I have been tasked with proving the following:

$a_n$ is bounded but not necessarily convergent

assume that $\lim{b_n} \to 0$

show $\lim{a_n b_n} \to 0$.

I started my proof listing what I know:

  1. If $a_n$ is bounded then there is some number $M \gt 0$ such that $|a_n| \le M$ for all $n \in \mathbb{N}$.
  2. $|b_n - 0| \lt \epsilon$.

My proof proceeded directly from this:

$|a_n b_n| = |a_n| |b_n| \lt \epsilon$

$|b_n| \lt \frac{\epsilon}{|a_n|}$

And since $|b_n - 0| = |b_n|$ can be made arbitrarily small by (2) above we have shown that for all $n \ge N$ ($N \in \mathbb{N}$),

$|a_n b_n| = |a_n| |b_n| \lt |a_n| \frac{\epsilon}{|a_n|} = \epsilon$.

Our limit therefore converges to 0 as desired.

I was fairly satisfied with this proof, I had considered $M$ being able to be replaced by $|a_n|$ (as I didn't see a reason I couldnt do this), so this took advantage of (1) as well.

The author went about it in a different but similar way starting from the same "base":

$|a_n b_n - 0| = |a_n| |b_n| \le M|b_n|$ from (1)

Because $(b_n) \to 0$ we can pick an $N$ such that:

$|b_N| \lt \frac{\epsilon}{K}$

Finally, we can conclude that for this choice of $N$,

$|a_n b_n - 0| \lt K |b_n| \lt K \frac{\epsilon}{K} = \epsilon$

The author seemed to have just "dropped" the $|a_n|$ from the last part where he does $K |b_n|$?

Other than that weirdness (and the ham fisted nature of my newbie proof), is there any fundamental differences in our approach? Am I not able to let $M = |a_n|$?


Your proof is fine, but it's hard to read. Here's a revised version:

Let $M$ be a positive number such that $|a_{n}|\leq M$ for all $n$. Then, $$ \left|a_{n}b_{n}\right| =\left|a_{n}\right|\left|b_{n}\right| =M\left|b_{n}\right|. $$ For each $\epsilon>0$, pick $N$ large enough such that $|b_{n}|<\epsilon/M$ whenever $n\geq N$. It follows that $|a_{n}b_{n}|<M(\epsilon/M)=\epsilon$ whenever $n\geq N$. Therefore, $a_{n}b_{n}\rightarrow0$.

  • 1
    $\begingroup$ Small typo $\leq$ in middle equation. $\endgroup$ Feb 16 '19 at 3:32
  • $\begingroup$ @JoaquinSan: +1 thanks, fixed! Technically not wrong... ;-) $\endgroup$
    – parsiad
    Feb 16 '19 at 3:33
  • $\begingroup$ Thank you for your answer. An answer below says my proof is wrong due to the dependence of n in my proof in the epsilon half of the inequality. If you could address this for me and why it's right or wrong (relative to your answer here) I will happily mark this answered. $\endgroup$
    – CL40
    Feb 16 '19 at 8:16
  • $\begingroup$ It's because you say let $M=|a_n|$ so it depends on $n$. You can fix it by taking $M=\sup_n|a_n|$. $\endgroup$
    – parsiad
    Feb 16 '19 at 21:08
  • $\begingroup$ I meant the second one, not the first. $M$ is an upper bound not necessarily on of the $a_n$. $\endgroup$ Feb 17 '19 at 6:46

Your mistake is when you put $\dfrac{\epsilon}{\vert{a_n} \vert}$ this expression depends on $n$ so your inequality is only valid for that $n$.

When you use the fact that $\vert a_n \vert \leq M, \forall n \in \mathbb{N}$ then that number $M$ can be used at your convenience to find values ​​that meet the definition of limit.

What you want to have is that $\vert a_n b_n \vert < \epsilon$ for very large values ​​of $n$. So if you give some $ \epsilon >0$ the value $\dfrac{\epsilon}{M} >0$ and applying the definition of limit you should that : $\exists$ $n_0 \in \mathbb{N} / n>n_0 : \vert b_n \vert < \dfrac{\epsilon}{M}$.

Now for $n>n_o : \vert a_n b_n \vert = \vert a_n \vert \vert b_n \vert < M. \dfrac{\epsilon}{M} = \epsilon $ and you have what you were looking for.


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.