Hessian equations in non-smooth domains (Q4700440)

This paper is concerned with the Dirichlet problem for \(k\)-Hessian equations \[ S_k(D^2u)=f\geq 0 \quad\text{in}\quad\Omega, \qquad u=h \quad\text{on}\quad\partial\Omega, \tag{\(*\)} \] where \(\Omega\) is a bounded domain in \({\mathbb R}^n\) and \(S_k(D^2u)\) denotes the \(k\)-th elementary symmetric function of the eigenvalues of \(D^2u\), \[ S_k(D^2u) = \sum_{1\leq i_1<\cdots <i_k \leq n} \lambda_{i_1}\cdots\lambda_{i_k}. \] Equations of this kind have been studied by many authors. \textit{L.~Caffarelli, L.~Nirenberg} and \textit{J.~Spruck} [Acta Math. 155, 261-301 (1985; Zbl 0654.35031)] showed that \((*)\) has a unique smooth \(k\)-convex solution if \(f,h,\partial\Omega\) are smooth with \(f>0\) in \(\overline\Omega\) and \(\Omega\) is uniformly \((k-1)\)-convex. Here \(u\) \(k\)-convex means that \(S_j(D^2u)\geq 0\) for each \(j=1,\dots,k\), while \(\Omega\) (uniformly) \((k-1)\)-convex means that at each point of \(\partial\Omega\) the \(m\)-th elementary symmetric function of the principal curvatures of \(\partial\Omega\) is nonnegative (positive) for \(m=1,\dots,k-1\). \textit{N. S.~Trudinger} [Commun. Partial Differ. Equ. 22, No. 7-8, 1251-1261 (1997; Zbl 0883.35035)] proved the existence and uniqueness of weak solutions of \((*)\) under the assumptions that \(f\in L^p(\Omega)\) for some \(p>n/2k\), \(h\in C(\partial\Omega)\), and \(\partial\Omega\in C^2\) is uniformly \((k-1)\)-convex. Here the authors prove the existence of viscosity solutions of \((*)\) under weaker regularity assumptions on \(\partial\Omega\). They define a weak notion of \(j\)-convexity for general domains in \({\mathbb R}^n\) which, roughly speaking, is that for every \(x\in\partial\Omega\) there is a \(j\)-convex (in the classical sense) domain touching \(\Omega\) at \(x\) from the outside. This definition is clearly very much in the spirit of viscosity theory. They then prove the existence of a unique viscosity solution of \((*)\) if \(h\) is constant, \(f\in C(\overline\Omega)\) and \(\Omega\) is uniformly \((k-1)\)-convex in the weak sense. Arbitrary \(h\in C(\partial\Omega)\) is permitted if \(\Omega\) is uniformly \((n-1)\)-convex in the weak sense.