Towards a counter-example to a conjecture of De Giorgi in high dimensions (Q2505612): Difference between revisions

Let \(u\in C^2({\mathbb R}^n)\) be a solution of \[ \Delta u = u^3-u, \quad |u|\leq 1, \quad \partial_nu>0. \leqno(*) \] A well known conjecture of \textit{E. De Giorgi} [Proceedings of the international meeting on Recent methods in nonlinear analysis. Rome, 1978, Collana Atti di Congressi. Bologna: Pitagora Editrice, 131--188 (1979; Zbl 0405.49001)] states that all the level sets \(\{u=\lambda\}\) are hyperplanes, at least if \(n\leq 8\). This conjecture has been proved in dimension \(n=2\) by \textit{N. Ghoussoub} and \textit{C. Gui} [Math. Ann. 311, No. 3, 481--491 (1998; Zbl 0918.35046)] and by \textit{L. Ambrosio} and \textit{X. Cabré} [J. Am. Math. Soc. 13, No. 4, 725--739 (2000; Zbl 0968.35041)]. It is still open for \(n\geq 4\). Here the authors show the following. Suppose there is a solution \(v\in C^2({\mathbb R}^{n-1})\) of \(\Delta v=v^3-v\) which is also a global minimizer of the corresponding functional \(\int[{1\over 2}|\nabla w|^2+{1\over 4}(w^2-1)^2]\), and \(v\) is symmetric with respect to \(x_i\) for each \(i=1,\cdots,n-1\). Then there is a solution of \((*)\) such that for some \(\lambda\), the set \(\{u=\lambda\}\) is not a hyperplane. This is an analogue, in the context of semilinear equations, of the result of \textit{E. Bombieri, E. De Giorgi} and \textit{E. Giusti} [Invent. Math. 7, 243--268 (1969; Zbl 0183.25901)] on the existence of nonplanar minimal graphs in \({\mathbb R}^n\) for \(n\geq 9\). It is a partial one, however, because it relies on the existence of a symmetric global minimizer, which has not been proved.