Sign up ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Suppose that $(a_n)$ and $(b_n)$ be sequences such that $\lim (a_n)=0$ and $\displaystyle \lim \left( \sum_{i=1}^n b_i \right)$ exists. Define $c_n = a_1 b_n + a_2 b_{n-1} + \dots + a_n b_1$. Prove that $\lim (c_n)=0$

share|cite|improve this question
Is it not $lim( \sum_1^n a_i=0)$? – Inceptio Apr 7 '13 at 15:43
@Inceptio, that is another (easier?) question. – vonbrand Apr 7 '13 at 15:54
@vonbrand: No. Was just trying use $AM-GM$ relation so that I could find bounds. – Inceptio Apr 7 '13 at 15:57

5 Answers 5

Counter example:if $a_n,b_n=\frac{(-1)^n}{\sqrt{n}}$,Then $\lim_{n\to \infty}a_n=0$ and $\sum_{i=1}^n\frac{(-1)^i}{\sqrt{i}}$ is convergent (Leibniz test) and we have $c_n=\sum_{k=1}^na_kb_{n-k+1}=(-1)^{n+1}\sum_{k=1}^n\frac{1}{\sqrt{(n-k+1)(k)}}$,now we consider $k(n-k+1)\le(n-k+1)(k+1)=(\frac{n}{2}+1)^2-(\frac{n}{2}-k)^2\le(\frac{n}{2}+1)^2$ and $\mid c_n \mid\ge\frac{2(n+1)}{n+2}$, hence $c_n$ isn't convergent to $0$.

share|cite|improve this answer

Suppose moreover that $\displaystyle \sum\limits_{k \geq 1} |b_k|<+ \infty$.

Let $\epsilon>0$ and $N \geq 0$ such that $\displaystyle \sum\limits_{k \geq N } |b_k| <\epsilon$. Also, let $M>0$ be such that $|a_k|<M$ for all $k \geq 0$.

Then $$\left| \sum\limits_{k=1}^n a_{n-k+1}b_k \right| \leq M \sum\limits_{k=N}^n |b_k| + \max\limits_{k \geq n-N+2} |a_k| \cdot \sum\limits_{k=1}^{N-1} |b_k|$$

For $n$ large enough, you get $$\left| \sum\limits_{k=1}^n a_{n-k+1}b_k \right| \leq (M+1)\epsilon$$

Therefore, it is true that $\lim\limits_{n\to + \infty} c_n=0$ when $\displaystyle \sum\limits_{n \geq 1} b_n$ is absolutely convergent.

share|cite|improve this answer

Suppose $b_i \geq 0$

I will give something of a hint.

Two observations will lead you to the answer.

$1)$ Since $\lim a_n=0$, $|a_n|$ is bounded

$2)$ The tail of $\sum _{n=0}^{\infty}b_n$ tend to zero. Which mean for $\epsilon >0$ you can find $n_0$ such that $\sum_{n=k}^{m}b_n<\epsilon \forall k,m \geq n_0$.

One good way to start is to use the condition $\lim a_n=0$ and try to make the most out of it. Try to find what part of $c_n$ is pushed to zero by $a_n$. When you find one the role of $a_n$ the other part should go to zero due to the conditions on $b_n$.

Of course this is a process, maybe the part you found is sent to zero by $a_n$ is not the optimal, which could imly the the rest of the $c_n$ cannot be sent to zero by $b_n$. Then you go back and try to increase to role of $a_n$.

Let's start: $\lim b_1 a_n=0$, also $\lim b_2 a_{n-1}=0$ and so on. That shows the right part of $c_n = a_1 b_n + a_2 b_{n-1} + \dots + a_n b_1$ goes to zero. A first good guess of what the role of $a_n$ is following: fix a $k$ then $\lim (b_1 a_n +b_2 a_{n-2}+ \cdots +b_ka_{n-k+1})=0 $

Now use to triangle inequality \begin{align} |c_n |&=|a_1 b_n + a_2 b_{n-1} + \dots + a_n b_1|\\ &\leq |(b_1 a_n +b_2 a_{n-2}+ \cdots +b_ka_{n-k+1})|+|b_{k+1}a_{n-k} + \ldots b_1 a_n|\\ &\leq |(b_1 a_n +b_2 a_{n-2}+ \cdots +b_ka_{n-k+1})|+\sup \{|a_n|\}(\sum_{n=k}^{m}b_n) \end{align}

share|cite|improve this answer
You are assuming that $\sum b_i$ converges absolutely, like Seirios. Is this correct? – robjohn Apr 9 '13 at 15:13
@PeterTamaroff: yes. I was just pointing out that this is the same assumption that Seirios made. In fact, Seirios' answer is a bit more inclusive since $b_i\ge0$ implies absolute convergence, but not vice-versa. – robjohn Apr 10 '13 at 20:30

Suppose $a_n\to0$ monotonically.

Let $B_k=\sum\limits_{j=k}^\infty b_j$. Since the sum of $b_j$ converges, $\bar{B}_n=\sup\limits_{k\ge n}\left|B_k\right|$ decreases monotonically to $0$.

Let $\epsilon\gt0$. Choose $N$ so that if $n\ge N$, $a_{n/2}\bar{B}_0\le\epsilon/4, a_0\bar{B}_{n/2}\le\epsilon/4$ $$ \begin{align} |c_n|& =\left|\sum_{k=0}^na_kb_{n-k}\right|\\ &=\left|\sum_{k=0}^na_k(B_{n-k}-B_{n-k+1})\right|\\ &=\left|a_nB_0-a_0B_{n+1}+\sum_{k=0}^{n-1}a_kB_{n-k}-\sum_{k=0}^{n-1}a_{k+1}B_{n-k}\right|\\ &=\left|a_nB_0-a_0B_{n+1}+\sum_{k=0}^{n-1}(a_k-a_{k+1})B_{n-k}\right|\\ &=\left|a_n\bar{B}_0-a_0\bar{B}_n+\sum_{k=0}^{n/2-1}(a_k-a_{k+1})B_{n-k}+\sum_{k=n/2}^{n-1}(a_k-a_{k+1})B_{n-k}\right|\\ &\le|a_n\bar{B}_0|+|a_0\bar{B}_n|+|a_0\bar{B}_{n/2}|+|a_{n/2}\bar{B}_0|\\ &\le\epsilon \end{align} $$ Thus, $c_n\to0$.

share|cite|improve this answer

I am sorry all, the correct one should be $b_n$ is positive. I have got my solution to this problem. WLOG $\displaystyle \sum_{i=1}^{\infty} b_i=1$. Given $\epsilon >0$. Then there exists $M$ such that for all $n>M$, $|a_n|< \epsilon$. Next, there exist $N$ such that for all $n \geq m>N$, $b_m+ \dots b_n < \epsilon/(|a_1|+ \dots |a_N|+1)$. Now, consider $n>M+N$, then $|c_n| \leq (|a_1|b_n+ \dots |a_N|b_{n-N+1})+(|a_{N+1}|b_{n-N}+ \dots |a_n|b_1) \leq \epsilon + \epsilon$. Since $\epsilon$ is arbitrary, $(c_n)$ goes to 0.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.