Causality and Cauchy horizons (Q1345157)

The author studies various notions of stability for Cauchy horizons. Here the topology with respect to which stability is considered is (mainly) the fine \(C^ 0\) Whitney topology for metrics. Denote the unperturbed spacetime by \((M,g_ 0)\). He calls a partial Cauchy surface \(S\) stable if for any nearby metric \(g\) \(S\) is still a partial Cauchy surface. He can then show that the Cauchy horizon \(H^ +(g_ 0,S)\) of a stable, partial Cauchy surface is locally stable at \(p \in H^ +(g_ 0,S)\), if either there exists a null generator of \(H^ +(g_ 0,S)\) which is not past imprisoned in any compact set or \(p\) is in the boundary of the chronology violating set. (The definition of locally stable is an intuitive (though complicated) one). This result has an analogue where, loosely speaking, the point \(p\) is replaced by a compact subset of the Cauchy horizon. Hence compact subsets of the Cauchy horizon can be controlled under such assumptions. He also shows that this stability is implied by the strong causality condition. Finally, he considers stability of the topological structure of the Cauchy horizon. If the Cauchy horizon is stable and compact then the original metric has a neighbourhood \(W(g_ 0)\) such that for all \(g \in W(g_ 0)\) with wider light cones than \(g_ 0\) the Cauchy horizon is homeomorphic to the original one. The author complements his results with many examples which suggest that stronger results would be hard to come by: Cauchy horizons can be very unstable in general.