I'm considering a possible alternate proof of an elementary theorem in functional analysis:

Let $(X, \|\cdot\|)$ be a normed space over $\mathbb{R}$ or $\mathbb{C}$. $X$ is Banach if and only if every absolutely convergent series in $X$ converges in $X$.

  1. Say $X$ is not Banach. That means there exists a non-convergent Cauchy sequence $(x_n)_{n=1}^\infty$. Is there a way to explicitly construct a non-convergent but absolutely convergent series in $X$ using the Cauchy series $(x_n)_{n=1}^\infty$?
  2. Conversely, say we have an absolutely convergent, but non-convergent series $\sum_{n=1}^\infty y_n$ in $X$. Is there a simple way to get one explicit non-convergent Cauchy series using this series?

My attempt:

  1. Let's try with defining a sequence $(z_n)_{n=1}^\infty$ with: \[z_1=x_1\] \[z_n=x_n-x_{n-1}, \forall n \geq 2\] Notice that $\sum_{n=1}^N z_n = x_N$ so $\sum_{n=1}^\infty z_n$ does not converge. But is it absolutely convergent? I have no idea.
  2. Consider the sequence $\left(\sum_{n=1}^N y_n\right)_{N=1}^\infty$. This sequence does not converge. We show that it is Cauchy: let $M \geq N \in \mathbb{N}$ \[\left\Vert\sum_{n=1}^M y_n - \sum_{n=1}^N y_n\right\Vert = \left\Vert\sum_{n=N+1}^M y_n\right\Vert \leq \sum_{n=N+1}^M \left\Vert y_n \right\Vert \leq \sum_{n=N+1}^\infty \left\Vert y_n \right\Vert \xrightarrow{M, N \to \infty} 0\] It seems that it is Cauchy. Is this correct?
  1. is correct. ,1. Is not.

In case of one, you first need to extract a subsequence out of your Cauchy sequence, or you might end up with something that does not work.

Using the fact that $x_n$ is Cauchy, you extract a subsequence $x_{\phi (n)}$, which we shall call $y_n$, such that $||y_{n+1}-y_n||\leq \frac{1}{2^n}$ (or any other positive sequence whose series converges).

This new sequence still does not converge, since a Cauchy sequence that has a partial limit is convergent. And we have $y_{n+1}-y_0=\sum \limits_{k=0}^n y_{k+1}-y_{k}$, so the rest works just like you described it.

| cite | improve this answer | |

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.