Kuttler–Sigillito Inequalities in Spectral Geometry

• Kuttler–Sigillito inequalities are a family of spectral comparison results relating eigenvalues of Laplace, Steklov, and biharmonic Steklov problems based on domain geometry.
• They employ variational methods, Rellich identities, and geometric bounds to derive key estimates, particularly for star-shaped domains and manifold settings.
• Extensions include adaptations to mixed boundary conditions and differential forms, offering practical insights for vibration analysis, heat diffusion, and shape optimization.

Searching arXiv for recent and foundational papers on Kuttler–Sigillito inequalities. Kuttler–Sigillito inequalities are a family of spectral comparison inequalities relating eigenvalues of Laplace, Steklov, and biharmonic Steklov problems on a fixed domain. In their classical form, they compare membrane-type spectra under Dirichlet or Neumann conditions with Steklov-type spectra, typically through geometric quantities such as the support function, radial extrema, boundary convexity, or curvature. Subsequent work has broadened the term to include geometric lower bounds for Steklov eigenvalues on star-shaped domains, manifold analogues derived from Rellich- and Reilly-type identities, mixed-boundary variants, and extensions to differential forms and fourth-order problems (Hassannezhad et al., 2017, Berge, 2021, Xiong, 2019).

1. Classical formulation and historical setting

The original Kuttler–Sigillito program, as described in later sources, concerns inequalities among eigenvalues of several boundary value problems on Euclidean domains, especially in dimension two. In the scalar setting, one formulation quoted in the modern literature is

$$q_1\,\sigma_1^2 \le \ell_1,\qquad \mu_1^{-1} \le \lambda_1^{-1} + (q_1\ell_1)^{-1/2},\qquad \mu_1^{-1} \le \lambda_1^{-1} + (q_1\sigma_1)^{-1},$$

where $\lambda_1$, $\mu_1$, and $\sigma_1$ denote Dirichlet, Neumann, and Steklov eigenvalues, and $q_1,\ell_1$ are biharmonic Steklov-type parameters. A separate chain, later extended to manifolds, takes the form

$$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$

These inequalities are representative of the classical KS perspective: different spectra are not studied in isolation, but as members of a comparison theory tied to domain geometry and variational structure (Hassannezhad et al., 2017, Assali, 10 Feb 2026, Assali, 7 Jul 2025).

The terminology also includes lower bounds for Steklov eigenvalues on star-shaped domains. In the planar Euclidean setting, Kuttler and Sigillito proved lower bounds for the first positive Steklov eigenvalue in terms of star-shapedness data, and later papers treated these estimates as the prototype for higher-dimensional and curved analogues. This suggests that “Kuttler–Sigillito inequalities” now denotes a broader class of comparison results rather than a single formula (Verma, 2018, Verma et al., 2019).

A historically important point is that some statements in the 1969 planar Steklov paper required correction. Modern work shows that the claim “no nodal line is a closed curve” needs simple connectivity, and that the ellipse bound stated there must be reinterpreted. These clarifications are part of the modern understanding of the KS literature rather than a peripheral erratum (DelaTorre et al., 31 Jul 2025).

2. Spectral problems and variational structure

The basic second-order Steklov problem on a bounded domain $\Omega$ is

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$

with discrete spectrum

$$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$

Its first positive eigenvalue admits the Rayleigh characterization

$$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$

This min–max structure is the starting point for essentially all KS-type arguments, because it permits transfer of test functions between different problems and exposes the role of geometric weights on the boundary (DelaTorre et al., 31 Jul 2025, Verma et al., 2019).

Two fourth-order Steklov problems introduced by Kuttler and Sigillito in 1968 are central to later generalizations. On a warped product manifold $\lambda_1$0 with metric $\lambda_1$1, the first problem is

$\lambda_1$2

and the second is

$\lambda_1$3

For the Euclidean ball $\lambda_1$4, the exact spectra are

$\lambda_1$5

These explicit formulas provide the model cases in which later lower bounds are shown to be optimal or rigid (Xiong, 2019).

The variational viewpoint extends without essential change to Dirichlet and Neumann Laplacians, mixed boundary problems, and biharmonic Steklov operators. In the scalar theory, this yields comparison inequalities by placing eigenfunctions of one problem into the Rayleigh quotient of another; in the form-valued theory, the same mechanism persists with $\lambda_1$6, $\lambda_1$7, and the Hodge Laplacian replacing the scalar gradient and Laplacian (Hassannezhad et al., 2017, Assali, 7 Jul 2025).

3. Star-shaped domains and geometric lower bounds

A characteristic KS theme is that star-shapedness converts geometric control into spectral control. On star-shaped domains, the relevant parameters are the inner and outer radii

$\lambda_1$8

the angle $\lambda_1$9 between the outward unit normal and the radial vector, and the slope parameter

$\mu_1$0

or its equivalent boundary-slope representation. In the Euclidean planar case, a KS-type lower bound for every Steklov eigenvalue can be written as

$\mu_1$1

with equality for disks. This formulation expresses the geometric content of the original estimate: radial thickness and maximal boundary tilt reduce the Steklov spectrum relative to the inscribed ball (Verma et al., 2019).

The same structure survives on hypersurfaces of revolution. If $\mu_1$2 has metric

$\mu_1$3

and $\mu_1$4 is star-shaped with respect to the pole, then for every $\mu_1$5,

$\mu_1$6

where

$\mu_1$7

Equality for some $\mu_1$8 forces $\mu_1$9, hence $\sigma_1$0 is a geodesic ball. On the paraboloid $\sigma_1$1, an analogous lower bound holds with the same $\sigma_1$2-factor and radial ratio $\sigma_1$3 (Verma et al., 2019).

A separate sharp extension concerns star-shaped bounded domains in $\sigma_1$4. There the first nonzero Steklov eigenvalue is bounded below by an explicit expression involving $\sigma_1$5, $\sigma_1$6, the angle bound $\sigma_1$7, the quantity

$\sigma_1$8

and the first Steklov eigenvalue of the geodesic ball $\sigma_1$9. Equality occurs if and only if the domain is a geodesic ball. This places the KS philosophy into a positively curved setting and shows that the Euclidean star-shaped estimates are not an artifact of flat geometry (Verma, 2018).

4. Manifold extensions and Rellich methodology

A decisive step in the modern theory is the replacement of Euclidean radial vector fields by distance-based fields on complete Riemannian manifolds. Let $q_1,\ell_1$0, $q_1,\ell_1$1, and

$q_1,\ell_1$2

On a bounded domain $q_1,\ell_1$3 with $q_1,\ell_1$4 boundary, define

$q_1,\ell_1$5

If $q_1,\ell_1$6 is star-shaped with respect to $q_1,\ell_1$7, then $q_1,\ell_1$8. The comparison geometry enters through the Riccati model function $q_1,\ell_1$9, determined by $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$0 with the standard singular asymptotics at the pole (Hassannezhad et al., 2017).

The analytic engine is a generalized Rellich identity. For a Lipschitz vector field $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$1 and $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$2,

$$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$3

With $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$4, this identity produces geometric weights $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$5 on the boundary and curvature-dependent bulk terms through Hessian and Laplacian comparison (Hassannezhad et al., 2017).

From this framework one obtains manifold extensions of classical KS comparisons. Among them are

$$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$6

for the biharmonic Steklov II spectrum $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$7, and the lower bound

$$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$8

where $$\mu_k\,\sigma_1 \le \ell_k,\qquad \mu_1\,\sigma_k \le \ell_k.$$9, with $\Omega$0 in the Euclidean case. There is also a two-sided Dirichlet–Neumann comparison

$\Omega$1

where $\Omega$2 is the multiplicity of $\Omega$3 and $\Omega$4 are explicit curvature-dependent constants. In $\Omega$5, these reduce to

$\Omega$6

The same method also yields an upper bound for biharmonic Steklov I in terms of $\Omega$7 and the generalized second moment of inertia $\Omega$8 (Hassannezhad et al., 2017).

5. Mixed boundary conditions and Rellich–Christianson identities

A major modern extension replaces pure boundary conditions by a partition $\Omega$9. For the mixed Neumann–Dirichlet problem,

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$0

with $$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$1, the min–max formula is taken over

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$2

and the inclusions of Sobolev spaces give

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$3

The mixed Steklov–Dirichlet and Robin–Dirichlet problems are defined on the same space, with spectra $$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$4 and $$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$5 (Berge, 2021).

Under a Ricci lower bound

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$6

for $$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$7 with $$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$8 boundary, and with

$$\Delta u = 0 \quad \text{in } \Omega,\qquad \partial_\nu u = \sigma u \quad \text{on } \partial\Omega,$$9

the principal mixed KS inequality is

$$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$0

For balls $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$1, this becomes

$$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$2

The mixed Robin–Dirichlet spectrum satisfies, for $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$3 and $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$4,

$$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$5

and, whenever differentiable at $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$6,

$$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$7

These statements extend KS-type comparisons from pure to mixed boundary conditions and from the first nonzero eigenvalue to all eigenvalues (Berge, 2021).

The same paper derives a Hadamard formula for simple mixed ND eigenvalues under smooth deformation: $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$8 For dilations $$0=\sigma_0<\sigma_1\le \sigma_2\le \cdots \nearrow +\infty.$$9, this yields the mixed ND Rellich identity

$$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$0

On a polytope $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$1, the same identity becomes a Rellich–Christianson formula expressed through signed distances from a point $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$2 to the supporting hyperplanes of the faces. This is a genuine mixed-boundary analogue of Christianson-type identities previously known in Dirichlet settings (Berge, 2021).

6. Fourth-order, warped-product, and differential-form extensions

On warped product manifolds $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$3 with strictly convex boundary and warping function $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$4, KS-type estimates acquire curvature-sensitive constants built from

$$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$5

For the classical Steklov spectrum, one has

$$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$6

under $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$7, and

$$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$8

under $$\sigma_1 = \min_{\substack{0\ne u\in H^1(\Omega)\ \int_{\partial\Omega}u\,ds=0}} \frac{\int_{\Omega} |\nabla u|^2\,dx}{\int_{\partial\Omega} u^2\,ds}.$$9, with equality if and only if $\lambda_1$00. For the first fourth-order KS problem,

$\lambda_1$01

when $\lambda_1$02, and for the second,

$\lambda_1$03

when $\lambda_1$04; there are corresponding dimension-dependent formulas for $\lambda_1$05 and $\lambda_1$06. In particular,

$\lambda_1$07

in the $\lambda_1$08 regime, confirming the Wang–Xia conjecture on warped product manifolds for $\lambda_1$09 and $\lambda_1$10. A notable methodological point is that the proof uses Reilly’s formula without discarding the Ricci term; instead, a positive piece of that term is extracted to cancel a negative contribution elsewhere (Xiong, 2019).

The KS framework has also been transplanted to differential forms. One line of work introduces three biharmonic Steklov problems with Neumann boundary conditions on $\lambda_1$11-forms, with positive spectra $\lambda_1$12, $\lambda_1$13, and $\lambda_1$14, ordered by

$\lambda_1$15

The principal comparison results are

$\lambda_1$16

with strictness for $\lambda_1$17,

$\lambda_1$18

and a biharmonic Steklov–Steklov inequality

$\lambda_1$19

The same theory yields

$\lambda_1$20

For $\lambda_1$21, these reduce to scalar biharmonic Steklov–Neumann comparisons (Assali, 7 Jul 2025).

A second form-valued development introduces a new biharmonic Steklov problem with Dirichlet-type boundary conditions, denoted BSD2, proves ellipticity by principal-symbol analysis and the Lopatinskii–Shapiro condition, and establishes curvature-dependent KS inequalities on forms. Under star-shapedness with respect to $\lambda_1$22, a Ricci lower bound, and the positivity assumptions $\lambda_1$23 and $\lambda_1$24, there is a strict lower bound for $\lambda_1$25 in terms of $\lambda_1$26, $\lambda_1$27, $\lambda_1$28, and the comparison function $\lambda_1$29. Under sectional curvature pinching $\lambda_1$30, there is also an upper bound

$\lambda_1$31

with explicit $\lambda_1$32, and the auxiliary comparison

$\lambda_1$33

This shows that the KS paradigm extends from scalar elliptic operators to Hodge-theoretic spectral problems (Assali, 10 Feb 2026).

7. Examples, corrections, applications, and limitations

Concrete model domains play a central role in assessing sharpness. For the half-disk $\lambda_1$34, the mixed ND spectrum is

$\lambda_1$35

while the mixed SD spectrum is

$\lambda_1$36

With $\lambda_1$37 equal to the semicircular arc, $\lambda_1$38, $\lambda_1$39, and $\lambda_1$40 in $\lambda_1$41, the mixed KS inequality gives

$\lambda_1$42

For the Robin–Dirichlet problem on the same domain, differentiating the dispersion relation yields

$\lambda_1$43

implying

$\lambda_1$44

On the square $\lambda_1$45 with mixed boundaries, the formulas for the ND and SD spectra make the KS bound fully explicit, while on hyperbolic balls the ball corollary forces $\lambda_1$46 as $\lambda_1$47, in agreement with known behavior (Berge, 2021).

A separate line of examples corrects the 1969 planar Steklov discussion. There exists a doubly connected planar domain for which a first Steklov eigenfunction has a closed nodal line homotopic to the boundary components, showing that the statement “no closed nodal line” fails without simple connectivity. For ellipses

$\lambda_1$48

the correct interpretation of the KS lower bound is

$\lambda_1$49

not $\lambda_1$50. A consistent upper test-function estimate is

$\lambda_1$51

so $\lambda_1$52 as $\lambda_1$53 with $\lambda_1$54 fixed, and $\lambda_1$55 is simple for every noncircular ellipse (DelaTorre et al., 31 Jul 2025).

The current theory also has clear structural limitations. In the mixed-boundary setting, the main KS lower bound becomes asymptotically trivial as $\lambda_1$56 because the SD and ND Weyl laws scale differently. Nontriviality typically requires $\lambda_1$57, as in strictly star-shaped geometry. The strongest results often assume $\lambda_1$58 boundary, though Lipschitz regularity suffices for some Rellich identities. In the form-valued theory, positivity assumptions such as $\lambda_1$59 and $\lambda_1$60 enter essentially, and a lower bound on $\lambda_1$61 in terms of $\lambda_1$62 remains conjectural. Despite these restrictions, KS-type inequalities have been identified as relevant to vibration analysis, acoustics, heat diffusion with partial insulation, inverse problems involving mixed data, and shape optimization (Berge, 2021, Assali, 10 Feb 2026).

Taken together, these developments show that Kuttler–Sigillito inequalities now constitute a substantial comparison framework in spectral geometry: they connect boundary operators of different order, survive under curvature and topology, admit mixed and form-valued analogues, and remain closely tied to Rellich-, Hadamard-, and Reilly-type identities.