English

Geometric properties of optimizers for the maximum gradient of the torsion function

Analysis of PDEs 2025-12-18 v2 Optimization and Control

Abstract

Consider J(Ω):=uΩ/ΩJ(\Omega):= \|\nabla u_\Omega\|_\infty/\sqrt{|\Omega|} and JP(Ω):=uΩ/P(Ω)J_P(\Omega):= \|\nabla u_\Omega\|_\infty/P(\Omega) , where Ω\Omega is a planar convex domain, uΩu_\Omega is the torsion function, P(Ω)P(\Omega) is the perimeter of Ω\Omega and Ω|\Omega| its area. We prove that there exist planar convex domains that maximize the functionals JJ and JPJ_P, and any maximizer has a C1C^1 boundary that contains a line segment on which uΩ|\nabla u_\Omega| attains its maximum.

Keywords

Cite

@article{arxiv.2512.09400,
  title  = {Geometric properties of optimizers for the maximum gradient of the torsion function},
  author = {Krzysztof Burdzy and Ilias Ftouhi and Phanuel Mariano},
  journal= {arXiv preprint arXiv:2512.09400},
  year   = {2025}
}