Oscillations in the Primordial Bispectrum: Mode Expansion

TL;DR

This paper develops an efficient Fourier mode expansion method to represent oscillatory primordial bispectra, enabling better constraints and detection of features from cosmological models with fewer modes.

Contribution

It introduces a Fourier basis for bispectrum expansion, reducing the number of modes needed and highlighting a resonance effect for improved detection of oscillatory signals.

Findings

Fourier basis reduces modes needed by 80% compared to polynomial basis.

Resonance effect depends on oscillation orientation, aiding shape and frequency extraction.

Fourier mode expansion can efficiently detect oscillatory bispectra in cosmological data.

Abstract

We consider the presence of oscillations in the primordial bispectrum, inspired by three different cosmological models; features in the primordial potential, resonant type non-Gaussianities and deviation from the standard Bunch Davies vacuum. In order to put constraints on their bispectra, a logical first step is to put these into factorized form which can be achieved via the recently proposed method of polynomial basis expansion on the tetrahedral domain. We investigate the viability of such an expansion for the oscillatory bispectra and find that one needs an increasing number of orthonormal mode functions to achieve significant correlation between the expansion and the original spectrum as a function of their frequency. To reduce the number of modes required, we propose a basis consisting of Fourier functions orthonormalized on the tetrahedral domain. We show that the use of Fourier mode functions instead of polynomial mode functions can lead to the necessary factorizability with the use of only 1/5 of the total number of modes required to reconstruct the bispectra with polynomial mode functions. Moreover, from an observational perspective, the expansion has unique signatures depending on the orientation of the oscillation due to a resonance effect between the mode functions and the original spectrum. This effect opens the possibility to extract informa- tion about both the frequency of the bispectrum as well as its shape while considering only a limited number of modes. The resonance effect is independent of the phase of the reconstructed bispectrum suggesting Fourier mode extraction could be an efficient way to detect oscillatory bispectra in the data.