We investigate primordial tensor non-Gaussianity in single field inflation, during a phase of nonattractor evolution when the spectrum of primordial tensor modes can be enhanced to a level detectable at interferometer scales. Making use of a tensor duality we introduced in [1], we analytically compute the full bispectrum of primordial tensor fluctuations during the non-attractor era. During this epoch the shape of the tensor bispectrum is enhanced in the squeezed limit, its amplitude can be amplified with respect to slow-roll models, and tensor non-Gaussianity can exhibit a scale dependence distinctive of our set-up. We prove that our results do not depend on the frame used for the calculations. Squeezed tensor non-Gaussianity induces a characteristic quadrupolar anisotropy on the power spectrum of the stochastic background of primordial tensor perturbations. As a step to make contact with gravitational wave experiments, we discuss the response function of a ground based Michelson interferometer to a gravitational wave background with such a feature.

We introduce the concept of stationary graviton non-Gaussianity (nG), an observable that can be probed in terms of 3-point correlation functions of a stochastic gravitational wave (GW) background. When evaluated in momentum space, stationary nG corresponds to folded bispectra of graviton nG. We determine 3-point overlap functions for testing stationary nG with pulsar timing array GW experiments, and we obtain the corresponding optimal signal-tonoise ratio. For the first time, we consider 3-point overlap functions including scalar graviton polarizations (which can be motivated in theories of modified gravity); moreover, we also calculate 3-point overlap functions for correlating pulsar timing array with ground based GW detectors. The value of the optimal signal-to-noise ratio depends on the number and position of monitored pulsars. We build geometrical quantities characterizing how such ratio depends on the pulsar system under consideration, and we evaluate these geometrical parameters using data from the IPTA collaboration. We quantitatively show how monitoring a large number of pulsars can increase the signal-to-noise ratio associated with measurements of stationary graviton nG. arXiv:1910.04758v1 [gr-qc]

The aim of this research is to use modern techniques in scalar field Cosmol-ogy to produce methods of detecting gravitational waves and apply them to current gravitational waves experiments and those that will be producing results in the not too distant future. In the first chapter we discuss dark matter and some of its candidates, specifically, the axion. We then address its relationship with gravitational waves. We also discuss inflation and how it can be used to detect gravitational waves. Chapter 2 concentrates on constructing a multi field system of axions in order to increase the mass range of the ultralight axion, putting it into the observation range of pul-sar timing arrays. Chapter 3 discusses non-attractor inflation which is able to enhance stochastic background gravitational waves at scales that allows them to be measured by gravitational wave experiments. Chapter 4 uses a similar method to chapter 3 and applies it to 3-point overlap functions for tensor, scalar and a combination of the two polarisations.