Range of the first three eigenvalues of the planar Dirichlet Laplacian
Michael Levitin, Rustem Yagudin
TL;DR
The paper addresses the problem of determining the admissible region for the first three Dirichlet Laplacian eigenvalue ratios on planar domains, namely $\left(\frac{\lambda_2}{\lambda_1}, \frac{\lambda_3}{\lambda_1}\right)$. It combines a survey of universal and isoperimetric bounds with extensive numerical experiments across diverse domain classes and employs perturbation theory to infer structural properties of potential maximizers. A key contribution is the identification that the maximal ratio $\frac{\lambda_3}{\lambda_1}$ among tested domains is $Y^*\approx 3.202$, attained by a dumbbell-shaped domain with $\lambda_3\approx\lambda_4$, and that the rectangle $R_{\sqrt{\frac{8}{3}}}$ is not the global maximizer; the work also establishes that a local maximizer would have to satisfy $\lambda_3=\lambda_4$ under plausible assumptions. These results narrow the search for spectral-maximizers in planar domains and suggest that the optimal shapes are simply connected with smooth boundaries near dumbbell geometries, with implications for spectral-geometry and shape-optimization problems.
Abstract
We conduct extensive numerical experiments aimed at finding the admissible range of the ratios of the first three eigenvalues of a planar Dirichlet Laplacian. The results improve the previously known theoretical estimates of M Ashbaugh and R Benguria. We also prove some properties of a maximizer of the ratio $λ_3/λ_1$.
