Hyponormality and subnormality for powers of commuting pairs of subnormal operators (Q883500)

Let \({\mathcal H}\) be a complex Hilbert space. Let \(\mathfrak{H}_0\) denote the class of all commuting pairs of subnormal operators on \({\mathcal H}\), and let \(\mathfrak{H}_\infty\) stand for the class of all subnormal pairs. For an integer \(k\geq1\), let \(\mathfrak{H}_k\) denote the class of all \(k\)-hyponormal pairs in \(\mathfrak{H}_0\). Obviously, the inclusions \[ \mathfrak{H}_\infty\subseteq\dots\subseteq\mathfrak{H}_k\subseteq\dots\subseteq \mathfrak{H}_2\subseteq\mathfrak{H}_1\subseteq\mathfrak{H}_0 \] always hold and are, in fact, strict [see J.~Funct.\ Anal.\ 229, No.\,2, 462--480 (2005; Zbl 1088.47011); Trans.\ Am.\ Math.\ Soc.\ 358, No.\,11, 5139--5159 (2006; Zbl 1115.47020)]. The Lifting Problem for Commuting Subnormals (LPCS) seeks necessary and sufficient conditions for a pair \({\mathbf T}=(T_1,T_2)\) of subnormal operators on \({\mathcal H}\) to guarantee the subnormality of the pair \({\mathbf T}=(T_1,T_2)\). In view of the above notations, the LPCS can be reformulated as follows: Find necessary and sufficient conditions for a pair \({\mathbf T}=(T_1,T_2)\in\mathfrak{H}_0\) that forces it to be in \(\mathfrak{H}_\infty\). In [J.~Oper.\ Theory 31, No.\,2, 219--228 (1994; Zbl 0832.47017)], \textit{E.\,Franks} proved that a pair \({\mathbf T}=(T_1,T_2)\) in \(\mathfrak{H}_0\) lies in \(\mathfrak{H}_\infty\) provided that \(p({\mathbf T})\) is subnormal for all polynomial \(p\) of degree at most \(5\). This result together with \textit{J.\,Stampfli}'s work [Pac.\ J.\ Math.\ 12, 1453--1458 (1962; Zbl 0129.08701); ibid.\ 17, 367--379 (1966; Zbl 0189.43902)] motivated the authors to consider the role of the powers of a pair in ascertaining its subnormality. As for every pair \({\mathbf T}=(T_1,T_2)\in\mathfrak{H}_\infty\) and all positive integers \(m\) and \(n\), the pair \({\mathbf T}^{(m,n)}:=(T_1^m,T_2^n)\) belongs to \(\mathfrak{H}_\infty\) and \(T_1^mT_2^n\) is a subnormal operator, the authors consider the following natural problem which suggests itself: Does the subnormality of both \({\mathbf T}^{(2,1)}=(T_1^2,T_2)\) and \({\mathbf T}^{(1,2)}=(T_1,T_2^2)\) ensure the subnormality of \({\mathbf T}\)? In this interesting paper, the authors first show that the class \(\mathfrak{H}_1\) is not invariant under squares. They construct a pair \({\mathbf T}=(T_1,T_2)\) in \(\mathfrak{H}_1\) such that \({\mathbf T}^{(2,1)}\not\in\mathfrak{H}_1\). Conversely, they find a pair \({\mathbf T}=(T_1,T_2)\) in \(\mathfrak{H}_0\) such that \({\mathbf T}^{(2,1)}\in\mathfrak{H}_1\) but \({\mathbf T}\not\in\mathfrak{H}_1\). They next prove that there is a large class of \(2\)-variable weighted shifts \({\mathbf T}=(T_1,T_2)\) for which the subnormality of both \({\mathbf T}^{(2,1)}=(T_1^2,T_2)\) and \({\mathbf T}^{(1,2)}=(T_1,T_2^2)\) does imply the subnormality of \({\mathbf T}\). Finally, they show that there exists a pair \({\mathbf T}=(T_1,T_2)\) in \(\mathfrak{H}_1\) such that \(T_1^mT_2^n\) is a subnormal operator for all positive integers \(m\) and \(n\) but \({\mathbf T}=(T_1,T_2)\not\in\mathfrak{H}_\infty\). This provides further evidence that the gap between the classes \(\mathfrak{H}_\infty\) and \(\mathfrak{H}_1\) is fairly large. Most of the examples and some of the proofs were obtained using calculations with the software tool \texttt{Mathematica}.