Singular Minimizing Free Boundaries

• Singular minimizing free boundaries are defined by variational problems where optimal free boundaries exhibit both smooth regions and singularities emerging from blow-up phenomena.
• The analysis employs monotonicity formulas, epiperimetric inequalities, and frequency gap methods to systematically classify and quantify the structure of these singularities.
• Results include precise dimension thresholds and stratification of the free boundary, offering insights into regularity theory in geometric, topological, and multisystem variational models.

A singular minimizing free boundary arises in variational problems where the minimizer (which may be a function, a surface, or a domain) displays regions—“free boundaries”—determined by optimality but not fixed a priori, and these free boundaries can develop singularities where classical regularity fails. In such problems, the regular set, where the free boundary is smooth (often analytic), coexists with a singular set of lower dimension where the geometry may become non-differentiable or display conical or multi-junction structure. These phenomena are ubiquitous in one-phase and obstacle-type problems, in minimizing surfaces under various geometric or topological constraints, in thin one-phase variational models, in degenerate perimeters, and for free boundaries arising from systems or higher-order energies. Recent research has established generic regularity results, precise dimension bounds, and explicit blow-up classification for the structure of these singular sets.

1. Structural Classification and Generic Regularity

In classical one-phase Alt–Caffarelli-type functionals, such as

$$\mathcal{J}(u; \Omega) = \int_\Omega |\nabla u|^2 + \chi_{\{u>0\}}\,dx,$$

the free boundary $\Gamma(u) := \partial\{u>0\} \cap \Omega$ can be stratified into regular and singular points. At regular points, after suitable blow-up rescaling, the solution converges to the half-plane solution $p_2(x) = \frac12(\max\{0, e\cdot x\})^2$; singular points are characterized by blow-up limits that are nontrivial full-space quadratic polynomials $p_2$, i.e., $p_2 \ge 0$, $\Delta p_2 = 1$ (Figalli et al., 2019).

Generic regularity theory, as developed in (Figalli et al., 2019, Fernández-Real et al., 2023), asserts that for parameterized families of minimizers (e.g., with monotone boundary data), the singular set generically has codimension at least $3$ in the free boundary, with Hausdorff measure zero in dimension $n-4$. This result solves Schaeffer's conjecture for $n\le 4$: for almost every parameter, the free boundary is entirely regular and $C^\infty$. Dimension-reduction and cleaning lemmas show that, except for a parameter set of measure zero, the singular set is considerably smaller than in worst-case scenarios.

2. Blow-up Analysis and Singular Set Stratification

The study of singular minimizing free boundaries relies heavily on stratified blow-up analysis. Around a singular point $x_0$, blow-ups take the form

$u_r(x) = r^{-2}u(x_0 + r x),\quad r \to 0^+\,,$

and converge to quadratic polynomials or more complicated homogeneous profiles. Further iterated blow-ups and frequency function analysis classify additional structure: it is possible to achieve higher-order expansions (cubic, quartic, etc.) outside sets of increasing codimension, with monotonicity formulas (Weiss, Almgren) and truncated frequency functions providing uniqueness of blow-ups at various degrees (Figalli et al., 2019). The singular set is stratified according to the rank of the kernel of the quadratic blow-up: for each $0\le m\le n-1$, the $m$-stratum is $C^1$ up to lower-dimensional subsets.

For functionals with higher-order operators (e.g., biharmonic), the natural scaling can change: in (Dipierro et al., 2018), for the minimization of $\int_\Omega |\Delta u|^2 + \chi_{\{u>0\}}$, singular blow-ups are $2$-homogeneous, and the free boundary nearby is either "rank-2 flat" (well-approximated by the zero set of a quadratic polynomial) or the solution exhibits quadratic growth. Rank-2 flat points stratify into finitely many symmetry types, and non-flat singular points form a set of Hausdorff dimension at most $n-2$.

3. Dimensional Thresholds and Quantitative Stratification

A key principle is the existence of a critical dimension below which minimizing free boundaries are generically smooth, and above which singularities can occur. In the obstacle problem, singularities may be as large as $(n-1)$-dimensional, yet generically are confined to a set of Hausdorff dimension at most $n-4$ (Figalli et al., 2019). In the one-phase problem for almost minimizers, the singular set has dimension at most $n-5$; for thin one-phase problems, the singular set has dimension at most $n-3$ (Silva et al., 2012, Engelstein et al., 2019). For capillary Bernoulli-type problems, the onset of singular boundary points occurs in codimension $7$, which is sharp due to the construction of area-minimizing capillary cones in high dimensions $n+1\ge8$ (Firester et al., 26 Jan 2026).

Quantitative stratification and monotonicity arguments, such as those based on Jones' $\beta$-numbers and density-drop techniques, further refine these results, producing Minkowski dimension and packing bounds on the singular set, as established in the thin one-phase variational framework (Engelstein et al., 2019).

4. Models and Explicit Singular Minimizing Cones

Several explicit models of singular minimizing free boundaries arise as homogeneous solutions (cones) to the underlying variational problems. For one-phase functionals, known nontrivial singular cones exist only in high dimensions $n\ge7$ (Silva et al., 2019). For capillary problems, area-minimizing capillary cones with $O(n-k)\times O(k)$ symmetry have been fully classified in (Firester et al., 26 Jan 2026): for every $n \ge 7$ and for appropriate contact angles, there exist cones with isolated apex singularities that are strict minimizers, interpolating between singular one-phase cones and halved Lawson cones. In the case of degenerate perimeter energies (weighted by distance to boundary), the singular set dimension depends on an effective dimension $n+a$, with the Simons cone model saturating the bound for $d_H$(Sing) (Gasparetto et al., 4 Mar 2025).

A comprehensive table of models and dimension thresholds:

Model Type	Dimension Threshold for Singularities	Prototype Singular Cone
Obstacle, one-phase	$n \ge 7$ (generic) / $n-5$	Non-flat 1-homogeneous Bernoulli cones
Thin one-phase	$n \ge 3$	Nontrivial $\alpha$-homogeneous cones
Capillary problem	$n+1 \ge 8$	Bi-orthogonal capillary cones
Degenerate perimeter	$n+a \ge 8 + \dots$	Rotationally-invariant minimal cones

5. Regularity Mechanisms: Monotonicity, Epiperimetric, and Frequency Gaps

A unifying mechanism for the analysis of singular free boundaries is provided by monotonicity formulas (Weiss-type, Almgren-type), which control the scaling behavior of blow-up sequences and characterize homogeneity. Epiperimetric inequalities are central to quantitative decay rates and uniqueness of blow-ups in both scalar and vectorial free boundary problems; these tools facilitate $C^{1,\beta}$ or even analytic regularity of the regular part of the free boundary (Andersson et al., 2013, Silva et al., 2021). For thin one-phase and spectral partition problems, frequency function gaps (e.g., jump from $N=1$ to $N \ge 3/2$) produce well-separated strata and force the codimension of the singular set (Novack et al., 2024).

In almost minimizer settings and degenerate perimeters, $\varepsilon$-regularity theorems ensure that any "flat" free boundary region is actually regular, provided the normalized energy is sufficiently small at that scale (Gasparetto et al., 4 Mar 2025, Figalli et al., 2023). Non-degeneracy lemmas and Harnack inequalities further preclude the clustering of singular points near the regular set and ensure optimal (often Lipschitz) growth away from the free boundary.

6. Special Geometries: Topological, Cooperative, and Systemic Free Boundaries

Topological constraints fundamentally impact the singularity structure. In variational free boundary problems with topological constraint (such as spanning a wire-frame), the regular part of the free boundary is analytic except on a codimension $2$ singular set; the models for the singularity are capacity-minimizing cones, such as $Y$-cones (triple junctions) in the plane and $T$-cones in space, closely connected with spectral partitions and segregation problems (Novack et al., 2024). In cooperative vectorial free boundary systems, blow-up limits are classified via monotonicity and epiperimetric tools, regular points correspond to half-plane solutions, and the singular set corresponds to those with higher energy (above the half-plane threshold) (Andersson et al., 2013, Silva et al., 2021).

For constraint maps into manifolds (Bernoulli-type functionals with target domains), singularities are even more restricted: under uniform convexity of the target, the free boundary is smooth, and all singularities are forced into the "coincidence" set (interior singularities of harmonic maps), aligning the analysis with classical results for harmonic maps (Figalli et al., 2023).

7. Borderline Regularity and Sharp Thresholds

Under minimal regularity assumptions on the potential (e.g., measurable $\sigma$), sign-changing minimizers can only achieve Log-Lipschitz regularity, which is sharp (Araújo et al., 20 Aug 2025). In the one-phase case, a structural gain arises: if $\sigma$ is continuous, minimizers achieve $C^1$ regularity at the free boundary, and a modulus of differentiability is determined explicitly by the modulus of $\sigma$. This represents a sharp threshold: no further differentiability is generically possible without stronger constraints on $\sigma$ (e.g., for $\sigma(t) = t_+^\gamma$, the regularity is $C^{1, \gamma/(2-\gamma)}$).

---

The current theory thus provides a comprehensive multidimensional framework for understanding the nature, dimension, and mechanism of singularities in minimizing free boundaries across a wide class of PDE-driven, geometric, and topologically constrained settings. The sharpness of dimension estimates, the interplay between variational structure and PDE theory, and the deep connections to minimal surface and spectral partition theories delineate the scope and limits of regularity in these nonlinear free boundary phenomena.