Prove that the series $\sum_{n=1}^{\infty} a_n b_n$ converges if the following conditions are met:

  1. series $\sum_{n=1}^{\infty} b_n$ converges,
  2. series $\sum_{n=1}^{\infty} (a_n - a_{n+1})$ absolutely converges.

I was thinking applying Abel's test, proving that if:

  1. $\sum_{n=1}^{\infty} b_n$ is a convergent series (given),
  2. {$a_n$} is a monotone sequence, and
  3. {$a_n$} is bounded.

Then the $\sum_{n=1}^{\infty} a_n b_n$ converges.

To prove the second one statement, we've to get following inequality: $a_{n+1} \leq a_n$. I've no idea how do we do that.

Third one, I think, is obtained by the fact, that the $\sum_{n=1}^{\infty} (a_n - a_{n+1})$ series converges, so $\lim_{n\to\infty} (a_n - a_{n+1}) = 0$. Is it correct?

  • 2
    $\begingroup$ Instead of using Abel's theorem itself, use its proof, or rather summation by parts. For a partial sum, beside the two terms that are left over when using summation by part and that tend to zero, since $\sum_{1}^{m} b_n\to B$ and $a_n-a_{n+1}\to0$, you also get a sum $\sum_{r}^{s} (a_{n}-a_{n+1})\sum_{1}^n b_k$. The sum you bound using triangle inequality by $\sum_{r}^{s} |a_n-a_{n+1}||\sum_{1}^{n}b_k|$ The factor $|\sum_{1}^{n}b_k|$ is bounded by some $M$. So you get the bound $M\sum_{r}^{s}|a_{n}-a_{n+1}|$ which can be small. $\endgroup$
    – plop
    Jun 16 at 14:49

There is no way of proving that $(a_n)_{n\in\Bbb N}$ is monotonic, since it may well not be.

It's not hard to prove that you always have, when $m,n\in\Bbb N$ and $m\geqslant n$,\begin{multline}\sum_{k=n}^na_kb_k=(a_n-a_{n+1})b_n+(a_{n+1}-a_{n+2})(b_n+b_{n+1})+\cdots+\\+(a_m-a_{m+1})\left(\sum_{k=n}^mb_k\right)+a_{m+1}\left(\sum_{k=n}^mb_k\right).\end{multline}Note that the sequence $(a_n)_{n\in\Bbb N}$ converges. In fact\begin{align}\sum_{n=1}^\infty|a_n-a_{n+1}|\text{ converges}&\implies\sum_{n=1}^\infty(a_n-a_{n+1})\text{ converges}\\&\iff(a_n)_{n\in\Bbb N}\text{ converges.}\end{align}Since it converges, it is a bounded sequence and therefore the sequence $\bigl(|a_n|\bigr)_{n\in\Bbb N}$ is bounded too.

Now, let $\varepsilon>0$. I will prove that there is a natural number $N$ such that if $m\geqslant n\geqslant N$, then$$\left|\sum_{k=n}^ma_kb_k\right|<\varepsilon,\tag1$$in order to apply the Cauchy's convergence test. Take $\varepsilon'\in(0,\varepsilon)$ and take $N\in\Bbb N$ such that$$m\geqslant n\geqslant N\implies\left|\sum_{k=n}^mb_k\right|<\frac{\varepsilon'}{\displaystyle\sum_{k=1}^\infty|a_k-a_{k+1}|+\sup_{n\in\Bbb N}|a_n|}.$$Actually, this doesn't make sense if every $a_n$ is $0$, but then the statement is trivially true. Now we have, thanks to the first equality of this proof, that, if $m\geqslant n\geqslant N$,\begin{align}\left|\sum_{k=n}^na_kb_k\right|&\leqslant\frac{\varepsilon'}{\displaystyle\sum_{k=1}^\infty|a_k-a_{k+1}|+\sup_{n\in\Bbb N}|a_n|}\left(\sum_{k=n}^m|a_k-a_{k+1}|+\sup_{n\in\Bbb N}|a_n|\right)\\&\leqslant\varepsilon'\\&<\varepsilon.\end{align}


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.