Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Given a bounded sequence $\pi = (\lambda_n)$ in $\mathbb{C}$ consider the continuous linear map $M_\pi:\ell^2\rightarrow \ell^2$ defined by $$M_\pi(x_n) = (\pi_nx_n)$$ a) determine the spectrum. b) Characterize the sequences $\pi$ for which $M_\pi$ is compact.

my try: a) The spectrum is defined where $(M_\pi - I\lambda)$ is invertible. $M_\pi$ is bounded iff $\pi$ is bounded. hence we get the solution $\sigma(M_\pi) = {\lambda_i} \;\; \forall i$ b) If $\pi_i \neq 0$ we have sequences $(x_n) = ne_n$ which has no cauchy subsequences, hence $M_\pi$ is only compact if $\pi_i = 0\;\; \forall i$

it feels like im missing something, is there any other $\pi$ that makes $M_\pi$ compact?

share|cite|improve this question
up vote 1 down vote accepted

Yes, there are. For (a) note, that the spectrum is closed. And the set you gave, namely $\Pi := \{\pi_n \mid n \in \mathbb N\}$ need be closed. As the spectrum is closed, we obtain $\overline\Pi \subseteq \sigma(M_\pi)$, we will now show the other inclusion: Let $\lambda \not\in \overline\Pi$, then $\mathrm{dist}(\lambda, \Pi)> 0$, define $\psi_n := (\pi_n - \lambda)^{-1}$, then $|\psi_n| \le \frac 1{\mathrm{dist}(\lambda,\Pi)}$, hence $\psi := (\psi_n)$ is bounded and $M_{\psi}$ is inverse to $M_\pi - \lambda$. So $\sigma(M_\pi) = \overline\Pi$.

For (b) note, that if $M_\pi$ is compact, by $M_\pi e_n = \pi_n e_n$ we need $(\pi_n e_n)$ to have a convergent subsequence, with 0 as only possible limit (as $\ell^2$-convergence implies coordinatewise convergene). Applying this argument to each subsequence, we obtain $\pi_n \to 0$ if $M_\pi$ is compact. On the other hand, suppose $\pi_n \to 0$. Consider for $n \in \mathbb N$ the sequence $$ \pi^{(n)} := (\pi_1, \ldots, \pi_n, 0, \ldots) $$and the corresponding multiplication operator $M_{\pi^{(n)}}$. $M_{\pi^{(n)}}$, having finite dimensional range, is compact, and for $x \in \ell^2$ we have \begin{align*}\def\norm#1{\left\|#1\right\|} \norm{M_\pi x - M_{\pi^{(n)}}x} &= \left(\sum_{k=n+1}^\infty |\pi_kx_k|^2\right)^{1/2}\\ &\le \sup_{k \ge n+1}|\pi_k| \cdot \left(\sum_{k=n+1}^\infty |x_k|^2\right)^{1/2}\\ &\le \sup_{k\ge n+1} |\pi_k| \cdot \|x\| \end{align*} Hence $M_{\pi^{(n)}} \to M_\pi$ in $L(\ell^2)$, and $M_\pi$, being the norm-limit of compact operators is compact. So $M_\pi$ is compact iff $\pi$ converges to zero.

share|cite|improve this answer
What is $\overline\Pi$ is that not equal to $\Pi$? – Johan Dec 10 '12 at 10:03
$\overline\Pi$ denotes the closure of $\Pi = \{\pi_n \mid n \in \mathbb N\}$, $\Pi$ is seldom closed ... exactly if $(\pi_n)$ contains its accumulation points. If, for example $\pi = (\frac 1n)$, then $\overline \Pi = \Pi \cup\{0\}$. – martini Dec 10 '12 at 10:06
I understand! Can you clarify a little bit why having finite dimensional range implies compactness? – Johan Dec 10 '12 at 10:23
Let $T\colon X \to Y$ a linear operator having finite dimensional range. Let $(x_n)$ a bounded sequence in $X$. Then $(Tx_n)$ is a bounded sequence in the finite dimensional space $TX \cong \mathbb K^d$. As we now from calculus, in $\mathbb K^n$, each bounded sequence has a convergent subsequence, so has $(Tx_n)$. – martini Dec 10 '12 at 10:26
Thanks! Looking more closely, how do we know that the $\psi$ is bounded in $\ell^2$, And what does it mean that we have an inverse to $M_\pi - \lambda$, it means that lambda is not in the spectrum, hence a contradictions? – Johan Dec 10 '12 at 12:14

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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