We continue the development, by reduction to a first-order system for the conormal
gradient, of
a priori estimates and solvability for boundary value problems of Dirichlet,
regularity, Neumann type for divergence-form second-order complex elliptic systems.
We work here on the unit ball and more generally its bi-Lipschitz images, assuming a
Carleson condition as introduced by Dahlberg which measures the discrepancy of the
coefficients to their boundary trace near the boundary. We sharpen our estimates by
proving a general result concerning
a priori almost everywhere nontangential
convergence at the boundary. Also, compactness of the boundary yields more
solvability results using Fredholm theory. Comparison between classes of solutions
and uniqueness issues are discussed. As a consequence, we are able to solve a long
standing regularity problem for real equations, which may not be true on
the upper half-space, justifying
a posteriori a separate work on bounded
domains.
Keywords
elliptic system, conjugate function, maximal regularity,
Dirichlet and Neumann problems, square function,
nontangential maximal function, functional and operational
calculus, Fredholm theory