Michael-Simon inequality for anisotropic energies close to the area via multilinear Kakeya-type bounds

1. Problem Statement and Context

The Core Problem:
The paper addresses the validity of the Michael–Simon inequality for anisotropic surface energies. In the classical isotropic setting (standard area functional), the Michael–Simon inequality is a fundamental tool derived from the monotonicity formula. It provides a Sobolev-type bound relating the mass of a varifold to its first variation (mean curvature). Specifically, for a rectifiable $k$-varifold $V$ in $\mathbb{R}^n$ with density $\Theta \geq \theta_0$, it asserts:
$$|V|(\mathbb{R}^n)^{k-1} \leq C(n, k, F) \theta_0^{-1} |\delta_F V|(\mathbb{R}^n)^k$$
where $\delta_F V$ is the first variation with respect to an anisotropic integrand $F$.

The Challenge:
In the anisotropic setting (where the energy depends on the orientation of the tangent plane via an integrand $F: Gr_k(\mathbb{R}^n) \to (0, \infty)$), the classical monotonicity formula fails. Consequently, standard tools used to prove compactness, regularity, and density bounds for stationary varifolds break down. While the Michael–Simon inequality is conjectured to hold for "good" anisotropic integrands (specifically those satisfying the Atomic Condition (AC)), a general proof has been elusive.

The Specific Question:
For which anisotropic integrands $F$ does the Michael–Simon inequality hold? The authors focus on the case $n=3, k=2$ (surfaces in 3D) and investigate integrands $F$ that are close to the isotropic area (i.e., $F$ is close to 1 in the $C^1$ topology) or satisfy specific structural conditions (like $\ell^p$ norms).

2. Methodology

The authors employ a novel combination of geometric measure theory and functional analysis, moving away from the traditional reliance on monotonicity.

A. Reduction to Vector Field Inequalities

The core strategy involves translating the geometric problem of varifolds into an analytical problem concerning vector fields on the plane.

1. Projection: They project the 3D varifold $V$ onto a 2D plane (say, the $xy$-plane).
2. Matrix Construction: They construct vector fields $S$ and $T$ derived from the first variation of the anisotropic energy. These fields are related to the matrix $B_F(\nu)$ associated with the integrand $F$.
3. Functional Inequality: The problem reduces to proving a specific inequality for these vector fields involving their divergences and a "min" term of their components.

B. Multilinear Kakeya-Type Bounds

The authors introduce a new nonlinear functional inequality for vector fields on $\mathbb{R}^2$, which serves as the paper's main technical engine.

• Theorem 3.1 (Main Inequality): For vector fields $S, T \in W^{1,1}(\mathbb{R}^2, \mathbb{R}^2)$, they define positive and negative parts of specific components ($S^P, S^N$, etc.). They prove:
  $$\int \chi \min\{S^P, T^P\} \leq C \|\chi\|_{L^2} \left( \int (|S| + |\text{div} S|) \right)^{1/2} \left( \int (|T| + |\text{div} T|) \right)^{1/2} + \text{error terms}$$
• Connection to Alberti's Rank-One Theorem: This inequality is viewed as a quantitative version of Alberti's rank-one theorem. It essentially states that if the "positive" parts of the vector fields overlap significantly, their divergences must be large.
• Smirnov's Decomposition: The proof utilizes Smirnov's decomposition of normal 1-currents into superpositions of curves. This allows them to interpret the divergence as the "boundary mass" of these curves and the determinant term as the "intersection count," leveraging the intuition that curves with non-negative determinant constraints intersect at most once.

C. Perturbation and Linearization

For the case where $F$ is close to the area:

1. They use a linear change of coordinates (Proposition 4.1) to ensure the diagonal entries of the associated matrix are non-negative.
2. They show that if $F$ is $C^1$-close to the isotropic area, the "error terms" in the functional inequality (specifically the negative parts $S^N, T^N$) become negligible.
3. This allows them to close the estimate and derive the Michael–Simon bound.

3. Key Contributions

1. Proof for Anisotropies Close to Area:
   The authors provide the first rigorous proof that the Michael–Simon inequality holds for convex integrands $F$ in $\mathbb{R}^3$ that are sufficiently close to the isotropic area in the $C^1$ topology. This confirms a conjecture for a large open set of integrands containing the standard area.

2. New Functional Inequality (Theorem 3.1):
   They establish a sharp, standalone inequality for vector fields on the plane (Theorem 3.1). This is a significant contribution to analysis, offering a quantitative refinement of Alberti's rank-one theorem and providing a new tool for studying anisotropic problems.

3. Extension to $\ell^p$ Norms:
   They prove the inequality for a specific class of integrands including $\ell^p$ norms ($p \in (1, \infty)$) in $\mathbb{R}^3$, provided they satisfy a specific symmetry condition regarding their gradients on coordinate planes (Theorem 1.5).

4. Equivalence with Compactness:
   The paper establishes a deep structural equivalence: The validity of the Michael–Simon inequality for a general integrand $F$ (satisfying the Atomic Condition) is equivalent to the compactness of rectifiable varifolds with a lower density bound.

   • If Michael–Simon fails, one can construct a "diffuse concentration" counterexample where varifolds converge to a Lebesgue measure despite having vanishing first variation and high density.

5. Simplification of Almgren's Work:
   The authors acknowledge that the result for $F$ close to area was previously claimed in an unpublished manuscript by Frederick J. Almgren Jr. Their work simplifies Almgren's highly technical proof by replacing intricate geometric constructions with the new functional inequality (Theorem 3.1).

4. Main Results

• Theorem 1.3: If $n=3, k=2$ and $F$ is a convex integrand close to the area in $C^1$, then the scale-invariant Michael–Simon bound holds:
  $$|V|(\mathbb{R}^3) \leq C(F) H^2(\{\theta > 0\})^{1/2} |\delta_F V|(\mathbb{R}^3)$$
  This implies the standard Michael–Simon inequality with a density lower bound.

• Theorem 1.4 (Regularity): As a consequence of Theorem 1.3 and Allard's regularity theorem, they establish that integer-rectifiable 2-varifolds with small first variation and density close to 1 are $C^{1,\alpha}$ graphs in the anisotropic setting (for $F$ close to area).

• Theorem 1.5: The inequality holds for strictly convex, even integrands (like $\ell^p$ norms) satisfying a specific directional derivative condition on coordinate planes.

• Proposition 1.9: The Michael–Simon inequality is equivalent to the compactness of rectifiable varifolds (under the Atomic Condition).

5. Significance

• Revitalizing Anisotropic Geometric Measure Theory: By proving Michael–Simon for a broad class of anisotropies, the authors restore a critical tool (compactness and density bounds) that was previously unavailable outside the isotropic case. This opens the door to studying regularity and singularities for a wider range of physical and geometric models (e.g., crystal growth, liquid crystals).
• Bridging Analysis and Geometry: The derivation of a geometric inequality from a purely functional inequality on vector fields (Theorem 3.1) demonstrates a powerful new methodology. It suggests that deep geometric properties of varifolds can be understood through the lens of multilinear Kakeya-type estimates and current decompositions.
• Resolving Open Questions: The work resolves a long-standing open question regarding the validity of Michael–Simon for anisotropies close to the area, confirming Almgren's intuition while providing a modern, simplified proof.
• Foundation for Future Regularity: The results provide the necessary compactness and density lower bounds required to apply Allard's regularity theory in anisotropic settings, which is a prerequisite for understanding the fine structure of minimizers of anisotropic energies.

In summary, this paper successfully generalizes a cornerstone of geometric measure theory to the anisotropic regime near the isotropic case, utilizing a sophisticated new functional inequality that links vector field analysis to the geometry of varifolds.