D.Canarutto:
An introduction to the geometry of singularities in general relativity.
La Rivista del Nuovo Cimento, 11 (1988), pp. 1-60.
Abstract
The main topics in space-time edge geometry are introduced in an essential
way, and their applications to the study of gravitational singularities
are discussed. A simple but modern introduction to the concepts of Differential
Geometry, which are most relevant to these topics, is also given.
Introduction to Part I: Fundamental concepts in Differential Geometry
In recent years, the importance of differential geometry in theoretical
physics has greatly increased, and field theories are now formulated largely
in terms of topics such as the theory of connections, principal bundles,
lagrangian and symplectic geometry, and so on. However, there are still
differences between the approach of physicists and that of mathematicians
to this crucial area of common interest. In fact, usually physicists prefer
to deal with coordinate calculations, thinking that a more abstract and
rigorous formalism is difficult to handle and unessential for physical
understanding, while mathematicians often consider the coordinate notation
as untidy and obscuring. In my opinion, the present state of the art enables
us to get over these differences. In fact, many ideas in differential geometry
can be introduced in a very clear way by adopting an abstract and general
framework, and this clearness turns out also to be undoubtedly very useful
in practice; in particular, we have now a powerful, neat and simple theory
of connections on fibred manifolds, which includes all the particular cases
(such as principal bundles and tensor bundles), and allows clearer understanding
and easier handling. On the other hand, the coordinate formalism can be
set in an absolutely rigorous way, and this increases both its power and
its reliability. This part I is intended to present to the physicist some
of the main ideas in this field, in an essential way and without boring
details. Its prerequisites are only standard undergraduate courses in analysis
and linear algebra, but of course it is not sufficiently expanded to give
a deep insight into the subject; however, it could be useful reading before
tackling a book on differential geometry.
Introduction to Part II: what is a space-time singularity?
It is well-known that the study of Einstein's equation of general relativity
leads to predict the existence of space-time "singularities"; in other
terms, it produces models of the physical world in which some observer
arrives, in a finite interval of proper time, to the "edge of the universe".
Moreover, this edge cannot be included in an extension of the space-time,
for example because the observer experiences unbounded curvature; the best
known examples are the Schwartzschild and Friedmann solutions. It had been
proposed (Lifshitz et al.) that singularities could arise from an excess
of imposed simmetry, and that they would not be present in physically more
realistic solutions. But, now, this seems not to be true; in particular,
Hawking and Penrose have proved theorems which say that singularities are
unavoidable when some general (and physically plausible) conditions are
satisfied. Then, relativistic singularities seem to have a kind of "stable"
character, and to be physically very important. Hence the problem of the
definition of singular space-time: we would like to have a general criterion
for deciding whether a given space-time has singularities and, possibly,
a classification of them. The natural thing to do is to try to imitate
the riemannian framework, where problems of this kind has a natural and
straightforward setting (§4.12). Essentially, we must find a suitable
criterion of completeness analogous to that of metric completeness (in
other terms: how can we decide whether a given space-time has "cuts" or
"holes"?) and then examine the problem of extendibility. As for completeness,
one might think of using the concept of geodesic completeness, which is
in fact sufficient in many space-times; however, this is not a satisfying
criterion from a general point of view. In fact, there are examples of
space-times which are geodesically complete, but admit world lines of finite
proper-time length and bounded acceleration which are inextendible, i.e.
have no endpoint in the space-time. In physical terms, this means that
a spaceship with an engine of finite power and a finite amount of fuel
could arrive, after a finite interval of proper time, at the "end of the
universe", and then would cease existing. It is clear that, in a physically
satisfying definition of completeness, such a space-time must turn out
to be incomplete. Thus, although this difficulty does not arise in the
commonly studied space-times, geodesic completeness is not sufficient for
the study of singularities. The construction of the b-completion (i.e.
bundle-completion ) seems to solve these problems in a satisfying manner.
A b-boundary is attached to any manifold with a connection; in particular,
it provides endpoints for all curves in a space-time which we would consider
incomplete from a physical point of view; hence, we may consider the singularities
of a space-time as points in its b-boundary. Thus we have the concept of
b-completeness (i.e. the emptiness of the b-boundary), which turns out
to be a generalization of that of geodesic completeness (b-completeness
implies g-completeness) and also of that of metric completeness (for riemannian
manifolds, the metric completion and the b-completion coincide). Thus,
in principle, the study of space-time singularities proceeds in two steps.
First, we should determine the b-boundary. Second, for each point of the
b-boundary, we should study extendibility; this is so because, analogously
to the riemannian case, the b-boundary does not distinguish between true
singularities and singularities which can be eliminated by extending the
space-time; in other words, a space-time obtained by removing points from
a b-complete one is b-incomplete, and its b-boundary is constituted exactly
by the topological boundary of the set of removed points. Of course, this
procedure may present difficult mathematical problems but, in principle,
is currently considered as the right setting for the study of space-time
singularities. In this part II we shall discuss these ideas and some of
their main developments.
|