Geometric Estimates for Solutions of Semilinear Equations with Singular Potentials

1. Problem Statement and Context

The paper investigates the qualitative behavior of nonnegative local minimizers for a specific class of elliptic functionals involving singular absorption terms and unbounded, sign-changing source terms.

The core equation studied is the semilinear elliptic equation:
$$\text{div}(A(x)\nabla u) = \gamma f(x) u^{\gamma-1} \quad \text{in } \{u > 0\} \cap B_1$$
where:

• $\gamma \in (0, 1)$: This creates a singular absorption term ($u^{\gamma-1} \to \infty$ as $u \to 0$).
• $A(x)$: A uniformly elliptic matrix with measurable coefficients.
• $f(x)$: The source term, which is allowed to be unbounded and sign-changing, belonging to the Lorentz space $L^{p, \infty}(B_1)$ with $p > n/2$.

Mathematical Context:
This problem interpolates between two classical free boundary regimes:

1. Bernoulli-type (Cavity): The limit as $\gamma \to 0$.
2. Obstacle-type: The limit as $\gamma \to 1$.

The Gap:
Previous literature (notably Alt and Phillips) established regularity and geometric properties for this equation under the assumption that $f$ is bounded and strictly positive. However, many physical models (combustion, porous media flow) involve highly irregular, unbounded, or sign-changing sources. The classical theory fails in these settings. This paper aims to extend the theory to the Lorentz-space setting ($f \in L^{p, \infty}$) without sign restrictions.

2. Methodology and Framework

The authors employ a combination of variational methods, multiscale analysis, and geometric measure theory.

A. Variational Framework

The solutions are defined as minimizers of the functional:
$$J_{A,f}(v) = \int_{B_1} \left( \frac{1}{2}\langle A(x)\nabla v, \nabla v \rangle + f(x)(v_+)^{\gamma} \right) dx$$
over the admissible set $K = \{v \in W^{1,2}(B_1) : v - g \in W^{1,2}_0(B_1)\}$.

B. Key Technical Tools

1. Lorentz Space Embeddings: The authors utilize the properties of $L^{p, \infty}$ (weak $L^p$) spaces. They establish specific embedding lemmas (Lemma 1) to handle the singular source term $f$ in energy estimates.
2. Approximation by Harmonic Functions: A central strategy involves comparing the minimizer $u$ to a solution $h$ of the homogeneous equation $\text{div}(A(x)\nabla h) = 0$. By quantifying the error $\|u - h\|$, they transfer regularity from the well-understood linear case to the singular nonlinear case.
3. Iterative Oscillation Control: To prove Hölder continuity, the authors use a multiscale iteration argument (Proposition 4.3). They rescale the problem and show that under smallness conditions on the source term, the oscillation of the solution decays geometrically.
4. Caccioppoli-type Estimates: Modified energy estimates are derived to handle the singular term $u^{\gamma-1}$, specifically by testing with $\eta^2 u^{1-\gamma}$ to derive bounds on the gradient.
5. Non-degeneracy via Mass Concentration: For the non-degeneracy results, the authors analyze the "mass" of the source term $f$ near the free boundary. They distinguish between points where $f$ is bounded below and points where $f$ is singular but has sufficient concentration of positive mass.

3. Key Contributions and Results

The paper provides a comprehensive extension of the Alt-Phillips theory to singular settings.

A. Existence and Boundedness (Section 3)

• Existence: Proven via the direct method in the calculus of variations, utilizing the coercivity of the functional in $H^1$ and the weak lower semicontinuity.
• Boundedness: The authors prove that minimizers are uniformly bounded ($L^\infty$) even when $f \in L^{p, \infty}$ and $p > n/2$. This is achieved via a De Giorgi-type iteration adapted to the singular source.

B. Universal Hölder Regularity (Section 4)

• Result: Minimizers are locally Hölder continuous ($C^{0, \beta}$).
• Significance: This establishes that the singular absorption term and the unbounded source do not destroy the basic continuity of the solution, provided the source is in the critical Lorentz space. This is the foundational step for analyzing the free boundary.

C. Sharp Regularity at the Free Boundary (Section 5)

• Result: The authors determine the optimal growth rate of solutions near the free boundary $\partial\{u > 0\}$.
• Optimal Exponent: The solution behaves like $|x - x_0|^\theta$ where:
  $$\theta = \frac{2 - n/p}{2 - \gamma}$$
• Significance: This generalizes the classical result (where $f$ is bounded, effectively $p=\infty$, yielding $\theta = 2/(2-\gamma)$). The paper proves that the singularity of $f$ (measured by $p$) directly reduces the regularity of the solution.

D. Non-degeneracy Estimates (Section 6)

Non-degeneracy ensures the solution does not vanish too slowly near the free boundary, preventing "flat" free boundaries.

• Theorem 3 (Positive Source): If $f \geq C_0 > 0$, the solution satisfies a lower bound of order $r^{2/(2-\gamma)}$.
• Theorem 4 (Singular/Sign-changing Source): This is a major novelty. Even if $f$ is unbounded or changes sign, if the positive mass of $f$ is sufficiently concentrated near a free boundary point $y_0$ (quantified by a measure condition on the set where $f$ is large), a sharp non-degeneracy estimate holds:
  $$\sup_{\partial B_{2r}(y_0)} u \geq \tilde{c} r^{(2 - n/p)/(2 - \gamma)}$$
• Positive Density: As a consequence, the free boundary has positive density (Theorem 7), meaning the positivity set $\{u > 0\}$ occupies a significant portion of any ball centered at a free boundary point.

E. Sharpness via Radial Example (Section 7)

The authors construct a radial solution $u(x) = |x|^\alpha$ with a source $f(x) \sim |x|^{-\beta}$. They demonstrate that:

1. $f \in L^{p, \infty}$ if and only if $\beta = n/p$.
2. The resulting exponent $\alpha$ matches exactly the theoretical optimal exponent $\theta = \frac{2 - n/p}{2 - \gamma}$.
   This confirms that the derived regularity estimates are sharp and cannot be improved.

4. Significance and Impact

1. Bridging the Gap: The work successfully bridges the gap between classical free boundary theory (bounded sources) and modern applications involving singular, irregular data.
2. Critical Integrability Scale: It identifies $L^{p, \infty}$ with $p > n/2$ as the critical integrability scale where the singular absorption term remains controlled, allowing for a well-posed free boundary problem.
3. Handling Sign-Changing Sources: By allowing $f$ to change sign, the results apply to a much broader class of physical models (e.g., combustion with varying fuel density) than previous theories.
4. Quantitative Precision: The paper provides explicit, sharp exponents for both regularity and non-degeneracy, showing precisely how the singularity of the source degrades the solution's smoothness.
5. Foundation for Future Work: These results lay the groundwork for further analysis of the free boundary's geometric structure (e.g., rectifiability, Hausdorff dimension) in singular settings, which was previously inaccessible due to the lack of regularity estimates.

In summary, Nascimento and Zhang have extended the robust theory of Alt-Phillips to a highly singular regime, proving that despite unbounded and sign-changing sources, the free boundary retains a rich geometric structure governed by the interplay between the absorption exponent $\gamma$ and the integrability of the source $p$.