Sharp Spectral Gap Estimates on Manifolds under Integral Ricci Curvature Bounds
Xavier Ramos Olivé, Shoo Seto, Malik Tuerkoen · 2025
Read Sharp Spectral Gap Estimates on Manifolds under Integral Ricci Curvature Bounds by Xavier Ramos Olivé, Shoo Seto, Malik Tuerkoen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-27083.txt.
Introduction
Xavier Ramos Olive, Shoo Seto, and Malik Tuerkoen prove sharp bounds on spectral gaps for curved spaces with limited curvature control. The spectral gap is the spacing between the lowest vibration modes of a space, a quantity tied to how fast heat and waves settle. The setting is compact manifolds, meaning finite curved spaces without boundary, where curvature is bounded only in an averaged, integral sense.
The results generalize classic estimates by Kroger and by Bakry and Qian to this weaker integral setting. The bounds are sharp, meaning no stronger universal bound of the same form can hold. The authors also confirm a conjecture posed in earlier work by Ramos and co-authors. For geometers, the paper extends reliable mode estimates to spaces where pointwise curvature control is unavailable.
Catalog record
| CREATOR | Xavier Ramos Olivé, Shoo Seto, Malik Tuerkoen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-27083
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Sharp Spectral Gap Estimates on Manifolds under Integral Ricci Curvature Bounds Thanks: S.S. is partially support by AMS-Simons PUI Grant. Thanks: M.T. is partially supported by an AMS-Simons Travel Grant Xavier Ramos Olivé, Shoo Seto, Malik Tuerkoen Abstract . We prove sharp spectral gap estimates on compact manifolds with integral curvature bounds. We generalize the results of Kröger [ 5 ] as well as of Bakry and Qian [ 2 ] to the case of integral curvature and confirm the conjecture in [ 10 ] for the case n ≥ 3 n\geq 3 . 1. Introduction Let M be an n -dimensional compact Riemannian manifold, possibly with nonempty convex and C^{2} boundary. We will consider the eigenvalue problem of the Laplacian on M , \displaystyle\Delta u+\lambda u=0\quad\textup{in }M, imposing Neumann boundary conditions (vanishing of the normal derivative) in the case of nonempty boundary. It is easily seen that the eigenvalues are nonnegative and that \lambda_{0}=0 is simple. By spectral theory of compact self-adjoint operators, there exists a sequence of eigenvalues \displaystyle\lambda_{0}=0<\lambda_{1}\leq\lambda_{2}\leq\dots\rightarrow+\infty. The study of the eigenvalues (or spectrum) has been an active area of research in geometric analysis and here we will focus on the first nonzero eigenvalue \lambda_{1} . In particular, obtaining a sharp quantitative lower bound in terms of geometric data has seen attention since Lichnerowicz [ 6 ] showing that for compact manifolds with a positive Ricci curvature lower bound the first nonzero eigenvalue is bounded below by the first nonzero eigenvalue on an n -sphere with radius matching the Ricci curvature bound, and Zhong and Yang [ 15 ] for compact manifolds with nonnegative Ricci curvature, the first nonzero eigenvalue is bounded below by \frac{\pi^{2}}{D^{2}} where D is the diameter of the manifold. These results can be summarized as \lambda_{1}(M)\geq\begin{cases}\lambda_{1}(S^{n}(1/\sqrt{K}))=nK&\text{ for }\Ric\geq(n-1)K>0,\\ \lambda_{1}(S^{1}(D/\pi))=\frac{\pi^{2}}{D^{2}}&\text{ for }\Ric\geq 0.\end{cases} The above two estimates have been unified by Bakry and Qian [ 2 ] (see also the earlier work by Kröger [ 5 ] ) in the following form: Theorem A . Let M be an n -dimensional compact Riemannian manifold (with possibly non-empty convex and C^{2} boundary), \Ric\geq(n-1)K , K\in\mathbb{R} and \diam(M)\leq D . Let \lambda_{1}(M) be the first nonzero eigenvalue of the Laplacian. Then (1) \displaystyle\lambda_{1}(M)\geq\lambda_{1}(n,K,D), where \lambda_{1}(n,K,D) is the first nonzero eigenvalue of the one-dimensional eigenvalue problem (2) \begin{cases}w^{\prime\prime}-T_{n,K}w^{\prime}+\lambda w=0&\textup{ on }[-\frac{D}{2}\frac{D}{2}]\\ w^{\prime}(\pm\frac{D}{2})=0,\end{cases} where T_{n,K} is given by (3) \displaystyle T_{n,K}(x)=\begin{cases}-(n-1)\sqrt{K}\tan\left(\sqrt{K}x\right),\quad\,\,\,\,\,&\textup{ if }K>0,\\ 0,\quad\quad\,\quad\quad\quad\quad\quad\quad\,&\textup{ if }K=0,\\ -(n-1)\sqrt{-K}\tanh\left(\sqrt{-K}x\right),&\textup{ if }K<0.\end{cases} Remark 1.1 . Bakry-Qian proved the result in a more general framework of symmetric elliptic operators on CD(R,n) spaces. Remark 1.2 . The above theorem unifies the results of Lichnerowicz and Zhong-Yang as it can be directly computed that \lambda_{1}(n,0,D)=\frac{\pi^{2}}{D^{2}} and \lambda_{1}(n,K,\frac{\pi}{\sqrt{K}})=nK when K>0 . In fact, we have an almost interpolation given by Shi and Zhang [ 11 ] : \displaystyle\lambda_{1}(n,K,D)\geq\max_{s\in(0,1)}4(s-s^{2})\frac{\pi^{2}}{D^{2}}+s(n-1)K. For the case K<0, Yang [ 13 ] derived the lower bound (4) \displaystyle\lambda_{1}(M)\geq\frac{\pi^{2}}{D^{2}}\exp\left(-c_{n}D\sqrt{(n-1)|K|}\right) where c_{n}=\max\{2,n-1\}. In this paper we generalize Theorem A to the setting of integral Ricci curvature. To this end, we let \rho(x) be the smallest eigenvalue of the Ricci tensor at x\in M and for a constant K\in\mathbb{R}, we let \rho_{K}(x) be the amount of Ricci curvature below (n-1)K at x , that is, (5) \displaystyle\rho_{K}(x)=\max\left\{-(\rho(x)-(n-1)K),0\right\}. We measure the amount of Ricci curvature below (n-1)K in an L^{p} sense with the following quantity (6) \displaystyle\overline{k}(p,K)=\left(\frac{1}{\textup{Vol}(M)}\int_{M}\rho^{p}\,d\textup{V}\right)^{\tfrac{1}{p}}. Note that \bar{k}(p,K)=0 if and only if \Ric\geq(n-1)K . Many classical results of geometric analysis under a pointwise Ricci lower bound have been generalized to integral curvature, c.f. [ 9 ] . The Lichnerowicz estimate with control on integral curvature has been established by Aubry [ 1 ] . In fact, Aubry showed that for K>0 (7) \displaystyle\lambda_{1}(M)\geq nK\left(1-C(n,p)\overline{k}(p,K)\right). The Zhong-Yang estimate under small integral curvature has been established by the first two authors along with G. Wei and Q.S. Zhang [ 10 ] . In particular, in [ 10 ] it was conjectured that one should be able to obtain an integral curvature version of ( 1 ) using an auxiliary function approach. In this work, we confirm this conjecture for the case n\geq 3 and generalize the results of Kröger and Bakry-Qian to the integral curvature setting. Our main theorem reads as follows. Theorem 1.3 . Let M be an n -dimensional compact Riemannian manifold (with possible non-empty convex and C^{2} boundary), n\geq 3 , and \diam(M)=D . Let \lambda_{1}(M) be the first non-trivial eigenvalue of the Laplacian and let p>\tfrac{n}{2} . Then for any \alpha\in(0,1) there exists \varepsilon_{0}=\varepsilon_{0}(n,p)>0 such that whenever \overline{k}(p,K)<\varepsilon_{0}, one has that \displaystyle\lambda_{1}(M)\geq\alpha{\lambda_{1}}(n,K,D), where \lambda_{1}(n,K,D) is the first non-trivial eigenvalue of the one-dimensional eigenvalue problem ( 2 ). To prove this theorem one shows a gradient comparison, that is, we aim to prove an estimate of the type |\nabla u|^{2}\leq(w^{\prime})^{2}\circ w^{-1}(u(x)), where u is the eigenfunction associated with \lambda_{1}(M) and w is the eigenfunction associated to \lambda_{1}(n,K,D). This approach was introduced by Kröger [ 5 ] and also used by Bakry and Qian [ 2 ] . Valtorta [ 12 ] extended this to the p -Laplacian framework (see also the work by Naber and Valtorta in [ 7 ] ). We extend the gradient comparison to the integral curvature setting. Our approach uses the technique introduced by Zhang and Zhu in [ 14 ] , which was successfully applied in [ 10 ] . The key idea here is to introduce an auxiliary function J that absorbs critical terms in the maximum principle calculation (see Definition 2.4 ). One of the key difficulties in our proof here is that one has to perturb parameters of the one-dimensional model to obtain a gradient comparison. These parameters will be close to the parameters that are used in the pointwise lower bound case. Finally, we will be able to obtain a comparison to the eigenvalue of the model ( 2 ) with the correct parameters since \lambda_{1}(n,K,D) is a continuous function with respect to n,K, and D , i.e., our estimate is sharp in the sense that it recovers Theorem A in the limit where {\rm Ric}\geq(n-1)K . Note that for the case K<0, under the assumptions of Theorem 1.3 , we get an integral curvature version of ( 4 ). In fact, for any \alpha\in(0,1), p>\tfrac{n}{2}, (with n\geq 3) there exists \varepsilon>0 such that whenever \overline{k}(p,K)<\varepsilon then one has that \lambda_{1}\geq\alpha\frac{\pi^{2}}{D^{2}}\exp\left(-c_{n}D\sqrt{(n-1)|K|}\right). It seems to us that the case n=2 does not permit such an approach to Theorem 1.3 and we are unable to obtain a sharp estimate. See the discussion at the end of the proof of Theorem 3 . 1.1. Overview of the Paper In Section 2 , we fix our notation and recall some results from previous work that will be of significance throughout this article. In Section 3 , we prove the key theorem, namely a gradient comparison theorem. The main difficulty there is to circumvent the need for a lower bound on the curvature, which we overcome by introducing the auxiliary function J which absorbs the integral curvature terms (see Lemma 2.5 ). In Section 4 , we show that there exists a one-dimensional model, whose maximum and minimum are the same as the first eigenfunction on the manifold. This is crucial to obtain a sharp spectral gap