No label defined (Q4342182)

The given paper deals with the number of non-real fixpoints of \(f\circ g\) or \(g\circ f\) for transcendental entire functions \(f\) and \(g\). The main result the author proves is that \(f\circ g\) or \(g\circ f\) has infinitely many fixpoints. In 1966 \textit{I. N. Baker} [Proc. Lond. Math. Soc., III. Ser. 16, 493-506 (1966; Zbl 0168.04101)] proved that for any transcendental entire function of order less than \({1\over 2}\) the second iterate of \(f\), i.e. \(f\circ f\), has infinitely many non-real fixpoints. Recently, the restriction on order in Baker's theorem has been removed by \textit{W. Bergweiler}, \textit{J. Clunie} and \textit{J. Langley} [Bull. Lond. Math. Soc. 27, No. 2, 148-154 (1995; Zbl 0822.30024)]. The arguments in the paper of Bergweiler et al. and also in the paper under review crucially depend on the use of some kind of Ahlfors covering theorem and on Wieman-Valiron theory as has been first used by Bergweiler in a paper on the existence of periodic points of entire functions [Complex Variables, Theory Appl. 17, No. 1/2, 57-72 (1991; Zbl 0748.30020)]. The key result to prove the above-mentioned main result of the given paper is some kind of extension of Wiman-Valiron-theory: An unbounded set \(E\subset(0,\infty)\) is called a Wiman-Valiron set for a given entire transcendental function \(f\) iff to \(z_0\) with \(|z_0|= r_0\in E\) and \(|f(z_0)|= M(r_0,f)\) (the maximum modulus) corresponds some \(k=k(z_0)\) such that \(f(ze^\tau)= e^{k\tau}f(z_0)(1+ w(\tau))\), where \(M(K/k,w)\to 0\) as \(r_0\to\infty\) in \(E\) for any fixed \(K>0\). Then the extension of Wiman-Valiron theory is given by (Theorem 2.2.1): Let \(f\) be a transcendental entire function such that \(\lim_{r\to\infty}\log M(r,f)/(\log r)^2=\infty\). Suppose that \(E\) is an unbounded set of \(r>0\) such that for some fixed \(\eta>1\), if \(r_0\in E\), then \(|f(z)|\geq 1\) \((r_0/\eta\leq|z|\leq\eta r_0)\). Then \(E\) is a Wiman-Valiron set for \(f\).