On the Dirichlet problem for a quasi-linear elliptic equation (Q1103103)

For a bounded domain \(\Omega \subset {\mathbb{R}}^ n\), \(n\geq 2\), with \(C^ 2\)-smooth boundary \(\partial \Omega\) consider the Dirichlet problem \[ (D)\quad Lu\equiv -\sum^{n}_{i,j=1}D_ i(a_{ij}(x,u)D_ ju)+c(x)=f(x)\text{ in \(\Omega\) and \(u=\phi\) on } \partial \Omega \] with real-valued u and weak boundary data \(\phi \in L^ 2(\partial \Omega)\) and suppose: (1) L is uniformly elliptic on \({\bar \Omega}\times {\mathbb{R}};\) (2) \(a_{ij}=a_{ji}\), \(a_{ij}\in C^ 0({\bar \Omega}\times {\mathbb{R}})\), \(a_{ij}(\cdot,u)\in C^ 1({\bar \Omega})\) for every \(u\in {\mathbb{R}}\) and \(D_ ia_{ij}\in L^{\infty}(\Omega \times {\mathbb{R}})\), \(i,j=1,...,n\); moreover there exist \(A_{ij}\in C^ 1({\bar \Omega})\) with \(\lim_{| u| \to \infty}a_{ij}(x,u)=A_{ij}(x)\) and \(\lim_{| u| \to \infty}D_ ia_{ij}(x,u)=D_ iA_{ij}(x)\), \(x\in {\bar \Omega}\), \(i,j=1,...,n;\) (3) \(f\in L^ 2(\Omega)\), \(c\in L^ 2(\Omega)\), c(x)\(\geq 0\) a.e. on \(\Omega\). The essential thing here is that only \(L^ 2(\partial \Omega)\)-boundary values \(\phi\) are prescribed which in general may not be regarded as the trace of some \(u\in W^ 1_ 2(\Omega)\); this precludes application of well-known existence theorems for the Dirichlet problem (D) that are valid in the case u-\(\phi\in\overset\circ W^ 1_ 2(\Omega)\). As was pointed out by the author and \textit{H. B. Thompson} [Bull. Aust. Math. Soc. 27, 1-30 (1983; Zbl 0518.35031)] an appropriate setting for studying Dirichlet problems with \(L^ 2\)-boundary values is provided by a suitable weighted generalized Sobolev space \(\tilde W^ 1_ 2(\Omega)\). Having given a precise meaning to the statement that some \(u \in W^ 1_ 2(\Omega)\) attains the boundary values phi in \(L^ 2(\partial\Omega)\) on \(\partial\Omega\) in a certain weak sense the author proves existence of a weak solution \(u \in \tilde W^ 1_ 2(\Omega)\) to (D) by first establishing existence of solutions \(u_ m\in \tilde W^ 1_ 2(\Omega)\) to the Dirichlet problem with smooth boundary values \(\phi_ m \in C^ 1(\partial\Omega)\) [where \((\phi_ m)\) tends to \(\phi\) in \(\tilde L^ 2(\partial\Omega)]\) and then inferring the boundedness of the sequence \((u_ m)\) in \(\tilde W^ 1_ 2(\Omega).\) The proof involves an embedding theorem for generalized Sobolev spaces by \textit{R. D. Meyer} [J. Math. Mech. 16, 739-760 (1967; Zbl 0149.093)] and some results from Chabrowski-Thompson (loc. cit.) and the author [Rocky Mt. J. Math. 16, 223-236 (1986; Zbl 0622.35022)]. As the author points out himself his approach is based on an argument somewhat similar to the one given in \textit{P. Hess} [Indiana Univ. Math. J. 23, 827-829 (1974; Zbl 0259.35036)].