Limiting behavior of principal eigenvalues for a class of mixed boundary value problems as the measure of the support domain goes to zero

1. Problem Statement

The paper investigates the asymptotic behavior of the principal eigenvalue, denoted as $\sigma_1[-\Delta, \beta, \Omega]$, associated with the linear eigenvalue problem:
$$\begin{cases} -\Delta u = \sigma u & \text{in } \Omega, \\ \frac{\partial u}{\partial n} + \beta(x)u = 0 & \text{on } \partial\Omega, \end{cases}$$
as the Lebesgue measure of the bounded domain $\Omega$ tends to zero ($|\Omega| \downarrow 0$).

Context and Motivation:

• Dirichlet Case: It is a classical result (Faber-Krahn inequality) that for Dirichlet boundary conditions ($u=0$), $\sigma_1 \to +\infty$ as $|\Omega| \to 0$.
• Neumann Case: For pure Neumann conditions ($\beta=0$), the principal eigenvalue is identically zero regardless of the domain size.
• Mixed Case: The behavior for general mixed boundary conditions (where $\beta(x)$ is a continuous function that may change sign) was an open problem. Specifically, it was unclear whether the eigenvalue always diverges to $+\infty$ or if it could diverge to $-\infty$ depending on the sign of $\beta$.

2. Methodology

The authors employ a combination of one-dimensional analysis, comparison principles, symmetry arguments, and perturbation theory.

A. One-Dimensional Prototype (Section 3)

The authors first analyze the 1D model on an interval $(0, L)$:
$$\begin{cases} -u'' = \sigma u, \\ -u'(0) + \beta_0 u(0) = 0, \\ u'(L) + \beta_\omega u(L) = 0. \end{cases}$$

• Technique: They derive explicit transcendental equations for the eigenvalues using trigonometric (for $\sigma > 0$) and hyperbolic (for $\sigma < 0$) functions.
• Analysis: By analyzing the intersection points of these functions (e.g., $\tan(s)$ vs. linear/hyperbolic terms) as $L \to 0$, they determine the limit of $\sigma_1$.
• Key Insight: The limit depends entirely on the sum of the boundary coefficients: $\beta_0 + \beta_\omega$.

B. Multidimensional Analysis (Section 4)

For the general $N$-dimensional case, the authors use comparison principles and monotonicity properties:

• Monotonicity: They utilize the fact that $\sigma_1$ is strictly increasing with respect to the boundary coefficient $\beta$ (Proposition 2.2).
• Faber-Krahn Inequality: They reduce the problem to a ball $B_R$ of the same volume as $\Omega$, as the ball minimizes the principal eigenvalue for a fixed volume.
• Radial Symmetry: On the ball, the problem reduces to a 1D ODE involving a convection term ($\frac{N-1}{r}u'$). They compare this to the 1D prototype without the convection term to establish bounds.
• Isoperimetric Inequality: To prove divergence to $-\infty$ when $\beta < 0$, they construct a supersolution using the solution to a Poisson problem and utilize the isoperimetric inequality to show that the ratio $\frac{\text{Area}(\partial\Omega)}{|\Omega|}$ diverges as $|\Omega| \to 0$.

C. Asymptotic Expansion (Section 5)

In the special case where $\beta > 0$ is constant, the authors perform a perturbation analysis (scaling the domain to the unit ball $B_1$) to derive the precise rate of divergence. They prove the mapping $R \mapsto \Sigma(R)$ is real analytic and compute the first-order term.

3. Key Contributions and Results

Theorem 1.1 (One-Dimensional Case)

The limit of the principal eigenvalue $\sigma_1$ as $L \downarrow 0$ is determined by the sum of the boundary parameters $\beta_0 + \beta_\omega$:

1. If $\beta_0 + \beta_\omega > 0$: $\lim_{L \downarrow 0} \sigma_1 = +\infty$.
2. If $\beta_0 + \beta_\omega = 0$: $\sigma_1 = -\beta_0^2$ (constant negative value).
3. If $\beta_0 + \beta_\omega < 0$: $\lim_{L \downarrow 0} \sigma_1 = -\infty$.

• Significance: This reveals that unlike the Dirichlet case, the eigenvalue can diverge to $-\infty$ if the "average" boundary condition is sufficiently negative.

Theorem 1.2 (Multidimensional Case)

For a general domain $\Omega \subset D$ with $\beta \in C(D)$:

1. If $\inf_D \beta > 0$: $\lim_{|\Omega| \downarrow 0} \sigma_1[-\Delta, \beta, \Omega] = +\infty$.
2. If $\sup_D \beta < 0$: $\lim_{|\Omega| \downarrow 0} \sigma_1[-\Delta, \beta, \Omega] = -\infty$.

• Significance: This resolves the open problem for multidimensional domains. It shows that the sign of the boundary coefficient dictates the direction of divergence. The result $\sigma_1 \to -\infty$ for negative $\beta$ is described as "surprising" given classical Dirichlet intuition.

Theorem 5.1 (Asymptotic Expansion)

For a ball $B_R$ with constant $\beta > 0$:
$$\lim_{R \downarrow 0} \left( R \cdot \sigma_1[-\Delta, \beta, B_R] \right) = \beta \frac{\text{Area}(\partial B_1)}{|B_1|}$$
This provides the exact leading-order term of the divergence: $\sigma_1 \sim \frac{C}{R}$ as $R \to 0$.

4. Significance and Implications

• Resolution of an Open Problem: The paper settles the behavior of principal eigenvalues for mixed boundary conditions as the domain shrinks, a topic previously unresolved even for constant coefficients.
• Counter-Intuitive Behavior: It demonstrates that shrinking a domain does not always lead to an increase in the principal eigenvalue. If the boundary condition is "absorbing" enough (negative $\beta$), the eigenvalue plunges to $-\infty$, indicating instability or a different spectral regime.
• Unified Framework: The work bridges the gap between Dirichlet (infinite positive limit) and Neumann (zero limit) cases, showing that mixed conditions occupy a spectrum where the sign of $\beta$ is the critical determinant.
• Applications: These results are relevant in mathematical physics and biology (e.g., reaction-diffusion systems, population dynamics) where domain size and boundary fluxes determine the survival or extinction of species. The divergence to $-\infty$ suggests that in small domains with negative boundary feedback, the system may exhibit unbounded growth or instability.

5. Conclusion

The paper rigorously characterizes the limiting behavior of principal eigenvalues for mixed boundary value problems. It establishes that the limit is $+\infty$ if the boundary coefficient is strictly positive, but surprisingly $-\infty$ if the coefficient is strictly negative. The authors provide explicit proofs using 1D prototypes, comparison principles, and isoperimetric inequalities, culminating in a precise asymptotic expansion for spherical domains.