scientific article; zbMATH DE number 1054236

From MaRDI portal
Publication:4351899





Let \(\mathbb D\) be the unit disc in \(\mathbb C, \mathbb T :=\partial\mathbb D\) its boundary and \(H^p(\mathbb T), p\geq1\) the classical Hardy spaces of holomorphic functions in \(\mathbb D\) with boundary values in \(L^p(\mathbb T)\). The starting point of this work is the following well known theorem of Beurling [\textit{W. Rudin}, Real and Complex Analysis, McGraw-Hill (1966; Zbl 0142.01701)]: Let \(E\) be a closed subspace of the Hardy space \(H^2(\mathbb T),\) invariant by multiplication by \(z,\) then there is an inner function \(\mathbf\Theta\in H^\infty(\mathbb T)\) such that: \(E = \mathbf\Theta\cdot H^2(\mathbb T)\). The notions of inner and outer functions are very important in one variable and the author's aim is to define analogous notions in several variables via the invariant subspaces by multiplication by polynomials. The author works in the following context: \(\Omega\) is a bounded pseudo-convex domain in \(\mathbb C^n\) smoothly bounded; \(\mathcal A(\Omega) :=\mathcal O(\Omega)\cap\mathcal C^\infty(\overline\Omega)\); \(g=(g_1,\dots,g_k)\in \mathcal A(\Omega)^k\); \(\mu\) is a probability measure on \(\overline\Omega\); and finally \(H^p(\mu)\) denotes the closure in \(L^p(\mu)\) of \(\mathcal A(\Omega)\). The main question treated here is: what conditions on the vector valued function \(g\in \mathcal A(\Omega)^k\) imply that \(g\cdot H^p(\mu)\) is dense in \(H^p(\mu)\)? If this is the case we may say that the ideal generated by the \(g_i\)'s in \(\mathcal A(\Omega)\) is an ``outer ideal for \(H^p(\mu)\). The author introduces a new notion of globally coherent sequence. Let \(\mathcal A\) be the sheaf associated to \(\mathcal A(\Omega)\) over \(\overline\Omega\), \(\mathcal W\) the sheaf associated to \(\mathcal C^\infty(\overline\Omega)\) and \(\mathcal F\) one of these sheaves. The (global) sequence \(g \in\mathcal A(\Omega)^k\) is said to be globally coherent of length \(p\) over \(\mathcal F\) if, for any \(z\in \overline\Omega\), \(g_z\cdot\mathcal F_z\) has a free resolution of length bigger than \(p + 1\) over \(\mathcal F:\) \[ \mathcal F_z^{k_p}\longrightarrow\dots\longrightarrow \mathcal F_z^{k_1}\longrightarrow \mathcal F_z^{k}{\overset {g}\longrightarrow} g_z\cdot\mathcal F_z\longrightarrow0 \] with \(\mathcal A_z\) arrows. A peak set P for \(\mathcal A(\Omega)\) is a set in \(\overline\Omega\) such that \(\exists\varphi\in\mathcal A(\Omega)\) with: \(\varphi|_P\equiv 1\) and \(\forall z\in\overline\Omega\setminus P, |\varphi(z)| < 1\); this implies by the maximum principle that \(P\in\partial\Omega\). As usual \(V(g)\) is the set of common zeroes of the \(g_i\)'s in \(\overline\Omega\). The main result of the paper is: Let \(g \in\mathcal A(\Omega)^k\) be such that: \(g\cdot\mathcal C^\infty(\overline\Omega)\) is closed in \(\mathcal C^\infty(\overline\Omega)\); \(g\) is globally coherent of length \(3n + 1\) over \(\mathcal W\); there is a peak set \(P\) for \(\mathcal A(\Omega)\) such that \(V(g)\cap\bar\Omega\subset P\); finally \(\mu(P) = 0\). Then \(g \cdot\mathcal A(\Omega)\) is dense in \(H^p(\mu), 1\leq p < \infty\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4351899)