On complexity of three-dimensional hyperbolic manifolds with geodesic boundary (Q1760537)

Let \(M\) be a compact \(3\)-manifold with nonempty boundary \(\partial M\). A compact \(2\)-dimensional subpolyhedron \(P \subset M\) is called a spine of \(M\) if \(M - P\) is homeomorphic to \(\partial M \times (0, 1]\). In particular, we say that \(P\) is almost simple if the link of each point of \(P\) can be embedded into a complete graph \(K_4\) with four vertices. A point in \(P\) is called a true vertex if its link is homeomorphic to \(K_4\). Then the complexity \(c(M)\) of \(M\) is equal to \(n\) if it has an almost simple spine \(P\) with \(n\) true vertices and has no almost simple spines with fewer true vertices. An almost simple spine \(P\) is said to be simple if the link of each point \(x \in P\) is homeomorphic to a circle (i.e. \(x\) is nonsingular), a circle with a diameter (i.e. \(x\) is a triple point) or \(K_4\) (i.e. \(x\) is a true vertex). Furthermore, if the connected components of the union of all nonsingular points (i.e. \(2\)-components of \(P\)) are open \(2\)-cells, and the connected components of the union of all triple points (i.e. triple lines of \(P\)) are open \(1\)-cells, then the simple spine is called a special spine. Let us denote by \(\mathcal{H}\) the set of connected orientable hyperbolic \(3\)-manifolds with nonempty totally-geodesic boundary. Then each \(M\) in \( \mathcal{H}\) has a special spine \(P\) such that the number of true vertices of \(P\) equals \(c(M)\) [\textit{S. V. Matveev}, Algorithmic topology and classification of 3-manifolds. Second edition. Algorithms and Computation in Mathematics 9. Berlin: Springer. (2007; Zbl 1128.57001)]. However it is difficult to show that the number of true vertices of the obtained spine is minimal, hence it is difficult to determine the complexity. In the paper under review, the authors split \(\mathcal{H}\) into subsets \(\mathcal{H}_{p, q}\) \((p, q \in \mathbb{N})\) so that \(\mathcal{H}_{p, q}\) consists of hyperbolic \(3\)-manifolds \(M \in \mathcal{H}\) such that \((1)\) \(\partial M\) has exactly \(q\) components and \((2)\) \(M\) has a special spine with \(p\) \(2\)-components and has no special spines with fewer \(2\)-components. By definition, if \((p, q) \neq (p', q')\) then \(\mathcal{H}_{p, q} \cap \mathcal{H}_{p', q'} \neq \emptyset\). Moreover, it follows from [\textit{R. Frigerio, B. Martelli} and \textit{C. Petronio}, J. Differ. Geom. 64, No. 3, 425-455 (2003; Zbl 1073.57010)] that \(\mathcal{H}_{p, q} = \emptyset\) when \(p < q\). As the authors observe, if \(M \in \mathcal{H}_{p, q}\), then \(c(M) = p - \chi(M)\). It is known by \textit{R. Frigerio, B. Martelli} and \textit{C. Petronio} [Pac. J. Math. 210, No. 2, 283--297 (2003; Zbl 1061.57017) and loc. cit.] that \(\mathcal{H}_{q, q}\) \((q \in \mathbb{N})\) contains an infinite family of hyperbolic \(3\)-manifolds whose complexities are determined. It is also known by the authors [Dokl. Math. 84, No. 1, 542--544 (2011); translation from Dokl. Akad. Nauk 439, No. 5, 727--729 (2011; Zbl 1234.57025)] that \(\mathcal{H}_{2, 1}\) contains an infinite family of hyperbolic \(3\)-manifolds whose complexities are determined. The authors give an infinite family of hyperbolic \(3\)-manifolds contained in \(\mathcal{H}_{3, 1}\) and determine the complexities of these manifolds. The proof is based on constructing \(3\)-manifolds analogous to the Paoluzzi-Zimmermann manifolds [\textit{L. Paoluzzi} and \textit{B. Zimmermann}, Geom. Dedicata 60, No. 2, 113--123 (1996; Zbl 0857.57009)] and calculating their \(\varepsilon\)-invariants.