Admissible convergence in Cartan-Hadamard manifolds. (Q5950488)

The classical Fatou convergence theorem for bounded holomorphic functions on the open unit disk was generalized by many authors for (locally) harmonic functions defined on non-compact globally symmetric spaces. The non-tangential approach domains to boundary points in the disc are essentially tubes of constant diameter around geodesic rays in the Poincaré metric. The admissible domains are such tubes which were introduced by the second author, and using these, the classical boundary convergence theorem of Fatou was generalized to non-compact globally symmetric spaces of rank one. The local Fatou theorem, i.e., the case of harmonic functions not defined on the whole space and the basic theory of harmonic \({\mathcal H}^p\) spaces was generalized to such symmetric spaces by the second author and Putz. Sullivan and Anderson independently considered complete, simply connected Riemannian manifolds with sectional curvature pinched between two negative constants. They showed that for any continuous function on the sphere at infinity the Dirichlet problem can be solved. This result was developed by Anderson and Schoen showing that, using tubes around geodesic rays as approach domains, Fatou's theorem is true in this class of Riemannian manifolds. Further results in this direction were obtained by \textit{A. Ancona} [Ann. Math. (2) 125, 495--536 (1987; Zbl 0652.31008)]. The local Fatou theorem was studied by Arai and Mouton. A large part of the \({\mathcal H}^p\) theory of harmonic functions was generalized by Arai to the class of complete, simply connected Riemannian manifolds with sectional curvature pinched between two negative constants. In connection with his work, two questions naturally arise. The first one is the converse of the result that the admissible domains which are near the boudary are contained in tubes by using a new definition of admissible ones, namely, whether his \({\mathcal H}^p\) spaces are the same as those defined by means of tubes. The second one is whether the hypothesis of some of Arai's results is always automatically satisfied since it is rather technical. In this article, the authors give an affirmative answers to both questions (proposition 1.3 and theorem 2.1). Further, the authors prove the equivalence of Arai's and the author's admissible domains with two other definitions involving a metric on the boundary (theorem 4).