Determination of a universal series (Q428951): Difference between revisions

Let \(\Omega\) be a domain and \(\zeta_0 \in \Omega\). A holomorphic function \(F\) on \(\Omega\) is called a universal (Taylor) series if the sequence of partial sums \[ S_N(F,\zeta_0)(z) = \sum_{n=0}^N \frac{F^{(n)}(\zeta_0)}{n!}(z - \zeta_0)^n, \quad N \in \mathbb{N}, \] of the Taylor series expansion has the following property: For every compact set \(K \subset \mathbb{C}\) with connected complement and \(K \cap \Omega = \emptyset\), and every function \(h : K \to \mathbb{C}\) continuous on \(K\) and holomorphic in the interior of \(K\), there exists a subsequence \(\lambda_n \subseteq \mathbb{N}_0\) such that \[ S_{\lambda_n}(F,\zeta_0)(z) \to h(z) \; \text{ uniformly on \(K\), as \(n \to \infty\).} \] Such universal series exist on simply connected domains and on some unbounded domains of arbitrary connectivity. By a careful examination of the proofs, amongst others of Runge's theorem on rational approximation, the authors show, that every Taylor coefficient of such a universal \(F\) may be determined by a finite program if \(\Omega\) has a nice structure, like a circle, a polygon, an open half plane, and so on. After a finite number of steps, only a polynomial is obtained, which -- as any polynomial -- can be completed to a power series and becomes universal or non-universal. To determine a specific universal Taylor series, all the Taylor coefficients are needed. Therefore, the authors speak of an infinite denumerable procedure.