The Sobolev class in Equation (19) in the published version is off by
one; it should be $W^{k-1,p}$ maps from the boundary instead of
$W^{k,p}$. Without the correction, the assertion on the next page of
the Fredholm property fails.
Back to my homepage .