From universality to generic non-extendability and total unboundedness in spaces of holomorphic functions (Q2318072)

For a domain \(\Omega \subset \mathbb{C}^d\), let \(H(\Omega)\) denote the space of holomorphic functions \(f:\Omega \to \mathbb{C}\). One definition of \(\Omega\) being a \textit{domain of holomorphy} is that there is \(f \in H(\Omega)\) which cannot be extended. An equivalent notion is that \(\Omega\) is a \textit{weak domain of holomorphy}, that is for every pair of balls \(B, b \subset \mathbb{C}^d\) such that \(\overline{b} \subset B \cap \Omega\) and \(B \cap \Omega^c \neq \emptyset\), there is \(h \in H(\Omega)\) whose restriction to \(b\) does not admit a bounded holomorphic extension to \(B\). The author's interest here is the extent to which these two notions are equivalent when the space \(H(\Omega)\) is replaced by a subspace \(X(\Omega)\). (The definitions of (weak) \(X(\Omega)\) domain of holomorphy are, as expected, the same as the corresponding cases for (weak) domain of holomorphy, the only difference being that the required function \(h\) must now be in \(X(\Omega)\).) It is noted that for the domain of holomorphy \(\Omega = D \backslash [0,1) \subset \mathbb{C}\), \(\Omega\) is not an \(X(\Omega) = A(\Omega)\) domain of holomorphy. It is essentially shown that the two notions of domain of holomorphy relative to \(X(\Omega)\) are equivalent provided that \(X(\Omega)\) is a complete topological vector space having a metric that is at least as strong as the compact-open topology. Similar results are proved for \textit{totally unbounded} functions in \(H(\Omega)\) (where \(f\) is said to be totally unbounded provided that for every ball \(B(\zeta)\) centered at a point of \(\partial \Omega\) and for every component \(Y \subset B(\zeta) \cap \Omega\), the restriction of \(f\) to \(Y\) is unbounded). A discussion is included to show how from universality one is naturally led to nonextendability and to the stronger notion of total unboundedness, as generic properties in spaces of holomorphic functions. As noted, similar results are known in the infinite dimensional setting. See, e.g., [the reviewer et al., ``Extendability and domain of holomorphy on infinite dimensional spaces'', Annal. Polon. Math (to appear)]. Reviewer's comment: In fact, some of the material in this paper is contained in work of \textit{T. R. Alves} [Stud. Math. 220, No. 2, 157--167 (2014; Zbl 1307.46036)] and also of \textit{L. Bernal-González} et al. [Linear Algebra Appl. 513, 149--159 (2017; Zbl 1364.46040)].