The Dirichlet problem for \(H\)-systems with small boundary data: blowup phenomena and nonexistence results (Q2492911)

The authors consider the Dirichlet problem for the prescribed mean curvature equation \[ \begin{aligned} \Delta u &= 2 H(u) u_x\wedge u_y \quad\text{ in } {\mathbb D}^2,\\ u &= \gamma \quad\text{ on }\partial{\mathbb D}^2, \end{aligned} \] where \({\mathbb D}^2\) is the two-dimensional open unit disc, \(H:{\mathbb R}^3\to{\mathbb R}\) is a given bounded smooth function, and \(\gamma\in H^1({\mathbb D}^2,{\mathbb R}^3) \cap C^0(\partial{\mathbb D}^2,{\mathbb R}^3)\). Solutions of this problem can be found formally as critical points of the energy functional \[ {\mathcal E}_H(u) = \frac12 \int_{{\mathbb D}^2} | \nabla u| ^2 + 2 {\mathcal V}_H(u) , \] where \( {\mathcal V}_H(u) \) is the \(H\)-volume functional. It is conjectured that the Dirichlet problem should admit a pair of solutions: one solution that minimizes the energy functional and another higher energy solution obtained from mountain-pass geometry. Earlier work of \textit{M. Struwe} [Math. Ann. 270, 441--459 (1985; Zbl 0582.58010)] has shown that if \(H\) is a nonzero constant and if the functional \({\mathcal E}_H\) has a nonconstant local minimum \(\underline u\), then the mountain-pass level for \({\mathcal E}_H\) is a critical level and hence there exists a weak solution \(\overline u\) of the Dirichlet problem that is distinct from \(\underline u\). In this paper, the authors show that the conclusion of Struwe's theorem is false when \(H\) is nonconstant, even if \(H\) is very close to constant. Their argument is based on considering \(H\)-bubbles, by which they mean nonconstant bounded solutions of \[ \begin{aligned} &\Delta \omega = 2H(\omega) \omega_x\wedge \omega_y \text{\;on\;}{\mathbb R}^2,\\ &\int_{{\mathbb R}^2} | \nabla \omega| ^2 < \infty.\end{aligned} \] There is also an appropriate notion of energy associated with \(H\)-bubbles. The authors say that a sequence \(\{u^n\}\subset H^1({\mathbb D}^2,{\mathbb R}^3)\) blows an \(H\)-bubble \(\omega\) if there exist sequences \(\{\zeta_n\}\subset {\mathbb D}^2\) and \(\{\epsilon_n\}\subset (0,1)\) such that, for some subsequence \(n_i\), it holds that \(\epsilon_{n_i}\rightarrow 0\), \((1/\epsilon_{n_i})\,\text{dist}(\zeta_{n_i},\partial{\mathbb D}^2) \rightarrow \infty\), and \(u^{n_i}(\epsilon_{n_i}z + \zeta_{n_i})\rightarrow \omega(z)\) in \(H^1_{loc}({\mathbb R}^2,{\mathbb R}^3)\). A crucial result of the authors shows that, for any appropriately bounded \(H\in C^1({\mathbb R}^3)\), if the mountain-pass level for \({\mathcal E}_H\) is a critical value for each \(\gamma^n\) in a sequence of boundary data \(\gamma^n\) shrinking to zero in the \(H^1\cap C^0\)-topology, then a minimal \(H\)-bubble must exist. Finally, the authors construct a class of nonconstant curvature functions \(H\) for which there exists no minimal \(H\)-bubble.