Simply invariant subspaces of \(H^2\) of some multiply connected regions (Q1909667)

Let \(\Omega\) be a finitely connected region in the plane. For simplicity, we will assume that \(\Omega\) is a circular domain, that is \(\Omega=\{\lambda:| \lambda| <1, | \lambda_j-a_j| >a_j,j=1,2\dots,n\}\), where \(r_j+| a_j| <1\) and \(| a_i-a_j| >r_i+r_j\) for \(i\neq j\). Thus, \(\Omega\) is obtained from the open unit disk \(D\) by removing closed disjoint disks \(\overline{D_j}\) centered at \(a_j\). The space \(H^2(\partial\Omega)\) can be viewed as the closure in \(L^2(\partial\Omega)\) of the space Rat\((\Omega)\) of rational functions with poles off \(\Omega\). A space \(M\subset H^2(\partial\Omega)\) is said to be rationally invariant if Rat\((\Omega)M\subset M\), and it is said to be simply invariant if \(\chi M\subset M\), where \(\chi(z)=z\) for \(z\in\Omega\). Rationally invariant subspaces are characterized via a generalization of Beurling's classical theorem (which applies to the case \(\Omega=D\)). Each rationally invariant subspace is determined by an inner function on \(\Omega\). Simply invariant subspaces are not as easily characterized. To see why, observe that if we set \(\Omega_1=\Omega\cup\overline{D_1}\), then \(H^2(\partial\Omega_1)\) can be viewed as a subspace of \(H^2(\partial\Omega)\), which is simply invariant but not rationally invariant. Other examples are obtained by considering rationally invariant subspaces of \(H^2(\partial\Omega_1)\). The natural question is whether all simply invariant subspaces can be obtained by `filling in holes' via some sort of analytic continuation, and then considering rationally invariant subspaces in the filled domain. This question was raised by \textit{H. L. Royden} [Pac. J. Math. 134, No. 1, 151-172 (1988; Zbl 0662.30034)] and proved to be true first for annuli by \textit{D. Hitt} [Pac. J. Math. 134, No. 1, 101-120 (1988; Zbl 0662.30035)] and for general \(\Omega\) by \textit{D. V. Yakubovich} [J. Sov. Math. 61, No. 2, 2046-2056 (1992; Zbl 0784.47038)]. The paper under review provides an alternate path to these results and, in addition, it provides spectral and multiplicity information about the multiplication operator by \(\chi\) on a simply invariant subspace. The authors make use of the concept of a nearly invariant subspace for the shift, first introduced in the work of \textit{D. Sarason} [Oper. Theory, Adv. Appl. 35, 481-493 (1988; Zbl 0687.47003)].