Fusion lemma and boundary structure (Q1195883)

Let \(K_ 1\) and \(K_ 2\) be a pair of compacts in the complex plane. This pair of sets has the fusion property if there exists some positive number \(\alpha=\alpha(K_ 1,K_ 2)\) such that for every pair \(r_ 1\), \(r_ 2\) of rational functions and every compact set \(K\) there exists some rational function \(r\) with \[ | r(z)-r_ j(z)|\leq \alpha \sup\{| r_ 1(w)-r_ 2(w)|:\;w\in K\cup(K_ 1\cap K_ 2)\}\quad\text{for all }z\in K_ j\cup K,\;j=1,2. \] The well-known theorem by Alice Roth states that every pair of disjoint compacts in \(\mathbb{C}\) has the fusion property. D. Gaier posed the question whether the assumption \(K_ 1\cap K_ 2\) of the ``fusion lemma'' may be replaced by a weaker condition. Denote by \(K^ c\) the complement of \(K\) and by \(\text{int }K\) its interior. A pair of compacts \(K_ 1\), \(K_ 2\) is called normal if (1) \((K_ 1\cup K_ 2)^ c\) is connected, (2) \(\text{int }K_ 1\) and \(\text{int }K_ 2\) are simply connected domains. The author proves: Theorem. For any normal pair \(K_ 1,K_ 2\subset\mathbb{C}\) the following conditions are equivalent: (i) \(K_ 1\), \(K_ 2\) have the fusion property, (ii) \(\partial K_ 1\cup \partial K_ 2=\partial(K_ 1\cup K_ 2)\). A similar result was obtained by \textit{A. A. Nersesyan} [Sov. J. Contemp. Math. Anal., Arm. Acad. Sci. 23, No. 2, 35-47 (1988; Zbl 0663.30038)].