Surface bundles versus Heegaard splittings (Q871724)

Throughout this review let \(M\) be a closed 3-manifold that fibers over \(S^1\) with monodromy \(\phi\), that is, \(F\) is a closed surface, \(\phi:F \to F\) a diffeomorphism, and \(M\) is obtained from \(F \times [0,1]\) by identifying \((p,0)\) with \((\phi(p),1)\) for all \(p \in F\). We will assume that \(M\) is a hyperbolic 3-manifold. By Thurston, this is equivalent to \(g(F) \geq 2\) and \(\phi\) being pseudo-Anosov. A Heegaard splitting for \(M\) is a decomposition of \(M\) into two handlebodies. For any positive integer \(n\) we can construct the cyclic cover of \(M\) dual to the fiber, which is a manifold that fibers over \(S^1\) with monodromy \(\phi^n\); note that \(M = M_1\). By taking two disjoint fibers and tubing them together once on each side one obtains a Heegaard surface for \(M_n\) of genus \(2g(F) + 1\). This is called the standard Heegaard splitting for \(M_n\). \textit{J.~H. Rubinstein} [Minimal surfaces in geometric 3-manifolds, Global theory of minimal surfaces. Proceedings of the Clay Mathematics Institute 2001 summer school, Berkeley, CA, USA, June 25--July 27, 2001. Providence, RI: American Mathematical Society (AMS). Cambridge, MA: Clay Mathematics Institute. Clay Mathematics Proceedings 2, 725--746 (2005; Zbl 1119.53042)] showed that for any positive integer \(h\) there exists \(N\) so that if \(n \geq N\), then any Heegaard surface for \(M_n\) of genus at most \(h\) is a stabilization of the standard Heegaard surface. \textit{M. Lackenby} [Invent. Math. 164, No. 2, 317--359 (2006; Zbl 1110.57015)] independently showed this for strongly irreducible Heegaard splittings; in particular, for sufficiently large \(n\) the Heegaard genus of \(M_n\) is exactly \(2g(F) + 1\). Both Rubinstein and Lackenby used minimal surfaces. Motivated by this, Bachman and Schleimer used the curve complex to study the Heegaard genus of \(M_n\). The curve complex of \(F\), denoted \(\mathcal{C}(F)\), is a finite dimensional complex whose vertices are isotopy classes of simple closed curves and \(k\) vertices span a \(k-1\) simplex if they can be isotoped to be disjoint. It is well known that the curve complex \(\mathcal{C}(F)\) is connected. By setting the length of each edge to be 1 and endowing every simplex with a Euclidean metric we obtain a metric space, and the distance between any two vertices is the minimal number of edges needed to get from one to the other. The monodromy \(\phi\) acts on the curve complex. The translation distance of \(\phi\), denoted \(d_{\mathcal{C}}(\phi)\), is defined to be the minimal translation distance that \(\phi\) moves any vertex of \(\mathcal{C}(F)\). The first theorem the authors prove is that if \(M\) admits an essential surface \(S\) that is not isotopic to a fiber, then \(d_{\mathcal{C}}(\phi) \leq -\chi(H)\). Next they show that if \(M\) admits a strongly irreducible Heegaard surface \(H\) then \(d_{\mathcal{C}}(\phi) \leq -\chi(H)\). (We note that the standard Heegaard surface is never strongly irreducible.) They conclude that if \(H\) is a Heegaard surface for \(M\) and \(d_{\mathcal{C}}(\phi) \geq -\chi(H)\) then \(H\) is a stabilization of the standard Heegaard surface. \textit{H. A. Masur} and \textit{Y. N. Minsky} [Invent. Math. 138, No. 1, 103--149 (1999; Zbl 0941.32012)] showed that \(\lim_{n \to \infty} d_{\mathcal{C}} (\phi^n) = \infty\) (in fact, the growth of \(d_{\mathcal{C}} (\phi^n)\) is linear in \(n\)). It is easy to see that the theorems of Lackenby and Rubinstein follow. The reviewer would like to add one comment. Masur and Minsky showed that there exists \(c\), so that \(d_{\mathcal{C}} (\phi^n) \geq n c\). The constant \(c\) depends only on \(g(F)\) and not on \(\phi\). This implies that \(N\) above depends only on \(h\) and \(g(F)\). By contrast, Lackenby and Rubinstein use the geometry of the manifold \(M\), and thus the bound that they get depends on \(h\) and \(\phi\).