Blow-Up Limits of Minimizers

• The paper demonstrates how rescaling variational minimizers near singularities reveals asymptotically homogeneous profiles critical for understanding local geometric and analytic behavior.
• It employs quantitative monotonicity formulas and epiperimetric inequalities to establish the uniqueness and classification of blow-up profiles across free boundary and nonlinear PDE problems.
• The findings extend to applications in obstacle problems, nonlinear Schrödinger equations, and geometric variational models, emphasizing implications for regularity and rigidity in singular structures.

A blow-up limit of minimizers arises from rescaling a sequence of variational minimizers near a putative singularity or critical point so that their energy profiles or geometric/analytic features are preserved in the limit. The study of blow-up limits is central to understanding singularities in free boundary problems, geometric variational problems, and nonlinear PDEs. Through this technique, singular points are replaced by tangent, typically simpler, global solutions that encode the local geometry or analytic behavior in a scaling-invariant fashion.

1. Definition and Construction of Blow-Up Limits

Given a variational minimizer (such as a solution to an elliptic PDE, an energy-minimizing function, or a minimal surface), the blow-up limit is produced by rescaling the function (or the set) around a point $x_0$—in the free boundary, singular set, or elsewhere—by a sequence of scales $r_n \downarrow 0$:

$u_{x_0, r_n}(x) = \frac{u(x_0 + r_n x)}{r_n^\mu}$

with a normalization exponent $\mu$ chosen according to scaling invariance of the problem (e.g., degree of homogeneity or mass constraint). A blow-up limit at $x_0$ is any local limit (in a suitable topology, typically $H^1_{\mathrm{loc}}$ or $C^\alpha_{\mathrm{loc}}$) of $(u_{x_0, r_n})$ as $n \to \infty$.

This construction applies in vector- and scalar-valued minimization problems with free boundaries (Philippis et al., 2021, Siclari et al., 31 Jan 2026), obstacle problems (Colombo et al., 2019), nonlinear Schrödinger equations (Phan, 2018, Phan, 2018, Dinh, 2019), geometric variational problems (minimal surfaces, Mumford–Shah, etc.) (Lemenant, 2014), and singular nonlinear equations (Tarantello, 2022, Mazumdar, 2017).

2. Classification and Uniqueness of Blow-Up Profiles

Blow-up analysis often reveals that minimizers are asymptotically approximated near a singular point by homogeneous, global, typically simpler objects: for scalar free boundary problems, these are often half-space solutions or homogeneous cones (Engelstein et al., 2018); for vectorial free boundary problems, they are linear maps $A x$ with matrix $A$ whose rank/topology classifies the singularity (Philippis et al., 2021, Siclari et al., 31 Jan 2026).

Uniqueness of blow-up limits is a central question: if all sequences $r_n\downarrow 0$ produce the same limit, the singularity is called (asymptotically) unique. This has been established almost everywhere for a variety of contexts:

• For the vectorial Bernoulli/Alt–Caffarelli free boundary problem, blow-ups are unique for $\mathcal{H}^{d-1}$-almost every two-phase singular point, and are always 1-homogeneous and linear (Philippis et al., 2021).
• In the modified planar Ericksen model (liquid crystals), the uniqueness of the tangent map (homogeneous blow-up) is proved using a Weiss monotonicity formula and analysis of Jacobi fields (Alper, 2017).
• The lower-dimensional obstacle problem admits uniqueness of the blow-up profile and a complete homogeneity classification almost everywhere (Colombo et al., 2019).

A central analytic tool for uniqueness is the monotonicity formula—typically of Weiss or Alt–Caffarelli–Friedman (ACF) type—often supplemented by an epiperimetric inequality that improves convergence to homogeneity (Engelstein et al., 2018, Philippis et al., 2021). An important insight is that rectifiability of the singular set (via the Naber–Valtorta criterion or Jones' $\beta$-numbers) implies almost everywhere uniqueness (Philippis et al., 2021).

3. Analytical Techniques: Monotonicity, Epiperimetric Inequalities, and Stratification

Quantitative monotonicity formulas are foundational:

• Weiss-type functionals control the energy defect as the scale shrinks and ensure (sub-)monotonicity, implying convergence to homogeneous profiles (Philippis et al., 2021, Alper, 2017, Colombo et al., 2019, Engelstein et al., 2018).
• The Alt–Caffarelli–Friedman monotonicity formula distinguishes between distinct blow-up profiles, often forcing equality (uniqueness) by functional invariance (Philippis et al., 2021, Siclari et al., 31 Jan 2026, Engelstein et al., 2018).
• Epiperimetric inequalities (sometimes logarithmic) yield rates of convergence of rescalings to homogeneous cones and can distinguish between integrable and non-integrable directions in the space of singularities (Engelstein et al., 2018, Colombo et al., 2019).

Blow-up limits lead to a stratification of the singular set according to, e.g., the rank of the limiting homogeneous solution. In the vectorial Bernoulli problem, the singular set decomposes as

$S_j = \{x_0 : \operatorname{rank} A_{x_0} = j\}$

with $S_j$ being $(d-j)$-rectifiable and of locally finite $\mathcal{H}^{d-j}$-measure (Philippis et al., 2021). The stratification captures both the dimension and the local geometry of singularities.

4. Profiles and Universality Across Problem Classes

The structure of blow-up limits is problem-dependent but reveals universal features:

• For free boundary functionals (Alt–Caffarelli, Bernoulli), the only possible blow-up limits at almost every two-phase singular point are one-homogeneous minimizers, i.e., linear functions $A x$, with rank and matrix structure determined by directional monotonicity (Philippis et al., 2021, Siclari et al., 31 Jan 2026).
• In nonlinear Schrödinger-type equations (mass-critical, supercritical, and Kirchhoff-type), blow-up minimizers, upon suitable rescaling and recentering, converge strongly to the unique optimizer of the corresponding Gagliardo–Nirenberg or Hardy–Sobolev inequality (Phan, 2018, Phan, 2018, Dinh, 2019, Yang et al., 26 Mar 2025).
• In geometric variational problems (e.g. Mumford–Shah, minimal surfaces in the Heisenberg group), the blow-up is a global minimizer in the ambient space, often a cone or a minimal surface model solution (Lemenant, 2014, Yu, 2024).
• Universality extends to the Lane–Emden profile (solution to a semilinear elliptic ODE), which arises in Chandrasekhar-type gravitational collapse, Hartree–Fock–Bogoliubov theory, and boson star models (Nguyen, 2017, Nguyen, 2019).

5. Implications for Regularity, Singular Set Structure, and Rigidity

Characterization of blow-up limits has immediate consequences for the structure and regularity theory of minimizers:

• Regularity: At points where the blow-up is flat (e.g., rank-1 linear map or half-space solution), the free boundary (or singular set) is $C^{1,\alpha}$ near such points (Engelstein et al., 2018, Colombo et al., 2019).
• Rectifiability: The stratification into rectifiable sets of prescribed codimension and the density structure of the singular set follow from monotonicity and $\beta$-number estimates connected via Naber–Valtorta theory (Philippis et al., 2021).
• Rigidity: Certain monotonicity results (e.g., in Mumford–Shah with the singular set in a half-plane) yield rigidity theorems: e.g., the only possible minimizer is the half-plane itself (Lemenant, 2014).

The blow-up method is also leveraged in problems with collapsing singularities (such as in Liouville equations with merging Dirac masses), where quantization of blow-up mass and explicit expansion of the blow-up sequence lead to geometric conclusions (as in mean curvature one surface construction) (Tarantello, 2022).

6. Open Problems and Extensions

While the structure and classification of blow-up limits are well understood in many scalar and low-rank settings, outstanding questions remain in higher-rank, vectorial, or non-Euclidean regimes:

• Complete classification of higher-rank one-homogeneous minimizers is typically open; for many free boundary and vectorial minimization problems, only special cases (e.g., rank 1) are fully described (Philippis et al., 2021, Siclari et al., 31 Jan 2026).
• In sub-Riemannian settings (such as the Heisenberg group), correction of earlier derivations has clarified the form of the limiting PDE for blow-ups, but the fine classification of all possible profiles remains in progress (Yu, 2024).
• For exotic or non-integrable cones in free boundary problems, the best possible rates of convergence and regularity at singularities may be logarithmic rather than algebraic, reflecting the underlying finite-dimensional analytic obstructions (Engelstein et al., 2018).

Blow-up analysis thus provides a powerful lens onto the small-scale, asymptotically universal, and geometric features of variational minimizers, dictating both singularity classification and large-scale structure across a broad spectrum of nonlinear and geometric problems.