Invariant subspaces of \({\mathcal H}^ p\) for multiply connected regions (Q1113320)

Let G be a bounded plane domain with smooth boundary and let \(H^ p(G)\) be the Hardy class of G with \(1\leq p\leq \infty\). The author studies closed \((weakly^*\) closed, if \(p=\infty)\) subspaces M of \(H^ p(G)\) which are invariant in the sense that zM\(\subseteq M\). The study of invariant subspaces in this sense was initiated by a classical work of \textit{A. Beurling} [Acta Math., Uppsala 81, 239-255 (1949; Zbl 0033.377)], who determined, among other things, closed invariant subspaces of \(H^ 2(\Delta)\) for the open unit disk \(\Delta\). A large amount of work followed. As for finitely connected domains G this was done by Sarason, Voichick and the reviewer in 1960's. They are concerned with so-called fully invariant subspaces, which means that r(z)M\(\subseteq M\) for any rational functions r(z) with poles in the exterior of G. The aim of the present author is to study subspaces of \(H^ p(G)\), which are invariant in the original sense. This seems to be a difficult problem. The author first describes a general theory of \(H^ p\) spaces on finitely connected domains and gives a brief proof to the theorem of Sarason, Voichick and the reviewer. The author describes a necessary condition for invariance in terms of analytic pseudo-continuation as follows: Let \(G_ j\) be any bounded component of the complement of \(\bar G.\) Then, either r(z)M\(\subseteq M\) for any rational function r(z) with poles in \(G_ j\) or else f/g for any \(f,g\in M\) has an analytic pseudo- continuation of class N in \(G_ j.\) Using this, he then obtains a characterization under a restrictive hypothesis (H): Let \(G_ 1\) be the union of bounded components of the complement of \(\bar G.\) M contains a function h whose boundary values on the boundary \(\partial G_ 1\) satisfies \(0<c\leq | h| \leq C<\infty\) a.e. Namely, he proves the following theorem: If M is a closed invariant subspace of \(H^ p(G)\), \(1<p<\infty\), satisfying (H), then there exists an inner function \(\phi\) on G and a measurable function \(\chi\) on \(\partial G_ 1\) such that \(| \chi |\) is equal to a constant a.e. on each component of \(\partial G_ 1\) and such that M is equal to \(\phi M_{\chi}\), where \(M_{\chi}\) consists of those \(f\in H^ p(G)\) such that \(f\chi\) are the boundary values of a function in \(H^ p(G_ 1)\). Further, he conjectures that all invariant subspaces are of the form \(hM_{\psi}\), where \(h\in H^ p(G)\) is a function whose outer part is continuous and non-vanishing on the outer boundary of G. It is mentioned in the introduction that the author had this paper held unpublished for fifteen years in order to settle this conjecture. Although the conjecture itself is far from being solved, \textit{D. Hitt} [Pac. J. Math. 134, 101-120 (1988; reviewed below)] has recently made a significant progress in Royden's program and settled the problem when G is an annulus. This is why this paper appeared after a long time.