Construction of a universal Laurent series (Q2492084): Difference between revisions

Let \(\Omega \subset {\mathbf C}\) be a finitely connected domain, that is, there is a \(k \in \{0,1, \dots \}\) such that the complement \({\mathbf C}_\infty \setminus \Omega\) of \(\Omega\) in the extended complex plane \({\mathbf C}_\infty := {\mathbf C} \cup \{\infty \}\) has \(k\) components \(A_0,A_1, \dots ,A_k\). It may be assumed that \(\infty \in A_0\), and points \(a_j \in A_j\) \((j=1, \dots ,k)\) can be fixed. Then every function \(f \in H(\Omega )\) (= the space of holomorphic functions in \(\Omega\), endowed with the compact-open topology) has a unique representation of the form \(f = f_0 + f_1 + \cdots + f_k\) with \(f_j \in H(A_j^c)\) \((j=0,1, \dots ,k)\) and \(\lim_{z \to \infty} f_j(z) = 0\) \((j=1, \dots ,k)\). Each \(f_j\) \((j=1, \dots ,k)\) has a Laurent expansion with center \(a_j\) for \(| z - a_j| \) large enough, say \(f_j(z) = \sum_{n=1}^\infty c_n(f_j)/(z - a_j)^n\). If \(\zeta \in A_0^c\) and \(z \in {\mathbf C} \setminus \{a_1, \dots ,a_k\}\), then one defines \(M_N(f,\zeta )(z) = \sum_{n=0}^N f_0^{(n)}(\zeta ) (z - \zeta)^n + \sum_{n=1}^N c_n(f_1)/(z - a_1)^n + \cdots + \sum_{n=1}^N c_n(f_k)/(z - a_k)^n\). A function \(f \in H(\Omega )\) is called a \textit{universal Laurent series} (ULS), if for every compact set \(K \subset \Omega^c \setminus \{a_1, \dots ,a_k\}\) with \(K^c\) connected and for every function \(h:K \to {\mathbf C}\) continuous on \(K\) and holomorphic in \(K^0\), there exists a sequence \((m(n))\) of natural numbers such that, for every compact set \(\Gamma \subset A_0^c\), \[ \lim_{n \to \infty} \sup_{\zeta \in \Gamma} \sup_{z \in K} | M_{m(n)} (f,\zeta )(z) - h(z)| = 0. \] \quad Costakis, Nestoridis and Papadoperakis have recently proved that the class of ULS is \(G_\delta\) and dense in \(H(\Omega )\). Their result extend, in turn, previous results on universality related to overconvergence of Taylor series due to Luh, Melas and Nestoridis, among others. \vskip .1cm In the present paper, the authors give a constructive proof of the existence of a ULS. This allows to know additional information about these functions;\ for instance, they can be constructed as to have Ostrowski gaps. In fact, the function constructed by the authors belongs to a class that is even more restrictive than that of ULS.