Tangential boundary behavior of Green potentials and contractive properties of \(L^ p\)-capacities (Q1821905)

Let \(H=\{x\in R^ n\); \(x_ n>0\}\), \(n\geq 2\), be the upper half space and G the Green function for H. The tangential behavior of Green potentials \(u(x)=\int_{H}G(x,y)\lambda (y)dy\) is considered under an \(L^ p\)-condition on \(\lambda\). Let \(P=\{x\); \(x_ n=0\}\), \(Q=\{x\); \(x_ n=1\}\) and \(x'=(x_ 1,...,x_{n-1})\). For \(\gamma\geq 1\), \(a\in P\) and \(b\in Q\), we denote by \(\Gamma\) (\(\gamma\),a,b) the arc in H joining b to a with tangency \(\gamma\) to the plane P so that if \(x\in \Gamma (\gamma,a,b)\), then x'-\(a'=x_ n^{1/\gamma}(b'-a')\). Let k be a nonnegative nonincreasing lower semicontinuous function on [0, \(+\infty)\). We define an \(L^ p\)-capacity for k as follows: \[ C_{k,p}(E)=\inf \{\| f\|^ p_ p;\quad \int k(| x- y|)f(y)dy\geq 1\quad on\quad E,\quad f\geq 0\}. \] Then the Bessel capacity \(B_{\alpha,p}\) with index (\(\alpha\),p) is defined by \(B_{\alpha,p}=C_{g_{\alpha},p}\), where \(g_{\alpha}\) is the Bessel kernel of order \(\alpha\). Let \(1\leq p\leq n/2\), \(2p-n<\beta \leq 2p-1\) and \(1\leq \gamma \leq (n-1)/(\beta -2p+n)\). Then it is shown that if \(u\not\equiv \infty\) and \(\int_{H}\lambda (y)^ p y_ n^{\beta} dy<\infty\), then there is a set \(V_{\gamma}\subset P\) with \((\beta - 2p+n)\gamma -\) dimensional Hausdorff measure zero, such that for each \(a\in P\setminus V_{\gamma}\) the set \(E_ a=\{b\in Q\); u(x) does not approach zero as \(x\to a\) along \(\Gamma\) (\(\gamma\),a,b)\(\}\) satisfies \(B_{2,\alpha}(E_ a)=0.\) This measurement of the exceptional set \(E_ a\), which is an improvement of \textit{J. M. G. Wu}'s Theorem 1 (i) [Indiana Univ. Math. J. 28, 895-911 (1979; Zbl 0449.31003)], is closely related to contractive properties of \(L^ p\)-capacities. Let T be a contraction mapping from \(R^ n\) to \(R^ n\) of the form \(Tx=(T_ 1x_ 1,...,T_ nx_ n)\), where \(T_ i\) (1\(\leq i\leq n)\) is a contraction mapping from R to R. Then it is proved that \(C_{k,p}(TE)\leq C_{k,p}(E)\) for all \(E\subset R^ n\). However, whether \(C_{k,p}(TE)\leq C_{k,p}(E)\) holds or not for a general contraction mapping T seems to be unknown.