Singular Minimal Surfaces

• Singular minimal surfaces are variational hypersurfaces defined by zero or prescribed mean curvature on their regular regions while inherently admitting nonempty singular sets.
• They serve as critical models in geometric measure theory, with applications spanning classical minimality, weighted curvature settings, and degenerate or Lorentzian ambient geometries.
• Their study impacts deformation theory and compactness, offering insights into tangent cones, stability conditions, and the rigidity or regularity of singular structures.

A singular minimal surface is not a single universally fixed object but a family of closely related notions centered on minimality in the presence of singular behavior. In geometric measure theory, the term usually refers to a stationary, stable, or area-minimizing hypersurface whose regular part has zero mean curvature and whose singular set is nonempty; in dimensions $n\ge 7$, such singularities are unavoidable in general and are modeled by minimal cones (Wang, 2020). In another major usage, especially in the work of López and its extensions, a singular minimal surface is a critical point of a weighted area functional and satisfies a prescribed mean-curvature equation of the form $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$, or the same equation with a different normalization of $H$ (López, 18 Jul 2025). Further variants occur in degenerate and Lorentzian ambient geometries, where “singular” may refer to the ambient metric, a weighted Euler–Lagrange law, or wave-front singularities of the immersion itself (Sato, 2018, Akamine et al., 17 Feb 2026).

1. Terminological scope and variational definitions

The main usages can be organized as follows.

Framework	Defining condition	Typical singular feature
Classical minimal hypersurface	$H\equiv 0$ on the regular part; stationarity for area	Nonempty singular set $\operatorname{Sing}(\Sigma)$
Weighted singular minimal surface	$$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$ or $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$ depending on convention	Degeneracy at $\langle \Phi,a\rangle=0$
Generalized ambient variants	Modified zero-mean-curvature or weighted equations in degenerate/Lorentzian settings	Ambient degeneracy or frontal singularities

For classical minimal hypersurfaces, let $(M^{n+1},g)$ be Riemannian and $\Sigma^n\subset M$ a two-sided immersed hypersurface with unit normal $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$0. The area functional is

$$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$1

and under a normal variation $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$2, the first variation is

$$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$3

Thus $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$4 is minimal iff $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$5. The second variation at a minimal hypersurface is

$$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$6

with Jacobi operator

$$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$7

Stability means the associated quadratic form is nonnegative on compactly supported variations (Wang, 2020).

In the weighted Euclidean theory, one fixes a unit vector $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$8 and considers immersed oriented surfaces $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$9 lying in a half-space $H$0. The weighted area functional

$H$1

has Euler–Lagrange equation

$H$2

in the convention $H$3, while several other papers use the averaged mean curvature and write $H$4 (López, 18 Jul 2025, López, 2018). The singularity is the plane $H$5, where the prescribed-curvature term becomes undefined.

2. Classical singular minimal hypersurfaces: singular sets, cones, and stability

For stable or area-minimizing hypersurfaces, regularity is dimension-dependent. In dimensions $H$6, the relevant objects are smooth; for $H$7, singularities can occur, and the Hausdorff dimension of $H$8 is at most $H$9. The standard model is the Simons cone in $H\equiv 0$0, which is area-minimizing and has an isolated singularity at the origin (Wang, 2020).

The local analysis near a singular point $H\equiv 0$1 is governed by tangent cones. Any tangent varifold is a stationary cone $H\equiv 0$2. When $H\equiv 0$3 is a strongly isolated singularity, the tangent cone has multiplicity $H\equiv 0$4 and singular set $H\equiv 0$5. In conic coordinates $H\equiv 0$6, the Jacobi operator on a regular minimal hypercone $H\equiv 0$7 splits as

$H\equiv 0$8

where $H\equiv 0$9. If $\operatorname{Sing}(\Sigma)$0 are the eigenvalues of $\operatorname{Sing}(\Sigma)$1, the radial exponents $\operatorname{Sing}(\Sigma)$2 solve

$\operatorname{Sing}(\Sigma)$3

and homogeneous Jacobi modes are $\operatorname{Sing}(\Sigma)$4. Strict stability of the cone means a spectral gap at the first mode; strict minimizing means the one-sided area minimizer has the fast asymptotic governed by $\operatorname{Sing}(\Sigma)$5 and yields a Hardt–Simon foliation on the complementary component (Wang, 2020).

The singular theory also has sharp spectral consequences on links in the sphere. For a stationary integral $\operatorname{Sing}(\Sigma)$6-varifold $\operatorname{Sing}(\Sigma)$7 with orientable regular part and the $\operatorname{Sing}(\Sigma)$8-structural hypothesis, if $\operatorname{Sing}(\Sigma)$9 is not totally geodesic then the first stability eigenvalue satisfies $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$0, with equality iff $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$1 is an integer multiple of a Clifford hypersurface $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$2 (Zhu, 2016). This extends Simons’ smooth estimate to the singular setting and sharpens the instability picture of nontrivial singular links.

3. Deformation theory, compactness, and generic regularity

A central question is whether singular minimal hypersurfaces persist under perturbation and how singularities behave in compactness limits. For a two-sided, locally stable minimal hypersurface $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$3 with optimal regularity, only strongly isolated singularities, trivial Jacobi kernel, and tangent cones that are both strictly minimizing and strictly stable, small $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$4 perturbations of the ambient metric preserve the existence of a nearby locally stable minimal hypersurface $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$5 with the same optimal regularity and small varifold distance from $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$6 (Wang, 2020). This is the singular analogue of the implicit-function picture familiar in the smooth nondegenerate case.

The same framework yields a generic desingularization mechanism. For scalar conformal perturbations $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$7, a residual set of perturbation directions $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$8 produces nearby closed embedded smooth minimal hypersurfaces $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$9 in the varifold sense. The mechanism is an associated Jacobi field on $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$0 whose asymptotic rate at each singular point is forced to be maximal; positivity of that field near the singularity gives barrier control and suppresses singularity formation in the deformed hypersurfaces (Wang, 2020).

The paper also proves a local homological minimizing theorem: under local stability and strictly minimizing tangent cones, a neighborhood of $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$1 admits a nested family of piecewise smooth mean convex neighborhoods, and $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$2 is homologically area-minimizing there and is the unique stationary varifold in that neighborhood up to multiplicity (Wang, 2020). In dimension $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$3, multiplicity-$$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$4 stable convergence $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$5 to a stable singular limit induces a nontrivial Jacobi field on $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$6; if that Jacobi field is positive near a singular point, then infinitely many members of the sequence are smooth near that point (Wang, 2020).

A complementary min-max result gives a quantitative trade-off between instability and singularity. For a one-parameter Almgren–Pitts min-max hypersurface $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$7,

$$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$8

where $$2H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$9 consists of isolated singular points with unique tangent cone that is non-area-minimizing on both sides (Chodosh et al., 2020). In dimension $\langle \Phi,a\rangle=0$0, this implies that for a dense set of metrics there exists a min-max hypersurface with at most one singular point, and for an open dense set of positive-Ricci metrics there exists a smooth one. This estimate makes precise the idea that, in a one-parameter min-max problem, either one spends the available complexity on Morse index or on a single worst-type singular point, but not both.

4. Polyhedral models, Y-singularities, and analytic boundary theory

Not all singular minimal objects are isolated-cone singularities. Near polyhedral cones $\langle \Phi,a\rangle=0$1, which model Y-lines, tetrahedral junctions, and higher-dimensional products, an improvement-of-flatness scheme yields $\langle \Phi,a\rangle=0$2-regularity for minimal varifolds sufficiently close to the cone. Under a no-hole condition on the singular set, the top strata of the singular set inherit $\langle \Phi,a\rangle=0$3 structure, and the sheets are $\langle \Phi,a\rangle=0$4 graphs over the cone wedges meeting with the correct $\langle \Phi,a\rangle=0$5 balance conditions (Colombo et al., 2017). This extends Allard-type regularity from planes to network-type singular models.

A related but more specialized theory treats Y-singular minimal surfaces in $\langle \Phi,a\rangle=0$6, where three smooth minimal sheets meet along a common curve with equal $\langle \Phi,a\rangle=0$7 angles. In the rotationally symmetric case where the singular set is a single circle, the second variation decomposes into fixed-interface Jacobi problems on the faces together with a Dirichlet-to-Neumann contribution from moving the singular circle. For the Y-catenoid, this computation gives Morse index $\langle \Phi,a\rangle=0$8 and nullity $\langle \Phi,a\rangle=0$9 (Matinpour, 2024). The result isolates the unstable mode as the axisymmetric junction-motion mode and shows that the first azimuthal modes are neutral.

Singular minimal hypersurfaces also support a global potential theory. Using skin transforms $(M^{n+1},g)$0, one can conformally unfold $(M^{n+1},g)$1 into a complete Gromov hyperbolic space of bounded geometry. In this unfolded metric, the singular set becomes the Gromov boundary and, for broad classes of elliptic operators, also the Martin boundary. Positive solutions then admit Martin representation

$(M^{n+1},g)$2

and one has boundary Harnack principles, Fatou-type non-tangential convergence in skin pencils, and Dirichlet solvability on the singular set (Lohkamp, 2015). This viewpoint treats $(M^{n+1},g)$3 not merely as a defect set but as a canonical boundary at infinity for the analytic geometry of the regular part.

5. Weighted singular minimal surfaces in Euclidean space

In the weighted Euclidean theory, singular minimal surfaces are governed by the prescribed-curvature equation

$(M^{n+1},g)$4

or its normalized variants, and are naturally posed in a half-space bounded by the singular plane $(M^{n+1},g)$5. This equation arises as the Euler–Lagrange equation for a power-law weighted area functional and is interpreted in some papers as a lowest-gravity-center condition (Erdur et al., 2019). The rigidity consequences are strong. If the Gauss curvature is constant, then the only such surfaces are planes parallel to $(M^{n+1},g)$6, spheres centered at $(M^{n+1},g)$7, and cylindrical singular minimal surfaces; constant negative Gauss curvature does not occur (López, 18 Jul 2025). The same paper classifies the cases of one constant principal curvature and of constant mean curvature, the latter reducing to planes and spheres.

Symmetry assumptions are especially restrictive. Helicoidal singular minimal surfaces are forced to be circular right cylinders: the axis must be orthogonal to the prescribed vector, $(M^{n+1},g)$8, and no noncylindrical helicoidal examples exist (López, 18 Jul 2025). For ruled surfaces, every singular minimal ruled surface in $(M^{n+1},g)$9 is cylindrical, hence either a plane parallel to the distinguished direction or an $\Sigma^n\subset M$0-catenary cylinder (Aydin et al., 2023). For singular minimal translation hypersurfaces with horizontal direction vector, the only possibilities are planes and $\Sigma^n\subset M$1-catenary cylinders, and an analogous conclusion holds for affine-translation graphs of the form $\Sigma^n\subset M$2 (Erdur et al., 2019).

The boundary-value theory is likewise rigid. For compact singular minimal surfaces with boundary, one has area and height estimates in terms of the boundary geometry, rotational symmetry for embedded solutions spanning a circle in a horizontal plane, and nonexistence results when two boundary curves are sufficiently far apart (López, 2018). A modified theory based on special semi-symmetric connections changes the classification: with a special semi-symmetric metric connection, the singular-minimal-and-minimal surfaces include generalized cylinders in addition to planes, whereas with a special semi-symmetric non-metric connection the only such surfaces are planes (Aydin et al., 2020). For translation surfaces in the same semi-symmetric setting, the V- and D-singular minimal solutions are again generalized cylinders or planes, now determined by ODEs of Abel or Emden–Fowler type rather than by the classical Levi-Civita equations (Erdur et al., 2020).

6. Degenerate and Lorentzian ambient geometries, and related singular variational models

Several papers transplant singular minimality to non-Euclidean ambient structures. In simply isotropic space, where the ambient metric is degenerate, the correct variational objects are relative arc length and relative area rather than Euclidean area. Singular minimal surfaces then satisfy graph equations such as

$\Sigma^n\subset M$3

depending on whether the reference plane is isotropic or non-isotropic. The corresponding rotational and parabolic-revolution families include logarithmic profile curves and explicit ODE reductions (Silva et al., 2022).

A different degenerate theory appears in the singular semi-Euclidean space $\Sigma^n\subset M$4, where the metric is $\Sigma^n\subset M$5 and the $\Sigma^n\subset M$6-direction is null. Here $\Sigma^n\subset M$7-minimal surfaces are defined by vanishing mean curvature with respect to the canonical flat connection, equivalently by harmonicity of the coordinate functions. For graphs $\Sigma^n\subset M$8, $\Sigma^n\subset M$9-minimality is simply $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$00. These surfaces admit a Weierstrass-type representation and correspond bijectively, up to planes and ambient symmetries, to spacelike flat zero-mean-curvature surfaces in Minkowski four-space (Sato, 2018).

In the Lorentzian Heisenberg group, timelike minimal surfaces are encoded by Lorentzian harmonic maps into the de Sitter two-sphere. Their singular set decomposes into loci $$H=\alpha \frac{\langle N,a\rangle}{\langle \Phi,a\rangle}$$01, and SUY-type criteria characterize cuspidal edges, swallowtails, and cuspidal cross caps directly in terms of the Gauss-map data. Explicit examples show all three singularity types occur (Akamine et al., 17 Feb 2026). This is a substantially different use of “singular minimal surface”: the ambient mean curvature equation remains minimal, but the immersion is treated as a frontal or front with generic wave-front singularities.

Related variational models show that singularity phenomena persist outside the strict minimal-surface equation. In a nonconvex domain arbitrarily close to the unit ball, the Simons cone can be a singular volume-constrained locally area-minimizing surface, showing that convexity is decisive for the regularity improvement observed in the ball (Sternberg et al., 2017). In higher codimension, limits of minimizing sequences of Lipschitz graphs for the minimal surface system can develop interior vertical parts and non-minimal interior portions projecting to an analytic hypersurface in the base, revealing a marked discrepancy between graph-based minimization and parametric area minimization (Mooney et al., 2024).

Taken together, these results show that singular minimal surfaces form a broad but coherent subject. In the classical hypersurface theory, the dominant questions are structure of the singular set, tangent-cone asymptotics, deformation under perturbation, and compactness. In the weighted Euclidean theory, the emphasis shifts to prescribed-curvature equations, rigidity under symmetry, and boundary behavior. In degenerate and Lorentzian settings, singularity may be built into the ambient metric or into the front structure of the immersion. Across these formulations, the common theme is that minimality remains variational, but singularity becomes an essential geometric datum rather than an exceptional pathology.