Extrapolation of the Dirichlet problem for elliptic equations with complex coefficients (Q785887): Difference between revisions

This paper is concerned with solvability of Dirichlet problems \((A_{ij}(x)u_j)_i +B_i(x)u_i = 0\) in \(\Omega\), \(u = f\) on \(\partial \Omega\). The coefficients and the unknown \(u\) are complex valued. The existence of \textit{energy solutions} for the above equation is controlled by the smoothness of the coefficients and that of the domain \(\Omega \in \mathbb{R}^n\). It is defined a notion of \textit{\(p\)-ellipticity} (for \(p>1\)) of the complex matrix valued function \(A:\Omega \to M_n(\mathbb{C})\), which for \(p=2\) reduces to the usual ellipticity condition for complex matrices. Let \(p_0\) be the supremum of those \(p>1\) for which \(A\) is \(p\)-elliptic. Regarding the domain \(\Omega\), which may be bounded or unbounded, it is required that it has the (somewhat technical) \textit{chord-arc} property. If in addition the vector valued coefficient \(B\) has a certain controlled growth behavior at the boundary of \(\Omega\), then if one can show that the \(L^q\) Dirichlet problem is solvable for some \(q \in (1, \frac{p_0(n-1)}{(n-2)})\) it follows from the main theorem that the \(L^p\) Dirichlet problem is solvable for all \(p\) in the interval \([q, \frac{p_0(n-1)}{(n-2)})\).