Spectral analysis of the biharmonic operator subject to Neumann boundary conditions on dumbbell domains (Q1686844)

The paper is devoted to a spectral analysis of the bi-harmonic operator in \(\mathbb R^2\) subject to Neumann boundary conditions on a domain which undergoes a singular perturbation. The domain \(\Omega\) is the union of two bounded smooth domains \(\Omega_L\) and \(\Omega_R\) with empty intersection satisfying the additional condition \[ (\Omega_L \cup \Omega_R) \cap \big\{(x,y)\in \mathbb R^2: (x,y)\in [0,1]\times [-1,1]\big\} =\emptyset. \] A family of domains \(\{\Omega_\varepsilon\}_{\varepsilon >0}\) is defined by \(\Omega_\varepsilon:=\Omega \cup R_\varepsilon \cup L_\varepsilon\), where \(R_\varepsilon \cup L_\varepsilon\) is a thin channel connecting \(\Omega_L\) and \(\Omega_R\). The family \(\{\Omega_\varepsilon\}_{\varepsilon >0}\) collapses to the limit set \[ \Omega_0:=\Omega \cup \big\{(x,y)\in \mathbb{R}^2: (x,y)\in [0,1]\times \{0\}\big\}\text{ for }\varepsilon \to +0. \] The authors consider the eigenvalue problem \[ \begin{aligned} & \Delta^2 u -\tau \Delta u + u =\lambda u \text{ in } \Omega_\varepsilon,\\ & (1-\sigma) \partial^2_n u + \sigma \Delta u =0 \text{ on } \partial\Omega_\varepsilon,\\ & \tau \partial_n u -(1-\sigma) \operatorname{div}_{\partial\Omega_\varepsilon}(D^2u \cdot n)_{\partial\Omega_\varepsilon} - \partial_n \Delta u=0\text{ on }\partial\Omega_\varepsilon, \end{aligned} \] where \(\tau \geq 0\) and \(\sigma \in (-1,1)\) are fixed parameters. The solution of the problem is understood in a weak sense basing on a variational formulation of the problem. A necessity which follows from the non-smooth behavior of \(\partial\Omega_\varepsilon\). The main goal of the authors is the description of the eigenvalues and eigenfunctions of the above problem for \(\varepsilon \to +0\). Roughly speaking the authors prove the following statements: {\parindent=0.7cm\begin{itemize}\item[{\(\bullet\)}] Either \(\lambda_n(\Omega_\varepsilon)\) tends to \(\omega_k\) for some \(k \geq 1\), in which case the corresponding eigenfunctions converge in \(\Omega\) to the eigenfunctions associated with \(\omega_k\) of the boundary value problem \[ \begin{aligned} & \Delta^2 w -\tau \Delta w + w =\omega_k w \text{ in } \Omega,\\ & (1-\sigma) \partial^2_n w + \sigma \Delta w =0 \text{ on } \partial\Omega,\\ & \tau \partial_n w -(1-\sigma) \operatorname{div}_{\partial\Omega}(D^2 w \cdot n)_{\partial\Omega} - \partial_n \Delta w=0\text{ on } \partial\Omega, \end{aligned} \] \item[{\(\bullet\)}] or \(\lambda_n(\Omega_\varepsilon)\) tends to \(\theta_l\) for some \(l \in \mathbb{N}\), in which case the corresponding eigenfunctions behave in \(R_\varepsilon\) like the eigenfunctions associated with \(\theta_l\) of the boundary value problem \[ \begin{aligned} &\Delta^2 v -\tau \Delta v + v =\theta_l v\text{ in } R_\varepsilon,\\ & (1-\sigma) \partial^2_n v + \sigma \Delta v =0\text{ on } \partial R_\varepsilon \setminus L_\varepsilon,\\ & \tau \partial_n v -(1-\sigma) \operatorname{div}_{\Gamma_\varepsilon}(D^2v \cdot n)_{\Gamma_\varepsilon} - \partial_n \Delta v=0\text{ on } \partial R_\varepsilon\setminus L_\varepsilon,\\ & \partial_n v=v=0\text{ on }L_\varepsilon. \end{aligned} \] \end{itemize}}