Steklov Spectrum in Riemannian Geometry

• Steklov spectrum is the set of eigenvalues from boundary problems that encode the manifold's geometric and analytic characteristics.
• It leverages the Dirichlet-to-Neumann map and min-max principles to relate eigenvalue behavior with boundary invariants and asymptotic properties.
• Applications include shape optimization, inverse problems, and spectral rigidity through conformal variations and stability under perturbation.

The Steklov spectrum is the set of eigenvalues arising from the Steklov boundary value problem on a compact Riemannian manifold with boundary, and encodes deep geometric and analytic properties of the underlying manifold and its boundary. The Steklov problem, originally formulated for the Laplace operator, prescribes the spectral parameter in the boundary condition rather than in the interior, distinguishing its spectral behavior from the more classic Dirichlet and Neumann problems. Over the past two decades, research has advanced the understanding of the Steklov spectrum in spectral geometry, geometric analysis, and inverse problems, with major contributions encompassing sharp asymptotics, boundary and topological invariants, isoperimetric inequalities, stability under perturbation, and inverse determination results.

1. Formulation and Basic Spectral Properties

Given a smooth compact Riemannian manifold (or orbifold) $(M, g)$ with nonempty boundary $\partial M$, the classical Steklov eigenvalue problem seeks nontrivial $u$: $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$ where $\Delta_g$ is the Laplace–Beltrami operator and $\partial_\nu$ is the outward normal derivative. The spectrum $\{ \sigma_j \}$ is discrete, real, and nonnegative: $$0 = \sigma_0 < \sigma_1 \leq \sigma_2 \leq \cdots \to +\infty.$$ Equivalently, $\sigma$ is a Steklov eigenvalue if there exists a nonzero solution $u$ so that the boundary value problem above is satisfied. The Dirichlet–to–Neumann (DN) map $\partial M$0 is defined by $\partial M$1, where $\partial M$2 is harmonic with $\partial M$3; its spectrum coincides with the set of Steklov eigenvalues.

The DN operator is a self-adjoint, positive, first-order elliptic pseudodifferential operator on $\partial M$4, and admits a min–max variational characterization: $\partial M$5 The Steklov spectrum also generalizes naturally to problems for Maxwell’s equations and other boundary conditions (Ferraresso et al., 2022).

2. Spectral Asymptotics and Invariants

Weyl Law and Sharpened Asymptotics

The counting function $\partial M$6 satisfies the Weyl law: $\partial M$7 where $\partial M$8 is the unit $\partial M$9-ball. In two dimensions, the spectral sequence is $u$0-close to a multiset determined by boundary component lengths (Girouard et al., 2013): $u$1 where $u$2 and $u$3 are the boundary lengths.

Boundary lengths and the number of boundary components are thus complete Steklov spectral invariants for smooth surfaces. In higher dimensions, only the total boundary measure and not finer topological invariants are spectrally determined (Girouard et al., 2013, Girouard et al., 2014).

Heat Trace Invariants

The heat trace asymptotics for the semigroup $u$4 yield local invariants on $u$5: $u$6 where $u$7, $u$8, and further $u$9 involve curvature and its derivatives (Polterovich et al., 2013). These coefficients appear in the nonlocal inverse spectral problem and in rigidity results for balls (Polterovich et al., 2013).

3. Conformal Geometry, Moduli, and Spectral Extremals

Maximization and Minimization under Conformal Variations

For a compact surface $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$0 with boundary, the normalized Steklov eigenvalues $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$1 are invariant under homotheties. The supremum over all metrics in a conformal class $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$2 is denoted $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$3.

Degenerations in conformal moduli correspond to pinching geodesics or boundary components; the limit of $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$4 splits as a sum over Steklov spectra on components and disks, following a precise partition of $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$5 (Medvedev, 2020). Friedlander–Nadirashvili–type invariants,

$$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$6

are shown to satisfy $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$7 for all compact surfaces with boundary, orientable or not.

Upper bounds on $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$8 for orientable $$\begin{cases} \Delta_g u = 0 & \text{in } M, \ \partial_\nu u = \sigma u & \text{on } \partial M, \end{cases}$$9 of genus $\Delta_g$0 with $\Delta_g$1 boundary components satisfy

$\Delta_g$2

with an explicit extension to non-orientable surfaces (Medvedev, 2020).

Conformal and Boundary-Rigidity

In dimension $\Delta_g$3, under the assumption that the boundary metric has Anosov geodesic flow with simple length spectrum, the Steklov spectrum determines the full boundary jet (i.e., all normal derivatives at the boundary) of the conformal factor (Florentin, 26 Jan 2025). In the real analytic setting, isospectrality within the conformal class implies equality of the metrics. Extended to Schrödinger operators, the boundary jet of the potential is likewise spectrally determined.

These results use deep microlocal techniques: wave trace invariants (Duistermaat–Guillemin singularities), X-ray transform injectivity, and Livšic rigidity for Anosov flows.

4. Boundary and Topological Detection

Polygonal and Orbisurface Cases

In the planar setting, for convex polygons, the full Steklov spectrum determines almost all triangles uniquely (admissible case: no angle is an odd rational multiple of $\Delta_g$4), with explicit finite bounds for non-congruent isospectral polygons (Dryden et al., 21 Apr 2026, Dryden et al., 2024). For regular $\Delta_g$5-gons, spectral determination holds within the class of all polygons. Distinction from smoothly bounded planar domains is achieved via the so-called characteristic polynomial of the domain, constructed from spectral data.

For two-dimensional Riemannian orbifolds, the Steklov spectrum determines the number of smooth and singular (reflector) boundary components and their lengths modulo a finite equivalence relation; for type II (reflector) boundaries with distinct lengths, all are spectrally determined (Arias-Marco et al., 2016). The Steklov spectrum does not detect interior singularities or the orbifold Euler characteristic in dimension two, nor does it always distinguish between certain geometric types (e.g., a flat disk vs. cone).

Determination of the Warping Factor

For Riemannian warped products of the form $\Delta_g$6 with metric $\Delta_g$7, the Steklov spectrum determines the warping function $\Delta_g$8 uniquely. Logarithmic stability results hold for small perturbations of the spectrum (Daudé et al., 2018). The high-frequency asymptotics reconstruct the Taylor expansion of the warping factor at the boundary.

5. Quantitative Spectral Inequalities and Shape Optimization

Isoperimetric and Geometric Inequalities

On any bounded domain in a manifold $\Delta_g$9 conformal to a non-negatively Ricci curved metric, the normalized eigenvalues are controlled in terms of the isoperimetric ratio: $\partial_\nu$0 with $\partial_\nu$1 (Colbois et al., 2011). In dimension $\partial_\nu$2, the isoperimetric ratio is absent, yielding saturation purely via topological complexity (genus, number of boundary components).

In three dimensions and above, there now exist examples of metrics with fixed boundary metric and fixed volume for which the first nonzero Steklov eigenvalue tends to infinity, constructed via warped product geometry (Girouard et al., 2024).

Discretization, Stability, and Domain Perturbation

The Steklov spectrum is spectrally stable under compactly convergent domain perturbations with control on perimeter and boundary regularity (Ferrero et al., 2021). For geometric discretizations of Riemannian manifolds with bounded geometry, the Steklov spectrum of the discretized graph with boundary can be compared multiplicatively with that of the manifold, enabling construction of surfaces with large Steklov gaps and other extremal phenomena (Colbois et al., 2016).

6. Inverse Spectral Problems and Spectral Rigidity

Inverse Results and Spectral Invariants

• Zeta-invariants $\partial_\nu$3 computed from the Steklov spectrum on the unit disk, defined via higher traces and depending on the Fourier coefficients of the boundary weight, are complete isospectral invariants and invariant under the boundary conformal group (Malkovich et al., 2014).
• For domains with Anosov boundary, knowledge of the Steklov spectrum determines analytic conformal factors and boundary jets, establishing rigidity under these dynamical and regularity assumptions (Florentin, 26 Jan 2025).

Open Problems and Extensions

Current questions involve fine behavior on nonsmooth domains, the spectral determination for higher eigenvalues and disconnected boundaries in higher dimensions, the isospectral compactness up to conformal reparametrization, positivity of higher zeta-invariants, and characterizing the geometric data encoded in further heat invariants. For the generalized Steklov–Helmholtz operator $\partial_\nu$4, new computational techniques such as the BIO-MOD discretization have enabled high-precision shape optimization and the study of nontrivial extremals under spectral constraints (Nigam et al., 8 Sep 2025).

7. Tables: Key Steklov Properties and Results

Aspect	Statement / Result	Reference
Mean curvature invariant	$\partial_\nu$5 appears as first heat coefficient	(Polterovich et al., 2013)
Asymptotic for $\partial_\nu$6-th eigenvalue ($\partial_\nu$7)	$\partial_\nu$8 for boundary length $\partial_\nu$9	(Girouard et al., 2013)
Conformal minimum	$\{ \sigma_j \}$0 for all surfaces with boundary	(Medvedev, 2020)
Isoperimetric bound ($\{ \sigma_j \}$1)	$\{ \sigma_j \}$2	(Colbois et al., 2011)
Polygonal uniqueness (generic triangles)	Steklov spectrum uniquely determines triangle, except for finite exceptions	(Dryden et al., 21 Apr 2026)
Polygon vs. smooth domain	No convex triangle or quadrilateral shares Steklov spectrum with a smooth domain	(Dryden et al., 21 Apr 2026)
Warped product uniqueness	Steklov spectrum determines warping function $\{ \sigma_j \}$3 for $\{ \sigma_j \}$4	(Daudé et al., 2018)
Inverse via X-ray + trace	Steklov spectrum rigid under analytic, Anosov dynamical hypotheses	(Florentin, 26 Jan 2025)

• (Girouard et al., 2013): "The Steklov spectrum of surfaces: asymptotics and invariants"
• (Polterovich et al., 2013): "Heat invariants of the Steklov problem"
• (Girouard et al., 2014): "Spectral geometry of the Steklov problem"
• (Medvedev, 2020): "Degenerating sequences of conformal classes and the conformal Steklov spectrum"
• (Florentin, 26 Jan 2025): "Steklov isospectrality of conformal metrics"
• (Dryden et al., 21 Apr 2026): "The Steklov spectrum of convex polygonal domains II"
• (Dryden et al., 2024): "The Steklov spectrum of convex polygonal domains I"
• (Daudé et al., 2018): "Stability in the inverse Steklov problem on warped product Riemannian manifolds"
• (Colbois et al., 2011): "Isoperimetric control of the Steklov spectrum"
• (Colbois et al., 2016): "The Steklov spectrum and coarse discretizations of manifolds with boundary"
• (Malkovich et al., 2014): "Zeta-invariants of the Steklov spectrum for a planar domain"
• (Girouard et al., 2024): "Large Steklov eigenvalues under volume constraints"
• (Nigam et al., 8 Sep 2025): "The spectrum of the Steklov-Helmholtz operator"
• (Ferraresso et al., 2022): "On a Steklov spectrum in Electromagnetics"
• (Ferrero et al., 2021): "Spectral stability of the Steklov problem"
• (Arias-Marco et al., 2016): "Spectral geometry of the Steklov problem on orbifolds"