Abstract. We review recent work and present new examples about the character of singularities in globally and regularly hyperbolic, isotropic universes. These include recent singular relativistic models, tachyonic and phantom universes as well as inflationary cosmologies.
IntroductionAn important question eventually arising in every study of the global geometric and physical properties of the universe is that of deciding whether or not the resulting model is geodesically complete. Geodesic completeness is associated with an infinite proper time interval of existence of privileged observers and implies that such a universe will exist forever. Its negation, geodesic incompleteness or the existence of future and/or past singularities of a spacetime, is often connected to an 'end of time' for the whole universe modeled by the spacetime in question. By now there exist simple singular cosmological models of all sorts, not incompatible with recent observations, in which an all-encompassing singularity features as such a catastrophic event.It is well known that in general relativity there are a number of rigorous theorems predicting the existence of spacetime singularities in the form of geodesic incompleteness under certain geometric and topological conditions (see, e.g., [1] for a recent review). These conditions can be interpreted as restrictions on the physical matter content as well as plausibility assumptions on the causal structure of the spacetime in question. Because all these assumptions are not unreasonable, the singularity theorems predicting the existence of spacetime singularities in cosmology and gravitational collapse have become standard ingredients of the current cosmological theory.However, such existence results cannot offer any clue about the generic nature of the singularities they predict. In addition, there are new completeness theorems (cf. [2]) which say that under equally general geometric assumptions, generic spacetimes are future (or past) geodesically complete. Among the chief hypotheses of the completeness theorems, except the usual causality ones also present in the singularity theorems, is the assumption that the space slice does not 'vibrate' too much as it moves forward (or backward) in time, and also the assumption that space does not curve itself too much in spacetime. It may well be that cosmological models in a generic sense are only mildly singular or even complete. It remains thus a basic open problem to decide about the character of the cosmological singularities predicted by the singularity theorems.In preparing to tackle such basic open questions more information is required, and one feels that perhaps different techniques are needed which the singularity theorems cannot provide. We