A distributional approach to nonlocal curvature flows
Filippo Cagnetti, Massimiliano Morini, Dario Reggiani
TL;DR
This work develops a nonlocal, distributional framework for curvature-driven flows by extending the distance-function approach to both Minkowski-type and $s$-fractional perimeters. It introduces nonlocal Cahn–Hoffman fields and a nonlocal divergence operator to formulate weak and level-set flows, and it proves key results including a comparison principle, the convergence of Almgren–Taylor–Wang minimizing movements, and global well-posedness for the nonlocal flows. The paper provides a unified treatment that covers crystalline-like and fractional-perimeter motions, and it demonstrates convergence to classical mean curvature flow as the nonlocal scale vanishes, with extensions to anisotropic variants. These contributions offer a robust analytic framework for nonlocal geometric evolutions applicable to image processing, materials science, and related fields, and they connect nonlocal models with established viscosity solutions in the fractional setting.
Abstract
In \cite{CMP17} a novel distributional approach has been introduced to provide a well-posed formulation of a class of crystalline mean curvature flows. In this paper, such an approach is extended to the nonlocal setting. Applications include the fractional mean curvature flow and the Minkowski flow; i.e., the geometric flow generated by the $(N-1)$-dimensional Minkowski pre-content.
