Abstract
Let $T, T'$ be weak contractions in the sense of Sz.-Nagy and Foiaş, with $m, m'$ as the minimal functions of their $C_0$ parts. Let $d$ be the greatest common inner divisor of $m$ and $m'$. This paper demonstrates that the space $I(T, T')$ of all operators intertwining $T$ and $T'$ is reflexive if and only if the model operator $S(d)$—the compression of the unilateral shift onto the space $H^2 \ominus dH^2$—is reflexive. In finite-dimensional contexts, this condition simplifies to the requirement that all roots of the greatest common divisor of the minimal polynomials of $T$ and $T'$ are simple. Furthermore, the paper provides an example illustrating that quasisimilarity does not preserve the hyperreflexivity of $I(T, T')$.
Introduction
In the study of linear operators on Hilbert spaces, the relationship between two operators $T$ and $T'$ is often characterized by the space of operators that "link" them. An operator $A$ is said to intertwine $T$ and $T'$ if $AT = T'A$. The set of all such operators, denoted $I(T, T')$, forms a subspace of the bounded linear operators. A central question in operator theory is determining when such a subspace is reflexive or hyperreflexive.
Reflexivity, in this context, refers to whether a subspace of operators is uniquely determined by its invariant subspaces. Hyperreflexivity is a stronger property involving a bounded distance estimate between an arbitrary operator and the subspace. While these properties have been established for specific operator classes, this work extends the theory to weak contractions, utilizing the framework of Sz.-Nagy-Foiaş dilation theory.
Methodology
The investigation relies on the classification of weak contractions and the properties of $C_0$ operators. Key methodological components include:
- Sz.-Nagy-Foiaş Dilation Theory: This provides a functional model for contractions, allowing them to be studied via shifts on Hardy spaces ($H^2$).
- Minimal Functions: For $C_0$ contractions, the minimal function $m$ serves as an analogue to the minimal polynomial in finite-dimensional linear algebra.
- Quasisimilarity: The paper leverages the fact that reflexivity is preserved under quasisimilarity, allowing for the reduction of complex operators to their Jordan models.
- Inner Function Divisors: The core proof hinges on the greatest common inner divisor ($d$) of the minimal functions of the operators involved.
Key Results and Insights
Reflexivity Criteria
The primary result (Theorem 4.2) establishes a definitive link between the reflexivity of the intertwining space and a specific model operator. Specifically, $I(T, T')$ is reflexive if and only if the model operator $S(d)$ is reflexive, where $d = \text{gcd}(m, m')$.
In the finite-dimensional case, this leads to a highly intuitive corollary: the space $I(T, T')$ is reflexive if and only if the roots of the greatest common divisor of the minimal polynomials of $T$ and $T'$ are simple. This aligns with known results for single operators where reflexivity is linked to the absence of repeated roots in the minimal polynomial.
The Limits of Hyperreflexivity
A significant contribution of this research is the distinction between reflexivity and hyperreflexivity regarding quasisimilarity. While reflexivity is a robust property preserved by quasisimilarity, hyperreflexivity is not.
In Section 5, the author constructs a counter-example using a direct sum of $2 \times 2$ matrices. By creating two quasisimilar operators where the "hyperreflexivity constant" of one space grows without bound, the author proves that hyperreflexivity can be lost even when the underlying operator structure remains quasisimilar.
Conclusion
This work provides a characterization of reflexivity for spaces of intertwining operators between weak contractions, reducing the problem to the study of a single model operator. The findings highlight the importance of the greatest common inner divisor of minimal functions. The discovery that hyperreflexivity is not preserved by quasisimilarity marks a clear boundary between these two structural properties and suggests that further research is needed to identify the precise conditions on inner functions that guarantee hyperreflexivity in infinite-dimensional spaces.
Summary updated