Nonlocal minimal graphs in the plane are generically sticky (Q2187285): Difference between revisions

The authors consider nonlocal minimal surfaces, first introduced in [\textit{L. Caffarelli} et al., Commun. Pure Appl. Math. 63, No. 9, 1111--1144 (2010; Zbl 1248.53009)], i.e. surfaces minimizing the fractional perimeter inside a, possibly unbounded, given set \(\Omega\). They consider graphical solutions, which have been deeply studied in [\textit{S. Dipierro} et al., Calc. Var. Partial Differ. Equ. 55, No. 4, Paper No. 86, 25 p. (2016; Zbl 1354.49088)] where it was shown that any minimal surface which is a graph outside a cylinder \(\Omega=B\times \mathbb{R}\), with \(B\subset \mathbb{R}^{n-1}\) smooth and bounded, it is a graph over the entire space. In such a setting it is known that solutions may possess discontinuities at the boundary of the cylinder and in order to obtain the boundary datum they need to stick in a vertical manner to the cylinder's walls. Such a phenomenon is known as ``stickiness''. In this paper, the authors restrict themselves to consider graphical solutions in \(\mathbb{R}^2\), i.e. to \(B\subset \mathbb{R}\). They prove that either a minimal graph is discontinuous at the boundary, thus producing the stickiness effect, or it is differentiable with Hölder continuous derivative, see Theorem~1.2. Employing this result they then show in Theorem~1.1 that the stickiness property is rather general for minimal graphs, in the sense that continuous solutions (which by Theorem~1.2 are actually differentiable with Hölder continuous derivative) are unstable: there always exists an arbitrarily small perturbation of the exterior datum which yields solutions which are discontinuous at the boundary.