Extension and rigidity of Perrin’s lower bound estimate for Steklov eigenvalues on graphs
Yongjie Shi 1, Chengjie Yu 2 · 2025
Read Extension and rigidity of Perrin’s lower bound estimate for Steklov eigenvalues on graphs by Yongjie Shi 1, Chengjie Yu 2 (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-2504-07361.txt.
Introduction
Yongjie Shi and Chengjie Yu extend a known lower bound for Steklov eigenvalues from plain graphs to weighted ones, and pin down exactly when the bound is tight. Their paper appeared in two thousand twenty-five. Steklov eigenvalues describe how a shape or network vibrates with its boundary free. Perrin's earlier estimate covered only unit-weighted graphs.
Catalog record
| CREATOR | Yongjie Shi 1, Chengjie Yu 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 11 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-07361
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Extension and rigidity of Perrin’s lower bound estimate for Steklov eigenvalues on graphs Thanks: 1 Research partially supported by GDNSF with contract no. 2025A1515011144 and NNSF of China with contract no. 11701355. Thanks: 2 Research partially supported by GDNSF with contract no. 2025A1515011144 and NNSF of China with contract no. 11571215. Yongjie Shi 1, Chengjie Yu 2 Abstract . In this paper, we extend a lower bound estimate for Steklov eigenvalues by Perrin [ 14 ] on unit-weighted graphs to general weighted graphs and characterise its rigidity. 2020 Mathematics Subject Classification Primary 39A12; Secondary 05C50 1. Introduction On a Riemannian manifold M with boundary, the Steklov operator sends the Dirichlet boundary data of a harmonic function to its Neumann boundary data. Steklov operator is a first order nonnegative self-adjoint elliptic pseudo-differential operator on \Sigma:=\partial M (see [ 19 , Chapter 7] ). The eigenvalues of the Steklov operator on M is called the Steklov eigenvalues of M . Such kinds of eigenvalues were first introduced by Steklov [ 18 ] when considering liquid sloshing. It was later found deep applications in geometry (see [ 2 , 3 , 4 ] ) and applied mathematics (see [ 12 ] ). For recent progresses of the topic, interested readers can consult the surveys [ 1 ] and [ 5 ] . In recent years, Steklov eigenvalues were introduced to discrete setting by Hua-Huang-Wang [ 10 ] and Hassannezhad-Miclo [ 7 ] independently. Although the notion is new, there are quite a number of works considering isoperimetric estimate (see [ 6 , 7 , 8 , 10 , 11 , 15 ] ), monotonicity (see [ 9 , 21 ] ), Lichnerowicz estimate (see [ 16 , 17 ] ) and extremum problems (see [ 22 , 13 ] ) of Steklov eigenvalues in the discrete setting. In this paper, we extend a lower bound for Steklov eigenvalues by Perrin [ 14 ] on unit-weighted graphs to general weighted graphs and characterise its rigidity. Let’s recall two lower bound estimates of Perrin [ 14 ] first. In [ 14 ] , Perrin obtained the following interesting lower bound for graphs equipped with unit weight. Theorem 1.1 (Perrin [ 14 ] ) . Let (G,B) be a connected finite graph with boundary such that E(B,B)=\emptyset and equipped with the unit weight. Then, (1.1) \sigma_{2}\geq\frac{|B|}{(|B|-1)^{2}d_{B}} where (1.2) d_{B}=\max\{d(x,y)\ |\ x,y\in B\}. Moreover, if the equality of ( 1.1 ) holds, then |B|=2 . Note that the lower bound in ( 1.1 ) is sharp. For example, the equality of ( 1.1 ) is attained when G is path with the two end vertices the boundary vertices. For graphs equipped with general weight, Perrin [ 14 ] obtained the following lower bound. Theorem 1.2 (Perrin [ 14 ] ) . Let (G,B,m,w) be a connected weighted finite graph with boundary such that E(B,B)=\emptyset . Then, (1.3) \sigma_{2}\geq\frac{w_{0}}{d_{B}V_{B}} where (1.4) V_{B}=\sum_{x\in B}m_{x}\mbox{ and }w_{0}=\min_{e\in E(G)}w_{e}. Note that when G is equipped with the unit weight in Theorem 1.2 , the lower bound ( 1.3 ) is weaker than ( 1.1 ) and is not sharp. So, Theorem 1.2 does not restore the lower bound in ( 1.1 ) when the graph is equipped with the unit weight and is not an appropriate extension of Theorem 1.1 for general weighted graphs. In this short note, we first obtain a more appropriate extension of Theorem 1.1 for general weighted graphs. Theorem 1.3 . Let (G,B,m,w) be a connected weighted finite graph with boundary. Then, (1.5) \sigma_{2}\geq\frac{w_{0}V_{B}}{(V_{B}-m_{0})^{2}d_{B}} where m_{0}=\min_{x\in B}m_{x} . It is clear that ( 1.5 ) is stronger than ( 1.3 ) and when the graph is equipped with the unit weight, ( 1.5 ) becomes ( 1.1 ). So, Theorem 1.3 is an appropriate extension of Theorem 1.1 to general weighted graphs. Secondly, we characterise the rigidity of ( 1.5 ). Theorem 1.4 . Let (G,B,m,w) be a connected weighted finite graph with boundary such that \sigma_{2}=\frac{w_{0}V_{B}}{(V_{B}-m_{0})^{2}d_{B}}. Then, we have the following conclusions: (1) |B|=2 and m_{x}=m_{0} for x\in B ; (2) There is a unique path P:v_{0}\sim v_{1}\sim\cdots\sim v_{d_{B}} with v_{0},v_{d_{B}}\in B . Moreover, w_{v_{i-1}v_{i}}=w_{0} for i=1,2,\cdots,d_{B} ; (3) G is a comb over P . Conversely, it is not hard to check that when the graph (G,B,m,w) satisfies (1)–(3) in Theorem 1.4 . Then the equality of ( 1.5 ) holds. For the definition of a comb, see Definition 2.2 . Finally, we would like to mention that the graphs with boundary we considered in this paper are more general than those in [ 14 ] . More precisely, we don’t assume that E(B,B)=\emptyset a priori. The rest of the paper is organized as follows. In Section 2, we introduce some preliminaries on Steklov operators and Steklov eigenvalues for graphs; In Section 3, we prove Theorem 1.3 and Theorem 1.4 . 2. Preliminary In this section, we introduce some preliminaries on Steklov operators and Steklov eigenvalues for graphs. We first introduce the notion of weighted graphs with boundary. Definition 2.1 . A quadruple (G,B,m,w) is called a weighted graph with boundary if (1) G=(V,E) is a simple graph and B\subset V ; (2) m:V\to\mathbb{R}^{+} and w:E\to\mathbb{R}^{+} . The set B is called the boundary of G , \Omega:=V\setminus B is called the interior of G , m is called the vertex-measure and w is called the edge-weight. When m\equiv 1 and w\equiv 1 , G is said to equip with the unit weight. For convenience, we also view w as a symmetric function on V\times V with w_{xy}=\left\{\begin{array}[]{ll}w(e)&e=\{x,y\}\in E\\ 0&\{x,y\}\notin E\end{array}\right. Let (G,B,m,w) be a connected weighted finite graph with boundary. Denote the space of functions on V as A^{0}(G) and the space of skew-symmetric functions \alpha on V\times V such that \alpha(x,y)=0 when x\not\sim y as A^{1}(G) . Equip A^{0}(G) and A^{1}(G) with the natural inner products: {\left\langle u,v\right\rangle}=\sum_{x\in V}u(x)v(x)m_{x} and {\left\langle\alpha,\beta\right\rangle}=\sum_{\{x,y\}\in E}\alpha(x,y)\beta(x,y)w_{xy}=\frac{1}{2}\sum_{x,y\in V}\alpha(x,y)\beta(x,y)w_{xy} respectively. For any u\in A^{0}(G) , define the differential du of u as du(x,y)=\left\{\begin{array}[]{ll}u(y)-u(x)&\{x,y\}\in E\\ 0&\mbox{otherwise.}\end{array}\right. Let d^{*}:A^{1}(G)\to A^{0}(G) be the adjoint operator of d:A^{0}(G)\to A^{1}(G) . The Laplacian operator on A^{0}(G) is defined as \Delta=-d^{*}d. By direct computation, \Delta u(x)=\frac{1}{m_{x}}\sum_{y\in V}(u(y)-u(x))w_{xy} for any x\in V . Moreover, by the definition of \Delta , it is clear that (2.1) {\left\langle\Delta u,v\right\rangle}=-{\left\langle du,dv\right\rangle} for any u,v\in A^{0}(G) . Moreover, for any u\in A^{0}(G) and x\in B , define the outward normal derivative of u at x as: (2.2) \frac{\partial u}{\partial n}(x):=\frac{1}{m_{x}}\sum_{y\in V}(u(x)-u(y))w_{xy}=-\Delta u(x). Then, by ( 2.1 ), one has the following Green’s formula: (2.3) {\left\langle\Delta u,v\right\rangle}_{\Omega}=-{\left\langle du,dv\right\rangle}+{\left\langle\frac{\partial u}{\partial n},v\right\rangle}_{B}. Here, for any set S\subset V , {\left\langle u,v\right\rangle}_{S}:=\sum_{x\in S}u(x)v(x)m_{x}. For each f\in\mathbb{R}^{B} , let u_{f} be the harmonic extension of f into \Omega : \left\{\begin{array}[]{ll}\Delta u_{f}(x)=0&x\in\Omega\\ u_{f}(x)=f(x)&x\in B.\end{array}\right. Define the Steklov operator \Lambda:\mathbb{R}^{B}\to\mathbb{R}^{B} as \Lambda(f)=\frac{\partial u_{f}}{\partial n}. By ( 2.3 ), {\left\langle\Lambda(f),g\right\rangle}_{B}={\left\langle du_{f},du_{g}\right\rangle} for any f,g\in\mathbb{R}^{B} . This implies that \Lambda is a nonnegative self-adjoint operator on \mathbb{R}^{B} . The eigenvalues of \Lambda is called the Steklov eigenvalues of (G,B,m,w) . Let 0=\sigma_{1}<\sigma_{2}\leq\cdots\leq\sigma_{|B|} be the eigenvalues of \Lambda . Here, \sigma_{1}=0 because nonzero constant functions are the corresponding eigenfunctions and \sigma_{2}>0 because we assume that G is connected. It is clear that \sigma_{2}=\min_{0\neq f\in\mathbb{R}^{B}:{\left\langle f,1\right\rangle}_{B}=0}\frac{{\left\langle du_{f},du_{f}\right\rangle}}{{\left\langle f,f\right\rangle}_{B}}. When i>|B| ,