Convolution of operators and applications (Q1108543)

The main results of this paper are characterizations of the Radon-Nikodym property and the analytic Radon-Nikodym property by means of convolution between functions and a special kind of operators, viz. the cone absolutely summing operators. Let \(L^ p(\) \(1<p<\infty)\) be the \(L_ p\)-space on the unit circle with normalized Lebesgue measure and B a Banach space. A bounded linear operator T from \(L_ p\) into B is called a positive 1-summing operator \((T\in \Lambda_ 1(L^ p,B))\) if there exists a constant \(C>0\) such that for every finite family \(\psi_ 1,...,\psi_ n\) of positive functions in \(L^ p\) the operator T satisfies \[ \sum^{n}_{i=1}\| T\phi_ i\| \leq C\sup \{\sum^{n}_{i=1}| \int \psi_ i(t)\phi (t)dt|:\| \phi \|_ q=1\} \] (with q the conjugate index of p). The author proves the following equivalences. Theorem 1. If \(\{g_ n\}^{\infty}_{n=1}\) is an approximate identity of positive \(L_ p\)-functions, then the following are equivalent. (i) B has the Radon-Nikodym property (i.e., every bounded linear operator from \(L^ 1\) into B is representable by a function in \(L^ 1(B))\) (ii) for every \(T\in \Lambda_ 1(L^ p,B))\) the convolutions \(g_ n*T\) converge to T in \(\Lambda_ 1(L^ p,B)).\) Theorem 2. If \(e_ n(t)=e^{-int}(n\in {\mathbb{Z}})\), then the following are equivalent. (i) B has the analytic Radon-Nikodym property (i.e., every B-valued bounded holomorphic function on the disc has limits at the boundary a.e.) (ii) for every \(T\in \Lambda_ 1(L^ p,B)\) with \(T(e_ n)=0\) for \(n<0\), the convolutions \(P_{r(n)}*T\) converge to T in \(\Lambda_ 1(L^ p,B)\) (where \(P_{r(n)}\) denote the Poisson kernels on the disc of a sequence r(n)\(\in (0,1)\) with r(n)\(\to 1)\).