Regularity of minimal surfaces with lower-dimensional obstacles (Q2209449)

In this paper the authors consider the classic Plateau problem with a thin obstacle. The main result is to show that optimal surfaces are \(C^{1,\frac{1}{2}-}\) regular near points on the free boundary of the contact set, in any dimension \(n\ge 2\). Moreover, as the authors pointed out, this regularity result is close to optimal, as any higher \(C^{1,\alpha}\) regularity with \(\alpha>1/2\) would run against the fact that solutions of the Signorini problem are only \(C^{1,\frac{1}{2}}\) regular. The Plateau problem involves finding the optimal surface \(E\) minimizing \[ \min \{ P(E;B_1): E\setminus B_1 = E_0 \setminus B_1 \}, \] where \(B_1\) is the unit ball, \(E_0\) is the boundary condition, and \(P(E;B_1)\) is the relative perimeter of \(E\) in \(B_1\). An ``obstacle'' is a requirement of the form \(E\supset \mathcal{O}\), where \(\mathcal{O}\) is a given set. Thus the Plateau problem with obstacle reads \[ \min \{ P(E;B_1):E\supset \mathcal{O},\ E\setminus B_1 = E_0 \setminus B_1 \}. \tag{1} \] A ``thin'' obstacle, as considered in this paper, is an obstacle \(\mathcal{O}\) of co-dimension 1. Such problem is the geometric version of the Signorini problem, and were studied by [\textit{E. De Giorgi}, Boll. Unione Mat. Ital., IV. Ser. 8, 80--88 (1973; Zbl 0289.49042)] in the 1970s, who established an existence result. However, regularity of optimal surfaces has been proven elusive, with some regularity results by \textit{M. De Acutis} [Rend. Semin. Mat. Univ. Padova 61, 133--144 (1979; Zbl 0514.49022)], \textit{D. Kinderlehrer} [Pac. J. Math. 37, 109--117 (1971; Zbl 0214.10503)], and \textit{E. Giusti} [Arch. Ration. Mech. Anal. 49, 41--56 (1972; Zbl 0257.49015)]. The key difficulty lies on the fact that near a typical point of the contact set the hypersurface \(\partial E\) consists of two surfaces that intersect transversally on \(\partial \mathcal{O}\), hence, after blow up, \(\partial E\) is not flat in general, and (1) cannot be considered as a perturbation of the Signorini problem. Also defining the solution to (1) is not trivial, as the relative perimeter does not see the obstacle. The authors overcome these issues by combining Savin's small perturbation method with a refined ``dichotomy approach'', which states that either \(\partial E\) is almost flat in \(B_1\), or it splits into two minimal surfaces that meet at an angle, and have full measure in a smaller ball. In the former case, the problem (1) is essentially a perturbation of the Signorini problem, for which regularity results are available. In the latter case, the authors rely on the \(C^{1,1}\) regularity for minimal surfaces, to improve the flatness of the two surfaces separately.