Quadratically hyponormal weighted shifts (Q915094)

For a hyponormal weighted shift \(Te_ j=\alpha_ je_{j+1}\) \((j=0,1,2,...)\) and a natural number k, the following assertions are shown to be equivalent: (i) T is k-hyponormal. (ii) The \(k\times k\) matrix \((a_{ij})^ k_{i,j=1}\) with \(a_{ij}([T^{*j}T^ i-T^ iT^{*j}]e_{n+j},e_{n+i})\) is positive for all \(n\geq -1.\) (iii) The matrix \((\beta^ 2_ n\beta^ 2_{n+i+j-2}-\beta^ 2_{n+i}\beta^ 2_{n+j})^ k_{i,j=1}\) is positive for each \(n\geq 0\), where \(\beta_ 0=1\), \(\beta_ k=\alpha_ 0\alpha_ 1...\alpha_{k-1}.\) (iv) The matrix \((\beta^ 2_{n+i+j-2})^{k+1}_{i,j=1}\) is positive for each \(n\geq 0.\) (An operator \(T\in B(H)\) is called k-hyponormal if the \(k\times k\) operator matrix \((A_{ij})\) with \(A_{ij}=T^{*i}T^ j-T^ jT^{*i}\) acting on the direct sum of k copies of the Hilbert space H is positive.) If T is 2-hyponormal then it is quadratically hyponormal, i.e., \(T+\lambda T^ 2\) is hyponormal for all \(\lambda\in {\mathbb{C}}\). It is also shown that if the unilateral weighted shift \(Te_ n=\alpha_{n+1}e_{n+1}\) is quadratically hyponormal and if \(\alpha_ n=\alpha_{n+1}=\alpha_{n+2}\) for some n, then \(\alpha_ 1=\alpha_ 2=\alpha_ 3=... \).