Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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

Show that the set $$ B = \left\lbrace(x_n) \in \ell^1 : \sum_{n\geq 1} n|x_n|\leq 1\right\rbrace$$ is compact in $\ell^1$. Hint: You can use without proof the diagonalization process to conclude that every bounded sequence $(x_n)\in \ell^\infty$ has a subsequence $(x_{n_k})$ that converges in each component, that is $\lim_{k\rightarrow\infty} (x_{n_k}^{(i)})$ exists for all i. Moreover, sequences in $\ell^1$ are obviously closed by the $\ell^1$-norm.

My try: Every bounded sequence $(x_n) \in \ell^\infty$ has a subsequence $(x_{n_k})$ that converges in each component. That is $\lim_{k\rightarrow\infty} (x_{n_k}^{(i)})$ exists for all i. .And all sequences in $\ell^1$ are bounded in $\ell^1$-norm. I want to show that every sequence $(x_n) \in B$, has an Cauchy subsequence. Choose an N and M such that for $l,k > M$ such that $|x_{n_k}^{(i)} - x_{n_l}^{(i)}| < \frac{1}{N^2}$ Then $$\sum_i^N |x_{n_k}^{(i)} - x_{n_l}^{(i)}| + \sum_{i = N+1} ^\infty |x_{n_k}^{(i)} - x_{n_l}^{(i)}| \leqslant \frac{1}{N} + \frac{1}{N+1} \sum_{i = N+1} ^\infty i|x_{n_k}^{(i)} - x_{n_l}^{(i)}| \leqslant \frac{3}{N+1}$$ It feels wrong to compine $M,N$ like this, is it? what can I do instead?

share|cite|improve this question
I don't understand the first sentence: you mean that a bounded sequence of real numbers has a convergent subsequence, right? But it will help if you can show that you can extract, when you work with sequences whose terms are in $\ell^1$, that will work for all coordinates. – Davide Giraudo Dec 30 '12 at 11:38
Thanks, I corrected it now, can you please see if its wrong now? – Johan Dec 30 '12 at 12:09

We can use and show the following:

Let $K\subset \ell^1$. This set has a compact closure for the $\ell^1$ norm if and only if the following conditions are satisfied:

  • $\sup_{x\in K}\lVert x\rVert_{\ell^1}$ is finite, and
  • for all $\varepsilon>0$, we can find $N$ such that for all $x\in K$, $\sum_{k\geqslant N}|x_k|<\varepsilon$.

These conditions are equivalent to precompactness, that is, that for all $r>0$, we can find finitely many elements $x^1,\dots,x^N$ such that the balls centered at $x^j$ and for radius $r$ cover $K$.

share|cite|improve this answer
Thanks! I have not seen this theorem, cant you prove it with the Hint? what is wrong with my try? – Johan Dec 30 '12 at 12:19

To prove compactness of $B$ you can use the following theorem:

A set $A \subseteq \ell_1$ is compact if and only if

  1. $A$ is closed
  2. $A$ is bounded
  3. $\sup_{x \in A} \sum_{j \geq n} |x_j| \to 0$ as $n \to \infty$

How to prove these properties of $B$?

  1. Let $X:= \{x \in \ell_1; \sum_j j \cdot |x_j|<\infty\}$. Define $$\|x\|_X := \sum_j j \cdot |x_j| \qquad (x \in X)$$ Then $(X,\|\cdot\|_X)$ is a normed space. Define $$T:X \to \ell_1, x=(x_j)_j \mapsto (j \cdot x_j)_j $$ Then $T$ is linear, surjective and isometric (in particular continuous) and therefore we conclude that $B=T^{-1}(B[0,1])$ is closed as a pre-image of a closed subset.
  2. Let $x=(x_n)_n \in B$, then $$\|x\|_1 = \sum_{n \geq 1} |x_n| \leq \sum_{n \geq 1} n \cdot |x_n| \leq 1$$ where we used the definition of $B$ in the last equation. Hence $$\sup_{x \in B} \|x\|_1 \leq 1$$ which means that $B$ is bounded.
  3. Use the following estimate: For all $x=(x_n)_n \in B$ we have $$\sum_{j \geq n} |x_j| = \frac{1}{n} \sum_{j \geq n} n \cdot |x_j| \leq \frac{1}{n} \cdot \sum_{j \geq n} j \cdot |x_j| \leq \frac{1}{n}$$

I think you mess things up. You wrote

Every convergent sequence $(x_n) \in \ell^\infty$ has a convergent subsequence $(x^{n_k})$. That converges component wise.

It doesn't make sense at all: If you consider a convergent sequence $(x_n)_{n} \in \ell_\infty$, then it's trivial that there exists a convergent subsequence (since the whole sequence is convergent). Probably you wanted to consider a sequence of sequences, i.e $(x^n)_n \subseteq \ell_\infty$ (i.e. $x^n \in \ell_\infty$). Similar for the next sentence:

And all sequences in $\ell^1$ are bounded in $\ell^1$-norm.

If you take one (!) sequence $x:=(x_n)_{n} \in \ell^1$, then (by definition) $\|x\|_1 < \infty$. But: If you consider a sequence of sequences, i.e. $(x^n)_n \subseteq \ell_1$, then it's not trivial that the sequence (of sequences) is bounded, i.e.

$$\sup_n \|x^n\|_1 < \infty$$

You have to differentiate between

  1. a sequence $x:=(x_n)_n$ which is an element of $\ell_1$, i.e. $\|x\|_1<\infty$
  2. a sequence in $\ell_1$, i.e. $(x^n)_{n \in \mathbb{N}} \subseteq \ell_1$ where $x^n \in \ell_1$ for all $n \in \mathbb{N}$.
share|cite|improve this answer
Great Theorem, but I have not seen it before... Is it hard to prove? I rewrote the text, there was some typos, sorry – Johan Dec 30 '12 at 12:16
@Johan It's not that difficult to prove. When I attended functional analysis, it was an exercise left to the students. Probably you'll find this theorem (and a proof) in most books about functional analysis. (And I recognized that you rewrote the text, but I'm still the same oppinion.) – saz Dec 30 '12 at 12:38
I found one, but I cannot manage to prove 1. can u give me a hint? – Johan Dec 30 '12 at 13:46
@Johan I added one. – saz Dec 30 '12 at 14:15
@Johan Sorry, there was a typo in the definition of $T$. – saz Dec 30 '12 at 15:15

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.