modified slightly to capture the fact that for statistically non-Gaussian initial conditions, short scale physics in one place can be related to large scale physics in some other place. This forces us to introduce new, fluid-like parameters in the theory on top of the standard speed of sound, viscosity, and related parameters. We show how the EFT of LSS outperforms Standard Perturbation Theory (SPT), which does not properly take into account the backreaction from short scales, in allowing us to extract information about PNG, and forecast the constraints near-future galaxy surveys could put on PNG. We also comment on the relevance and subtleties involved in properly accounting for inevitable theoretical errors in forecasts. Finally we investigate the reach of any standard perturbative treatment to the evolution of the matter density field. We do this by means of a study in one spatial dimension, for which some exact results are known. We show that perturbation theory converges for a range of Fourier space observables, but find that it is asymptotic for at least some real space observables such as the two point correlation function. We argue that this asymptotic behavior is related to the theory’s inability to describe the evolution of statistically rare events in which the density field is very large even on large scales. This suggests that there might be a floor to how well any perturbative treatment can perform. We provide a tentative estimate of the size of this floor as a function of scale.