4

The question asks me to prove that if $s_n ≤ b$ for all but finitely many $n$, then $\lim s_n ≤ b$ where $(s_n)$ be a sequence that converges. . Here is how I did it but im not sure if its entirely correct. I used proof by contradiction.

Suppose $\lim s_n>b$ and $s_n \leq b$ for all but finitely many $n$. Let $S=\lim s_n$. By definition we have for every $n>N$ which implies $|s_n-s|<\epsilon$. Let $\epsilon= b-2s+s_n$. Now we get $s_n-s<\epsilon=b-2s+s_n$ which simplifies to $s<b=\lim s_n<b$ which is a contradiction.

user60887
  • 2,935
  • 1
    If the only assumption is that $s_n\le b$ for all but finitely many $n$, we do not know that $\lim s_n$ need exist. Also, do you mean $\lim_{n\rightarrow \infty} s_n$? – vadim123 Jun 04 '13 at 21:01
  • You should define $\epsilon$ before you define $N$, since $N$ is not independent of $\epsilon$. – Thomas Andrews Jun 04 '13 at 21:02
  • So should it go Suppose $lims_n>b$ and $s_n≤b$ for all but finitely many n. Let $s=lims_n$ and let $ϵ=b−2s+s_n$. By definition there exists a natural number $N$ such that $n>N$ and $|s_n−s|<ϵ$. Now we get $s_n−s<ϵ=b−2s+s_n$ which simplifies to $s<b=lims_n<b$ which is a contradiction?. – user60887 Jun 04 '13 at 21:13

2 Answers2

1

Let $$S=\{n\in\mathbb N\quad|\quad s_n>b\}$$ This set is finite subset of $\mathbb{N}$ so let $n_0=\max S$ hence we have $s_n\leq b\quad \forall n>n_0$ and we conclude.

1

Your idea of proof by contradiction might be better modified as follows:

If the limit of $s_n$ is greater than $b$, it can be written as $b+c$ for some $c \in \mathbb{R}^{+}$.

Since this is the limit of the sequence, we can find infinitely many $s_n$ that are arbitrarily close to $b+c$; in particular, we can find infinitely many $s_n$ that are greater than $b$, which contradicts the given that there are only finitely many such $s_n$.

Therefore, our supposition that the limit is greater than $b$ must have been incorrect, hence the limit is less than or equal to $b$ as desired. QED

  • How do you say that there can be infinitely many $s_n$ near $b+c$? Is there a certain theorem or statement that you can supply for this? – sprajagopal Aug 18 '14 at 07:02