Three-spheres theorem for harmonic functions (non-concentric case)

1. Problem Statement and Motivation

The paper addresses the generalization of Hadamard's Three-Circle Theorem from complex analysis to the realm of harmonic functions in $\mathbb{R}^n$ ($n \geq 2$).

• Classical Context: Hadamard's theorem states that for a holomorphic function, the maximum modulus on a circle is a log-convex function of the logarithm of the radius. This has analogs for harmonic functions on concentric spheres (Landis, Agmon, Korevaar).
• The Gap: Existing generalizations largely focus on concentric spheres or balls. The authors aim to establish a direct analog for non-concentric, non-touching spheres in $\mathbb{R}^n$.
• Goal: To derive a "Three-Spheres Theorem" in a weighted $L^2$-norm for harmonic functions defined on a domain containing three correlated, non-concentric spheres. This result is intended to facilitate new uniqueness theorems and "propagation of smallness" results.

2. Methodology

The authors employ a sophisticated geometric and analytic approach combining inversion techniques, Kelvin transforms, and log-convexity properties.

A. Geometric Construction: Correlated Spheres

The core innovation is the definition of "correlated spheres."

• Two balls $B_{x,r}$ and $B_{\tilde{x},\tilde{r}}$ inside the unit ball $B$ are defined as correlated if their centers are codirected (lie on the same ray from the origin) and satisfy a specific algebraic relation (Eq. 6):
  $$(1 + |x|^2 - r^2)|\tilde{x}| = |x|(1 + |\tilde{x}|^2 - \tilde{r}^2)$$
• This correlation ensures that the family of balls generated by varying the center along a ray can be mapped to a family of concentric balls via a specific inversion.

B. Inversion and Kelvin Transform

• Inversion Map: The authors utilize a conformal inversion $\phi$ centered at a point $a$ (where $|a| > 1$) with a specific radius $\rho$. This map is chosen such that:
  1. It maps the unit sphere $S$ to itself ($\phi(B) = B$).
  2. It maps the family of non-concentric correlated balls $\{B_{x_t, r_t}\}$ to a family of concentric balls $\{B_{r^*_t}\}$ centered at the origin.
• Kelvin Transform: For a harmonic function $f$, the authors consider its Kelvin transform $f^*$ with respect to this inversion. Since the Kelvin transform preserves harmonicity, $f^*$ is harmonic in the transformed domain.

C. Log-Convexity and Derivation

• Reduction to Concentric Case: By transforming the problem into the concentric setting via $\phi$, the authors apply the known log-convexity of the $L^2$-norm for harmonic functions on concentric spheres (a result by Korevaar and Meyers).
• Change of Variables: They carefully track the change of measure under the inversion. The standard surface measure $ds$ transforms into a weighted measure $ds_a$ involving the distance to the inversion center $a$.
• Differentiation: Using a specific parametrization of the ball centers and radii, they derive differential formulas (Eqs. 12-13) linking the derivative of the volume integral to surface integrals, allowing them to establish the convexity inequality in the original non-concentric coordinates.

3. Key Contributions and Results

A. The Three-Spheres Theorem (Theorem 1)

The main result establishes an inequality for a harmonic function $f \in h^2(B^n)$ on three correlated spheres $S_{x,r}$, $S_{\tilde{x},\tilde{r}}$, and the boundary sphere $S$.

• Inequality:
  $$\int_{S_{x,r}} |f|^2 ds_a \leq \left( \int_{S_{\tilde{x},\tilde{r}}} |f|^2 ds_a \right)^\beta \left( \int_{S} |f|^2 ds_a \right)^{1-\beta}$$
• Weighted Measure: The measure $ds_a$ is defined as $ds_a = \frac{|y-a|^2 + 1 - |y|^2}{|y-a|^4} ds$.
• Exponent: The exponent $\beta$ is explicitly calculated based on the geometric parameters of the spheres and the inversion center $a$.
• Significance: This provides a direct analog of Hadamard's theorem for non-concentric geometry, extending previous work which was limited to concentric cases or different norms.

B. Three-Balls Theorem (Theorem 2 & Statement 1)

The authors extend the result from spheres to balls (volume integrals).

• They derive an inequality for the weighted $L^2$-norm over balls.
• To remove the specific weighted measure $d\mu_a$ and return to the standard Lebesgue measure $dx$, they employ an embedding technique. They embed $\mathbb{R}^n$ into $\mathbb{R}^{n+5}$ and extend the harmonic function $u(x)$ to $\tilde{u}(x, y) = u(x)$. This allows them to bound the weighted integral by standard integrals, resulting in an inequality (Eq. 29) with a constant factor depending on the dimension and the ratio of radii.

C. Uniqueness and Propagation of Smallness (Theorem 3)

The paper culminates in applications to uniqueness problems.

• Theorem 3: If a harmonic function $u$ on $\mathbb{R}^n$ satisfies a growth condition $A_2(r, u) \leq \phi(r)$ and is "small" (bounded by $\epsilon_m$) on a sequence of balls $B_{x_m, r_m}$ where the centers go to infinity, then $u \equiv 0$ under specific decay conditions on $\epsilon_m$.
• Mechanism: The proof uses a contradiction argument. If $u$ is not zero, the log-convexity derived in the previous sections forces the norm to grow too fast to satisfy the decay condition given in the hypothesis (Eq. 38).
• Result: This establishes a uniqueness theorem for harmonic functions on $\mathbb{R}^n$ based on their behavior on a sequence of non-concentric, non-touching balls.

4. Significance and Impact

• Generalization: The paper successfully bridges the gap between the classical concentric three-spheres theorems and the more complex non-concentric geometry, which is more relevant to practical applications where domains are irregular.
• New Tools: The introduction of "correlated spheres" and the specific inversion technique provides a new toolkit for analyzing elliptic equations in non-symmetric domains.
• Applications: The results directly contribute to the theory of propagation of smallness (how smallness of a solution in a small region forces smallness in a larger region) and uniqueness (determining if a function is zero based on local data). These are fundamental in inverse problems, control theory, and mathematical physics.
• Historical Context: The paper notes that the core results were originally part of a 1999 Master's thesis and a 1999 journal publication, but this expanded version (2026) adds significant new uniqueness results (Sections 5-6) that were previously unpublished.

In summary, Arakelian and Matevosyan provide a rigorous geometric framework to extend Hadamard's classical convexity principles to non-concentric settings, yielding powerful new uniqueness theorems for harmonic functions in $\mathbb{R}^n$.