Definition – Limit of a sequence

Contents
1. Definition

(!) Convergence and divergence both direct to this page.

Definition

Let s_n be a sequence. If there exists an L\in{\mathbb{R}} such that

\forall \epsilon >0,~ \exists N>0 \text{~s.t~} n>N \implies |s_n - L| < \epsilon.

Then we write \lim_{n \rightarrow \infty} s_n = L, and we say that s_n converges. If no such L exists, then we say that s_n diverges.

Theorem – Every convergent sequence is bounded.

Proof – Let (s_n) be a sequence. We have from the definition of convergence that for every \epsilon > 0 , there exists an N>0 such that

n>N \implies |s_n - L| < \epsilon\quad \bf{(1)}

Where L is the limit of (s_n) . Now, \bf{(1)} can be rewritten as

n>N \implies -\epsilon < s_n - L <\epsilon \implies -L - \epsilon < s_n < L + \epsilon

However, this gives two inequalities, firstly -L < s_n + \epsilon \implies -L \leq s_n, and s_n < L + \epsilon \implies s_n \leq L. These two inequalities prove that (s_n) is bounded above/below by -L and L respectively.

Theorem – If (s_n) is a convergent sequence with limit L, then every subsequence of (s_n) converges to L.

Theorem – Every convergent sequence, (s_n) , has a convergent subsequence.

Proof – We have two separate proofs based on whether (s_n) is finite or infinite. If (s_n) is finite, then we have that there is an infinite number of i\in\mathbb{N} such that s_i = x for some x. Thus, (s_n) converges to x because for all n>N_i, |s_n - L| = 0 < \epsilon.

Now, if (s_n) is infinite, since it is bounded by assumption, there exists at least one accumulation point, x (often referred to as a limit point) of (s_n). But then, we may construct consecutive neighborhoods N(x,\epsilon). But then (s_n) \rightarrow L.

Theorem – Every unbounded sequence, (s_n), contains a monotone subsequence that has either \pm\infty as its limit.

Proof – We already know that an unbounded increasing sequence has limit \infty, and that an unbounded decreasing sequence has limit -\infty, so it just remains to construct such a subsequence from (s_n). We may do so as a consequence of the hypothesis. We may easily choose one element to be in the subsequence. Then, assume we already have taken k elements in the subsequence, (s_{n_1}, ..., s_{n_k}). Since (s_n) is infinite and unbounded, it follows that there exists another element, and if it were the highest element in (s_n), then (s_n) would be bounded. It follows that another element s_{n_{k+1}} exists such that s_{n_k} \leq s_{n_{k+1}}. But this completes the proof, as we have shown that such a subsequence may be constructed.

A similar method may be used if we assume that (s_n) is unbounded below.