7
$\begingroup$

Let $a_n$ be any sequence converging to $a$ when $n \to \infty$.

Can you rewrite $a_n$ so that it is the sum of two other sequences? $$a_n=b_n + c_n,$$ with $b_n=b$ for every $n \in \mathbb{N}$ and $c_n\to 0$ as $n\to \infty$.

In other words: Is a converging sequence ($a_n$) actually a null sequence ($c_n$) "shifted" by a constant ($b$)?

Or is there any counterexample where one is not allowed to do so?

$\endgroup$
  • 2
    $\begingroup$ You probably mean $b=a$, otherwise $c_n$ cannot converge to $0$. $\endgroup$ – Yves Daoust Jan 2 '16 at 15:57
  • 1
    $\begingroup$ "a null sequence" isn't defined. $\endgroup$ – Yves Daoust Jan 2 '16 at 15:57
  • 3
    $\begingroup$ @YvesDaoust ProofWiki (proofwiki.org/wiki/Definition:Null_Sequence) would disagree with you. Although I, myself, would have upon hearing "null sequence" out of context would have assumed it meant {a_i} where all a_i = 0. In context, I had no such impression however. $\endgroup$ – fleablood Jan 2 '16 at 16:42
  • $\begingroup$ @fleablood: Ok, my bad. $\endgroup$ – Yves Daoust Jan 2 '16 at 16:43
  • 1
    $\begingroup$ Well, I wasn't familiar with the term either. I was going to chastise about how "obviously" a null sequence would mean only zero terms and ... $\endgroup$ – fleablood Jan 2 '16 at 17:07
13
$\begingroup$

Set $b_n = a$ and $c_n = a_n - a$ for all $n \in \mathbb{N}$. Then $a_n = b_n + c_n$ for all $n \in \mathbb{N}$ and $c_n = a_n - a \to a-a = 0$ for $n \to \infty$.

$\endgroup$
  • $\begingroup$ Yeah. It's pretty easy when you can just define c<sub>n</sub> to be the set that satisfies this condition. $\endgroup$ – Daniel Jan 2 '16 at 20:19
  • $\begingroup$ Of course, this only works in spaces where there is actually a subtraction operation. Converging sequences also exist in metric spaces with no such concept! But granted, the question pretty much asserts a vector space structure. $\endgroup$ – leftaroundabout Jan 3 '16 at 17:29
  • $\begingroup$ Any topologcial abelian group should do the trick. (We don’t really need abelian, but then we deal with different sequences, depending on which side we choose for our construction.) $\endgroup$ – Jendrik Stelzner Jan 3 '16 at 17:34
6
$\begingroup$

Yes, you can do that. Simply take $b_n=a,c_n=a_n-a$. By basic properties of limits $$\lim\limits_{n\rightarrow\infty}c_n=\lim\limits_{n\rightarrow\infty}(a_n-a)=(\lim\limits_{n\rightarrow\infty}a_n)-a=a-a=0$$

$\endgroup$
4
$\begingroup$

Let $(a_n)$ be a sequence converging towards $a$ and let define the following sequences: $$\forall n\in\mathbb{N},b_n:=a,c_n:=a_n-a.$$ One has: $$\forall n\in\mathbb{N},a_n=b_n+c_n$$ and $(c_n)$ is converging towards $0$.

$\endgroup$
3
$\begingroup$

For a constant $b$ any number $a$ (whether it's a term in a sequence or not) can be written as $a = b + c$ where $c = a - b$ so any sequence $\{a_n\}$ can be written as $\{b + c_n\}$ where $c_n = a_n - b$ and in particular if $\lim a_n = a$ then the sequence can be written as $\{a + c_n\}$ where $c_n = a_n - a$. And clearly $\lim \{a_n\} = \lim \{a + c_n\} = a$.

So your question boils down to does $\lim\{b + c_n\} = b + \lim\{c_n\}$? And therefore if $b = a = \lim a_n$ does $\lim c_n = 0$.

This should be a basic proposition early on in the study of convergent sequence and the answer is: yes.

$|a - a_n| = |(a -b) - (a_n - b)| = |(a - b) - c_n|$. So whatever $\epsilon$, $N$, $n > N$ crap that you can say about $a$ and $a_n$ can also be said about $(a-b)$ and $c_n$.

So if $c_n = a_n - b$ then $\{a_n\} \rightarrow a \iff \{c_n\} \rightarrow (a - b)$.

$\endgroup$
2
$\begingroup$

You are allowed to add, subtract, multiply or divide all terms of a converging sequence with a constant, and you get another converging sequence that converges to $L+C,L-C,L\cdot C, L/C$. $C/L$ also works, provided the original sequence has no zero.

In your case, if $a_n\to a$, then $c_n:=a_n-a\to 0$ and $c_n+a\to a$. You can also have $c_n+b\to a$, provided that $c_n:=a_n-b$, and then $c_n\to a-b$.

$\endgroup$
  • $\begingroup$ Original sequence must have no zero and the limit has to not be zero. $\endgroup$ – Wojowu Jan 2 '16 at 16:03
  • $\begingroup$ @Wojowu: let's say that the sequence has no zero, even at infinity ;-) $\endgroup$ – Yves Daoust Jan 2 '16 at 16:04

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.