Geometry of the smallest 1-form Laplacian eigenvalue on hyperbolic manifolds (Q1632238)

The well-known Cheeger-Buser inequalities give precise upper and lower bounds for the first positive eigenvalue of the Laplace operator on a Riemannian manifold, depending on the so-called Cheeger constant of the manifold defined using the geometry of its codimension-1 submanifolds. In this paper the authors establish a geometric lower bound for the first positive eigenvalue \(\lambda_1^1\) of the Hodge-Laplace operator on differential 1-forms on compact manifolds of constant curvature \(-1\) (hyperbolic manifolds). This is related the following problem: let \(M_0\) be a hyperbolic manifold, does there exist \(c, \alpha > 0\) such that \(\lambda_1^1 \geq cd^{-\alpha}\) for all \(d \geq 1\) and all degree \(d\) covers of \(M_0\)? The latter problem is in particular motivated by the application to homology growth, see [\textit{N. Bergeron} and \textit{A. Venkatesh}, J. Inst. Math. Jussieu 12, No. 2, 391--447 (2013; Zbl 1266.22013)]. The interesting part of the 1-form spectrum lies in that on co-closed 1-forms; the authors thus work with the smallest eigenvalue \(\lambda_1^1(M)_{d^*}\) on this subspace. The main estimate proven by the authors depends on the notion of stable area of a closed geodesic: if \(\gamma\) is a loop in \(M\) then one may define its area as the least area of a surface in \(M\) which it bounds, if such a surface exists (in other words if it is trivial in integral homology). For a curve with multiplicity we can define a similar notion, and the stable area is then defined by \[ \mathrm{sArea}(\gamma) := \inf_{m \geq 1} \frac{\mathrm{Area}(\gamma^m)}m \] which makes sense when \(\gamma\) is trivial in rational homology. The main bound in the paper (Theorem 1.3) then implies that : \[ \lambda_1^1(M)_{d^*} \geq c \cdot d^{-5/2} \cdot \frac 1{1 + \sup_{\gamma} \frac{\mathrm{sArea}(\gamma)}{\ell(\gamma)}}. \] Here \(M\) is a cover of degree \(d\) of a fixed closed hyperbolic manifold \(M_0\), \(c\) is a constant depending only on the geometry of \(M_0\), and the sup is taken over all elements of \(\pi_1(M)\) which are trivial in rational homology. The statement in the paper is in terms of geometric terms rather than \(d\) and in many cases it will give a better estimate than \(d^{-5/2}\) (for example, for an expanding sequence it will give \(d^{-3/2+\varepsilon}\) for any \(\varepsilon\)). The proof of this result is essentially combinatorial but not easily explained. It uses min-max type arguments with an analysis of the shape of fundamental domains, Stokes' theorem and Sobolev bounds to control the integrals on submanifolds. The paper also contains results relating the spectral gap on 1-forms on a manifold to that on certain totally geodesic submanifolds (Theorem 1.9): if \(M_0 \subset N_0\) is totally geodesic then \[ \lambda_1^1(M)_{d^*} \geq c \cdot d^{-5/2} \cdot (d')^{-2}\lambda_1^1(N') \] where \(c, d, M\) are as in the previous statement and \(N'\) is a cover of \(N_0\) with certain properties with respect to \(M_0, M\) and \(d'\) its degree. As the authors remark, higher-dimensional hyperbolic manifolds often have a uniform spectral gap on 1-forms (see [\textit{N. Bergeron} and \textit{L. Clozel}, Spectre automorphe des variétés hyperboliques et applications topologiques. Paris: Société Mathématique de France (2005; Zbl 1098.11035)]), if it was possible to control \(d'\) then this would give a solution to the problem mentioned above. However the techniques used to produce \(N'\) (which are always available when \(M_0\) is arithmetic, see [\textit{N. Bergeron} et al., J. Lond. Math. Soc., II. Ser. 83, No. 2, 431--448 (2011; Zbl 1236.57021)]) do not permit this at present.