Conformal optimization of eigenvalues on surfaces with symmetries
Denis Vinokurov
TL;DR
The paper develops a unified framework for conformal optimization of Laplace- and Steklov-eigenvalues on surfaces with finite-group symmetries, proving existence and regularity of Γ-invariant maximizers that arise from Γ-equivariant harmonic maps (or free-boundary harmonic maps). It provides a streamlined bubbling-tree approach in 2D that reduces maximizing metrics to a finite combination of harmonic maps and bubbles, and derives sharp upper and lower bounds for the eigenvalue functionals. The results yield complete, explicit classifications of equivariant maxima on the sphere and disk for all finite Γ, resolving open questions about the sharpness of the Hersch–Payne–Schiffer inequality and the maximization of Steklov eigenvalues by the standard disk under symmetry. Significantly, the work both simplifies prior methods in conformal eigenvalue optimization and delivers concrete configurations (disjoint unions of spheres or disks) that realize the maxima in symmetric settings. This advances understanding of optimal metrics under symmetry constraints and deepens the connection between spectral optimization and harmonic map theory in two dimensions.
Abstract
Given a conformal action of a discrete group on a Riemann surface, we study the maximization of Laplace and Steklov eigenvalues within a conformal class, considering metrics invariant under the group action. We establish natural conditions for the existence and regularity of maximizers. Our method simplifies previously known techniques for proving existence and regularity results in conformal class optimization. Finally, we provide a complete solution to the equivariant maximization problem for Laplace eigenvalues on the sphere and Steklov eigenvalues on the disk, resolving open questions posed by Arias-Marco et al. (2024) regarding the sharpness of the Hersch-Payne-Schiffer inequality and the maximization of Steklov eigenvalues by the standard disk among planar simply connected domains with $n\text{-rotational}$ symmetry.
