Steklov eigenvalues for the \(\infty\)-Laplacian (Q996935): Difference between revisions

The very interesting paper under review deals with the Steklov eigenvalue problem for the \(\infty\)-Laplacian operator \[ \Delta_\infty u=\sum_{i,j=1}^N \,\frac{\partial u}{\partial x_j} \frac{\partial^2 u}{\partial x_j\,\partial x_i} \frac{\partial u}{\partial x_i}\quad \text{(in viscosity sense)} \] taken as \(\lim_{p\to\infty}\Delta_p\) of the \(p\)-Laplacian \(\Delta_pu=\text{div\,}(|\nabla u|^{p-2}\nabla u).\) Precisely, consider the problem \[ \leqno{(1)}\qquad\qquad\qquad \begin{cases} -\Delta_pu=0 & \text{ in}\;\Omega,\\ |\nabla u|^{p-2}\,\frac{\partial u}{\partial\nu}=\lambda|u|^{p-2}u& \text{ on}\;\partial\Omega \end{cases} \] where \(\Omega\) is a bounded and smooth domain of \(\mathbb R^N\) and \(\nu\) stands for the outer normal to the boundary. Let \(\lambda_{k,p}\) be the \(k\)-th variational eigenvalue of \((1)\) as defined in \textit{J. Fernández Bonder}and \textit{J. D. Rossi} [J. Math. Anal. Appl. 263, No. 1, 195--223 (2001; Zbl 1013.35038)]. The main result of the paper justifies the passage to the limit \(\lim_{p\to\infty}\lambda_{k,p}.\) The following result is proved. For the first eigenvalue of \((1)\) it holds \[ \lim_{p\to\infty}\lambda_{1,p}^{1/p}=\lambda_{1,\infty}=0 \] with eigenfunction given by \(u_{1,\infty}=1.\) The second eigenvalue satisfies \[ \lim_{p\to\infty}\lambda_{2,p}^{1/p}=\lambda_{2,\infty}=\frac{2}{\text{diam}(\Omega)}. \] Moreover, given eigenfunctions \(u_{2,p}\) of \((1)\) with eigenvalues \(\lambda_{2,p}\) and normalized by \(\|u_{2,p}\|_{L^\infty(\partial\Omega)}=1,\) there exists a sequence \(p_i\to\infty\) such that \(u_{2,p_i}\to u_{2,\infty}\) in \(C^\alpha(\overline\Omega).\) The limit \(u_{2,\infty}\) is a solution of \[ \leqno{(2)}\qquad\qquad\qquad \begin{cases} \Delta_\infty u=0 & \text{ in}\;\Omega,\\ \Lambda(x,u,\nabla u)=0& \text{ on}\;\partial\Omega \end{cases} \] in the viscosity sense, where \[ \Lambda(x,u,\nabla u)=\begin{cases} \min\{|\nabla u|-\lambda_{2,\infty}|u|,\partial u/\partial\nu\} & \text{if}\;u>0,\\ \max\{\lambda_{2,\infty}|u|-|\nabla u|,\partial u/\partial\nu\} & \text{if}\;u<0,\\ \partial u/\partial\nu & \text{if}\;u=0. \end{cases} \] For the \(k\)-th variational eigenvalue \(\lambda_{k,p}\) of \((1)\) with eigenfunction \(u_{k,p}\) normalized by \(\|u_{k,p}\|_{L^\infty(\partial\Omega)}=1,\) every sequence tending to infinity has a subsequence \(p_i\) such that \[ \lim_{i\to\infty}\lambda_{k,p_i}^{1/p_i}=\lambda_{\ast,\infty} \] and \(u_{k,p_i}\to u_{\ast,\infty}\) in \(C^\alpha(\overline\Omega)\) where \((u_{\ast,\infty},\lambda_{\ast,\infty})\) is a solution of (2).