On Weiss Almost Monotonicity Formula

1. Problem Statement and Context

The paper addresses the regularity theory of free boundary problems associated with energy functionals of the Alt–Phillips type with variable coefficients. Specifically, the author investigates almost minimizers of the functional:
$$J(u, \Omega) = \int_{\Omega} \left( \frac{1}{2} \langle A(x)\nabla u, \nabla u \rangle + \delta(x) u^{\gamma(x)} \right) dx$$
where:

• $u \geq 0$ is the solution.
• $A(x)$ is a uniformly elliptic matrix-valued function.
• $\delta(x)$ and $\gamma(x)$ are variable coefficients.
• The exponent $\gamma(x)$ lies in $(0, 1)$.

The Challenge:
Classical regularity results for free boundaries (specifically the classification of blow-up limits and the structure of the singular set) rely heavily on Weiss-type monotonicity formulas. However, standard Weiss formulas require the coefficients to be sufficiently smooth (typically $W^{1,q}$ with $q > d$) to allow for differentiation of the energy functional.
In many physical and geometric applications, coefficients may only possess low regularity (e.g., Dini continuity). Under these weaker assumptions, the standard monotonicity formula fails because the error terms generated by the variable coefficients cannot be controlled.

The Goal:
To establish a Weiss-type almost-monotonicity formula for this class of functionals under minimal regularity assumptions on the coefficients, thereby enabling the classification of blow-up limits and the extension of free boundary regularity results to this broader setting.

---

2. Methodology

The core of the paper relies on a "freezing coefficients" strategy combined with optimal regularity estimates.

A. The "Freezing" Strategy

Instead of analyzing the functional with variable coefficients directly, the author compares the original functional $J$ with a frozen functional $J_{x_0}$ at a free boundary point $x_0$:
$$J_{x_0}(w) = \int \left( \frac{1}{2} \langle A(x_0)\nabla w, \nabla w \rangle + \delta(x_0) w^{\gamma(x_0)} \right) dx$$
The strategy involves:

1. Testing against competitors: Using the almost-minimality of $u$ to compare $J(u)$ with $J(w)$, where $w$ is a suitable competitor (specifically, a homogeneous extension of $u$ from the boundary of a ball).
2. Back-and-forth comparison: Comparing the original functional $J$ with the frozen functional $J_{x_0}$ to quantify the deviation caused by the variation of coefficients.
3. Error Quantification: The deviation from strict monotonicity is shown to be an integrable error term $g(r)$ that depends on the moduli of continuity of the coefficients ($A, \delta, \gamma$) and the gauge function $\varrho$ of the almost-minimizer.

B. Regularity Assumptions

The paper significantly weakens the required regularity compared to previous works (e.g., [4]). The main result holds under the following Dini-type continuity assumptions:

• $\omega_A + \omega_\delta + \varrho + |\ln(\cdot)|\omega_\gamma \in C_{0, \text{Dini}}([0, 2])$.
• Specifically, $\delta \in C^{0, \text{Dini}}$ and $\gamma \in C^{0, \text{log-Dini}}$.
• This is a substantial improvement over the $W^{1,q}$ ($q>d$) requirement in prior literature.

C. Blow-up Analysis

The almost-monotonicity formula implies that the Weiss energy $A_{u, x_0}(r)$ has a limit as $r \to 0$. This limit is used to prove that any blow-up limit $u_0$ (obtained by rescaling $u$ near a free boundary point) is homogeneous with a specific exponent $\beta(x_0) = \frac{2}{2-\gamma(x_0)}$.

---

3. Key Contributions and Results

1. The Weiss Almost-Monotonicity Formula (Theorem 3.1)

The paper proves that for an almost minimizer $u$, the functional:
$$A_{u, x_0}(r) = r^{-(d+2(\beta(x_0)-1))} J_{x_0}(u, B_r(x_0)) - \frac{1}{2\beta(x_0)} r^{-(d+2\beta(x_0)-1)} \int_{\partial B_r(x_0)} u^2 d\mathcal{H}^{d-1}$$
satisfies:
$$\frac{d}{dr} A_{u, x_0}(r) \geq -g(r)$$
where $g \in L^1(0, 1)$. The function $g(r)$ explicitly captures the error due to the non-constant coefficients.

2. Homogeneity of Blow-up Limits (Corollary 3.1)

As a direct consequence of the almost-monotonicity, the paper establishes that all blow-up limits of local minimizers are homogeneous functions of degree $\beta(x_0)$. This is a crucial step for classifying the free boundary.

3. Regularity of the Free Boundary (Theorem 4.1)

The author applies the new formula to the Alt–Phillips problem with varying coefficients. The main regularity result states that under the Dini assumptions:

• The regular set $\mathcal{R}(u)$ (where the free boundary is a $C^{1,\alpha}$ surface) is locally open in the free boundary.
• The singular set $\Sigma(u)$ has Hausdorff dimension $\dim_H(\Sigma(u)) \leq d-3$.
• The free boundary has finite $(d-1)$-dimensional Hausdorff measure.

This extends the full regularity program previously established in [4] (which required $W^{1,q}$ coefficients) to a much more natural setting of Dini-continuous coefficients.

4. Extensions to Other Problems (Section 5)

The methodology is shown to be robust and applicable to:

• Alt–Caffarelli One-Phase Problem: With variable coefficients $A(x)$ and $Q(x)$.
• Two-Phase Problems: Both for Alt–Caffarelli and Alt–Phillips types, where the functional involves terms for both $u>0$ and $u<0$.
• The paper provides the corresponding almost-monotonicity formulas for these cases, provided optimal growth estimates hold.

---

4. Significance and Impact

• Lowering Regularity Barriers: The paper bridges a gap between the theoretical requirements for monotonicity formulas and the practical regularity of coefficients in many applications. By reducing the requirement from $W^{1,q}$ to Dini continuity, it makes the powerful tools of Weiss monotonicity applicable to a wider class of non-smooth problems.
• Classification of Singularities: By proving the homogeneity of blow-up limits under low regularity, the paper ensures that the classification of free boundary points (regular vs. singular) remains valid even when coefficients are not smooth.
• Methodological Innovation: The "freezing" technique combined with optimal regularity estimates offers a template for analyzing other non-differentiable or variable-coefficient free boundary problems where standard differentiation of the energy fails.
• Future Directions: The work opens avenues for investigating the weakest possible regularity assumptions for free boundary regularity and suggests potential extensions to $p$-Laplacian type problems.

In summary, Sobral's work provides a rigorous analytical framework to handle variable-coefficient free boundary problems with minimal smoothness, ensuring that the fine structure of the free boundary (regularity and dimension of singular sets) can be established using almost-monotonicity arguments.