The Nth diameter problem in the class \(\Sigma\) (Q1095281)

For a continuum E in the plane by the Nth diameter \(d_ N(E)\) is meant the quantity \[ \max (\prod_{i\neq j}| w_ i-w_ j|)^{(N(N- 1))^{-1}}, \] where the maximum is taken over all sets of N points \(w_ 1,...,w_ N\) in E. The problem discussed in this paper is to determine the maximum of \(d_ N(E)\) when E is the complement of the image of \(| z| >1\) under a function in \(\Sigma\). The case \(N=2\) is just the diameter theorem. The case \(N=3\) was solved by \textit{G. Golusin} [Mat. Sb., Nov. Ser. 21(63), 83-115 (1947; Zbl 0044.305)] who showed that the (essentially unique) extremal configuration consists of three equal segments emanating from a point at equal angles. The reviewer [Math. Nachr. 39, 349-356 (1969; Zbl 0172.099)] proved the stronger result that this configuration maximizes also the sum of the distances. This result was rediscovered many years later by \textit{P. T. Mocanu} and \textit{D. Ripeanu} [Lect. Notes Math. 743, 323-330 (1979; Zbl 0419.30001)]. The case \(N=4\) was dealt with by \textit{G. V. Kuzmina} [Tr. Mat. Inst. Steklova 139, 240 p. (1980)]. A standard variational argument shows that an extremal configuration is the complement E of a domain on the sphere admissible with respect to the quadratic differential \[ Q(w)dw^ 2=-\sum_{i\neq j}[(w-w_ i)(w-w_ j)]^{-1}dw^ 2. \] The points \(w_ 1,...,w_ N\) are obviously Fekete points for E. A result of \textit{R. Kühnau} [J. Reine Angew. Math. 234, 216-220 (1969; Zbl 0185.328)] shows that all zeros of \(Q(w)dw^ 2\) lie in the convex hull of \(w_ 1,...,w_ N\). The authors show that all zeros of \(Q(w)dw^ 2\) as well as all trajectories with them as end points lie in E. They use a counting argument but actually this is an immediate consequence of the reviewer's basic structure theorem. The authors consider the following auxiliary coefficient result. Let E be an extremal configuration for the Nth diameter problem and let \(Q(z)dz^ 2\) have a zero of order k at a non-Fekete point of E. Let \({\mathcal E}_ 0\) consist of equal line segments in a subset of the directions of E joined at this zero and rotated to lie in the directions of the \((k+2)th\) roots of unity. Then for F meromorphic and univalent in the complement of \({\mathcal E}_ 0\) with Laurent expansion at infinity beginning with the term w, if log F'(w) has the expansion \(\sum^{\infty}_{\nu =2}C_{\nu}w^{-\nu}\) about infinity, \({\mathcal R} C_{k+2}\geq 0\). An analogous result is given for Fekete points. By constructing examples the authors use these results to show that for an extremal configuration all \(w_ j\) are actually simple poles of \(Q(w)dw^ 2\) and that \(Q(w)dw^ 2\) cannot have a zero of order \(k\geq 2\) if \(k+2\) is not prime. They conjecture that only simple zeros can occur. This has been affirmed in a latter paper of \textit{Y. J. Leung} [Complex Variables 9, 227-239 (1987)]. Finally the authors give some results on coefficients of Bloch functions.