# Homework Help: Convergent series

1. Dec 3, 2013

### mahler1

The problem statement, all variables and given/known data.

Let $\{a_n\}_{n \in \mathbb N}$ a sequence of real numbers such that $lim_{n \to \infty} a_n=0$ and let $b_n=a_n+2a_{n+1}-a_{n+2}$.

Prove that $\sum_{n=1}^{\infty} a_n$ is convergent iff $\sum_{n=1}^{\infty} b_n$ is convergent.

The attempt at a solution.

Honestly, I don't have a clue how to prove this. I know that if $\{a_n\}_{n \in \mathbb N}$ is convergent, then for a given $ε>0$, there exists $N : m,n>N$ (suppose $n>m$) $\implies |\sum_{i=1}^ n a_i -\sum_{i=1}^ m a_i|=|a_{m+1}+...+a_n|<\epsilon$. I've tried to relate this to partial sums of the series $\sum_{n=1}^{\infty} b_n$ but I couldn't conclude anything. With the other implication I am also stuck. And I don't see how to use the fact that $lim_{n \to \infty} a_n=0$.

2. Dec 3, 2013

### Dick

You really shouldn't be stuck on the forward direction. If $\sum_{n=1}^{\infty} a_n=L$ then what are $\sum_{n=1}^{\infty} a_{n+1}$ and $\sum_{n=1}^{\infty} a_{n+2}$? So it should be pretty easy to guess what $\sum_{n=1}^{\infty} b_n$ is and prove it using partial sums. The reverse direction is the harder one. That's where you need $lim_{n \to \infty} a_n=0$. If not then a counterexample is $a_n=1$ for all n. Then $\sum_{n=1}^{\infty} b_n$ converges but $\sum_{n=1}^{\infty} a_n$ doesn't.

3. Dec 3, 2013

### mahler1

It's true, the forward direction wasn't difficult:

Lets prove that $\sum_{n=1}^{\infty} b_n=2L-a_1+a_2$

By hypothesis, $\sum_{n=1}^{\infty} a_n$ is convergent. Call its limit $L$, then given $\epsilon>0$, there exists $N_{\epsilon} : \forall n\geq N_{\epsilon} \implies |\sum_{i=1}^n a_i -L|<\dfrac{\epsilon}{3}$.

Then, for $n\geq N_{\epsilon}$, $|\sum_{i=1}^n b_i -(2L-a_1+a_2)|=|\sum_{i=1}^n a_i + 2\sum_{i=1}^n a_{i+1}-\sum_{i=1}^n a_{i+2}-(2L-a_1+a_2)|$. By an index change of the form $j=i+1$ and $k=i+2$, we have

$|\sum_{i=1}^n a_i + 2\sum_{i=1}^n a_{i+1}-\sum_{i=1}^n a_{i+2}-(2L-a_1+a_2)|=|\sum_{i=1}^n a_i + 2\sum_{j=1}^{n+1} a_j-2a_1-\sum_{k=1}^{n+2} a_k+ a_1+a_2-(2L-a_1+a_2)|=|\sum_{i=1}^n a_i-L + 2\sum_{j=1}^{n+1} a_j-2L+L-\sum_{k=1}^{n+2} a_k|\leq |\sum_{i=1}^n a_i-L|+2|\sum_{j=1}^{n+1} a_j-L|+|\sum_{k=1}^{n+2} a_k-L|<\dfrac{\epsilon}{3}+\dfrac{2\epsilon}{3}+\dfrac{\epsilon}{3}=\dfrac{4\epsilon}{3}$.

This proves $\sum_{n=1}^{\infty} b_n$ is convergent and that its limit is $2L-a_1+a_2$

I haven't retried the other direction. Now I'll put the matter in my hands (brain). Thanks!!!

Last edited: Dec 3, 2013
4. Dec 3, 2013

### Dick

Sort of, but you can't just reindex series like that without paying some attention to the terms you dropping by doing that. I think $\sum_{n=1}^{\infty} b_n=2L-a_0+a_1$.

Last edited: Dec 3, 2013
5. Dec 3, 2013

### mahler1

You're right, but I think the limit is $2L-a_1+a_2$, I suppose you were considering the first term of the sequence to start at $n=0$ and not $1$. Now I correct my post. By the way, I still don't know how to prove the other implication, could you give me one more hint?

6. Dec 3, 2013

### Dick

Yes, you are right. I was starting at n=0. Sorry. And I haven't even thought about going the other way yet. But I'm not going to do that until tomorrow. Yawn! That gives you some time to think about it and beat me to the punch.

7. Dec 3, 2013

### mahler1

Thanks for all the suggestions and corrections. I am going to bed now and see if I have some sort of brain illumination while sleeping.

8. Dec 4, 2013

### haruspex

it's essentially the same. What is the difference between the two sums taken from 1 to r?

9. Dec 4, 2013

### mahler1

Your question led me to this:

Lets call $s_n=\sum_{k=1}^n a_k$, $t_n=\sum_{k=1}^n b_k$. I'll prove that $\lim_{n \to \infty} 2s_n-t_n=a_1-a_2$. From this, one immediately could deduce that $\{s_n\}_{n \in \mathbb N}$ converges iff $\{t_n\}_{n \in \mathbb N}$ converges.

$2s_n-t_n=2\sum_{k=1}^n a_k-\sum_{k=1}^n b_k=2\sum_{k=1}^n a_k-\sum_{k=1}^n a_k-2\sum_{k=1}^n a_{k+1}+\sum_{k=1}^n a_{k+2}=2\sum_{k=1}^n a_k-\sum_{k=1}^n a_k-2\sum_{k=1}^n a_{k}-2a_{n+1}+2a_1+\sum_{k=1}^n a_{k}+a_{n+2}+a_{n+1}-a_1-a_2=2s_n-s_n-2s_n+s_n+(a_{n+2}-a_{n+1}+a_1-a_2)=a_{n+2}-a_{n+1}+a_1-a_2$.

Then, $\lim_{n \to \infty} 2s_n-t_n=\lim_{n \to \infty} a_{n+2}-a_{n+1}+a_1-a_2=a_1-a_2$.

It follows that $\{s_n\}_{n \in \mathbb N}$ converges iff $\{t_n\}_{n \in \mathbb N}$ converges.

10. Dec 4, 2013

Looks good.