Monotonicity for the fractional semi-linear problem in a half space

1. Problem Statement

The paper investigates the qualitative properties of positive solutions to the fractional semi-linear equation in the upper half-space $\mathbb{R}^n_+ = \{x = (x', x_n) \in \mathbb{R}^n : x_n > 0\}$:
$$\begin{cases} (-\Delta)^s u(x) = f(u(x)), & x \in \mathbb{R}^n_+, \\ u(x) = 0, & x \in \mathbb{R}^n \setminus \mathbb{R}^n_+, \end{cases}$$
where $0 < s < 1$, $n \geq 2s$, and $(-\Delta)^s$ is the fractional Laplacian defined via a singular integral. The nonlinearity $f$ is assumed to be locally Lipschitz continuous ($f \in C^{0,1}_{loc}([0, \infty))$) with $f(0) \geq 0$.

The Core Objective: To prove that any non-negative, non-trivial classical solution $u$ is strictly increasing in the $x_n$-direction (i.e., $\frac{\partial u}{\partial x_n} > 0$ in $\mathbb{R}^n_+$).

Key Constraint: The authors aim to remove the standard assumption that the solution $u$ must be globally bounded in $\mathbb{R}^n$. Instead, they only assume the solution is bounded on each slab $\Omega_\lambda = \{x \in \mathbb{R}^n_+ : 0 < x_n < \lambda\}$ for any $\lambda > 0$.

---

2. Methodology

The proof relies on the Method of Moving Planes, adapted to handle the non-local nature of the fractional Laplacian and the lack of global boundedness. The methodology introduces several novel technical tools:

A. Boundary Regularity under Local Boundedness

Standard approaches to boundary regularity for fractional equations often require global boundedness to construct supersolutions.

• Innovation: The authors decompose the solution $u$ into a potential part $v$ (driven by $f$) and an $s$-harmonic part $w$ (driven by boundary data).
• Technique: They establish a boundary Hölder regularity estimate ($u(x) \leq C x_n^s$) assuming only that $u$ and $f$ are bounded near the boundary.
• Mechanism: They use an iterative bootstrapping argument. Starting with a rough estimate, they substitute the regularity back into the Poisson integral representation of the harmonic part to gain higher regularity, repeating the process until $s$-Hölder continuity is achieved.

B. Multiple Narrow Region Principle

The method of moving planes requires starting the plane from the boundary ($x_n=0$).

• Innovation: Previous "narrow region principles" applied only to a single narrow strip. The authors establish a Multiple Narrow Region Principle applicable to a union of finitely many disjoint narrow slabs.
• Significance: This allows the moving plane procedure to be initiated even when the domain geometry or solution behavior creates multiple potential "narrow" regions simultaneously, ensuring the plane can start moving from $x_n=0$.

C. Averaging Effects

A major difficulty in unbounded domains is that the minimum of the difference function $w_\lambda(x) = u(x^\lambda) - u(x)$ may not be attained, preventing the standard contradiction arguments.

• Innovation: The authors introduce Averaging Effects (Theorems 1.7 and 1.9).
• Mechanism: They prove that if a solution (or an antisymmetric function) is positive in a bounded region $D$, this positivity "diffuses" or "averages out" to a larger surrounding region $D_R$, regardless of how far $D$ is from the boundary.
• Application: This effect is applied four times during the moving plane process:
  1. To propagate positivity from the interior to the boundary.
  2. To handle the antisymmetric function $w_\lambda$.
  3. To normalize the solution sequence and derive a contradiction if the plane stops.
  4. To ensure the normalized sequence does not vanish, leading to a contradiction with the strong maximum principle.

---

3. Key Results

Theorem 1.1 (Main Monotonicity Result)

Under the assumption that $u$ is bounded on every slab $\Omega_\lambda$ (Condition BDS) and $f$ is locally Lipschitz with $f(0) \geq 0$, any non-negative non-trivial solution satisfies:
$$\frac{\partial u}{\partial x_n}(x) > 0, \quad \forall x \in \mathbb{R}^n_+.$$
This result holds without assuming global boundedness of $u$.

Theorem 1.2 (Boundary Regularity)

If $u$ is a non-negative solution to $(-\Delta)^s u = f(x)$ in $\mathbb{R}^n_+$ with $u=0$ outside, and both $u$ and $f$ are locally bounded near the boundary, then $u$ is $s$-Hölder continuous up to the boundary ($u(x) \leq C x_n^s$). This improves upon existing results by removing the global boundedness requirement.

Theorem 1.3 (Monotonicity for Uniformly Continuous Solutions)

If $u$ is uniformly continuous (a slightly stronger condition than slab-boundedness in some contexts, but weaker than global boundedness), the strict monotonicity holds.

Theorem 1.5 (Multiple Narrow Region Principle)

Establishes a maximum principle for antisymmetric functions in a union of finitely many narrow slabs, allowing the moving plane method to start effectively.

Theorem 1.7 & 1.9 (Averaging Effects)

Proves that positivity in a local region implies a uniform lower bound in a surrounding region, independent of the location of the original region. This is crucial for handling unbounded domains where the plane moves to infinity.

---

4. Significance and Contributions

1. Relaxation of Global Boundedness: This is the most significant contribution. Previous literature (e.g., Fall, Weth, Barrios, et al.) required solutions to be globally bounded ($u \in L^\infty(\mathbb{R}^n)$) to apply the moving plane method. This paper demonstrates that local boundedness on slabs is sufficient. This opens the door to studying solutions that may grow at infinity, which are common in physical applications but previously intractable for monotonicity analysis.

2. Novel Regularity Theory: The derivation of boundary Hölder regularity under only local boundedness near the boundary is a substantial technical advancement. It bypasses the need for global supersolutions, which are difficult to construct for non-local operators without global bounds.

3. New Tools for Non-local Operators:

   • The Multiple Narrow Region Principle generalizes a fundamental tool in the moving plane method.
   • The Averaging Effect provides a robust alternative to the "blow-up" or "limiting sequence" arguments that typically require higher regularity ($C^{2s+\epsilon}$) to converge. The averaging effect works with merely uniform continuity.

4. Broad Applicability: The techniques developed are not limited to the specific equation studied. The authors suggest these tools (especially the averaging effects and the relaxed regularity assumptions) will be powerful for studying other qualitative properties (e.g., symmetry, classification, Liouville theorems) of fractional equations in unbounded domains.

Conclusion

Chen, Guo, and Wu successfully extend the monotonicity theory for fractional semi-linear equations to a much broader class of solutions by fundamentally rethinking the requirements for the moving plane method. By replacing global boundedness with slab-boundedness and introducing the averaging effect, they overcome the non-locality challenges that have historically restricted such results.