Area Eigenvalue in Spectral Geometry

• Area eigenvalue is a scale-invariant measure that combines eigenvalues with area, ensuring consistent spectral comparisons under rescaling.
• It is defined differently for Laplace, Dirac, and Robin operators, reflecting unique normalizations and geometric constraints in each context.
• The concept extends to holography and extrinsic geometry, linking geometric optimization with spectral analysis and physical state quantization.

In spectral geometry, “area eigenvalue” most often denotes a scale-invariant quantity obtained by combining an eigenvalue with area, so that the resulting expression is unchanged by homothetic rescaling; in holographic settings, the phrase also appears in the literal sense of an eigenvalue of an area operator acting on fixed-area states. On closed surfaces with the Laplace–Beltrami operator, the natural normalization is $\lambda_k(M,g)\operatorname{Area}(M,g)$; for Dirac operators on spin surfaces, the natural two-dimensional normalization is $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$; for Robin problems, the boundary parameter must be rescaled simultaneously, and the scale-invariant quantity is $\lambda_j(\Omega;\alpha/L)\,|\Omega|$ (Kang, 6 Jun 2025, Karpukhin et al., 2023, Girouard et al., 2019).

1. Scale invariance and the basic normalizations

The central reason area enters these problems is scaling. For Laplace eigenvalues on a surface, if $g\mapsto c\,g$, then $\lambda_k(cg)=c^{-1}\lambda_k(g)$ while $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$, so $\lambda_k(g)\operatorname{Area}(M,g)$ is unchanged (Kang, 6 Jun 2025). For Dirac operators on a compact oriented spin surface, if $g\mapsto c^2g$, then $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$ and $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$, so the invariant quantity is $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$0 (Karpukhin et al., 2023). For the first Dirichlet Laplacian eigenvalue on planar domains, $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$1 and $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$2, making $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$3 the natural scale-invariant comparison quantity (Nitsch, 2014).

Setting	Scale-invariant quantity	Source
Laplace–Beltrami on surfaces	$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$4	(Kang, 6 Jun 2025)
Dirac on spin surfaces	$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$5	(Karpukhin et al., 2023)
Dirichlet Laplacian on planar domains	$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$6	(Nitsch, 2014)
Robin Laplacian on planar domains	$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$7	(Girouard et al., 2019)

This already shows that “area eigenvalue” is not a single universal formula. The normalization depends on the operator and, in Robin problems, on the scaling of the boundary parameter. In magnetic Neumann problems, fixed area remains the natural constraint, but the field strength $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$8 introduces an additional scale, and the effective parameter can be regarded as $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$9 (Fournais et al., 2017). A closely related but different development is the torsion-based comparison for the lowest magnetic Neumann eigenvalue, where fixed-area optimization is proved for ellipses in a moderate-field regime (Kachmar et al., 2023).

2. Closed surfaces: normalized Laplace and Dirac spectra

For Dirac operators on compact oriented spin surfaces, the area-normalized quantity is

$\lambda_j(\Omega;\alpha/L)\,|\Omega|$0

A central problem is, for a fixed conformal class $\lambda_j(\Omega;\alpha/L)\,|\Omega|$1,

$\lambda_j(\Omega;\alpha/L)\,|\Omega|$2

The sphere furnishes the sharp model case: $\lambda_j(\Omega;\alpha/L)\,|\Omega|$3 equivalently $\lambda_j(\Omega;\alpha/L)\,|\Omega|$4, with equality if and only if $\lambda_j(\Omega;\alpha/L)\,|\Omega|$5 is homothetic to the standard round metric (Karpukhin et al., 2023). On the torus, the same paper proves sharp conformal-class minimization results for many conformal classes: if $\lambda_j(\Omega;\alpha/L)\,|\Omega|$6 is a spin structure on $\lambda_j(\Omega;\alpha/L)\,|\Omega|$7, $\lambda_j(\Omega;\alpha/L)\,|\Omega|$8, and $\lambda_j(\Omega;\alpha/L)\,|\Omega|$9 is the unit-area flat metric in the class, then

$g\mapsto c\,g$0

with the flat metric as the unique smooth minimizer (Karpukhin et al., 2023).

For the Laplace–Beltrami operator on closed surfaces, the basic functional is

$g\mapsto c\,g$1

The topological supremum

$g\mapsto c\,g$2

is central in the existence theory of extremal metrics. One key monotonicity theorem shows that if $g\mapsto c\,g$3 is obtained from $g\mapsto c\,g$4 by attaching a cylinder or a cross cap, then there exists a smooth metric $g\mapsto c\,g$5 on $g\mapsto c\,g$6 such that

$g\mapsto c\,g$7

and, as a consequence, there exists a maximizing metric for the normalized first eigenvalue on any closed surface of fixed topological type (Matthiesen et al., 2019).

Several sharp and explicit bounds are known in low topology. For compact orientable surfaces of genus $g\mapsto c\,g$8,

$g\mapsto c\,g$9

improving the Yang–Yau bound $\lambda_k(cg)=c^{-1}\lambda_k(g)$0 (Ros, 2020). For the real projective plane, the second non-zero Laplace eigenvalue satisfies

$\lambda_k(cg)=c^{-1}\lambda_k(g)$1

and the value is attained only in the limit by a singular metric realized as a union of the projective plane and the sphere touching at a point, with area ratio $\lambda_k(cg)=c^{-1}\lambda_k(g)$2 (Nadirashvili et al., 2016). On the torus, the second non-zero eigenvalue in a fixed conformal class admits an explicit upper bound depending on the moduli parameters $\lambda_k(cg)=c^{-1}\lambda_k(g)$3, and there is a uniform bound

$\lambda_k(cg)=c^{-1}\lambda_k(g)$4

for all unit-area torus metrics (Kang, 6 Jun 2025).

3. Planar domains, area-normalized Dirichlet problems, and triangles

For planar Dirichlet problems, fixed area is the standard normalization because $\lambda_k(cg)=c^{-1}\lambda_k(g)$5 alone is not meaningful under dilation. In the regular polygon problem, the scale-invariant quantity is $\lambda_k(cg)=c^{-1}\lambda_k(g)$6, and the paper on regular polygons makes explicit that the conjectural monotonicity

$\lambda_k(cg)=c^{-1}\lambda_k(g)$7

for equal-area regular $\lambda_k(cg)=c^{-1}\lambda_k(g)$8- and $\lambda_k(cg)=c^{-1}\lambda_k(g)$9-gons is not proved there; what is proved is strict monotonicity at fixed circumradius and a near-monotonicity estimate for the area-normalized quantity (Nitsch, 2014). The same work recovers the Faber–Krahn lower bound for regular polygons,

$$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$0

with the disk value approached as $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$1 (Nitsch, 2014).

A different area-based method appears for geodesic balls. If $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$2 is a geodesic ball with $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$3, then the first Dirichlet eigenvalue $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$4 admits a sharp upper bound computable only from the area function of geodesic spheres

$$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$5

The construction replaces the metric by a rotationally symmetric metric preserving the area of each geodesic sphere, and equality holds if and only if the inward mean curvature of every geodesic sphere is radial (Gimeno et al., 2021). This is an area-based symmetrization of the metric tensor rather than of the domain.

Triangles supply the sharpest current fixed-area Dirichlet inequalities in the data. The first Dirichlet eigenvalue satisfies the classical triangle Faber–Krahn bound

$$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$6

with equality for the equilateral triangle, and the recent sharp result strengthens this to

$$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$7

again with equality if and only if $$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$8 is equilateral (Endo et al., 5 May 2026). Equivalently,

$$\operatorname{Area}(M,cg)=c\,\operatorname{Area}(M,g)$$9

The same paper also proves the sharp Cheeger-type inequality

$\lambda_k(g)\operatorname{Area}(M,g)$0

with equality only for the equilateral triangle (Endo et al., 5 May 2026).

Area-constrained eigenvalue minimization also appears for the fully nonlinear Pucci supremum operator. In a specific explicitly solvable family of planar domains, the principal eigenvalue is minimal, for fixed area, at the most symmetric member of the family (Birindelli et al., 2013).

4. Robin and magnetic fixed-area eigenvalue problems

For the lowest Robin eigenvalue on triangles in the attractive regime $\lambda_k(g)\operatorname{Area}(M,g)$1, the fixed-area question becomes a maximization problem. The conjectured reverse isoperimetric inequality is

$\lambda_k(g)\operatorname{Area}(M,g)$2

for the equilateral triangle $\lambda_k(g)\operatorname{Area}(M,g)$3 of the same area, but the paper proves this only partially: the equilateral triangle is a strict local maximizer for all $\lambda_k(g)\operatorname{Area}(M,g)$4, with $\lambda_k(g)\operatorname{Area}(M,g)$5, and there are additional global-in-shape results in weak- and strong-coupling regimes (Krejcirik et al., 2022). The same work emphasizes that the natural dimensionless parameter is $\lambda_k(g)\operatorname{Area}(M,g)$6, reflecting the fixed-area scaling.

A different Robin normalization is needed for higher eigenvalues. For the third Robin eigenvalue on simply-connected planar domains, the scale-invariant quantity is

$\lambda_k(g)\operatorname{Area}(M,g)$7

For $\lambda_k(g)\operatorname{Area}(M,g)$8, this quantity is strictly bounded above by the corresponding value for the disjoint union of two equal disks of the same total area, and equality is achieved only asymptotically by a degenerating simply-connected dumbbell sequence (Girouard et al., 2019). This is a Robin analogue of the disconnected optimizers familiar from higher Neumann eigenvalues.

For the magnetic Neumann Laplacian with constant magnetic field, the fixed-area extremal question asks whether the disk maximizes the lowest eigenvalue. The full statement remains open for simply connected domains, but it is proved in two asymptotic regimes: for sufficiently small $\lambda_k(g)\operatorname{Area}(M,g)$9, via the torsional rigidity coefficient in the weak-field expansion, and for sufficiently large $g\mapsto c^2g$0, via the semiclassical expansion involving maximal boundary curvature (Fournais et al., 2017). The same paper gives the universal area-based upper bound

$g\mapsto c^2g$1

The torsion-function approach sharpens this in another direction. For a bounded, convex, $g\mapsto c^2g$2-smooth planar domain with $g\mapsto c^2g$3, the lowest magnetic Neumann eigenvalue satisfies

$g\mapsto c^2g$4

where $g\mapsto c^2g$5 is the torsion function and $g\mapsto c^2g$6, $g\mapsto c^2g$7 are geometric quantities built from its level sets (Kachmar et al., 2023). For ellipses, the geometric factor is exactly $g\mapsto c^2g$8, and if

$g\mapsto c^2g$9

then among ellipses of fixed area the disk uniquely maximizes the lowest magnetic Neumann eigenvalue (Kachmar et al., 2023).

5. Extremals, degeneration, and geometric correspondences

Area-eigenvalue problems are closely tied to geometric structures behind extremal metrics. For Dirac operators on spin surfaces, conformal criticality of $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$0 is characterized by eigenspinors $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$1 satisfying

$\lambda_k(c^2g)=c^{-1}\lambda_k(g)$2

and such data define a harmonic map

$\lambda_k(c^2g)=c^{-1}\lambda_k(g)$3

Globally critical metrics satisfy a stronger Euler–Lagrange equation involving the energy-momentum tensor, and the associated projective map is a quaternionic branched minimal immersion into $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$4 (Karpukhin et al., 2023).

For normalized Laplace eigenvalues on closed surfaces, maximizing metrics are induced by possibly branched minimal immersions into spheres by first eigenfunctions, and this is one reason the existence of maximizers is linked to strict topological monotonicity of $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$5 (Matthiesen et al., 2019). On $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$6, the sharp value $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$7 is not attained by a smooth metric; instead it is attained in the limit by a singular configuration $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$8 with area ratio $\lambda_k(c^2g)=c^{-1}\lambda_k(g)$9, an explicit example of bubbling in higher-eigenvalue optimization (Nadirashvili et al., 2016).

Degeneration can also decrease the normalized quantity. For surfaces obtained by attaching a collapsing flat handle or flat cross cap, there are sharp asymptotics for $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$0: with the right symmetry assumptions on the first eigenspace, the construction can strictly increase $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$1, but without those symmetry conditions, in the resonant regime the first eigenvalue drops by order $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$2 while the area gain is only order $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$3, so the normalized first eigenvalue strictly decreases (Matthiesen et al., 2019). This sharp contrast clarifies that area gain alone does not control the sign of the variation.

A recurring regularity theme is that extremals need not be everywhere smooth. In the Dirac conformal-class problem, if

$$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$4

then there exists a minimizer $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$5, smooth outside possibly at most $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$6 conical singularities, where $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$7 is the genus (Karpukhin et al., 2023). Similar singular or limiting extremals appear in Laplace and Robin problems.

6. Area as an operator eigenvalue and extrinsic variants

In holographic conformal field theory, “area eigenvalue” can mean precisely the eigenvalue of a state-dependent area operator. For a pure geometric state $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$8 and subsystem $$\operatorname{Area}(M,c^2g)=c^2\operatorname{Area}(M,g)$$9, the reduced density matrix $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$00 is decomposed into fixed-$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$01 sectors, and the corresponding fixed-area states are eigenstates of

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$02

They satisfy

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$03

and the Ryu–Takayanagi formula takes the operator form

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$04

The fluctuation of $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$05 in the geometric state $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$06 is suppressed in the semiclassical limit $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$07 (Guo, 2021).

An extrinsic differential-geometric use of the language appears for the area Jacobi operator of a compact complex curve $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$08 in a Kähler surface. If $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$09 denotes its first eigenvalue, then

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$10

where $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$11 is the infimum of the ambient Ricci curvature (Xie, 26 Feb 2026). In the Kähler–Einstein case $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$12 with $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$13, this becomes

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$14

and equality is achieved for all curves of genus $\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$15; the first eigenspace then has dimension

$\lambda_k(M,g,S)\operatorname{Area}(M,g)^{1/2}$16

(Xie, 26 Feb 2026).

Taken together, these developments show that “area eigenvalue” has at least two precise meanings in current research. In spectral geometry, it denotes the scale-invariant spectral quantities obtained by coupling eigenvalues to area under fixed-area optimization. In holography and extrinsic geometry, it can denote an actual eigenvalue of an operator built from area. In both senses, area is not an auxiliary normalization; it is the quantity that makes the spectral problem geometrically meaningful.