Let $X:=\{ x \in \ell^{1}:\vert\vert\vert x \vert \vert \vert< \infty\}$, and that $\vert\vert\vert x \vert \vert \vert=\sum\limits_{j=1}^{\infty}j\vert x_{j}\vert$

$a.$ Show that $(X, \vert\vert\vert \cdot \vert \vert \vert)$ is a Banach space

$b.$ Show that $(X, \vert\vert \cdot \vert \vert_{1})$ is not a Banach space.


$a.$ Let $(x^{n})_{n\in\mathbb N}\subset X$ be a cauchy sequence of sequences. This means for an arbitrary $\epsilon > 0$ there exists $N \in \mathbb N$ so that for any $n> m \geq N$: $\vert\vert\vert x^{n}-x^{m} \vert \vert \vert=\sum\limits_{i=1}^{\infty}i\vert x_{i}^{n}-x_{i}^{m}\vert<\epsilon$

thus the $\vert x_{i}^{n}-x_{i}^{m}\vert < \frac{\epsilon}{i}\leq\epsilon$,thus we have a Cauchy sequence $(x_{i}^{n})_{n \in \mathbb N}$ for a arbitrary, but fixed $i \in \mathbb N$. Now my thought (and only one at that), is to use that fact that $(x_{i}^{n})_{n \in \mathbb N}$ is a cauchy sequence oa a complete space (when equipped with the euclidean norm), therefore a pointwise limit $x_{i}:=\lim\limits_{n \to \infty}x_{i}^{n}$ exists so that we have an ideal candidate for a limit in $(X, \vert\vert\vert \cdot \vert \vert \vert)$, namely $x:=(x_{i})_{i \in \mathbb N}$

But the more I think about it, the less it makes sense that $\vert\vert\vert x \vert \vert \vert<\infty$ yet $\vert \vert x \vert \vert_{1}=\infty$ given the definitions of the norms, and that $j\vert x_{j}\vert\geq\vert x_{j}\vert$ for $j \in \mathbb N$

Any help is greatly appreciated


For (a), note that $\ell_1\to X$, $(a_n)\mapsto (n^{-1}a_n)$ is an isometry $(\ell_1,\lVert-\rVert_1)\to(X,\lVert\!\lvert-\rvert\!\rVert)$ makes the question rather trivial, but let's copy the proof of $\ell_1$ being complete here.

To show that $(x_i)\in X$, given $\varepsilon>0$ arbitrary, for every $N\in\mathbb{N}$, we have $$\require{color} \sum_{j=1}^N j\lvert x_j\rvert\leq\underbrace{\color{red}\sum_{j=1}^Nj\lvert x_j-x_j^{(m)}\rvert}_{<\varepsilon} + \underbrace{\color{blue}\sum_{j=1}^N j\lvert x_j^{(m)}\rvert}_{\leq M} $$ where $M=\sup\{\lVert\!\lvert x^{(m)}\rvert\!\rVert\colon m\in\mathbb{N}\}<\infty$ (since $(x^{(m)})$ is Cauchy in $X$) and the red term is $<\varepsilon$ for sufficiently large $m$ since we are only taking finitely many componentwise limit $j=1,2,\dots,N$. Therefore, $$ \sum_{j=1}^Nj\lvert x_j\rvert\leq M+\varepsilon $$ for every $N$. Now take $N\to\infty$ shows $x\in X$.

Of course, finally we need to show $x^{(n)}\to x$. Can you do that?

For (b), note that the space of eventually zero sequences $c_{00}$ is contained in $X$. So $$ x^{(n)}_j=\begin{cases} j^{-2} & j<n\\ 0 & \text{otherwise} \end{cases} $$ is a sequence in $X$. This sequence clearly converges to the sequence $(j^{-2})$ in $\ell_1$ since $\sum_j j^{-2}<\infty$, but this sequence is not in $X$.


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.