# Explain the steps for finding the upper bound of a sequence

For example we have the two sequences

$A(n) = \dfrac{3n^2+n+2}{n^2-n+1}$

and

$B(n) = 1+(-1/2) + (-1/2)^2 + (-1/2)^3 +\cdots+ (-1/2)^{n-1}$

I can't figure out the steps to work out the upper bound of these two sequences. I've searched so much but I couldn't find any clear examples of that type of questions about bounded sequences.

-
So, I have edited this question and rendered equation in $\LaTeX$, Hope you'll implement them in your next posts –  user21436 Dec 27 '11 at 17:58
Thanks i didn't know how to do it –  Vaios Argiropoulos Dec 27 '11 at 17:59
If you have an upper bound on a sequence, then any larger number is also an upper bound for the sequence. Do you want the least upper bound? –  Dylan Moreland Dec 27 '11 at 18:18
Yeah i know that. I want some steps two prove both "general" bound and least upper bound. –  Vaios Argiropoulos Dec 27 '11 at 18:23

For $A(n)$, you probably know (or can believe) that $A(n)\to 3$ as $n\to\infty$. Hence $A(n)\le4$ for large $n$. More precisely, write $A(n) = 3+\dfrac{4n-1}{n^2-n+1}$ and find $N$ such that $4n-1\le n^2-n+1$ for all $n\ge N$. Then $A(n)\le 4$ for all $n\ge N$.

-
how did you figure out 4n-1 ? –  Parhs Sep 4 '14 at 6:12
@Parhs, polynomial long division. –  lhf Sep 4 '14 at 10:57

For the first, it may help to look at $A(n+1) - A(n)$ and show that $A(n)$ is decreasing for $n \ge 2$.

For the second, do you know a formula for the sum of a geometric series?

-

Generally, you can make a conjecture about the upper bound of the sequence (try looking at it like a regular function and taking the limit). For example, you can guess that $A(n) \rightarrow 3$ as $n \rightarrow \infty$. Then you must show, by definition of the upper bound of a sequence, that

$\forall\epsilon>0$, $\exists s(\epsilon)\in A(n): inf[A(n)]-s(\epsilon)<\epsilon$

,where $inf[A(n)]$ is the upper bound. Does that make intuitive sense? If you take $\epsilon$ to be given, it will help you to draw a picture on the number line to see exactly what you're looking for. As a further hint, it will help you to know that there is an Archimedean property that says $\forall\epsilon\in\mathbb{R},\exists n\in\mathbb{N}:\frac{1}{n}<\epsilon$

-