Heegaard splittings, the virtually Haken conjecture and property \((\tau)\) (Q2491161)

This paper is motivated by the following three well-known conjectures: The virtually Haken conjecture. A closed orientable irreducible 3-manifold \(M\) with infinite fundamental group is virtually Haken. That is, \(M\) has a finite sheeted cover that contains an essential surface. The positive virtual \(b_1\) conjecture. A closed orientable irreducible atoroidal 3-manifold \(M\) with infinite fundamental group has an infinite cover with positive first Betti number. That is, \(\pi_1(M)\) has a finite index subgroup with infinite abelianization. The virtually fiber conjecture. A compact orientable irreducible 3-manifold \(M\) with \(\partial M\) a (possibly empty) collection of tori, whose fundamental group is infinite and contains no \(\mathbb{Z}\times\mathbb{Z}\), is finitely covered by a surface bundle over the circle. It is well known that the third conjecture implies the second and the second implies the first. The author relates these conjectures to the spectral geometry of \(M\). More precisely, the first result of this paper (Theorem 1.1) relates the spectral geometry of a group \(G\) to the existence of finite index subgroups \(G_i< G\) with finite abelianization. The author considers all finite index subgroups \(\{G_i\}\) and shows that (for some \(i\)) \(G_i\) has infinite abelianization if and only if the sequence \(\lambda_1(X_i)[G: N(G_i)]^2[G: G_i]^2\) has a bounded subsequence. (By \(X_i\) we mean the Schreier coset graph of \(G/G_i\), which is the same as the Cayley graph of \(G/G_i\) when \(G_i= (1)\).) The author gives three other equivalent statements, two using the Cheeger constant \(h(X_i)\). All the notation (including the first eigenvalue of the Laplacian \(\lambda_1\) and the Cheeger constant) is explained well in Section 2. When \(G\) is the fundamental group of a closed orientable Riemannian manifold, let \(M_i\) be the cover of \(M\) corresponding to the subgroup \(G_i\). Then the term \(\lambda_1(X_i)\) can be replaced by \(\lambda_1(M_i)\). This allows the author to restate the positive virtual \(b_1\) conjecture in terms of the spectral geometry of \(M_i\). The author introduces two new 3-manifold invariants. Given a compact orientable 3-manifold \(M\) let \(\{M_i\}\) be a collection of finite sheeted covers of \(M\). For simplicity of presentation we concentrate on the case where \(\{M_i\to M\}\) are all the finite sheeted covers. Let \(d_i\) denote the degree of \(M_i\to M\). Let \(\chi_-\) denote minus the characteristic of a minimal genus Heegaard surface (that is, \(\chi_-(M)= 2g(M)- 2\) where \(g(M)\) is the Heegaard genus of \(M\)). Similarly, let \(\chi^{sh}_-\) denote minus the Euler characteristic of a minimal genus strongly irreducible Heegaard surface for \(M\), if one exists, and \(\infty\) otherwise. The author defines the infimal Heegaard gradient of \(M\) to be \(\text{inf}_i\,{\chi_-(M_i)\over d_i}\). The infimal strong Heegaard gradient is defined to be \(\liminf_{i}\,{\chi^{sh}_-(M_i)\over d_i}\). The author makes the following two conjectures: The Heegaard gradient conjecture. A compact orientable hyperbolic 3-manifold has zero Heegaard gradient if and only if it is a virtual fiber. The strong Heegaard gradient conjecture. Any closed orientable hyperbolic 3-manifold has strong Heegaard gradient zero. In a sequence of fascinating theorems the author relates these conjectures and a conjecture of Lubotzky and Sarnak about Property \((\tau)\) to the virtually Haken and the virtually positive \(b_1\) conjecture. We refer the reader to the introduction of the paper for a clear explanation of these results. As an example we explain a few of them here. In Corollary 1.4 the author shows that if \(M\) is a closed orientable negatively curved 3-manifold, and there is some collection of covers \(\{M_i\}\) (of degree \(d_i\)) such that \(\text{inf\,}\chi_-(M_i)/\sqrt{d_i}= 0\), then \(M\) satisfies the positive \(b_1\) conjecture. In Theorem 1.7 the author shows that if the covers \(\{M_i\}\) fail to have Property \((\tau)\) and they have positive strong Heegaard gradient, then \(M\) is virtually Haken. In Theorem 1.11 the author shows that if \(M\) is a compact orientable hyperbolic manifold, and \(\{M_i\}\) are the cyclic covers dual to a fixed non-trivial element \(z\in H_2(M,\partial M)\), then the infimal Heegaard gradient of \(\{M\}\) is zero if and only if \(z\) represents a virtual fiber. For the special case of arithmetic hyperbolic 3-manifolds, Lubotzky has shown that these manifolds in fact do have property \((\tau)\) with respect to their congruence covers, and the author is able to show that in this case the Heegaard genus grows linearly in the degree of the cover. Since this paper relates different fields (group theory, spectral geometry, and 3-manifolds) it is fair to assume that most readers will find at least some aspect new. We would like to emphasize that this paper is exceptionally well written and serves as an excellent introduction to these ideas, which have already been developed further by several authors, a trend that is likely to continue.