My friend and I are a little bit confused about this, the question we are working on asks us to find the eigenvalues of $R$, the right shift operator. The main point of contention is that $\sigma(R)$ is non-empty, and so should surely(?) contain an eigenvalue of $R$. Does the non-emptiness of $\sigma(R)$ only apply in finite dimensional Hilbert spaces? A detailed reply to clear up the confusion would be appreciated.

  • 2
    $\begingroup$ $\sigma(R)$ is the set of points $\lambda\in\Bbb C$ so that $R-\lambda$ is not invertible. Unlike in finite dimensional spaces, $\lambda\in\sigma(R)$ does not imply that $\lambda$ is an eigenvalue of $R$. Indeed $R$ has no eigenvalues, but $\sigma(R)$ is the entire (closed) unit disk in $\Bbb C$. $\endgroup$ – s.harp Feb 5 '17 at 19:10

For a bounded linear operator on a complex Hilbert space, the spectrum is always non-empty. But it may lack eigenvalues.

If $H=\ell^2(\mathbb N)$ and $R$ is the shift $$ R(a_1,a_2,\ldots)=(0,a_1,s_2,\ldots), $$ then $R$ has no eigenvalues. It's spectrum, though, is $\{\lambda:\ |\lambda|\leq1\}$. The easiest way to see it is to consider the left shift $$ L(a_1,a_2,\ldots)=(a_2,a_3,\ldots). $$ For any $\lambda$ with $|\lambda|<1$, one can check that $\lambda$ is an eigenvalue of $L$ with eigenvector $(\lambda,\lambda^2,\lambda^3,\ldots)$. As the spectrum is closed and $\|L\|\leq1$, one deduces that $$\sigma(L)=\{\lambda:\ |\lambda|\leq1\}.$$ Then $$ \sigma(R)=\{\bar\lambda:\ \lambda\in\sigma(L)\}=\{\lambda:\ |\lambda|\leq1\}. $$


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.