Extension theorem and representation formula in non-axially-symmetric domains for slice regular functions (Q6057874)

This work is part of a recent development in the theory of \emph{slice-regular} quaternionic functions, initiated by Gentili and Struppa in 2006. This development concerns a known extension phenomenon and started with \textit{X. Dou} and \textit{G. Ren}'s paper [``Riemann slice-domains over quaternions I'', Preprint, arXiv:1808.06994 [math.CV] (2018). Let \(\mathbb{H}=\mathbb{R}+i\mathbb{R}+j\mathbb{R}+k\mathbb{R}\) denote the real algebra of quaternions and let \(\Omega\subseteq\mathbb{H}\). Under suitable hypotheses on \(\Omega\), slice-regular functions \(f:\Omega\to\mathbb{H}\) resemble holomorphic complex functions of one complex variable in many aspects, while exhibiting some peculiarities due to the non-commutative setting. The theory was initially developed in the case when \(\Omega\) is a (suitably defined) \emph{slice domain} and has axial symmetry with respect to the real axis \(\mathbb{R}\). Because of a known extension formula, for about a decade the symmetry hypothesis was incorrectly believed not to be restrictive. In other words, it was commonly believed that a regular function on a slice domain \(\Omega\) would automatically extend to the \emph{symmetric completion} \(\widetilde\Omega\) of \(\Omega\), i.e., to the smallest symmetric subset of \(\mathbb{H}\) including \(\Omega\). In 2018, Dou and Ren constructed in [loc. cit., \S2] a clever example of a slice domain \(\Omega\) and of regular function \(f:\Omega\to\mathbb{H}\) such that \(f\) cannot be extended to \(\widetilde\Omega\). This was the start of a new phase in the theory of regular functions, which focused on exploiting the extension formula while taking into account the subtleties revealed by this counterexample. In this new phase, some works in literature studied regular functions on slice domains without the symmetry hypothesis, proved a local extension theorem for these functions and established sufficient conditions on \(\Omega\) to guarantee global extension to \(\widetilde\Omega\). In the article under consideration, the authors choose a different approach. They drop the hypothesis that \(\Omega\) be an open subset of \(\mathbb{H}=\mathbb{R}^4\) and adopt the \emph{slice topology} defined in [loc. cit., Definition 3.1]. This topology is strictly finer than the Euclidean topology and also strictly finer than the \emph{\(\sigma\)-topology} (previously constructed in the context of power series expansions of regular functions). In a sense, it is the finest topology that allows a natural generalization of the concept of slice-regular function, performed in [loc. cit., Definition 4.1]. In this setup, the present work proves very comprehensive versions of some important properties of regular functions: the Identity Principle, the Representation Formula and the Extension Lemma. To prove the last two results, the authors generalize to the context of slice topology the notions of \emph{stem function} and \emph{slice function}, due to [\textit{R. Ghiloni} and \textit{A. Perotti}, Adv. Math. 226, No. 2, 1662--1691 (2011; Zbl 1217.30044)]. To prove the Representation Formula, the authors also use the concept of \emph{path-slice function}. The present work includes a variant of Dou and Ren's counterexample. It ends with a first study of the \emph{domains of slice regularity}, defined in analogy with the concept of domain of holomorphy of complex functions.