[Paper Review] Operator Theory and Complex Geometry

Ronald G. Douglas|ArXiv.org|Oct 9, 2007

Holomorphic and Operator Theory23 references3 citations

TL;DR

This paper establishes a deep connection between multivariable operator theory and complex geometry by analyzing Hilbert modules through holomorphic vector bundles and curvature invariants. It proves that essentially reductive Hilbert modules over the ball algebra containing a pure isometric submodule must be isomorphic to weighted Hardy modules, with the submodule arising from an inner function, extending subnormality and rigidity results in multivariable operator theory.

ABSTRACT

One approach to multivariate operator theory involves concepts and techniques from algebraic and complex geometry and is formulated in terms of Hilbert modules. In these notes we provide an introduction to this approach including many proofs. We are particularly interested in examples related to hermitian holomorphic vector bundles and we study submodules and reducing submodules in such cases. We go into some detail concerning a problem of Zhu on the reducing subspaces of powers of the Bergman shift as well as more recent work of the author and J. Sarkar on proper submodules which are unitarily equivalent to the orginal. Although the basic results are not new, there is some novelty in the details and the organization of the material.

Motivation & Objective

• To develop a geometric approach to multivariable operator theory using Hilbert modules and complex geometry.
• To understand the structure of Hilbert modules over function algebras via holomorphic vector bundles and curvature invariants.
• To investigate when submodules of Hilbert modules are isometrically isomorphic to the full module, particularly in the context of the Hardy and Bergman spaces.
• To establish conditions under which essentially reductive Hilbert modules are subnormal or isomorphic to weighted Hardy modules.
• To extend classical results on invariant subspaces and inner functions to higher dimensions and more general domains.

Proposed method

• Represent Hilbert modules as spaces of holomorphic functions on the unit ball or disk with module multiplication by coordinate functions.
• Use the anti-holomorphic family of one-dimensional subspaces spanned by reproducing kernels to construct Hermitian anti-holomorphic line bundles.
• Analyze curvature invariants and module resolutions to relate operator-theoretic properties to geometric structures.
• Apply the theory of Šilov modules and inner functions to characterize submodules and isometric isomorphisms.
• Use the fact that essential normality of module operators implies commutativity of their boundary operators almost everywhere on the unit sphere.
• Leverage the $C^*$-algebra structure of the boundary operators to deduce subnormality and extend modules to the $L^2$-boundary space.

Experimental results

Research questions

• RQ1 Under what conditions is a Hilbert module over the algebra of holomorphic functions on a domain essentially reductive and subnormal?
• RQ2 When can a submodule of a Hilbert module be isometrically isomorphic to the full module, and what geometric or operator-theoretic conditions force this?
• RQ3 How do curvature invariants and holomorphic vector bundles classify Hilbert modules in multivariable operator theory?
• RQ4 What role do inner functions play in characterizing isometric submodules of weighted Hardy and Bergman modules?
• RQ5 Can the absence of joint normal extensions for commuting subnormal operators be explained geometrically through module structure?

Key findings

• An essentially reductive Hilbert module over $A(\Omega)$ containing a pure isometrically isomorphic submodule is subnormal.
• If $\mathcal{R}$ is an essentially reductive quasi-free Hilbert module over $A(\mathbb{B}^n)$ with a pure isometric submodule $\mathcal{M}$, then $\mathcal{R} \cong H^2_\mathcal{E}(\mathbb{B}^n)$ and $\mathcal{M} = \theta H^2_\mathcal{E}(\mathbb{B}^n)$ for an inner function $\theta$ and finite-dimensional $\mathcal{E}$.
• No proper submodule of $L^2_a(\mu)$ on a bounded domain $\Omega$ is isometrically isomorphic to $L^2_a(\mu)$, generalizing results for the Bergman space.
• The restriction of $M_{z_1}$ on $H^2_n$ to cyclic subspaces generated by $z^\ell_2$ is unitarily equivalent to a weighted Bergman shift, showing $M_{z_1}$ is subnormal.
• The coordinate multiplication operators on the $n$-shift space are not jointly subnormal, confirming Arveson's result via geometric module theory.
• The $C^*$-algebra generated by the boundary operators of an essentially reductive module contains no non-zero compact operators, implying normality a.e. on the boundary.