“…So, Causality can be identified to conformal geometry in Lorentzian signature (nevertheless, a subtler modification of the notion of Causality has been recently introduced by García-Parrado and Senovilla [57], see also [56]). Remarkably, one can associate a (conformally invariant) causal boundary to every sufficiently well behaved spacetime, this allows to describe the possible asymptotics of timelike curves in a subtle way, ordering them by inclusion of their pasts [50] As a remarkable difference with the Riemannian case, both, completeness and incompleteness are C r unstable, for every r ∈ N, even in the case of Lorentzian metrics on a torus [99], but some results on stability can be still obtained in the Lorentzian case, as C 1 -fine stability in the globally hyperbolic case, see [17,28]. The interplay between these particularities, causality and some physical interpretations, have the effect that, in Lorentzian Geometry, singularity theorems (which ensure incompleteness rather than completeness) play an important role, see below.…”