Multiplication operators on the Bergman space (Q2348236): Difference between revisions

If \(f\) is an inner function on the unit disk, the multiplication operator \(M_f\) on the classical Hardy space \(H^2\) is an isometry. It then follows from general functional analysis that the reducing subspaces of \(M_f:H^2\to H^2\) are in a one-to-one correspondence with the closed subspaces of \(H^2\ominus fH^2\). Thus the reducing subspace lattice of \(M_f\) on \(H^2\) induced by an inner function is well understood. In particular, either \(M_f:H^2\to H^2\) is irreducible (when \(f\) is a Möbius map) or it has infinitely many reducing subspaces. The situation for the Bergman space \(A^2\) on the unit disk is much different. For an arbitrary inner function \(f\), it is not even clear if \(M_f:A^2\to A^2\) has any nontrivial reducing subspace at all. Elementary examples suggest that \(M_f:A^2\to A^2\) has far fewer reducing subspaces than \(M_f:H^2\to H^2\) does. For example, it was shown in the reviewer's paper [J. Lond. Math. Soc., II. Ser. 62, No. 2, 553--568 (2000; Zbl 1158.47309)] that, for a Blaschke product \(B\) with two zeros, the operator \(M_B:A^2\to A^2\) has exactly two nontrivial reducing subspaces. Furthermore, these two reducing subspaces were explicitly computed. I then made the innocent conjecture that if \(B\) is a Blaschke product of \(n\) zeros, then the operator \(M_B:A^2\to A^2\) should have exactly \(n\) minimal reducing subspaces. I also included the conjecture in one of my NSF proposals. A reviewer of the proposal wrote that such a problem appeared too elementary for an NSF proposal and it would be more appropriate for a beginning PhD student. Naturally, the proposal was not funded. I visited Fudan University in (approximately) 2002 and gave a talk based on the paper mentioned above. Somehow the talk got a group of analysts in Shanghai really interested in multiplication operators on the Bergman space induced by finite Blaschke products. The (extended) Fudan group led by Kunyu Guo kept working on the problem, making steady progress over the years when the number of zeros of \(B\) increased from \(3\) to \(8\). What they soon realized was that my conjecture above was false in general. The structure of the reducing subspace lattice for \(M_B:A^2\to A^2\) is much more delicate and a full understanding can be achieved only with the help of the theories of Riemann surfaces and von Neumann algebras. Eventually, a beautiful theory emerged from this study, with the paper [\textit{R. G. Douglas} et al., J. Funct. Anal. 263, No. 6, 1744--1765 (2012; Zbl 1275.47071)] putting a temporary end to the story: For a Blaschke product \(B\) of order \(n\), the (commutant) von Neumann algebra \(\mathcal A\) generated by \(M_B:A^2\to A^2\) is abelian of dimension \(q\), where \(q\) is the number of connected components of the Riemann surface \(B^{-1}\circ B\). Moreover, \(q\) coincides with the number of minimal reducing subspaces for \(M_B:A^2\to A^2\). The book under review, written by two of the main players in the field, is a summary of the developments in the past fifteen years or so about reducing subspaces of multiplication operators on the Bergman space plus a few other closely related topics. The material is well organized, the presentation is smooth, and the mathematics is simply beautiful. I would recommend the book to anyone who is interested in the Bergman space and multiplication operators on it. Finally, I want to mention that, as far as I know, the discovery of the distinguished reducing subspace for \(M_B:A^2\to A^2\) first appeared in the paper by \textit{M. Stessin} and the reviewer [Can. J. Math. 55, No. 2, 379--400 (2003; Zbl 1048.30021)], not the paper [Zbl 1158.47309] as was suggested on page 92 of the book.