The Friedman-Lemaitre-Robertson-Walker (FLRW) cosmological models are based on the assumptions of large-scale homogeneity and isotropy of the distribution of matter and energy. They are usually taken to have spatial sections that are simply connected; they have finite volume in the positive curvature case, and infinite volume in the null and negative curvature ones. I want to call the attention to the existence of an infinite number of models, which are based on these same metrics, but have compact, finite volume, multiply connected spatial sections. Some observational implications are briefly mentioned.

This paper deals with two aspects of relativistic cosmologies with closed (compact and boundless) spatial sections. These spacetimes are based on the theory of General Relativity, and admit a foliation into space sections S(t), which are spacelike hypersurfaces satisfying the postulate of the closure of space: each S(t) is a 3-dimensional, closed Riemannian manifold. The discussed topics are: (1) A comparison, previously obtained, between Thurston's geometries and Bianchi-Kantowski-Sachs metrics for such 3-manifolds is here clarified and developed. (2) Some implications of global inhomogeneity for locally homogeneous 3-spaces of constant curvature are analyzed from an observational viewpoint.

The cosmic crystallography method of Lehoucq et al. [1] produces sharp peaks in the distribution of distances between the images of cosmic sources. But the method cannot be applied to universes with compact spatial sections of negative curvature. We apply to these a second order crystallographic effect, as evidenced by statistical parameters.

The original Casimir effect results from the difference in the vacuum energies of the electromagnetic field, between that in a region of space with boundary conditions and that in the same region without boundary conditions. In this paper we develop the theory of a similar situation, involving a scalar field in spacetimes with compact spatial sections of negative spatial curvature.

In this talk work done by our group on cosmic topology is reviewed. It ranges from early attempts to solve a famous controversy about quasars through the multiplicity of images, to quantum cosmology in this context and an application to QED renormalization.

A few years ago, Cornish, Spergel and Starkman (CSS), suggested that a multiply connected ``small'' Universe could allow for classical chaotic mixing as a pre-inflationary homogenization process. The smaller the volume, the more important the process. Also, a smaller Universe has a greater probability of being spontaneously created. Previously DeWitt, Hart and Isham (DHI) calculated the Casimir energy for static multiply connected flat space-times. Due to the interest in small volume hyperbolic Universes (e.g. CSS), we generalize the DHI calculation by making a a numerical investigation of the Casimir energy for a conformally coupled, massive scalar field in a static Universe, whose spatial sections are the Weeks manifold, the smallest Universe of negative curvature known. In spite of being a numerical calculation, our result is in fact exact. It is shown that there is spontaneous vacuum excitation of low multipolar components.

We clarify and develop the results of a previous paper on the birth of a closed universe of negative spatial curvature and multiply connected topology. In particular we discuss the initial instanton and the second topology change in more detail. This is followed by a short discussion of the results.

The results of a previous paper by the author [Astrophys. J. 470, 43 (1996) are improved, through a more reliable technique for finding the location of sources whose images are on the surface of last scattering.

We present a modified version of the cosmic crystallography method, especially useful for testing closed models of negative spatial curvature. The images of clusters of galaxies in simulated catalogs are ``pulled back'' to the fundamental domain before the set of distances is calculated.

We try to apply the cosmic crystallography method of Lehoucq, Lachieze-Rey, and Luminet to a universe model with closed spatial section of negative curvature. But the sharp peaks predicted for Einstein-de Sitter closed models do not appear in our hyperbolic example. So we turn to a variant of that method, by subtracting from the distribution of distances between images in the closed model, the similar distribution in Friedmann's open model. The result is a plot with much oscillation in small scales, modulated by a long wavelength quasi-sinusoidal pattern.

We propose a modified form of the spontaneous birth of the universe by quantum tunneling. It proceeds through topology change and inflation, to eventually become a universe with closed spatial sections of negative curvature and nontrivial global topology.

Universes with multiconnected spatial sections predict multiple images of cosmic sources. A confusing terminology exists in the naming of these images as real ones vs. ghosts. Here an attempt is made to clarify the situation.

A few corrections and comments are made upon a previously published paper by the author (Gen. Rel. Gravit. 24, 199 (1992)), on the subject of cosmological models with compact spatial sections.

We briefly summarize the idea of cosmological models with compact, flat spatial sections. It has been suggested that, because of the COBE satellite's maps of the microwave background, such models cannot be small in the sense of Ellis, and hence are no longer interesting. Here we use Lehoucq et al.'s method of cosmic crystallography to show that these models are physically meaningful even if the size of the spatial sections is of the same order of magnitude as the radius of the observational horizon.

A numerical integration is made, starting with a Kantowski-Sachs (KS) model with an added initial perturbation at time one-tenth the age of the universe, and evolving up to the present. According to a prediction by Barrow (1989), the ratio between the fluctuations of the average matter density and those of the metric tensor varies with the inverse squared wavelength of the fluctuation. In this paper we take an extravantly large value for this wavelength (20% of the horizon's radius), to see that even then that ratio remains greater than unity. So we may be reasonably confident of models assuming density homogeneity, despite the fact that the scale of observed structure in the universe has been increasing lately.