I was solving this Functional Analysis problem, but I'm not sure I'm correct on this one, the problem is:

Let $\big(E,\|\bullet\|\big)$ be a normed vector space, and $(x_n)_n$ a sequence in $E$ such that $\sum_{n=1}^\infty x_n$ converges. Show that $$\left\|\sum_{n=1}^\infty x_n\right\| \leq \sum_{n=1}^\infty\|x_n\|$$

My solution:

We know that the following inequality holds for any $m\in\mathbb{N}$ (just using the triangular inequality):

$$\left\|\sum_{n=1}^m x_n\right\| \leq \sum_{n=1}^m \|x_n\|$$

Since the left side converges, the inequality still holds in the limit $m\rightarrow\infty$, hence we have the result.

My doubts:

The only thing I am not totally sure is my last statement, because I can't know if the right side converges, so how can I compare the two things?

But then I thought I can use the following argument:

The sequence $\left(\sum_{n=1}^m \|x_n\|\right)_m$ is clearly a monotone (non-decreasing) real sequence, so if it is bounded it converges and my result is correct, the other option is that the sequence is unbounded, in which case

$$\sum_{n=1}^\infty \|x_n\|=\infty$$

And then it makes sense to write:

$$\left\|\sum_{n=1}^\infty x_n\right\| < \infty$$

Can someone tell me if my reasoning is correct? Or if another way would be better to solve this problem?

  • 1
    $\begingroup$ Thank you for a question that is well written, and you also give your own thoughts. +1 $\endgroup$ – Eff Sep 9 '17 at 13:00

If $\sum_{n=1}^\infty \|x_n\|=\infty$ then it is true. So let us assume that $\sum_{n=1}^\infty \|x_n\|<\infty$. So, $$ \left\|\sum_{n=1}^m x_n\right\| \leq \sum_{n=1}^m \|x_n\| \le \left\|\sum_{n=1}^m x_n\right\| \leq \sum_{n=1}^m \|x_n\|=l $$ Since, norm is a continuous function so $$ \lim_{m\to \infty}\left\|\sum_{n=1}^m x_n\right\|=\left\|\lim_{m\to\infty} \sum_{n=1}^m x_n\right\| =\left\|\sum_{n=1}^\infty x_n\right\| \le \lim_{m\to \infty} l=l $$


You're quite right that the right hand side need not converge : a simple example in the reals is $x_n = \frac{(-1)^n}{n}$ where the series converges, but not absolutely. So you can indeed make the distinction between $+\infty$ and finite for the right hand series. $+\infty$ is no problem for the inequality, and the finite case follows from the triangle inequality, as said. Or just use (in your first version) that in the extended reals $x_n \le y_n$, $x\to x, y\to y$ also implies $x \le y$ (this holds in all ordered topological spaces, as $\{(p,q) \in X^2: p \le q \}$ is closed in $X^2$). The right hand limit always exists by monotonicity (and completeness of $[-\infty, +\infty]$), he left hand side as $\sim_n x_n$ converges and $\|.\|$ is continuous.


Using the continuity of $\|\cdot\|$, we have: $$\left\|\sum_{n=1}^\infty x_n\right\| = \left\|\lim_{m\to\infty}\sum_{n=1}^mx_n\right\| = \lim_{m\to\infty}\left\|\sum_{n=1}^m x_n\right\| \leq \lim_{m\to\infty}\sum_{n=1}^m \left\|x_n\right\| = \sum_{n=1}^\infty \left\|x_n\right\|$$


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.