The anisotropic Bernstein problem (Q6139247)

Let \(\Phi\) be a uniformly elliptic integrand, that is, a one-homogeneous function on \(\mathbb{R}^{n+1}\) positive and smooth on \(\mathbb{S}^n\), and such that the set \(\{\Phi <1\}\) is uniformly convex. The anisotropic area functional of an oriented hypersurface \(\Sigma\) is \[ A_\Phi(\Sigma) = \int_\Sigma \Phi(\nu) \mathrm{d}A, \] being \(\nu\) the unit normal to \(\Sigma\). The anisotropic Bernstein problem asks if it holds true that critical points of the anisotropic area functional that are graphs of entire functions (i.e., defined on \(\mathbb{R}^N\)) are hyperplanes. The isotropic case (i.e., \(\Phi(x) = \|x\|\)), was solved in a series of breakthrough papers by Bernstein, Fleming, De Giorgi, Almgren and Bombieri-De Giorgi-Giusti, proving this to be true if and only if \(n\le 7\). For general choices of \(\Phi\) this fact has been proved true for \(n=2,3\) [\textit{H. B. Jenkins}, Arch. Ration. Mech. Anal. 8, 181--206 (1961; Zbl 0143.14804)], [\textit{L. Simon}, Math. Z. 154, 265--273 (1977; Zbl 0388.49026)] while false for \(n\ge 6\) [\textit{C. Mooney}, J. Eur. Math. Soc. (JEMS) 24, No. 12, 4353--4361 (2022; Zbl 1505.35232)]. The paper under consideration concludes the investigation of the problem showing this fact to be false for \(n\ge 4\) (see Theorem~1.1). The proofs rely on the maximum principle and on exploiting the fact that constructing entire graphical minimizers in \(\mathbb{R}^{n+1}\) is closely linked to the existence of singular minimizers in \(\mathbb{R}^n\).