20
$\begingroup$

Let $\Sigma^2$ be an orientable compact surface of genus $gen(\Sigma)\geq2$, and denote by $\mathcal M(\Sigma)$ the moduli space of hyperbolic metrics on $\Sigma$, i.e., Riemannian metrics of constant curvature $-1$. Recall that, from Teichmüller theory, this is a finite-dimensional subspace of the space of all metrics on $\Sigma$, and actually has dimension $6gen(\Sigma)-6$. For a given metric $g$, denote by $Spec(\Delta_g)=\{0=\lambda_0<\lambda_1\leq\lambda_2\leq\dots\}$ the spectrum of the Hodge-Laplacian of $g$, acting on $C^\infty(\Sigma)$.

Question 1: Given a natural number $n\in\mathbb N$, does there exist a hyperbolic metric $g\in\mathcal M(\Sigma)$ such that $n\notin Spec(\Delta_g)$?

I have attempted various ``deformation'' arguments to try to prove that if some $g\in\mathcal M(\Sigma)$ has $n$ as an eigenvalue, then a small perturbation $g'\in\mathcal M(\Sigma)$ would no longer have that eigenvalue, but without success... Instead of avoiding any given natural number, one can try the (seemingly) easier task:

Question 2: Does there exist a hyperbolic metric $g\in\mathcal M(\Sigma)$ such that $n\in Spec(\Delta_g)$ only for finitely many numbers $n\in\mathbb N$?

Or even more,

Question 3: Does there exist a hyperbolic metric $g\in\mathcal M(\Sigma)$ such that $\mathbb N\not\subset Spec(\Delta_g)$?

Although the answers to Q2 and Q3 seem to be "obviously yes", I have not been able to find a rigorous argument to prove that. I have tried to argue by contradiction, to show that if $Spec(\Delta_g)$ contains infinitely many natural numbers (or all of them), then we would somehow violate Weyl's formula (see this post) which says that $\lambda_k\sim \frac{k}{gen(\Sigma)-1}$ as $k\to+\infty$, but again without success (even in the case $gen(\Sigma)=2$).

$\endgroup$
  • $\begingroup$ The internet claims that it is possible to understand the spectrum of the Laplacian on a modular curve via Eisenstein series, but I haven't been able to find any details about this. $\endgroup$ – Qiaochu Yuan Oct 31 '13 at 20:45
  • $\begingroup$ I know about a result which claims that given any finite series of N numbers and a smooth manifold, you can find a metric on it, such that the first N eigenvalues of the Laplacian is given by this series. However, restricting to your finite-dimensional space, this is probably not true anymore, and one needs other techniques. $\endgroup$ – Matthias Ludewig Nov 1 '13 at 6:47
14
$\begingroup$

I was informed by Sugata Mondal at the MPI that Scott Wolpert proved the following result in his 1994 Annals paper Disappearance of cusp forms in special families:

Theorem 5.14. The eigenvalues of the Laplacian above $\tfrac14$ on a closed hyperbolic surface vary nontrivially under analytic deformations.

That is, if $g_t$ is a real-analytic path of hyperbolic metrics on $\Sigma$, then no real-analytic branch $\lambda_k(t)$ of an eigenvalue of the Laplacian $\Delta_{g_t}$ can be constant. Recall that by the Kato Selection Theorem, up to relabeling the indices $k$'s, the functions $\lambda_k(t)$ are real-analytic.

I also contacted Scott Wolpert, who confirmed that this indeed answers all my above questions. In particular, he had called Q1 the "middle C embarrassment question", since prior to the above result it could have been that middle C on the piano is a frequency for every hyperbolic Laplacian. Finally, note that his result also clearly allows to prescribe the genus of the closed hyperbolic surface $\Sigma$ at the same time as avoiding any given real number $\lambda>\tfrac14$ in its spectrum.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.