A not-so-strange term coming from somewhere

Technical Summary: Homogenisation of Laplace's Equation in Periodically Perforated Domains with Scaled Robin Conditions

Problem Statement
The paper investigates the asymptotic behaviour of solutions to Laplace's equation (with a potential term) in a bounded domain $\Omega \subset \mathbb{R}^n$ ($n \ge 3$, with extensions to $n=2$) periodically perforated by a set of holes $P_\varepsilon$. The holes are defined as $P_\varepsilon = \varepsilon \mathbb{Z}^n + \varepsilon H_\varepsilon$, where $H_\varepsilon$ represents a rescaled reference hole within the unit cell.

The governing equation is:
$$-\Delta u_\varepsilon + \alpha u_\varepsilon = f \quad \text{in } \Omega_\varepsilon := \Omega \setminus P_\varepsilon$$
subject to a Robin boundary condition on the hole boundaries $\Gamma_\varepsilon = \partial P_\varepsilon \cap \Omega$:
$$\partial_\nu u_\varepsilon + \frac{\beta}{\mu_\varepsilon} u_\varepsilon = \frac{g}{\mu_\varepsilon} \quad \text{on } \Gamma_\varepsilon$$
and Dirichlet conditions on the outer boundary $\partial \Omega$.

The core novelty lies in the scaling of the Robin coefficient. The parameter $\mu_\varepsilon$ is chosen to be proportional to the total surface area of the perforations, $\sigma(\Gamma_\varepsilon)$. Specifically, $\mu_\varepsilon$ satisfies $\lim_{\varepsilon \to 0} \frac{|\Omega|\mu_\varepsilon}{\sigma(\Gamma_\varepsilon)} = 1$. This scaling ensures that the surface energy term in the associated variational functional is of the same order as the bulk energy, creating a critical regime where surface and bulk effects compete.

Methodology
The authors employ a homogenisation framework based on the analysis of energy minimisers and trace inequalities. The methodology proceeds through several key steps:

1. Variational Formulation: The problem is framed as the minimisation of a functional $F_\varepsilon(u)$ comprising bulk energy (gradient and potential terms) and a surface energy term scaled by $1/\mu_\varepsilon$.
2. Abstract Setting: The authors establish a general theory for a family of holes satisfying abstract geometric and functional assumptions (C1–C5). These include Lipschitz regularity, compactness, and the existence of uniform extension operators.
3. Spectral Analysis (Steklov-type Problems): A central technical tool is the analysis of a specific eigenvalue problem (denoted $EV_{\varepsilon, \kappa}$) on the perforated unit cell. This is a Steklov-type problem where the spectral parameter $\lambda(\varepsilon, \kappa)$ appears both in the differential equation and the boundary condition:
   $$-\Delta v_{\varepsilon, \kappa} + \kappa \lambda(\varepsilon, \kappa) v_{\varepsilon, \kappa} = 0 \quad \text{in } T^n \setminus H_\varepsilon$$
   $$\partial_\nu v_{\varepsilon, \kappa} = \lambda(\varepsilon, \kappa) v_{\varepsilon, \kappa} \quad \text{on } \partial H_\varepsilon$$
   The authors study the asymptotic limit of the scaled eigenvalues $\lambda^*(\kappa) = \lim_{\varepsilon \to 0} \varepsilon^{-2}\lambda(\varepsilon, \kappa)$.
4. Trace Inequalities: The convergence analysis relies on establishing trace inequalities that bound the surface integral of functions on $\Gamma_\varepsilon$ by their bulk energy. The optimal constants in these inequalities are linked to the eigenvalues of the spectral problem mentioned above.
5. Corrector Construction: To prove strong $L^2$ convergence, the authors construct correctors $w_\varepsilon$ based on the minimisers of the spectral problem. These correctors are used to test the weak formulation of the problem and identify the limit equation.

Key Contributions and Results

• Identification of a Critical Regime: The paper identifies a specific scaling for the Robin parameter (inverse total surface area) that bridges the gap between the classical Neumann limit (vanishing coupling) and the Dirichlet limit (infinitely stiff inclusions).
• Nonlinear "Strange Term": The main result (Theorem 2.8) demonstrates that the homogenised equation contains an additional zeroth-order term:
  $$-\Delta u_0 + \alpha u_0 + \beta \kappa^*(\beta) u_0 = f + \kappa^*(\beta) g \quad \text{in } \Omega$$
  The coefficient $\beta \kappa^*(\beta)$ is the "strange term." Unlike classical results where the strange term is linear or constant, here the coefficient depends nonlinearly on the Robin parameter $\beta$.
• Characterisation via Spectral Problem: The parameter $\kappa^*(\beta) \in (0, 1)$ is uniquely determined by the equation $-\lambda^*(\kappa^*(\beta)) = \beta$, where $\lambda^*(\kappa)$ is the limit of the scaled eigenvalues of the Steklov-type problem.
• Interpolation of Limits:
  • As $\beta \to 0$ (Neumann limit), $\beta \kappa^*(\beta) \to 0$, recovering the standard Laplacian without a strange term.
  • As $\beta \to +\infty$ (Dirichlet limit), $\beta \kappa^*(\beta)$ converges to the classical capacitary strange term $\Lambda^*_{Cap}$ (Cioranescu-Murat term), recovering the known result for Dirichlet holes.
• Explicit Formulas for Spherical Holes: In the specific model case where holes are spherical ($D = B^n$) and scaled as $r_\varepsilon \sim \varepsilon^{n/(n-2)}$, the authors derive an explicit formula for the strange term:
  $$\kappa^*(\beta) = \frac{\sigma_n(n-2)}{\sigma_n(n-2) + \beta}$$
  leading to a strange term coefficient of $\frac{\beta \sigma_n(n-2)}{\sigma_n(n-2) + \beta}$.
• Convergence Rates: Under additional regularity assumptions (domain $\Omega \in C^2$, bounded source terms), the paper provides an error estimate for the convergence of $u_\varepsilon$ to the homogenised solution $u_0$ (Proposition 2.9).
• Robustness to Surface Area: The paper includes an example (Section 6) where the surface area of the holes $\sigma(\Gamma_\varepsilon)$ tends to infinity as $\varepsilon \to 0$. Despite the Robin coefficient $\beta/\mu_\varepsilon$ vanishing, the large surface area compensates, and a non-trivial strange term persists in the limit.

Significance and Claims
The authors claim that this work provides a unified framework for homogenisation problems on perforated domains. By identifying the critical scaling where surface and bulk energies are balanced, they show that the effective behaviour is not merely a transition between Neumann and Dirichlet extremes but involves a distinct, nonlinear interaction characterised by a Steklov-type spectral problem.

The paper emphasizes that the resulting "strange term" interpolates continuously between the two classical regimes. The authors note that while the existence of such a term was known for Dirichlet conditions, the specific nonlinear dependence on the Robin parameter and its characterisation via a spectral problem with the parameter in both the equation and boundary condition is a new contribution. The work is motivated by applications in continuum mechanics, such as modelling composite materials with surface energy effects (e.g., liquid crystal colloids), but the primary contribution is mathematical, offering a rigorous asymptotic analysis and convergence rates for this specific critical scaling.