On Cauchy integrals of logarithmic potentials and their multipliers (Q2367819)

This may be viewed as a sequel to the paper `Multipliers of families of Cauchy-Stieltjes transforms, by \textit{R. A. Hibschweiler} and \textit{T. H. MacGregor} [Trans. Am. Math. Soc. 331, No. 1, 377-394 (1992; Zbl 0769.30029)]. Let \[ k_ \alpha(z)=(1-z)^{-\alpha}\quad(\alpha>0);\;k_ 0(z)=1+ \log 1/(1-z). \] The authors study the family \({\mathcal F}_ \alpha\) of analytic functions \(f(z)\) in the unit disk \(\Delta\) which may be represented as \[ f(z)= \int_{\partial \Delta}k_ \alpha(\overline xz)d \mu(x) \] for some probability measure \(\mu\); there is an obvious definition of norm. Here the case \(\alpha=0\) is given special scrutiny; it is also Möbius invariant. The `multipliers' of \({\mathcal F}_ \alpha\) are denoted \(M_ \alpha\). In the cited paper, the authors show that \(M_ \alpha\subset M_ \beta\) if \(0<\alpha<\beta\), but whether these containments are strict is not known. Here the authors show that \(M_ 0 \neq M_ \alpha\) when \(\alpha>0\) by producing a function in \(H^ 1 \backslash M_ 0\). This depends on the (known) fact that there are \(H^ 1\) functions which are rather unbounded on the positive axis. The proofs depend on careful analysis of the kernels involved and Banach space theory. There is a misprint on p. 615 in the formula for \(Sf\); the first term needs an \(n\) in the denominator.