Optimal eigenvalues on a metric graph with densities
Kiyan Naderi, Noema Nicolussi · 2025
Read Optimal eigenvalues on a metric graph with densities by Kiyan Naderi, Noema Nicolussi (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-2512-20508.txt.
Introduction
Kiyan Naderi and Noema Nicolussi study vibration-like operators on graphs whose edges have lengths and weights. Their Laplacians live on finite metric graphs endowed with generalized densities, measures of finite mass. The approach unifies discrete graph Laplacians, Kirchhoff Laplacians, and Dirichlet-to-Neumann operators under one roof.
The main question is spectral optimization: which density makes an eigenvalue as small as possible? Unlike domains and manifolds, a minimizer for the k-th eigenvalue provably exists, while maximization turns out to be meaningless here. The first optimal eigenvalue admits a transparent geometric characterization through the resistance metric. Higher optimal eigenvalues obey a Weyl law, linking their growth to the graph's geometry. The results guide how to distribute mass on a network to tune its spectrum.
Catalog record
| CREATOR | Kiyan Naderi, Noema Nicolussi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 114 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-20508
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Optimal eigenvalues on a metric graph with densities Thanks: This research was funded in whole or in part by the Austrian Science Fund (FWF) [Grant DOI: 10.55776/PAT4632823]. For open access purposes, the authors have applied a CC BY public copyright license to any author-accepted manuscript version arising from this submission. Kiyan Naderi, Noema Nicolussi Abstract . We introduce and study Laplacians on a finite metric graph endowed with generalized densities, that is, measures of finite mass. One important motivation is that this setting provides a common framework for several interesting classes of operators: discrete graph Laplacians, Kirchhoff Laplacians and Dirichlet-to-Neumann operators on graphs. Our main interest lies in spectral optimization with respect to the underlying measure. In contrast to the setting of domains and manifolds, we prove that a minimal k k -th eigenvalue exists, whereas the corresponding maximization problem has no meaning. We then establish connections between these optimal eigenvalues and the geometry of the metric graph, including a transparent geometric characterization of the first optimal eigenvalue via the resistance metric, and a Weyl law for the higher optimal eigenvalues. 2020 Mathematics Subject Classification Primary 81Q35 ; 35R02 ; Secondary 35P15 ; 35P20 1. Introduction Let X be a Euclidean domain or Riemannian manifold and consider the eigenvalue problem (1.1) \Delta f=\lambda f\mu, where \Delta is the Laplacian, \lambda the spectral parameter and \mu a suitable positive measure on X . A natural question is to optimize the eigenvalues of ( 1.1 ) when \mu varies in a fixed class of measures. Such problems have a long tradition and we only point out some variants which have been studied. M. G. Krein considered ( 1.1 ) for the Dirichlet Laplacian on a compact interval and measures of fixed mass with an L^{\infty} -density satisfying upper and lower bounds [ 41 ] . He was able to explicitly compute the densities which minimize and maximize the k -th eigenvalue. The analogous problem for higher dimensional domains is sometimes referred to as inhomogeneous or composite media (or membranes in dimension n=2 ) and has received much attention, see the overview in [ 30 , Chapter 9] and [ 9 , 15 , 16 , 23 ] . In the setting of Riemannian manifolds, the optimization problem appeared recently for measures with smooth densities in [ 12 , 13 , 36 , 48 ] . For Riemannian surfaces, the corresponding maximization problem coincides with the one for conformal eigenvalues due to conformal invariance of the energy form [ 11 ] . Concerning more general measures, eigenvalue optimization for Riemannian surfaces was studied by Kokarev in [ 39 ] and appeared more recently in [ 35 ] . On higher dimensional manifolds, eigenvalues associated to general measures were also considered in [ 26 ] . This paper concerns the above optimization problem on a compact metric graph G (i.e., a finite graph whose edges are considered as intervals of certain lengths). The first task consists in giving a meaning to the eigenvalue equation ( 1.1 ) on G . To this end, to each finite Borel measure \mu on G , we associate a Laplacian H_{\mu}\colon\operatorname{dom}(H_{\mu})\subseteq L^{2}(G,\mu)\to L^{2}(G,\mu) which is a self-adjoint operator in the L^{2} -space L^{2}(G,\mu) . The construction is based on considering the energy form q(f)=\int_{G}|f^{\prime}(x)|^{2}\,dx in L^{2}(G,\mu) (see Section 3 for details). It turns out that several well-studied classes of operators fall into this framework (see Section 3.2 ). If the support of \mu coincides with the vertex set V of G , then the operator H_{\mu} is a weighted discrete Laplacian (Example 3.8 ). On the other hand, if \mu has a positive density with respect to the Lebesgue measure dx on G , then H_{\mu} turns out to be a weighted Kirchhoff Laplacian (a.k.a. quantum graph [ 5 , 42 ] , see Example 3.11 ). Discrete Laplacians and Kirchhoff Laplacians are known to enjoy non-trivial relations [ 7 , 20 , 40 , 45 , 50 ] and, thus, putting them into the above common framework is also conceptually interesting. As another special case, weighted Dirichlet-to-Neumann operators on G correspond to measures \mu supported on a proper subset of the vertex set V (Example 3.9 ). For a general measure \mu , the Laplacian H_{\mu} acts on each interval edge as a measure-derivative f\mapsto-\frac{d}{d\mu}\frac{d}{dx}f (see Appendix A ). In the special case of only one interval, such generalized differential operators are also known as Krein strings [ 33 ] . Each Laplacian H_{\mu} has non-negative, purely discrete spectrum. Counting multiplicities, we enumerate its non-zero eigenvalues as an infinite, increasing sequence (\lambda_{k}(H_{\mu}))_{k=1}^{\infty} , with convention \lambda_{k}(H_{\mu}):=+\infty for k\geq\dim(L^{2}(G,\mu)) . Our main interest is then to optimize the eigenvalue \lambda_{k}(H_{\mu}) , k\geq 1 , with respect to the measure \mu . For obvious scaling reasons, we impose the normalization condition \mu(G)=1 , that is, we restrict to the class \mathcal{N} of probability measures on G . It turns out that the maximization problem is trivial in the sense that (Corollary 4.2 ) \sup_{\mu\in\widetilde{\mathcal{N}}}\lambda_{k}(H_{\mu})=+\infty for all weak- \ast -dense sets of measures \widetilde{\mathcal{N}}\subset\mathcal{N} . However, the minimization problem is non-trivial with \inf_{\mu\in\widetilde{\mathcal{N}}}\lambda_{k}(H_{\mu})=\min_{\mu\in\mathcal{N}}\lambda_{k}(H_{\mu})>0 for every dense set \widetilde{\mathcal{N}}\subset\mathcal{N} . The above claims follow from a continuity result (Theorem 4.1 ), which asserts that the k -th eigenvalue \lambda_{k}(H_{\mu}) depends continuously on \mu in the weak- \ast topology. Let us stress that this behavior is actually opposite to the situation for higher dimensional manifolds. Indeed, in dimension n\geq 2 , the k -th eigenvalue on the weak- \ast -dense class of probability measures with smooth, positive densities remains uniformly bounded above, but can be made arbitrarily small [ 12 , 13 , 29 ] . On the other hand, it was already observed in [ 14 ] that the behavior in dimension n=1 is opposite. One explanation is that, in higher dimensions, the dependence of the eigenvalues \lambda_{k}(\mu) on \mu is only upper semi-continuous, see e.g. [ 39 , Proposition 1.1] and [ 26 , Proposition 4.1] . Motivated by the above, we define the k -th optimal eigenvalue of a compact metric graph G as \lambda_{k}^{\min}(G):=\min_{\mu\in\mathcal{N}}\lambda_{k}(H_{\mu}). Let us stress the following physical interpretation of \lambda_{k}^{\min}(G) . The measure \mu can be interpreted as a mass distribution in a weighted wave equation \mu\cdot\partial_{tt}u(x,t)=\Delta u(x,t). After separating the variables, we obtain precisely the eigenvalue equation ( 1.1 ). Thus, the above minimization problem is equivalent to distributing the mass on an oscillating graph in order to obtain the smallest possible k -th natural frequency. Spectral optimization on metric graphs has so far been investigated mainly in terms of the shape of the graph. Related optimization in terms of metrics, i.e., edge lengths, can be found, e.g., in [ 2 ] . To the best of our knowledge, our approach to optimization with regard to measures has not yet been considered. In this regard, we would also like to draw attention to the recent preprint [ 8 ] , in which optimization simultaneously with regard to metrics and smooth measures is studied. The remainder of the paper is devoted to the study of the optimal eigenvalues. The first optimal eigenvalue \lambda_{1}^{\min}(G) The optimal eigenvalue \lambda_{1}^{\min}(G) displays several interesting properties. For instance, in Section