0
$\begingroup$

Let $T\in\mathscr{B(\mathcal{H})}$ where $\mathcal{H}$ is an infinite dimensional seperable Hilbert Space and $k\in\mathbb{N}\cup\{\infty\}$. Now we define k-rank numerical range of $T$ denoted by $\Lambda_k(T)$ is defined as $$\Lambda_k(T):=\{\lambda\in\mathbb{C}: PTP=\lambda P, \text{ for some orthogonal projection } P \text{ of rank } k\}$$ or equivalently we can write $$\lambda\in\Lambda_k(T) \text{ iff there exists an orthonormal set } \{f_j\}_{j=1}^k \text{ s.t. } \langle Tf_j,f_r\rangle=\lambda\delta_{j,r} \text{ for } j,r\in\{1,2\ldots, k\}$$ where $\delta_{j,r}$ is Kronecker delta. Clearly $\Lambda_1(T)=W(T)$ i.e. $\Lambda_k(T)$ is a genaralization of numerical range $W(T)$.


Question: Let $T\in\mathscr{B(\mathcal{H})}$ be a normal operator. Show that if $\Re(e^{i\theta}\lambda)\in\Lambda_k(\Re(e^{i\theta}T))\text{ for all }\theta\in[0,2\pi)\text{ then }\lambda\in\Lambda_k(T)$


Comments: I could able to prove this using some big result of k-rank numerical range of an operator in terms of intersection of half-planes. But unfortunately I could not able to prove it using just definition of k-rank numerical range (written in the beginning) which I feel one could. This result is easy to see for numerical range $W(T)=\Lambda_1(T)$.

Any Hints or comments is highly appreciated. Instead of showing $\lambda\in\Lambda_k(T)$, if one show $\lambda\in\overline{\Lambda_k(T)}$ is also highly appreciable.

$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.