The Lindelöf principle in several complex variables (Q865334)

The most classical formulation of the Lindelöf principle on the disc is as follows: Let \(f\) be a bounded, holomorphic function on the unit disc \(\mathbb D\subseteq\mathbb C\). Suppose that the radial limit \[ \lim_{r\to1-0} f(re^{i\theta})\equiv\lambda\in\mathbb C \tag{1} \] exists at the boundary point \(e^{i\theta}\). Then in fact \(f\) has nontangential limit \(\lambda\) at \(e^{i\theta}\). Thus we have a sort of Tauberian theorem: for bounded holomorphic functions, radial convergence implies nontangential convergence. It is of interest to have a result of this nature in several complex variables. Pioneering work on the Lindelöf principle in several complex variables was done by E. Chirka. Then Cima and Krantz established a Lindelöf principle for holomorphic functions of several variables [\textit{J. A. Cima} and \textit{S. G. Krantz} [Duke Math. J. 50, 303--328 (1983; Zbl 0522.32003)]. That result was new and optimal in the following senses: (i) It was proved for normal functions, in a sense the most natural function space for which to consider a Lindelöf principle. (ii) It was formulated in terms of the Kobayashi metric, thus providing an ``optimal'' result in terms of the intrinsic Levi geometry of the domain. But the result of Cima-Krantz has certain drawbacks. Notable among these is that, whereas the most natural mode of boundary convergence in the several variable setting is admissible convergence, that used in the Cima-Krantz results is hypoadmissible convergence -- a strictly weaker concept. The Cima-Krantz result is in fact false for admissible convergence. In addition, when the radial curve, as in (1), is replaced by a fairly arbitrary curve -- especially by a curve with a significant complex tangential component -- then a rather unsatisfying result is obtained. In the present paper, the author introduces new techniques that address the shortcomings of the Cima-Krantz result and produce a sharp version of the Lindelöf principle. Inspired by the classical result, as improved later by Lehto and Virtanen, this new result meshes closely with the well-established Fatou theorems of Koranyi and Stein. In particular, this is a Lindelöf principle for admissible approach regions. Let \(\mathbb B^n\) be the unit ball in \(\mathbb C^n\). For \(P=(p_1,p_2,\dots,p_n)\in\partial\mathbb B^n\) and \(\alpha>1\), let the admissible approach region of aperture \(\alpha\) at \(P\) be \(\mathcal A_\alpha(P) =\left\{z\in\mathbb B^n: | 1-z\cdot \bar P| <\alpha(1-| z| )\right\}.\) Here \(z\cdot \bar P = \sum_jz_j\bar P_j\). This region is of nontangential shape in the complex normal direction at \(P\) and of parabolic shape in the complementary, complex tangential direction. The author formulates a more sophisticated version of the Lindelöf principle on the ball as follows: Let \(f\) be a bounded, holomorphic function on the unit ball \(\mathbb B^2\). Let \[ T = \left\{(s +i0,t + i0): s, t\in\mathbb R, 0<s <1, 0 < | t| < \sqrt{2 -2s}\right\}. \] Suppose that \(\rho : T\to\mathbb R^2\) is a \(C^2\) function, with bounded first and second derivatives, such that (writing \(\rho(s, t) = (\rho_1(s, t), \rho_2(s, t))\)) \[ \mathcal T = \left\{(s + i\rho_1(s, t), t + i\rho_2(s, t)): (s,t)\in T\right\} \] is a two-dimensional, totally real manifold in \(\mathbb B^2\). Let \({\mathbf 1} = (1 + i0, 0 + i0)\). Suppose that \(\lim_{\mathcal T\ni z\to\mathbf1}\! f(z)\!\equiv\!\lambda\in\mathbb C\) exists. Then, for any \(\alpha\!>\!1\), \(\lim_{\mathcal A_\alpha({\mathbf 1})\ni z\to{\mathbf 1}} f(z) =\lambda.\) The author further adapts the new principle to the Levi geometry of various domains. In common with the Cima-Krantz result the author proves the result not only for bounded holomorphic functions but also for normal functions. The results in this paper improve on earlier results of Chirka, Cima-Krantz, Abate, and Abate-Tauraso.