Convergent Sequences

Definition 1. A sequence of real numbers (sn ) is said to converge to a real number s if

∀ε > 0, ∃N ∈ N, such that n > N implies |sn − s| < ε. (1)

When this holds, we say that (sn ) is a convergence sequence with s being its limit, and write
sn → s or s = limn→∞ sn . If (sn ) does not converge, then we say that (sn ) is a divergent
sequence.

We first show that one sequence (sn ) can not have two different limits. Suppose sn → s
and sn → t. Let ε > 0. Then 2ε > 0. Since sn → s, by definition there is N1 ∈ N such that
for n > N1 , |sn − s| < 2ε . Since sn → t, by definition there is N2 ∈ N such that for n > N2 ,
|sn − t| < 2ε . Here we use N1 and N2 in the two statements because the N coming from the
two limits may not be the same. Let N = max{N1 , N2 }. If n > N , then n > N1 and n > N2
both hold. So |sn − s| < 2ε and |sn − t| < 2ε , which by triangle inequality imply that
ε ε
|s − t| ≤ |sn − s| + |sn − t| < + = ε.
2 2
Now |s−t| < ε holds for every ε > 0. We then conclude that |s−t| = 0 (for otherwise |s−t| > 0,
we then get a contradiction by choosing ε = |s − t|). So s = t, and the uniqueness holds.
We will use the following tools to check whether a sequence converges or diverges.

1. the definition

2. basic examples

3. limit theorems

4. boundedness and subsequences.

We have stated the definition. Now we consider some examples.

Example 1. Let s ∈ R. If sn = s for all n, i.e., (sn ) is a constant sequence, then lim sn = s.

Proof. For any given ε > 0 we simply choose N = 1. If n > N , then |sn − s| = 0 < ε.
1
Example 2. We have n → 0.
1
Proof. Let ε > 0. By Archimedean property, there is N ∈ N such that N < ε. If n > N , then

1 1 1
| − 0| = < < ε.
n n N

Example 3. The following two sequences are divergent

1
 (i) (sn ) = ((−1)n ) = (−1, 1, −1, 1, −1, 1, . . . );

(ii) (sn ) = (n) = (1, 2, 3, 4, 5, 6, . . . ).

Proof. (i) We use the notation of subsequence and statement that will be proved later. Suppose
n1 < n2 < n3 < · · · is a strictly increasing sequence of indices, then (snk ) is a subsequence of
(sn ). We will prove a theorem, which asserts that, if (sn ) converges to s, then any subsequence
of (sn ) also converges to s. The sequence (sn ) = ((−1)n ) contains two constant sequences
(1, 1, 1, . . . ) (with nk = 2k) and (−1, −1, −1, . . . ) (with nk = 2k −1), which converge to different
limits. So the original (sn ) can not converge.
(ii) We use the following theorem. If (sn ) is convergent, then it is a bounded sequence. In
other words, the set {sn : n ∈ N} is bounded. So an unbounded sequence must diverge. Since
for sn = n, n ∈ N, the set {sn : n ∈ N} = N is unbounded, the sequence (n) is divergent.

Remark 1. This example shows that we have two ways to prove that a sequence is divergent:
(i) find two subsequences that convergent to different limits; (ii) show that the sequence is
unbounded. Note that the (sn ) in (i) is bounded and divergent. The (sn ) in (ii) is divergent,
but lim sn actually exists, which is +∞, and its every subsequence also tends to +∞. We will
define that limit later.

Now we state some limit theorems.

Theorem 1 (Theorem 9.1). Every convergent sequence is bounded.

Proof. Let (sn ) be a sequence that converges to s ∈ R. Applying the definition to ε = 1, we see
that there is N ∈ N such that for any n > N , |sn − s| < 1, which then implies that |sn | ≤ |s| + 1.
Let
M = max{|s1 |, |s2 |, . . . , |sN |, |s| + 1}.
The maximum exists since the set is finite. Then for any n ∈ N, |sn | ≤ M (consider the case
n ≤ N and n > N separately), i.e., −M ≤ sn ≤ M . So {sn : n ∈ N} is bounded.

Theorem 2 (Theorem 9.3). If (sn ) converges to s and (tn ) converges to t, then (sn + tn )
converges to s + t.

Proof. Let ε > 0. Then 2ε > 0. Since sn → s, there is N1 ∈ N such that for n > N1 , |sn −s| < 2ε .
Since tn → t, there is N2 ∈ N such that for n > N2 , |tn − t| < 2ε . Let N = max{N1 , N2 }. If
n > N , then n > N1 and n > N2 both hold, and so |sn − s| < 2ε and |tn − t| < 2ε , which together
imply (by triangle inequality) that
ε ε
|(sn + tn ) − (s + t)| ≤ |sn − s| + |tn − t| < + = ε.
2 2
So we have the desired convergence.

Theorem 3 (Theorem 9.4). If (sn ) converges to s and (tn ) converges to t, then (sn ·tn ) converges
to s · t.

2
Discussion. We need to bound |sn tn − st| from above for big n. We write

sn tn − st = sn tn − sn t + sn t − st = sn (tn − t) + t(sn − s).

By triangle inequality, we get

|sn tn − st| ≤ |sn (tn − t)| + |t(sn − s)| = |sn ||tn − t| + |t||sn − s|.

Since tn → t and sn → s, we know that |tn − t| and |sn − s| can be arbitrarily small if we choose
n big enough. Thus, if |sn | and |t| are not too big, then we can control the sum on the RHS
(righthand side). In fact, the size of |sn | can be controlled because of Theorem 9.1.

Proof. Since (sn ) is convergent, by Theorem 9.1, there is M > 0 such that |sn | ≤ M for every
ε
n. We may choose M big such that M ≥ |t|. Let ε > 0. Then 2M > 0. Since sn → s, there
ε
is N1 ∈ N such that for n > N1 , |sn − s| < 2M . Since tn → t, there is N2 ∈ N such that for
ε
n > N2 , |tn − t| < 2M . Let N = max{N1 , N2 }. If n > N , then n > N1 and n > N2 both hold,
ε ε
and so |sn − s| < 2M and |tn − t| < 2M , which together with |sn | ≤ M and |t| ≤ M imply that

|sn tn − st| ≤ |sn (tn − t)| + |t(sn − s)| = |sn ||tn − t| + |t||sn − s|
ε ε
≤ M |tn − t| + M |sn − s| < M +M = ε.
2M 2M

Corollary 1. If (sn ) converges to s, k ∈ R, and m ∈ N, then (ksn ) converges to ks and sm

n
converges to sm .
Proof. For the sequence (ksn ), we apply Theorem 9.4 to the sequence (tn ) with tn = k for all
n. For the sequence (sm
n ) we use induction. In the induction step, note that sn
m+1 = s ∗ sm
n n
and apply Theorem 9.4 to tn = sn m

Corollary 2. If (sn ) converges to s and (tn ) converges to t, then (sn − tn ) converges to s − t.

Proof. We write sn + tn = sn + (−1)tn and apply Theorem 9.3 and the previous corollary.

From this corollary we see that sn → s iff sn − s → 0. By the Theorem below, the latter
statement is equivalent to that |sn − s| → 0.
Theorem 4. (a) Suppose two sequences (sn ) and (tn ) satisfy that tn → 0 and |sn | ≤ |tn | for
all but finitely many n. Then sn → 0.
(b) For any sequence (sn ), sn → 0 if and only if |sn | → 0.
Proof. (a) Let N0 ∈ N be such that |sn | ≤ |tn | for n > N0 . Let ε > 0. Since tn → 0, there is
N1 ∈ N such that for n > N1 , |tn − 0| < ε. Let N = max{N0 , N1 }. For n > N , |sn | ≤ |tn | and
|tn − 0| < ε, which imply that |sn − 0| = |sn | ≤ |tn | = |tn − 0| < ε.
(b) From (a) we know that if |sn | = |tn | for all n, then sn → 0 iff tn → 0. We then apply
this result to tn = |sn | and use that ||sn || = |sn |.

3
Lemma 1 (Lemma 9.5). If (sn ) converges to s such that s 6= 0 and sn 6= 0 for all n, then
(1/sn ) converges to 1/s.
Discussion. We need to bound |1/sn − 1/s| from above for big n. We write
1 1 s − sn |sn − s|
− = = .
sn s sn s |sn ||s|
Since sn → s, |sn − s| can be arbitrarily small if we choose n big enough. Thus, if |sn | and |s|
are not too close to 0, then we can control the size of the RHS. This means that we need a
positive lower bound of the set {|s1 |, |s2 |, . . . }.

Proof. Since s 6= 0, we have |s| |s|

2 > 0. Since sn → s, applying the definition to ε = 2 , we get
N ∈ N such that for n > N , |sn − s| < |s| 2 , which then implies by triangle inequality that
|sn | ≥ |s| − |sn − s| > |s| − 2 = 2 . Let m = min{|s1 |, |s2 |, . . . , |sN |, |s|
|s| |s|
2 }. Then m exists and is
positive since the set is a finite set of positive numbers.
Let ε > 0. Then m|s|ε > 0. Since sn → s, there is N 0 ∈ N such that n > N 0 implies that
|sn − s| < m|s|ε, which together with |sn | ≥ m for all n implies that
1 1 |sn − s| |sn − s| m|s|ε
− = ≤ < = ε.
sn s |sn ||s| m|s| m|s|

Theorem 5 (Theorem 9.6). Suppose (sn ) converges to s and (tn ) converges to t. If s 6= 0 and
sn 6= 0 for all n, then (tn /sn ) converges to t/s.
Proof. By Lemma 9.5, (1/sn ) converges to 1/s. Applying Theorem 9.4 to the sequences (1/sn )
and (tn ), we get the conclusion.
3
Example 4. Derive lim 3n+1 4n +3n
7n−4 and lim n3 −6

Solution. We write
3n + 1 3 + 1/n 4n3 + 3n 4 + 3 ∗ (1/n)2
= , = .
7n − 4 7 + (−4) ∗ 1/n n3 − 6 1 + (−6) ∗ 1/n
We have shown that lim 1/n = 0. So (i) lim(3 + 1/n) = 3 + 0 = 3 and lim(7 + (−4) ∗ 1/n) =
7 + (−4) ∗ 0 = 7, which imply that lim 3n+1 7n−4 = lim(3 + 1/n)/ lim(7 + (−4) ∗ 1/n) = 3/7; (ii)
lim(4 + 3 ∗ (1/n)2 ) = 4 + 3 ∗ 02 = 4 and lim(1 + (−6) ∗ 1/n) = 1 + (−6) ∗ 0 = 1, which imply
3
that lim 4nn3+3n
−6
= lim(4 + 3 ∗ (1/n)2 )/ lim(1 + (−6) ∗ 1/n) = 4.

We now state some theorems about the relation between limits and orders.
Theorem 6 (Exercise 8.9). (a) If (sn ) converges to s, and there is N0 ∈ N such that sn ≥ 0
for all n > N0 , then s ≥ 0.
(b) Suppose (sn ) converges to s and (tn ) converges to t. If there N0 ∈ N such that sn ≤ tn
for all n > N0 , then s ≤ t.

4
Proof. (a) We prove by contradiction. Suppose s < 0. Let ε = |s| = −s > 0. Since sn → s,
there is N ∈ N such that for n > N , |sn − s| < ε, which implies that sn < s + ε = 0. Let
n = max{N, N0 } + 1. Then n > N0 and n > N . From n > N0 we get sn ≥ 0; from n > N we
get sn < 0. This is the contradiction.
(b) Applying (i) to the sequence (tn − sn ) we conclude that its limit t − s is nonnegative.

For x ∈ [0, ∞) and n ∈ N, the power root x1/n is defined as the unique y ∈ [0, ∞) such that
yn = x. The uniqueness of such y follows from the fact that if 0 ≤ y1 < y2 , then y1n < y2n . The
existence follows from the “Intermediate Value Theorem” for continuous function f (x) = xn ,
which will be stated and proved later. We now just accept the existence of x1/n for any
1/n 1/n
x ∈ [0, ∞). It is clear that 0 ≤ x1 < x2 implies that 0 ≤ x1 < x2 . We restrict our attention
to [0, ∞) although in the case that n is an odd number, we can also define x1/n for x < 0.
√
When n = 2, x1/2 is often written as x. We have the following theorem.
√
Theorem 7 (Example 5). Suppose (sn ) converges to s and sn ≥ 0 for all n. Then ( sn )
√
converges to s.
√ √
Discussion We want to bound | sn − s| from above for big n. It is useful to note the equality
√ √ √ √ √ √
( sn − s)( sn + s) = ( sn )2 − ( s)2 = sn − s.

Taking absolute value, we get

√ √ √ √
| sn − s| · | sn + s| = |sn − s|.
√
If s > 0, then
√ √ |sn − s| |s − s|
| sn − s| = √ √ ≤ n√ .
sn + s s
√ √
Proof. By Theorem 6, s ≥ 0. First suppose s > 0. Then s > 0. Let ε > 0. Then sε > 0.
√
Since sn → s, there is N ∈ N such that for n > N , |sn − s| < sε, which implies that
√
√ √ |sn − s| |sn − s| sε
| sn − s| = √ √ ≤ √ < √ = ε.
sn + s s s
√
We leave the proof in the case s = 0 as an exercise. Note that for x ≥ 0, x < ε iff x2 < ε.