The analysis of the CMB anisotropies is a rich source of cosmological informations. In our study, we simulated the signals produced by the relics of a first order phase transition occured during an inflationary epoch in the early Universe. These relics are bubbles of true vacuum that leave a characteristic non-Gaussian imprint on the CMB. We use different statistical estimators in order to evaluate this non-Gaussianity. We obtain some limits on the allowed values of the bubble parameters comparing our results with the experimental data. We also predict the possibility to detect this signal with the next high resolution experiments.

We calculate analytically and numerically the Dyer-Roeder distance in perfect fluid quintessence models and give an accurate fit to the numerical solutions for all the values of the density parameter and the quintessence equation of state. Then we apply our solutions to the estimation of $H_{0}$ from multiple image time delays and find that the inclusion of quintessence modifies sensibly the likelihood distribution of $H_{0}$, generally reducing the best estimate with respect to a pure cosmological constant. Marginalizing over the other parameters ($\Omega_{m}$ and the quintessence equation of state), we obtain $H_{0}=71\pm 6$ km/sec/Mpc for an empty beam and $H_{0}=64\pm 4$ km/sec/Mpc for a filled beam. We also discuss the future prospects for distinguishing quintessence from a cosmological constant with time delays.

The existence of large scale voids in several galaxy surveys suggests the occurence of an inflationary first order phase transition. This process generates primordial bubbles that, before evolving into the present voids, leave at decoupling a non-Gaussian imprint on the CMB. I this paper we evaluate an analytical expression of the collapsed three point correlation function from the bubble temperature fluctuations. Comparing the results with COBE-DMR measures, we obtain upper limits on the allowed non-Gaussianity and hence on the bubble parameters.

We investigate to which extent a time variation of the gravitational constant or other fundamental constants affects the best fit of the Hubble diagram of type Ia supernovae. In particular, we show that a slow increase of $G$ in the past, below experimental constraints, can reconcile the SNe Ia observations with an open zero-$\Lambda$ universe.

We generalize in several ways the results existing in the literature: a) we make use of an exact general relativistic solution for a spherical, nearly empty cavity in the matter dominated era to evaluate the null geodesics and the Sachs-Wolfe effect; b) we evaluate the magnitude of the adiabatic fluctuations of the photon-baryon plasma; c) we study the influence of the shell profile; and d) we take into account the finite thickness of the last scattering surface (LSS) and the influence of its position with respect to the void center. We find empirically an analytic approximation to the Sachs-Wolfe effect for all crossing geometries and we derive an upper limit of $\approx$ 25 $h^{-1}$ Mpc for the comoving radii of voids sitting on the LSS in order to achieve compatibility with COBE's data. As a nearly empty void has an overcomoving expansion of a factor of $\approx$ 4 between decoupling and the present, the maximum allowed size at present is $\approx$ 100 $h^{-1}$ Mpc. On the other hand, the smallness of the comoving size relative to the sound horizon reduces strongly the adiabatic effect by Silk damping and makes it negligible. Most of the signature of primordial voids comes therefore from metric effects and consists of subdegree spots blue or red depending on whether the center lies beyond or within the LSS. In conclusion we refine and confirm earlier constraints on a power law void spectrum originated in an inflationary phase transition and capable of generating the observed large scale structure.

We calculate analytically the bubble nucleation rate in a model of first order inflation which is able to produce large scale structure. The computation includes the first-order departure from the thin-wall limit, the explicit derivation of the pre-exponential factor, and the gravitational correction. The resulting bubble spectrum is then compared with constraints from the large scale structure and the microwave background. We show that there are models which pass all the constraints and produce bubble-like perturbations of interesting size. Furthermore, we show that it is in principle possible to reconstruct completely the inflationary two-field potential from observations.

It is already understood that the increasing observational evidence for an open Universe may be reconciled with inflation if our horizon is contained inside one single huge bubble nucleated during the inflationary phase transition. In the scenario we present here, the Universe consists of infinitely many superhorizon bubbles, like our own, the distribution of which can be made to peak at $\Omega_0\approx 0.2$. Therefore, unlike the existing literature, we do not have to rely upon the anthropic principle nor upon special initial conditions.

Following a suggestion by Gasperini and Veneziano, that String Cosmology can be reconciled with Inflation and, hence, with the Standard Big Bang, we display an analytical solution which possesses four interesting properties: (1) it is non-singular; (2) it distinguishes the dynamics of the external scale factor, $a(t)$, from that of the internal one, $b(t)$; (3) it exhibits a non-monotonic behavior of $a(t)$; and (4) it stabilizes both Newton's constant and $b(t)$ (the latter to a finite, non-vanishing value). The interest of the non-monotonic evolution of $a(t)$ consists in the fact that it contains three phases of accelerated expansion, contraction and expansion before the final decelerated expansion which eventually becomes the Standard Big Bang. The total number of e-folds of the three accelerated eras can be calculated and tuned to fit the requests of observational astronomy.

A generic outcome of theories with scalar-tensor coupling is the existence of inflationary attractors, either power-law or de Sitter. The fluctuations arising during this phase are Gaussian and their spectrum depends on the wavenumber $k$ according to the power-law $k^{1/(1-p)}$, where $p$ is the inflationary power-law exponent. We investigate to which extent these properties depend on the coupling function and on the potential. We find the class of models in which viable attractors exist. Within this class, we find that the cosmic expansion and the scaling of the fluctuation spectrum are independent of the coupling function. Further, the analytical solution of the Fokker-Planck equation shows that the deviations from Gaussianity are negligible.