Table of Contents
Fetching ...

A sharp lower bound on the small eigenvalues of surfaces

Renan Gross, Guy Lachman, Asaf Nachmias

TL;DR

This work proves a sharp, universal lower bound for the small Laplacian eigenvalues on compact hyperbolic surfaces in terms of the genus $g$ and the geometric bottleneck quantity $I(S)$. Central to the approach is a geometric-extremal-length framework: by conformally stretching collars to cylinders and bounding the extremal length between small discs via a reciprocal-injectivity-weight, the authors bound the Dirichlet energy of a spectral kernel and derive a quantitative eigenvalue bound $\lambda_k \ge \frac{c k^2}{I(S) g^2}$. This eigenvalue control yields a global heat-kernel estimate on average, $\frac{1}{\mathrm{Vol}(S)} \int_S |p_t(x,x) - 1/\mathrm{Vol}(S)|dx \le C \sqrt{\frac{I(S)}{t}}$ for all $t\ge 1$, with universal constants; both results are shown to be optimal in the sense that reversed inequalities hold with different constants on certain surfaces. The paper blends extremal-length techniques with spectral kernel methods and thin-thick geometry to connect injectivity-radius bottlenecks to spectral data, yielding robust, geometry-driven bounds applicable across all genera.

Abstract

Let $S$ be a compact hyperbolic surface of genus $g\geq 2$ and let $I(S) = \frac{1}{\mathrm{Vol}(S)}\int_{S} \frac{1}{\mathrm{Inj}(x)^2 \wedge 1} dx$, where $\mathrm{Inj}(x)$ is the injectivity radius at $x$. We prove that for any $k\in \{1,\ldots, 2g-3\}$, the $k$-th eigenvalue $λ_k$ of the Laplacian satisfies \begin{equation*} λ_k \geq \frac{c k^2}{I(S) g^2} \, , \end{equation*} where $c>0$ is some universal constant. We use this bound to prove the heat kernel estimate \begin{equation*} \frac{1}{\mathrm{Vol}(S)} \int_S \Big| p_t(x,x) -\frac{1}{\mathrm{Vol}(S)} \Big | ~dx \leq C \sqrt{ \frac{I(S)}{t}} \qquad \forall t \geq 1 \, , \end{equation*} where $C<\infty$ is some universal constant. These bounds are optimal in the sense that for every $g\geq 2$ there exists a compact hyperbolic surface of genus $g$ satisfying the reverse inequalities with different constants.

A sharp lower bound on the small eigenvalues of surfaces

TL;DR

This work proves a sharp, universal lower bound for the small Laplacian eigenvalues on compact hyperbolic surfaces in terms of the genus and the geometric bottleneck quantity . Central to the approach is a geometric-extremal-length framework: by conformally stretching collars to cylinders and bounding the extremal length between small discs via a reciprocal-injectivity-weight, the authors bound the Dirichlet energy of a spectral kernel and derive a quantitative eigenvalue bound . This eigenvalue control yields a global heat-kernel estimate on average, for all , with universal constants; both results are shown to be optimal in the sense that reversed inequalities hold with different constants on certain surfaces. The paper blends extremal-length techniques with spectral kernel methods and thin-thick geometry to connect injectivity-radius bottlenecks to spectral data, yielding robust, geometry-driven bounds applicable across all genera.

Abstract

Let be a compact hyperbolic surface of genus and let , where is the injectivity radius at . We prove that for any , the -th eigenvalue of the Laplacian satisfies \begin{equation*} λ_k \geq \frac{c k^2}{I(S) g^2} \, , \end{equation*} where is some universal constant. We use this bound to prove the heat kernel estimate \begin{equation*} \frac{1}{\mathrm{Vol}(S)} \int_S \Big| p_t(x,x) -\frac{1}{\mathrm{Vol}(S)} \Big | ~dx \leq C \sqrt{ \frac{I(S)}{t}} \qquad \forall t \geq 1 \, , \end{equation*} where is some universal constant. These bounds are optimal in the sense that for every there exists a compact hyperbolic surface of genus satisfying the reverse inequalities with different constants.
Paper Structure (14 sections, 15 theorems, 108 equations, 3 figures)

This paper contains 14 sections, 15 theorems, 108 equations, 3 figures.

Key Result

Theorem 1

There exists a universal constant $c>0$ such that for any compact hyperbolic surface ${S}$ of genus $g\geq 2$ and every $k\in \{1, \ldots, 2g-3\}$,

Figures (3)

  • Figure 1: The domain $D$ as a subset of $\tilde{{S}}$. Note that the uniform thickness in the middle part is guaranteed by the injectivity radius lower bound of $\tilde{{S}}$. The paths $\gamma_q$ start at distances in $[-\alpha r_x/\widehat{\mathrm{Inj}}\left({x} \right), \alpha r_x/\widehat{\mathrm{Inj}}\left({x} \right)]$ from $x$, fan out to a spread of $\alpha$, then fan back towards $y$.
  • Figure 2: The domain $D$ as a subset of $\tilde{{S}}$. In this case there is no middle part corresponding to the main term $\mathrm{dist}_w\left(x, y \right)$ in \ref{['eq:main_extremal_length']}.
  • Figure 3: A schematic of $S$ for $n=6$.

Theorems & Definitions (33)

  • Theorem 1
  • Theorem 2
  • Remark
  • Definition 3: Extremal length
  • Definition 4: Reciprocal-injectivity weight
  • Theorem 5
  • Remark 6
  • Lemma 7: Collar lemma
  • Proposition 8: Injectivity radius estimate
  • proof
  • ...and 23 more