Isoperimetric inequalities and sharp upper bounds for Aharonov-Bohm eigenvalues on surfaces

Marco Michetti, Luigi Provenzano, Alessandro Savo · 2026

Read Isoperimetric inequalities and sharp upper bounds for Aharonov-Bohm eigenvalues on surfaces by Marco Michetti, Luigi Provenzano, Alessandro Savo (2026) 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-2604-11718.txt.

Introduction

Michetti, Provenzano, and Savo bound the first eigenvalue of the magnetic Laplacian, with no magnetic field, on simply connected compact surfaces, tying it to a limit on Gaussian curvature. Among spherical regions of fixed area, they show the value peaks for a geodesic disk with the magnetic pole at its centre. For a sphere pierced at two points, the peak instead comes when those points sit opposite each other.

Catalog record

CREATORMarco Michetti, Luigi Provenzano, Alessandro Savo
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE67733
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.11718

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Isoperimetric inequalities and sharp upper bounds for Aharonov-Bohm eigenvalues on surfaces Thanks: The first author acknowledges support of INDAM-GNAMPA. The second and third authors acknowledge support of INDAM-GNSAGA. The second author acknowledges support of the INDAM-GNSAGA 2025 project “Analisi geometrica e teoria spettrale su varietà riemanniane ed hermitiane” and of the the project “Perturbation problems and asymptotics for elliptic differential equations: variational and potential theoretic methods” funded by the European Union – Next Generation EU and by MUR-PRIN-2022SENJZ3. The authors would like to thank the Isaac Newton Institute for Mathematical Sciences, Cambridge, for support and hospitality during the programme Geometric spectral theory and applications, where work on this paper was undertaken. This work was supported by EPSRC grant EP/Z000580/1. Marco Michetti, Luigi Provenzano, Alessandro Savo Abstract . We consider the first eigenvalue of the magnetic Laplacian with zero magnetic field on simply connected compact surfaces and we establish isoperimetric inequalities and upper bounds in terms of a bound on the gaussian curvature. As a corollary, we prove that among all simply connected spherical domains of fixed area, the first eigenvalue is maximal for a geodesic disk with the pole of the magnetic potential at its center; also, for the sphere punctured at two points, the first eigenvalue is maximal when the punctures are antipodal. 1991 Mathematics Subject Classification 35P15, 49Rxx, 58J50, 81Q10 1. Introduction In this paper we consider the first eigenvalue, or ground state energy , of the magnetic Laplacian on simply connected domains of the sphere and, more generally, on compact, simply connected, Riemannian surfaces. Let (M,g) be a compact Riemannian manifold (of any dimension, with or without boundary). A magnetic potential is a smooth real 1 -form A on M . It defines a magnetic differential d^{A} acting on smooth, complex-valued functions as follows: d^{A}u=du-iuA, and a magnetic codifferential \delta^{A} , acting on 1 -forms \omega as \delta^{A}(\omega)=\delta\omega+i\langle A,\omega\rangle_{g} . Here \delta is the usual codifferential associated with the metric g , which in \mathbb{R}^{n} with the Euclidean metric is -{\rm div} . The magnetic Laplacian acts on smooth complex-valued functions as follows: \Delta_{A}u=\delta^{A}d^{A}u. When \partial M\neq\emptyset , we consider magnetic Neumann boundary conditions d^{A}u(N)=0 , which in terms of the dual vector field \nabla^{A}u:=\nabla u-iuA^{\sharp} , read \langle\nabla^{A}u,N\rangle=0 , where N is the unit outer normal to \partial M . Here A^{\sharp} denotes the vector field dual to the 1 -form A ; the expression \nabla^{A}u denotes the magnetic gradient of u . Then \Delta_{A} defines a compact self-adjoint operator on L^{2}(M,dv_{g}) , hence it has a non-negative spectrum made of eigenvalues of finite multiplicity diverging to +\infty : 0\leq\mu_{1}(M,g,A)\leq\mu_{2}(M,g,A)\leq\cdots\leq\mu_{k}(M,g,A)\leq\cdots\nearrow+\infty. As usual, the eigenvalues are characterized by the min-max principle: \mu_{k}(M,g,A)=\min_{{\begin{subarray}{c}U\subset H^{1}_{A}(M)\\ {\rm dim}U=k\end{subarray}}}\max_{0\neq u\in U}\frac{\int_{M}|d^{A}u|_{g}^{2}dv_{g}}{\int_{M}|u|^{2}dv_{g}}, where H^{1}_{A}(M) is the magnetic Sobolev space , namely, the space of complex-valued functions in L^{2}(M,dv_{g}) with d^{A}u in L^{2}(M,dv_{g}) . If A is smooth, this space coincides with the usual Sobolev space H^{1}(M,dv_{g}) . Here and in what follows we drop the dependence of the eigenvalue on the metric when it is clear from the context, and if (M,g) is a Riemannian manifold, we write simply \mu_{k}(M,g,A)\doteq\mu_{k}(M,A). Analogously, one can define the magnetic Dirichlet eigenvalues: 0<\lambda_{1}(M,g,A)\leq\lambda_{2}(M,g,A)\leq\cdots\leq\lambda_{k}(M,g,A)\leq\cdots\nearrow+\infty which are characterized by (1.1) \lambda_{k}(M,g,A)=\min_{{\begin{subarray}{c}U\subset H^{1}_{0,A}(M)\\ {\rm dim}U=k\end{subarray}}}\max_{0\neq u\in U}\frac{\int_{M}|d^{A}u|_{g}^{2}dv_{g}}{\int_{M}|u|^{2}dv_{g}}, where H^{1}_{0,A}(M) is the subspace of H^{1}_{A}(M) of functions with zero boundary trace. Also in this case we set \lambda_{k}(M,g,A)\doteq\lambda_{k}(M,A) . The magnetic Dirichlet eigenvalues will appear in the proof of the main results. 1.1. Gauge invariance. A fundamental feature of the magnetic Laplacian is gauge invariance . Recall that the flux of a 1 -form A around a closed loop c in M is defined as \frac{1}{2\pi}\oint_{c}A . We have the following fundamental theorem which characterizes potentials that can be “ gauged-away ” (see [ 17 ] , see also [ 6 , Appendix 5] ). Theorem 1.1 . a) Assume that A,A^{\prime} are such that A-A^{\prime} is closed and \frac{1}{2\pi}\oint_{c}(A-A^{\prime})\in\mathbb{Z} for any simple closed curve c in M . Then \mu_{k}(M,A)=\mu_{k}(M,A^{\prime}) . b) One has \mu_{1}(M,A)=0 if and only if A is closed and \frac{1}{2\pi}\oint_{c}A\in\mathbb{Z} for any simple closed curve c in M . In particular, if the flux of A around any closed curve in M is an integer (in particular, if A is exact) then the magnetic spectrum coincides with the usual Laplacian spectrum for Neumann boundary conditions, which we simply denote \mu_{k}(M) : that is \mu_{k}(M,A)=\mu_{k}(M)\quad\text{for all $k$}. According to this numbering \mu_{1}(M)=0 so that the first positive Neumann eigenvalue is then \mu_{2}(M) . We remark that, classically, the 2 -form B\doteq dA is the magnetic field associated to A . We then remark that the magnetic field does not determine the spectrum; in particular, the magnetic field could be zero and yet the lowest eigenvalue (ground state energy) is positive; this happens when the flux of A around some curve is not an integer. 1.2. Aharonov-Bohm potentials The scope of the present paper is precisely to estimate the ground state energy when the magnetic field is zero on the regular part of singular potentials called Aharonov-Bohm potentials , which are the object of intensive work, both from the physical and the mathematical point of view. On a simply connected domain \Omega , an Aharonov-Bohm potential with pole at x_{0}\in\Omega is a closed smooth 1 -form on \Omega\setminus\{x_{0}\} . The circulation of A around \partial\Omega (and around any loop enclosing x_{0} ) is a real number \nu which is called the flux of A . In \mathbb{R}^{2} , an example of Aharonov-Bohm potential with pole at the origin and flux \nu is A^{(\nu)}_{0}=\nu\left(-\frac{y}{x^{2}+y^{2}}dx+\frac{x}{x^{2}+y^{2}}dy\right). We consider magnetic Neumann boundary conditions, so that the spectrum is discrete (see e.g., [ 8 ] ). By what we said above \mu_{1}(\Omega,A^{(\nu)}_{0})>0 if and only if \nu\notin\mathbb{Z} . We remark that any two Aharonov-Bohm potentials with pole at x_{0} and flux \nu differ by an exact form, hence give rise to the same spectrum. In this paper we are interested in shape optimization results for the first Aharonov-Bohm eigenvalue of compact, simply connected surfaces (with or without boundary). 1.3. State of the art Punctured domains in \mathbb{R}^{2} . The following isoperimetric inequality has been proved in [ 8 ] : for any bounded, smooth domain \Omega\subset\mathbb{R}^{2} and any flux \nu we have (1.2) \mu_{1}(\Omega,A^{(\nu)}_{x_{0}})\leq\mu_{1}(D,A^{(\nu)}_{0}) for all x_{0}\in\Omega . Here D is the disk centered at 0 with the same area: \left|\Omega\right|=\left|D\right| . Equality holds if and only if \Omega=D and x_{0}=0 . In fact, the isoperimetric inequality ( 1.2 ) holds for any domain (not necessarily simply connected). The proof uses the classical argument of Weinberger for the second eigenvalue of the Neumann Laplacian [ 19 ] . Note that in polar coordinates (r,\theta) around 0 , A^{(\nu)}_{0}=\nu d\theta . The result holds also for bounded, smooth domains of the hyperbolic plane \mathbb{H}^{2} , with the same proof. Spherical domains. On the two-dimensional round sphere \mathbb{S}^{2} , fix a point x_{0} and consider polar coordinates (r,\theta) based at x_{0} , where r is the geodesic distance from x_{0} . Then A_{x_{0}}^{(\nu)}\doteq\nu d\theta defines a smooth 1 -form on \mathbb{S}^{2}\setminus\{x_{0},-x_{0}\} which is singular at the two poles: x_{0} and -x_{0} , and defines a Aharonov-Bohm potential on \mathbb{S}^{2} . We note that the

Read the complete text