I have a problem with the spectrum of this operator:

$(Tx)_1 = x_2$

$(Tx)_2 = x_1$

$(Tx)_n = \frac{1}{n}x_{n+1}$

with $n\ge3$

Find the $||T||$, the point spectrum $\sigma_P(T)$ and $\sigma_P(T^{\dagger})$ and the residual spectrum $\sigma_{\rho}(T)$ and $\sigma_{\rho}(T^{\dagger})$.

For the $||T||$ I have found:

$||T|| = 1$

Then for the point spectrum I try with:

$\lambda x_1 = x_2$

$\lambda x_2 = x_1$

$\lambda x_n = \frac{1}{n}x_{n+1}$

I found easily that some eigenvalues are $\lambda_n =0,\pm1$, but I have a problem, when I study the case $\lambda \neq\lambda_n$, I found the eigenvector:

$v_{\lambda} = (x_1,\lambda x_1, x_3, 3\lambda x_3, 3\cdot4\lambda^2x_3,....)$

But I don't understand the condition that the $\lambda$ have to satisfies for $v_{\lambda} \in \ell_2 $

For adjoint $T^{\dagger}$ I have found:

$\sigma_P(T^{\dagger}) = \{\lambda = \pm 1 \}$

$\sigma_{\rho}(T^{\dagger}) = \{z = 0\}$

It is correct?


The operator consists of two unrelated parts: reflection in the plane $(x_1,x_2)$, which contributes norm $1$ and eigenvalues $\pm 1$, and scaled backward shift, which contributes the eigenvalue $0$. There are no other eigenvalues because if $\lambda \notin \{0,\pm 1\}$ then $x_3,x_4,\dots $ must be nonzero and satisfy $x_{n+1} = n\lambda x_n$ for all $n$. Since $|n\lambda|>1$ for sufficiently large $n$, such a sequence $x$ tends to infinity, so it can't be in $\ell_2$.

The residual spectrum of $T$ is empty: for any $\lambda$, the range of $T-\lambda I$ contains the space $c_{00}$ of the sequences that are eventually zero; therefore, the range is dense.


The adjoint $T^*$ consists of a planar reflection combined with a scaled forward shift. You correctly found its eigenvalues to be $\pm 1$ and its residual spectrum to be $\{0\}$, since $\operatorname{ran} T^* \subset \{x_3=0\}$ and for $\lambda\ne 0$, the range of $T-\lambda I$ contains $c_{00}$.

  • $\begingroup$ Thanks for the explanation, do you know any references, that have this kind of exercise with the relative solution? $\endgroup$ – Saimon Jun 5 '16 at 14:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.