Negatively oriented ideal triangulations and a proof of Thurston's hyperbolic Dehn filling theorem (Q1978136)

Let \(M\) be a compact orientable 3-manifold whose interior has a complete hyperbolic structure. Then its boundary consists of finitely many tori, and Thurston's hyperbolic Dehn surgery theorem states that, excluding a finite number of Dehn fillings for each boundary component, any 3-manifold obtained by Dehn filling on \(M\) (i.e. by gluing solid tori along the boundaries) has again a complete hyperbolic structure (obtained as the completion of an incomplete hyperbolic structure on the interior of \(M\), after a small deformation of its original complete structure). Preceding Thurston's big hyperbolization theorem for Haken 3-manifolds, this is of fundamental importance for the theory of 3-manifolds and one of the most beautiful results in the area which has suscitated numerous applications and extensions. In the present carefully written paper, a complete proof of this theorem is given using ideal triangulations. Such a proof has been given by Thurston in his Princeton Notes for the case of the complememt of the figure-8 knot, using its decomposition into two ideal tetrahedra; for the general case, a proof is sketched in Thurston's Notes using representation varieties. In the present paper the ideal triangulation approach is used, starting from the decomposition of \textit{D. B. A. Epstein} and \textit{R. C. Penner} of a complete cusped hyperbolic 3-manifold into ideal polyhedra [J. Differ. Geom. 27, No. 1, 67-80 (1988; Zbl 0611.53036)] and the analysis of \textit{W. D. Neumann} and \textit{D. Zagier} of the volumes of the Dehn filled manifolds using ideal triangulations [Topology 24, 307-332 (1985; Zbl 0589.57015)]. It does not seem to be known, however, whether every cusped hyperbolic 3-manifold admits an ideal triangulation. In the present paper, the Epstein-Penner decomposition into ideal polyhedra is refined to a ``partially flat'' ideal triangulation, i.e. one in which some of the tetrahedra degenerate into flat quadrilaterals with distinct vertices. When deforming the complete hyperbolic structure, it becomes inevitable to consider negatively oriented tetrahedra, i.e. to consider nontrivial overlappings between the tetrahedra. This is one of the main technical points of the paper, and it is shown how to associate a hyperbolic structure to such a deformed triangulation. What remains open in the present approach is the proof of smoothness near the complete structure in the variety of deformations which follows in the representation approach from some cohomology computations (see chapter 8.8 of the book of \textit{M. Kapovich} [Hyperbolic manifolds and discrete groups, Prog. Math. 183 (2001; Zbl 0958.57001)]). So smoothness is avoided in the present paper which uses instead some classical tools from the theory of stratifications and analytic spaces for the proof that there exists a neighbourhood of the original complete structure all of whose points correspond to hyperbolic Dehn fillings. For a proof using smoothness of the representation variety, see the appendix of the monograph of \textit{M. Boileau} and \textit{J. Porti} [Geometrization of 3-orbifolds of cyclic type, Astérisque 272 (2001; Zbl 0971.57004)].