A small remark on Bernstein's theorem (Q6077859)

Let \(f_0(P):=\sqrt{1+|P|^2}\), \(P\in\mathbb{R}^2\). \textit{S. Bernstein} proved in [Math. Z. 26, 551--558 (1927; JFM 53.0670.01)], that if \(u=u(x)\), \(x=(x_1,x_2)\), is a smooth solution of the non-parametric minimal surface equation \(\mathrm{div}(D f_0(\nabla u))=\mathrm{div}(\nabla u/(\sqrt{1+|\nabla u|^2}))=0\) defined on the whole plane, then it must be an affine function; that is, \(u(x)=ax_1+bx_2+c\) for some \(a,b,c\in\mathbb{R}\). The authors investigate the question of replacing \(f_0\) with a splitting-type function \(f(P)=f_1(p_1)+f_2(p_2)\), \(P=(p_1,p_2)\in\mathbb{R}^2\), which is also of linear growth (the precise assumptions on \(f_1\) and \(f_2\) are ommited here). A function \(u\in C^1(\mathbb{R}^2)\), \(u=u(x_1,x_2)\), is called unbalanced if and only if both of the following two conditions hold: \[ \limsup_{|x|\rightarrow\infty}\frac{|\partial_iu(x)|}{1 + |\partial_ju(x)|}=\infty\qquad i,j=1,2,\quad i\neq j. \] The main theorem states that for balanced solutions, a result analogous to Bernstein's theorem is true. An elementary counterexample shows that without assumptions of this type on the solution, Bernstein's theorem cannot be carried over to the splitting case. Some modifications of the above theorem are also included.