Symmetries of cosmological Cauchy horizons with non-closed orbits (Q2299357)

The authors continue their research on the cosmic censorship conjecture. An earlier part of this research is published in [\textit{V. Moncrief} and \textit{J. Isenberg}, Commun. Math. Phys. 89, 387--413 (1983; Zbl 0523.53057)]. The relevant new result is formulated in their theorem 1: ``Let \( V \) be a real analytic, time orientable, vacuum spacetime which admits a compact, connected Cauchy horizon \( N \) that separates \( V \) into open Lorentzian submanifolds \( V_+ \) and \( V_- \) of which one is globally hyperbolic and the other acausal. Assume that \( N \) is realized as a level set of some analytic function \( \tau: V \rightarrow \mathbb{R} \) having no critical points in a neighborhood of \( N \). The vector field \( X = \operatorname{grad} \tau \) will therefore be non-vanishing on this neighborhood, null on the hypersurface \( N \) and thus tangent to its null geodesic generators and will naturally induce (by restriction of \( X \) to \( N \)) a corresponding tangent vector field \( X \) on the Cauchy horizon itself. In the cases referred to here as `non-ergodic' the null generators of \( N \) are either closed curves or densely fill 2-tori embedded in \( N \) and every such generator is either complete in both the directions of \( X \) and \( -X \) (the `degenerate' case) or else every generator is incomplete in one direction (say that of \( X\)) and complete in the opposite direction (the `non-degenerate' case). Compact, non-degenerate, non-ergodic Cauchy horizons in analytic, vacuum spacetimes \(V\) are Killing horizons in that there always exists a non-trivial, analytic Killing field \(Y\), globally defined on a full neighborhood of the horizon manifold \( N \subset V \) which, when restricted to \(N\), is everywhere tangent to the null generators of this hypersurface. \(Y\) extends (at least smoothly) to a Killing field defined throughout the maximal Cauchy development of the globally hyperbolic region of \( V\) whose Cauchy horizon is \(N\).'' The proof is quite lengthy but well-organized. Additionally, many details of papers of others to this topic have been commented.