Optimisation of Steklov transmission eigenvalues and minimal surfaces

Abstract

In the present paper, we study the variational properties of Steklov transmission eigenvalues, which can be seen as eigenvalues of the sum of two Dirichlet-to-Neumann operators on two different sides of a given curve contained in a surface. Inspired by the analogous results for Laplacian and Steklov eigenvalues, we show that critical metrics for this problem correspond to the so-called stationary configurations in the Euclidean ball, i.e. pieces of minimal surfaces inside the ball whose normals at the boundary sum up to a vector normal to the boundary sphere. They exhibit strong similarities with free boundary minimal surfaces, and for that reason we call them free curve minimal surfaces. Furthermore, we study the maximisation problem for these eigenvalues in two contexts. First, in the unconstrained setting, we show that there are no smooth maximal metrics. Second, for the first nontrivial eigenvalue, we explicitly characterize rotationally symmetric maximal metrics on the sphere as those arising from the configurations of stacked catenoids with flat caps.