Two disks maximize the third Robin eigenvalue: positive parameters
Hanna N. Kim, Richard S. Laugesen
TL;DR
This work proves a sharp upper bound for the third Robin eigenvalue $\\lambda_3$ on simply-connected planar domains of fixed area, showing it is maximized (in the degenerating limit) by two equal disks for all Robin parameters in $[-4\\pi,4\\pi]$ and extending the positive-range result $(0,4\\pi]$ via a degree-theoretic method. The authors build a four-parameter family of trial functions $u_{w,C}$ from disk eigenfunctions through conformal maps, hyperbolic caps, and fold/cap transformations, and then prove the existence of a trial function orthogonal to the first two eigenfunctions using a vanishing vector-field argument grounded in degree theory for reflection-symmetric maps on $\mathbb{S}^3$. A key technical contribution is a streamlined degree-theory lemma (special case of Karpukhin-Stern) to certify the required orthogonality, enabling extension of Girouard-Laugesen’s approach to positive Robin parameters. The result resolves an open problem and provides a robust framework for sharp spectral bounds via conformal and topological methods with potential applications to related eigenvalue optimization problems. The interplay of Möbius transformations, hyperbolic caps, and degree theory highlights a powerful blend of complex analysis, geometry, and topology in spectral optimization.
Abstract
The third eigenvalue of the Robin Laplacian on a simply-connected planar domain of given area is bounded above by the corresponding eigenvalue of a disjoint union of two equal disks, for Robin parameters in $[-4π,4π]$. This sharp inequality was known previously only for negative parameters in $[-4π,0]$, by Girouard and Laugesen. Their proof fails for positive Robin parameters because the second eigenfunction on a disk has non-monotonic radial part. This difficulty is overcome for parameters in $(0,4π]$ by means of a degree-theoretic approach suggested by Karpukhin and Stern that yields suitably orthogonal trial functions.
