Abstract:We study topological lower bounds on the number of small Laplacian eigenvalues on hyperbolic surfaces. We show there exist constants $a,b>0$ such that when $(g+1)<a\frac{n}{\log n}$, any hyperbolic surface of genus-$g$ with $n$ cusps has at least $b\frac{2g+n-2}{\log\left(2g+n-2\right)}$ Laplacian eigenvalues below $\frac{1}{4}$. We also show that, under certain additional constraints on the lengths of short geodesics, the lower bound can be improved to $b\left(2g+n-2\right)$ with the weaker condition $(g+1)<an$.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the topological lower bound problem regarding the number of small Laplace eigenvalues on hyperbolic surfaces with multiple cusps. Specifically, the author studied the number of small Laplace eigenvalues of hyperbolic surfaces with high genus and a large number of cusps under specific conditions, and attempted to find the lower bound of the number of these eigenvalues.
### Main problems and conclusions of the paper
1. **Research background**:
- A hyperbolic surface \(X\) is a connected, finite - area hyperbolic surface, characterized by genus \(g\) and \(n\) cusps.
- The discrete eigenvalues in the spectrum of the Laplace operator \(\Delta_X\) within the interval \([0,\frac{1}{4})\) are called small eigenvalues.
- Previous research has shown that for any hyperbolic surface with signature \((g, n)\), there are at most \(2g + n - 2\) small eigenvalues (the result of Otal and Rosas).
2. **Research purpose**:
- This paper aims to find the topological lower bound of the number of small eigenvalues on hyperbolic surfaces with a large number of cusps.
- Specifically, the author attempts to prove that when the genus \(g\) of the surface and the number of cusps \(n\) satisfy certain conditions, the number of small eigenvalues is almost linearly dependent on the volume of the surface.
3. **Main conclusions**:
- **Theorem 1.3**: There exist constants \(a,b>0\) such that when \(g + 1 < a\frac{n}{\log n}\), any hyperbolic surface with signature \((g, n)\) has at least \(b\frac{2g + n}{\log(2g + n)}\) small eigenvalues.
\[
\text{When } g + 1 < a\frac{n}{\log n}\text{, any hyperbolic surface with signature }(g, n)\text{ has at least }b\frac{2g + n}{\log(2g + n)}\text{ small eigenvalues.}
\]
- **Theorem 1.4**: Under the additional assumption of the length of short closed geodesics, the logarithmic factor can be removed and the restriction on the signature \((g, n)\) can be relaxed. Specifically, there exists a constant \(c_1>0\), if \(g + 1 < c_1n\), then for any \(c_2>0\), there exists \(c_3>0\) such that a hyperbolic surface \(X\) satisfying the following condition has at least \(c_3(2g + n - 2)\) small eigenvalues:
\[
\sum_{\gamma\in P(X),\ell(\gamma)<1}\frac{\log\left(\frac{1}{\ell(\gamma)}\right)}{\ell(\gamma)}< c_2(2g + n - 2)
\]
### Method overview
The author combines geometric and analytical methods to prove the above conclusions. Specific steps include:
1. **Geometric preliminaries**: Prove that hyperbolic surfaces with a large number of cusps contain many embedded pants with controllable boundary lengths, and these pants have at least one cusp.
2. **Effective prime geodesic theorem**: Use Naud's result to control the number of closed geodesics on each pair of pants.
3. **Selberg trace formula**: Compare the contributions on the spectral side with those on the geometric side through the trace formula, thereby deriving the lower bound of the number of small eigenvalues.
### Summary
The main contribution of this paper is to find the topological lower bound of the number of small eigenvalues on hyperbolic surfaces with a large number of cusps, which provides a new perspective for further understanding the spectral properties of hyperbolic surfaces.