Sequence Series
Sequence Series
UNIT FOUR
SEQUENCES AND SERIES
4.1 Sequences
Definition: A sequence is a function whose domain is the set 𝑁 of natural numbers. If 𝑓 is such a
sequence, let 𝑓(𝑥) = 𝑥𝑛 denote the value of the sequence 𝑓 at 𝑛 ∈ ℕ. In this case we denote the
sequence 𝑓 by (𝑥)∞𝑛=1 of simply (𝑥𝑛 ).
An infinite sequence is an unending set of real numbers which are determined according to some rule.
A sequence is normally defined by giving a formula for the 𝑛𝑡ℎ term.
Examples
𝑛 1 2 3
1) (𝑛+1) is the sequence (2 , 3 , 4 . … )
2) (−1𝑛 ) is the sequence (−1, 1, −1, 1, … . )
3) (2𝑛 ) is the sequence (2, 4, 8, … . . )
𝑥𝑛 +𝑥𝑛+1
We can also use recursive formulas e.g. 𝑥𝑛+1 = were 𝑥1 = 0 and 𝑥2 = 1, then the terms
3
1 4 7
of the sequence (𝑥𝑛 ) are (0, 1, 3 , 7 , 27 , … . . ).
Remark
(1) The order of the terms of the sequence is an important part of the definition of the sequence.
For example, the sequence (1, 5, 7, … . ) is not the same as the sequence (1, 7, 5, … . ).
(2) There is a distinction between the terms of a sequence and the values of a sequence. A
sequence has infinitely many terms while its values may or may not be finite.
(3) It is not necessary for the terms of a sequence to be different. For example, (1, 2, 2, 2, 2, … … . )
is a particularly good sequence.
Exercise 4.1
Write down the first five terms of the following sequences
𝑛2 _2𝑛 𝑐𝑜𝑠 𝑛𝜋 𝑛𝜋
a) ( ) b) ( ) c) (𝑠𝑖𝑛 )
3𝑛 𝑛2 2
It is easy to see that a sequence (𝑥𝑛 ) is bounded if and only if there is a positive real number 𝑀such
that |𝑥𝑛 | ≤ 𝑀 for all 𝑛 ∈ ℕ.
24
REAL ANALYSIS 1 Issa Ndungo
Examples
1 1
a) The sequence (𝑛) is bounded since 0 < 𝑛 ≤ 1 for all 𝑛 ∈ ℕ.
b) The sequence (−1𝑛 ) is not bounded above and is not bounded below.
1
c) The sequence (𝑛 + 𝑛) is bounded below by 2 but is not bounded above.
Note that if a sequence does not converge to a real number, it is said to diverge.
Definition: A sequence (𝑥𝑛 ) is said to diverge to ∞ denoted by 𝑥𝑛 → ∞ as 𝑛 → ∞ if for any
particular real number 𝑚 there is an 𝑁 ∈ ℕ such that 𝑥𝑛 > 𝑚 for all 𝑛 ≥ 𝑁.
Similarly, (𝑥𝑛 ) diverges to −∞ denoted by 𝑥𝑛 → −∞ as 𝑛 → ∞ if for any particular real number
𝑘 there is an 𝑁 ∈ ℕ such that 𝑥𝑛 < 𝑘 for all 𝑛 ≥ 𝑁.
Example 4.1
1. Show that a sequence (𝑥𝑛 ) converges to zero if and only if the sequence (|𝑥𝑛 |) converges to zero.
Solution
Assume that the sequence (𝑥𝑛 ) converges to zero. Then given 𝜖 > 0, there exists a natural number 𝑁
(which depends on 𝜖) such that |𝑥𝑛 − 0| < 𝜖 ∀𝑛 ≥ ℕ.
Now for all 𝑛 ≥ 𝑁 we have ||𝑥𝑛 | − 0| = |𝑥𝑛 | < 𝜖 That is the sequence (|𝑥𝑛 |) converges to zero.
For the converse, assume that the sequences (|𝑥𝑛 |) converges to zero. That is 𝜖 > 0, there exists a
natural number 𝑁 (which depends on 𝜖) such that:
||𝑥𝑛 | − 0| = |𝑥𝑛 | < 𝜖 ∀𝑛 ≥ ℕ. It follows that the sequence 𝑥𝑛 converges to zero
1
2. Show that 𝑙𝑖𝑚 𝑛 = 0
𝑛→∞
Solution
1
Let 𝜖 > 0 be given. We can find a 𝑁 ∈ ℕ such that |𝑛 − 0| < 𝜖 ∀ 𝑛 ≥ 𝑁 . By Archimedean property,
1
there is an 𝑁 ∈ ℕ such that 0 ≤ 𝑁 < 𝜖.
1 1 1
Thus if 𝑛 ≥ 𝑁 then we have that |𝑛 − 0| = 𝑛 ≤ 𝑁 < 𝜖.
1
That is 𝑙𝑖𝑚 𝑛 = 0.
𝑛→∞
25
REAL ANALYSIS 1 Issa Ndungo
1
3. Show that 𝑙𝑖𝑚 (1 − )=1
𝑛→∞ 2𝑛
Solution
1
Let 𝜖 > 0 be given. We need to find an 𝑁 ∈ ℕ such that |(1 − 2𝑛) − 1| < 𝜖 ∀ 𝑛 ≥ 𝑁
1 1 1
Noting that: |(1 − 2𝑛) − 1| = 2𝑛 = (1+1)𝑛 and
26
REAL ANALYSIS 1 Issa Ndungo
Theorem 4.1
Let (𝑠𝑛 ) and (𝑡𝑛 ) be sequences of real numbers and let 𝑠 ∈ ℝ if for some positive real number 𝑘 and
some 𝑁1 ∈ ℕ we have |𝑠𝑛 − 𝑠| ≤ 𝑘|𝑡𝑛 | for all 𝑛 ≥ 𝑁1 and if 𝑙𝑖𝑚 𝑡𝑛 = 0 then 𝑙𝑖𝑚 𝑠𝑛 = 𝑠.
𝑛→∞ 𝑛→∞
Proof
𝜖
Let 𝜖 > 0 be given. Since 𝑡𝑛 → 0 as 𝑛 → ∞, there exist an 𝑁𝑛 ∈ ℕ such that |𝑡𝑛 | < 𝑘 for all 𝑛 ≥ 𝑁2 .
𝜖
Let 𝑁 = 𝑚𝑎𝑥(𝑁1 , 𝑁2 ), then for all 𝑛 ≥ 𝑁 we have |𝑠𝑛 − 𝑠| ≤ 𝑘|𝑡𝑛 | < 𝑘 . 𝑘 = 𝜖.
Thus 𝑙𝑖𝑚 𝑠𝑛 = 𝑠. ∎
𝑛→∞
Example 4.2
𝑛
Show that 𝑙𝑖𝑚 √𝑛 = 1
𝑛→∞
Solution
𝑛 𝑛
Since √𝑛 ≥ 1 for each 𝑛 ∈ ℕ, there is a nonnegative real number 𝑎𝑛 such that √𝑛 = 1 + 𝑎𝑛 . Thus,
by binomial theorem we have:
𝑛
𝑛 𝑘 𝑛(𝑛 − 1)𝑎𝑛2 𝑛(𝑛 − 1)𝑎𝑛2
𝑛 = (1 + 𝑎𝑛 )𝑛 = ∑ ( ) 𝑎𝑛 = 1 + 𝑛𝑎𝑛 + 𝑛
+ ⋯ + 𝑎𝑛 ≥ 1 +
𝑘 2 2
𝑘=1
2
𝑛(𝑛−1)𝑎𝑛 2 2
Therefore, 𝑛 − 1 ≥ hence 𝑎𝑛2 ≤ 𝑛 or 𝑎𝑛 = √𝑛 for all 𝑛 ≥ 2
2
𝑛 2 2 𝑛
Now since | √𝑛 − 1| = |𝑎𝑛 | = 𝑎𝑛 ≤ √𝑛 and 𝑙𝑖𝑚 √𝑛 = 0 we have by theorem 4.1 that 𝑙𝑖𝑚 √𝑛 = 1
𝑛→∞ 𝑛→∞
Proof
Let 𝜖 > 0 be given. Then there exist natural numbers 𝑁1 and 𝑁2 such that
𝜖 𝜖
|𝑠𝑛 − ℓ1 | < for all 𝑛 ≥ 𝑁1 and |𝑠𝑛 − ℓ2 | < for all 𝑛 ≥ 𝑁2 .
2 2
Thus, |ℓ1 − ℓ2 | = 𝜖 and since 0 ≤ |ℓ1 − ℓ2 | < 𝜖 holds for every 𝜖 > 0, we have ℓ1 − ℓ2 = 0 and so
ℓ1 = ℓ2 . Thus, a sequence (𝑠𝑛 ) converges to only and only one limit (the limit of a sequence (𝑠𝑛 ) is
unique) ∎
Proposition 4.1
A sequence (𝑥𝑛 ) converges to ℓ ∈ ℝ if and only if for each 𝜖 > 0, the set {𝑛: 𝑥𝑛 ∉ (ℓ − 𝜖, ℓ + 𝜖} is
finite.
27
REAL ANALYSIS 1 Issa Ndungo
Theorem 4.3
Every convergent sequence of real numbers is bounded.
Proof
Let (𝑠𝑛 ) be a sequence of real numbers which converges to 𝑠, then with 𝜖 = 1 there exists an 𝑁 ∈ ℕ
such that |𝑠𝑛 − 𝑠| < 1 ∀ 𝑛 ≥ 𝑁
By the triangle inequality, we have that: |𝑠𝑛 | ≤ |𝑠𝑛 − 𝑠| + |𝑠| ≤ 1 + |𝑠| for all 𝑛 ≥ 𝑁
Let 𝑀 = 𝑚𝑎𝑥{|𝑠1 |, |𝑠2 |, … |𝑠𝑁 |, |𝑠| + 1}. Then |𝑠𝑛 | ≤ 𝑀 for all 𝑛 ∈ 𝑁. That is the sequence (𝑠𝑛 ) is
bounded. ∎
The converse of Theorem 4.3 is not necessarily true. There are sequences which are bounded but do
not converge. E.g. the sequences (−1𝑛 ) is bounded but not convergent.
Theorem 4.4 (Squeeze theory on limits)
Suppose that (𝑠𝑛 ), (𝑡𝑛 ) and (𝑢𝑛 ) are sequences such that 𝑠𝑛 ≤ 𝑡𝑛 ≤ 𝑢𝑛 for all 𝑛 ∈ 𝑁. If
𝑙𝑖𝑚 𝑠𝑛 = ℓ = 𝑙𝑖𝑚 𝑢𝑛 , then 𝑙𝑖𝑚 𝑡𝑛 = ℓ.
𝑛→∞ 𝑛→∞ 𝑛→∞
Proof.
Let 𝜖 > 0 be given. Then there exist 𝑁1 , 𝑁2 ∈ ℕ such that:
|𝑠𝑛 − ℓ| < 𝜖 for all 𝑛 ≥ 𝑁1 and |𝑢𝑛 − ℓ| < 𝜖 for all 𝑛 ≥ 𝑁2
That is ℓ − 𝜖 < 𝑠𝑛 < ℓ + 𝜖 for all 𝑛 ≥ 𝑁1 and ℓ − 𝜖 < 𝑢𝑛 < ℓ + 𝜖
Let 𝑁 = 𝑚𝑎𝑥{𝑁1 , 𝑁2 }. Then for all 𝑛 ≥ 𝑁, we have
ℓ − 𝜖 < 𝑠𝑛 ≤ 𝑡𝑛 ≤ 𝑢𝑛 < ℓ + 𝜖 and consequently |𝑡𝑛 − ℓ| < 𝜖 for all 𝑛 > 𝑁.
That is 𝑙𝑖𝑚 𝑡𝑛 = ℓ. ∎
𝑛→∞
Example 4.3
𝑛𝜋
𝑐𝑜𝑠
2
1. Show that 𝑙𝑖𝑚 =0
𝑛→∞ 𝑛2
Solution
𝑛𝜋 𝑛𝜋 𝑛𝜋
𝑐𝑜𝑠 𝑐𝑜𝑠 1 1 𝑐𝑜𝑠
2 2 2
Since 𝑜 ≤ | − 0| = | | ≤ 1. 𝑛2 and 𝑙𝑖𝑚 𝑛2 = 0 it follows that 𝑙𝑖𝑚 =0
𝑛2 𝑛2 𝑛→∞ 𝑛→∞ 𝑛2
2. Show that for every 𝑥, with |𝑥| < 1 𝑙𝑖𝑚 = 𝑛𝑥 𝑛 = 0.
𝑛→∞
Solution
Without loss of generality, assume that 𝑥 ≠ 0 and 𝑛 > 1. Since |𝑥| < 1, then there is a positive real
1
number 𝑎 such that |𝑥| = 1 + 𝑎.
1 𝑛(𝑛−1)𝑎2
Then 𝑥 𝑛 = (1 + 𝑎)𝑛 = (1 + 𝑎𝑛 )𝑛 = ∑𝑛𝑟=1(𝑛𝑟)𝑎𝑟 ≥ for some 𝑎 > 0 then
2
2 2
|𝑥 𝑛 | ≤ this implies that |𝑛𝑥 𝑛 | ≤ and so,
𝑛(𝑛−1)𝑎2 (𝑛−1)𝑎2
−2 2
⇒ 2
≤ 𝑛𝑥 𝑛 ≤
(𝑛 − 1)𝑎 (𝑛 − 1)𝑎2
−2 2
Since 𝑙𝑖𝑚 (𝑛−1)𝑎2 = 0 = 𝑙𝑖𝑚 (𝑛−1)𝑎2 . We have by the squeeze theorem, that 𝑙𝑖𝑚 = 𝑛𝑥 𝑛 = 0
𝑛→∞ 𝑛→∞ 𝑛→∞
28
REAL ANALYSIS 1 Issa Ndungo
Exercise 4.3
𝑥𝑛
1) Show that for any 𝑥 ∈ ℝ, 𝑙𝑖𝑚 = =0
𝑛→∞ 𝑛!
1
2) Show that 𝑙𝑖𝑚 = 𝑛! = 0
𝑛→∞
𝑛
3) Show that 𝑙𝑖𝑚 = 2𝑛 = 0
𝑛→∞
𝑡ℎ 2 4 8 16 32
4) Find the 𝑛 term of the sequence (1 , 3 , 5 , , ,…)
7 9
Theorem 4.5
Let 𝑆 be a subset of ℝ which is bounded above. Then there exists a sequence (𝑠𝑛 ) in 𝑆 such that
𝑙𝑖𝑚 𝑠𝑛 = 𝑠𝑢𝑝𝑆.
𝑛→∞
Proof
Let 𝑐 = 𝑠𝑢𝑝𝑆. By the characterization of supremum in Theory 3.4, for each 𝑛 ∈ 𝑁 there exists 𝑠𝑛 ∈
1 1
𝑆such that 𝑐 − 𝑛 < 𝑠𝑛 ≤ 𝑐. Since 𝑙𝑖𝑚 (𝑐 − 𝑛) = 𝑐 = 𝑙𝑖𝑚 𝑐, we have by squeeze theorem that
𝑛→∞ 𝑛→∞
𝑙𝑖𝑚 𝑠𝑛 = 𝑐 = 𝑠𝑢𝑝𝑆 ∎
𝑛→∞
ii) 𝑙𝑖𝑚 𝑠𝑛 𝑡𝑛 = 𝑠𝑡
𝑛→∞
𝑠 𝑠
iii) 𝑙𝑖𝑚 𝑡𝑛 = 𝑡 if 𝑡𝑛 ≠ 0 for all 𝑛 ∈ ℕ and 𝑡 ≠ 0
𝑛→∞ 𝑛
Proof
𝜖
i) Let 𝜖 > 0 be given. Then there exist 𝑁1 and 𝑁2 in ℕ such that: |𝑠𝑛 − 𝑠| < 2 for all 𝑛 ≥ 𝑁1 and
𝜖
|𝑡𝑛 − 𝑡| < for all 𝑛 ≥ 𝑁2 .
2
Therefore, 𝑙𝑖𝑚 𝑠𝑛 𝑡𝑛 = 𝑠𝑡
𝑛→∞
Remark
An increasing sequence (𝑠𝑛 ) is bounded below by 𝑠1 a decreasing sequence is bounded below by 𝑡1
it therefore follows that an increasing sequence is bounded if and only if it is bounded above and a
decreasing sequence is bounded if an only if it is bounded below.
Examples
1. The sequence (1, 1, 2, 3, 5, … . ) is increasing
2. The sequence (3, 1, 0, 0, −3, … . ) is decreasing
3. The sequence (𝑛2 ) is strictly increasing
4. The sequence (−𝑛) is strictly decreasing
30
REAL ANALYSIS 1 Issa Ndungo
Theorem 4.7
Let a sequence (𝑠𝑛 ) be a bounded sequence.
i) If (𝑠𝑛 ) is monotonically increasing then it converges to its supremum.
ii) If (𝑠𝑛 ) is monotonically decreasing then it converges to its infimum.
Proof
Let 𝑠1 = 𝑠𝑢𝑝𝑠𝑛 and 𝑠2 = 𝑖𝑛𝑓𝑠𝑛 and take 𝜖 > 0
i) Since 𝑠𝑢𝑝𝑠𝑛 = 𝑠1 there exists 𝑠𝑛0 such that 𝑠1 − 𝜖 < 𝑠𝑛0 .
Since 𝑠𝑛 is increasing then 𝑠1 − 𝜖 < 𝑠𝑛0 < 𝑠𝑛 < 𝑠1 < 𝑠1 + 𝜖
⇒ 𝑠1 − 𝜖 < 𝑠𝑛 < 𝑠1 + 𝜖 for all 𝑛 > 𝑛0
⇒ |𝑠𝑛 − 𝑠1 | < 𝜖 for all 𝑛 > 𝑛0
⇒ 𝑙𝑖𝑚 𝑠𝑛 = 𝑠1
𝑛→∞
ii) Since 𝑖𝑛𝑓𝑠𝑛 = 𝑠2 there exists 𝑠𝑛1 such that 𝑠𝑛1 < 𝑠2 + 𝜖.
Since 𝑠𝑛 is decreasing then 𝑠2 − 𝜖 < 𝑠𝑛 < 𝑠𝑛1 < 𝑠2 < 𝑠2 + 𝜖
⇒ 𝑠2 − 𝜖 < 𝑠𝑛 < 𝑠1 + 𝜖 for all 𝑛 > 𝑛1
⇒ |𝑠𝑛 − 𝑠2 | < 𝜖 for all 𝑛 > 𝑛1 . ⇒ 𝑙𝑖𝑚 𝑠𝑛 = 𝑠2 ∎
𝑛→∞
Theorem 4.8
A monotone sequence converges if and only if it is bounded.
Proof
We already proved in Theorem 4.3 that every convergent sequence is bounded. To prove the converse
let (𝑠𝑛 ) be a bounded increasing sequence and let 𝑆 = {𝑠𝑛 |𝑛 ∈ ℕ}. Since S is bounded above it has a
supremum, 𝑠𝑢𝑝𝑆 = 𝑠 say. We claim that 𝑙𝑖𝑚 𝑆 = 𝑠. Let 𝜖 > 0 be given, by the characterization of
𝑛→∞
supremum, there exist 𝑠𝑁 ∈ 𝑆 such that 𝑠 − 𝜖 < 𝑠𝑁 ≤ 𝑠𝑛 < 𝑠 + 𝜖 for all 𝑛 ≥ 𝑁. Thus, |𝑠𝑛 − 𝑠| < 𝜖
for all 𝑛 ≥ 𝑁
The proof for the case when the sequence (𝑠𝑛 ) is decreasing is similar. ∎
Example 4.4
𝑛+1
Show that ( ) is a convergent sequence
𝑛
Solution
𝑛+1
We show that ( ) is monotone and bounded. Its convergence will ben follow from theorem 4.8.
𝑛
31
REAL ANALYSIS 1 Issa Ndungo
𝑥+1 −1
Another proof for monotone: Consider 𝑓(𝑥) = , 𝑓 ′ (𝑥) = < 0 for all 𝑥 ∈ [1, ∞). Thus, 𝑓 is
𝑥 𝑥2
𝑛+1 𝑛+2
decreasing on [1, ∞). Therefore 𝑓(𝑛) > 𝑓(𝑛 + 1) i.e. > 𝑛+1 for all 𝑛 ∈ ℕ.
𝑛
𝑛+1 𝑛+1
Boundedness: is bounded below by 1. So is a convergent sequence by Theorem 4.8.
𝑛 𝑛
4.6 Subsequences
If the terms of the sequence (𝑠𝑛 ) are contained in other sequences (𝑡𝑛 ) then (𝑠𝑛 ) is a subsequence of
(𝑡𝑛 ).
Definition:
Let (𝑠𝑛 ) be a sequence of real numbers and let (𝑛𝑘 )𝑘 ∈ ℕ be a sequence of natural numbers
such that 𝑛1 < 𝑛2 < ⋯ . Then the sequence (𝑠𝑛𝑘 ) is called a subsequence of (𝑠𝑛 ). That is a
subsequence (𝑠𝑛𝑘 ) of sequence (𝑠𝑛 ) is strictly increasing function 𝜙: 𝑘 ↦ 𝑠𝑛𝑘 .
Example:
1 1 1 1
Let (𝑠𝑛 ) be the sequence (1, 2, 2 , 3, 3 , … ) then (1, 2 , 3 , … . . ) and (1, 2, 3, … ) are subsequences of
(𝑠𝑛 ).
Theorem 4.9
Let (𝑠𝑛 ) be a sequence which converges to 𝑠. Then any subsequences of (𝑠𝑛 ) converges to 𝑠.
Proof
Let (𝑠𝑛𝑘 ) be a subsequence of (𝑠𝑛 ) and let 𝜖 > 0 be given. Then there is an 𝑁 ∈ ℕ such that
|𝑠𝑛 − 𝑠| < 𝜖 for all 𝑛 ≥ 𝑁. Thus when 𝑘 ≥ 𝑁 we have that 𝑛𝑘 ≥ 𝐾 ≥ 𝑁 and so |𝑠𝑛𝑘 − 𝑠| < 𝜖 for all
𝐾 ≥ 𝑁. Thus, 𝑙𝑖𝑚 𝑠𝑛𝑘 = 𝑠. ∎
𝑛→∞
Solution
𝑛+1 𝑚+1 𝑚−𝑛 𝑚+𝑛
For all 𝑛, 𝑚 ∈ ℕ, |𝑠𝑛 − 𝑠𝑚 | = | − | = | 𝑚𝑛 | < .
𝑛 𝑚 𝑚𝑛
𝑚+𝑛 2𝑚 2
Therefore if 𝑚 ≥ 𝑛 then, |𝑠𝑛 − 𝑠𝑚 | < 𝑚𝑛
< 𝑚𝑛 = 𝑛
32
REAL ANALYSIS 1 Issa Ndungo
1 𝜖
Let 𝜖 > 0 be given. Then there is an 𝑁 ∈ ℕ such that < . Thus, for all 𝑛 ≥ 𝑁 we have
𝑁 2
𝑛+1 𝑚+1 𝑚−𝑛 2 2
|𝑠𝑛 − 𝑠𝑚 | = | − | = | 𝑚𝑛 | < 𝑛 ≤ 𝑁 < 𝜖. Hence, (𝑠𝑛 ) is Cauchy. ∎
𝑛 𝑚
Theorem 4.11
Every convergent sequence (𝑠𝑛 ) is a Cauchy sequence.
Proof
𝜖
Assume that (𝑠𝑛 ) converges to 𝑠. Then given 𝜖 > 0 there exists an 𝑁 ∈ ℕ such that |𝑠𝑛 − 𝑠| < 2 for
all 𝑛 ≥ 𝑁. Now for all 𝑛, 𝑚 ≥ 𝑁 we have that:
|𝑠𝑛 − 𝑠𝑚 | = |(𝑠𝑛 − 𝑠 + 𝑠 − 𝑠𝑚 )| ≤ |𝑠𝑛 − 𝑠| + |𝑠 − 𝑠𝑚 |
𝜖 𝜖
= |𝑠𝑛 − 𝑠| + |𝑠𝑚 − 𝑠| < 2 + 2 = 𝜖
Theorem 4.12
Every Cauchy sequence (𝑠𝑛 ) is bounded.
Proof
Let 𝜖 = 1, then there exists 𝑁 ∈ ℕ such that |𝑠𝑛 − 𝑠𝑚 | < 1 for all 𝑛, 𝑚 ≥ 𝑁. Choose a 𝑘 ≥ 𝑁 and
observe that |𝑠𝑛 | = |𝑠𝑛 − 𝑠𝑘 + 𝑠𝑘 | ≤ |𝑠𝑛 − 𝑠𝑘 | + |𝑠𝑘 | < 1 + 𝑠𝑘
Let 𝑀 = 𝑚𝑎𝑥{|𝑠1 |, |𝑠2 |, … . |𝑠𝑁 |, |𝑠𝑘 | + 1}.
Then (𝑠𝑛 ) < 𝑀 for all 𝑛 ≥ 𝑁 and therefore, (𝑠𝑛 ) is bounded. ∎
Theorem 4.13
Every Cauchy sequence (𝑠𝑛 ) of real numbers converges.
Proof
Let (𝑠𝑛 ) be Cauchy, by Theorem 4.12, (𝑠𝑛 ) is bounded and therefore by Bolzano Weierstrass
theorem, (𝑠𝑛 ) has a subsequence (𝑠𝑛𝑘 ) that converges to some real number ℓ.
We claim that (𝑠𝑛 ) converges to ℓ. Let 𝜖 > 0 be given. Then there exist 𝑁1 , 𝑁2 ∈ ℕ such that:
𝜖
|𝑠𝑛 − 𝑠𝑚 | < for all 𝑛. 𝑚 ≥ 𝑁1 and
2
𝜖
|𝑠𝑛𝑘 − ℓ| < for all 𝑛, 𝑚 ≥ 𝑁2
2
33
REAL ANALYSIS 1 Issa Ndungo
Theorem 4.11 and theorem 4.13 combined together gives the Cauchy’s convergent criterion for
sequences: “A sequence (𝒔𝒏 ) of real numbers converges if and only if it is Cauchy.”
Example 4.6
−1𝑛
1) Use Cauchy criterion to show that the sequence ( ) converges.
𝑛
Solution
−1𝑛 −1𝑛
We need to show that the sequence ( ) in Cauchy. To that end 𝜖 > 0 and 𝑠𝑛 = ( ). then, for all
𝑛 𝑛
(−1)𝑛 (−1)𝑚 1 1 1 1 2
𝑛, 𝑚 ∈ ℕ with 𝑚 ≥ 𝑛 |𝑠𝑛 − 𝑠𝑚 | = | − | ≤ 𝑛 + 𝑚 ≤ 𝑛 + 𝑛 = 𝑛. Now there is an 𝑁 ∈ ℕ
𝑛 𝑚
2 (−1)𝑛 (−1)𝑚 2
such that 𝑁 < 𝜖. Thus, for all 𝑛 ≥ 𝑁, we have |𝑠𝑛 − 𝑠𝑚 | = | − | ≤ 𝑁 < 𝜖.
𝑛 𝑚
−1𝑛
Thus ( ) is a Cauchy sequence and so it converges.
𝑛
1 1 1
2) Show that the sequence 𝑠𝑛 = 1 + 2 + 3 + ⋯ + 𝑛 diverges.
Solution
It suffices to show that (𝑠𝑛 ) is not a Cauchy sequence. Now for 𝑛, 𝑚 ∈ ℕ with 𝑛 > 𝑚, we have:
1 1 1 1 1 1 1 1 1
|𝑠𝑛 − 𝑠𝑚 | = |(1 + + + ⋯ + ) − (1 + + + ⋯ + )| = | + 𝑚+2 + ⋯ + 𝑛|
2 3 𝑛 2 3 𝑚 𝑚+1
1 1 1
= 𝑚+1 + 𝑚+2 + ⋯ + 𝑛
1 1 1 𝑛−𝑚
>>⏟ +𝑛 + ⋯..+𝑛 =
𝑛 𝑛
𝑛−𝑚 𝑡𝑖𝑚𝑒𝑠
Exercise 4.5
Show that every subsequence of a bounded sequence is bounded.
34
REAL ANALYSIS 1 Issa Ndungo
𝑆𝑛 = 𝑎1 + 𝑎2 + 𝑎3 + ⋯ + 𝑎𝑛 .
If the sequence of partial sum (𝑆𝑛 ) converges to 𝑆 then, the series ∑∞
𝑛=1 𝑎𝑛 converges. The limit 𝑆 is
called the sum of the series.
𝑆 = 𝑎1 + 𝑎2 + 𝑎3 + ⋯ + 𝑎𝑛 + ⋯ so 𝑆 = ∑∞
𝑛=1 𝑎𝑛 .
Example 4.7
1 1 1 1
a) The series ∑∞
𝑛=1 2𝑛 = 2 + 4 + 8 + ⋯ has the following partial sums
1 1 1 3 1 1 1 1 2𝑛 −1
𝑆1 = 2 , 𝑆1 = 2 + 4 = 4 , … . , 𝑆𝑛 = 2 + 4 + 8 + ⋯ + 2𝑛 = 2𝑛
2𝑛 −1
Because 𝑙𝑖𝑚 = 1. It follows that the series converge and its sum is 1.
𝑛→∞ 2𝑛
1 1 1 1 1
b) The 𝑛𝑡ℎ partial sum of the series ∑∞
𝑛=1 (𝑛 − 𝑛+1) = (1 − 2) + (2 − 3) + ⋯ is given by
1 1 1
𝑆𝑛 = 1 − 𝑛+1 because the 𝑙𝑖𝑚 (𝑛 − 𝑛+1) = 1. Thus, the series converge and its sum is 1.
𝑛→∞
c) The series ∑∞
𝑛=1 1 = 1, 1, 1, 1, 1, … diverges because 𝑆𝑛 = 𝑛 and the sequence of partial sums
diverges.
NB. The series in Example 4.7b) is a telescoping series of the form (𝑏1 − 𝑏2 ) + (𝑏2 − 𝑏3 ) +
(𝑏3 _𝑏4 ) + ⋯ . (𝑏𝑛−1 − 𝑏𝑛 ) + (𝑏𝑛 − 𝑏𝑛+1 ).
Because the sum of a telescoping series is given by 𝑆𝑛 = 𝑏1 − 𝑏𝑛+1 , it follows that a telescoping
series will converge if and only if 𝑏𝑛 approaches a finite number as 𝑛 → ∞. Moreover, if the series
converges its sum is 𝑆𝑛 = 𝑏1 − 𝑙𝑖𝑚 𝑏𝑛+1 .
𝑛→∞
Solution
2 2 2 2
Using partial fractions, we write 𝑎𝑛 = 4𝑛2 −1 = (2𝑛−1)(2𝑛+1) = 2𝑛−1 + 2𝑛+1 .
1 1 1 1 1 1 1
The partial sums is 𝑆𝑛 = (1 − 3) + (3 − 5) + ⋯ + (2𝑛−1 − 2𝑛+1) = 1 − 2𝑛+1 so the series
1 1
converge and its sum is 𝑙𝑖𝑚 𝑆𝑛 = 𝑙𝑖𝑚 (1 − 2𝑛−1
) = 1 − 𝑙𝑖𝑚 2𝑛−1 = 1 − 0 = 1.
𝑛→∞ 𝑛→∞ 𝑛→∞
35
REAL ANALYSIS 1 Issa Ndungo
Proof
It is easy to see that the series diverge if 𝑟 = ±1. If 0 < |𝑟| < 1 then
𝑆𝑛 = 𝑎 + 𝑎𝑟 + 𝑎𝑟 2 + … + 𝑎𝑟 𝑛−2 + 𝑎𝑟 𝑛−1 . Let us multiply this equation by 𝑟 to yield:
𝑟𝑆𝑛 = 𝑎𝑟 + 𝑎𝑟 2 + 𝑎𝑟 3 + … + 𝑎𝑟 𝑛−1 + 𝑎𝑟 𝑛
Subtracting 𝑟𝑆𝑛 from 𝑆𝑛 we get:
(𝑟 − 1)𝑆𝑛 = 𝑎 − 𝑎𝑟 𝑛
𝑎(1−𝑟 𝑛 )
𝑆𝑛 = 1−𝑟
, with 𝑟 ≠ 1
𝑎(1−𝑟𝑛 )
When |𝑟| ≥ 1, it follows that 𝑟 𝑛 → ∞ as 𝑛 → ∞ that is 𝑙𝑖𝑚 = ∞ . Thus, the series diverge
𝑛→∞ 1−𝑟
𝑎(1−𝑟𝑛 ) 𝑎
When |𝑟| < 1, it follows that 𝑟 𝑛 → 0 as 𝑛 → ∞ that is 𝑙𝑖𝑚 = . Thus, the series converge
𝑛→∞ 1−𝑟 1−𝑟
𝑎
and the sum is 1−𝑟.
Example 4.8
3 1 𝑛
a) The geometric series ∑∞
𝑛=0 2𝑛 = 3 (2)
1 1 1
= 3(1) + 3 (2) + 3((2) + ⋯ has 𝑟 = 2 and 𝑎 = 3. Because 0 <
𝑎 3
|𝑟| < 1 then the series converge and its sum is 𝑆𝑛 = = 1 = 6.
1−𝑟 1−
2
3 𝑛 3 9 3
b) The geometric series ∑∞𝑛=0 (2) = 1 + 2 + 4 + ⋯ has 𝑟 = 2. Since |𝑟| ≥ 1 then the series
diverge.
c) Use a geometric series to write 0. ̅̅
08̅̅ as a ratio of two integers.
Solution
8 8 8
̅̅̅̅, we write 0.08080808080808 … … … . = 2 + 4 + 6 + ⋯
For the repeated decimal 0. 08 10 10 10
8 1 𝑛
= ∑∞
𝑛=0 (102 ) (102 )
8 1
For this series we have 𝑎 = 102 and 𝑟 = 102 so the series converge and the sum is given by
8
𝑆𝑛 = 0. ̅̅
08̅̅ = 𝑎 = 102
1 = 99
8
1−𝑟 1− 2
10
36
REAL ANALYSIS 1 Issa Ndungo
Proof
Assume that ∑∞
𝑛=1 𝑎𝑛 = 𝑙𝑖𝑚 𝑆𝑛 = ℓ then because 𝑆𝑛 = 𝑆𝑛−1 + 𝑎𝑛 and 𝑙𝑖𝑚 𝑆𝑛 = 𝑙𝑖𝑚 𝑆𝑛−1 = ℓ
𝑛→∞ 𝑛→∞ 𝑛→∞
it follows that ℓ = 𝑙𝑖𝑚 𝑆𝑛 = 𝑙𝑖𝑚 (𝑆𝑛−1 + 𝑎𝑛 ) = 𝑙𝑖𝑚 (𝑆𝑛−1 ) + 𝑙𝑖𝑚 (𝑎𝑛 ) = ℓ + 𝑙𝑖𝑚 𝑎𝑛
𝑛→∞ 𝑛→∞ 𝑛→∞ 𝑛→∞ 𝑛→∞
Example 4.9
1) For the series ∑∞ 𝑛
𝑛=0 2 we have:
𝑙𝑖𝑚 2𝑛 = ∞ so, the limit of the 𝑛𝑡ℎ term is not zero so the series diverge.
𝑛→∞
𝑛! 𝑛! 1
2) For the series ∑∞
𝑛=1 2𝑛!+1 we have 𝑙𝑖𝑚 2𝑛!+1 = 2 ≠ 0 so, the series diverge.
𝑛→∞
Proof
Begin by partitioning the interval [1, 𝑛] into 𝑛 − 1 unit interval as illustrated on Figure 4.1 a and
4.1b. The total area of the inscribed rectangles and the circumscribed rectangles are as follows:
∑𝑛𝑖=2 𝑓(𝑖) = 𝑓(2) + 𝑓(3) + ⋯ 𝑓(𝑛) inscribed rectangles
∑𝑛=1
𝑖=1 𝑓(𝑖) = 𝑓(1) + 𝑓(2) + ⋯ 𝑓(𝑛 − 1) circumscribed rectangles
37
REAL ANALYSIS 1 Issa Ndungo
The exact area under the graph i𝑓 from 𝑥 = 1 to 𝑥 = 𝑛 lies between the inscribed area and the
circumscribed area.
𝑦
𝑦
Inscribed rectangles
Circumscribed rectangles
∑𝑛𝑖=2 𝑓(𝑖) = 𝐴𝑟𝑒𝑎
∑𝑛−1
𝑖=1 𝑓(𝑖) = 𝐴𝑟𝑒𝑎
1 2 3 4 𝑛−1 𝑛 x 1 2 3 4 𝑛−1 𝑛 x
Figure 4.1a Figure 4.1b
𝑛
∑𝑛𝑖=2 𝑓(𝑖) ≤ ∫1 𝑓(𝑥) 𝑑𝑥 ≤ ∑𝑛=1
𝑖=1 𝑓(𝑖) (1)
Using the 𝑛𝑡ℎ partial sum, 𝑆𝑛 = 𝑓(1) + 𝑓(2) + ⋯ 𝑓(𝑛) we write (1) as
𝑛 ∞
𝑆𝑛 − 𝑓(1) ≤ ∫1 𝑓(𝑥) 𝑑𝑥 ≤ 𝑆𝑛−1 , assuming that ∫1 𝑓(𝑥) 𝑑𝑥 to ℓ it follows that for 𝑛 ≥ 1
𝑆𝑛 − 𝑓(1) ≤ ℓ ⇒ 𝑠𝑛 = ℓ + 𝑓(1)
Consequently 𝑠𝑛 is bounded and monotonic and by Theorem 4.7, it converges. So ∑ 𝑎𝑛 converges.
For the other direction proof;
∞ 𝑛
Assume that the improper integral ∫1 𝑓(𝑥) 𝑑𝑥 diverges, then ∫1 𝑓(𝑥) 𝑑𝑥 approaches infinity as
𝑛
𝑛 → ∞ and the inequality 𝑆𝑛−1 ≥ ∫1 𝑓(𝑥) 𝑑𝑥 implies that (𝑆𝑛 ) diverges and so ∑ 𝑎𝑛 diverges. ∎
Example 4.10
𝑛
Apply the integral test to the series ∑∞
𝑛=1 𝑛2 +1
Solution
𝑥 −𝑥 2 +1
𝑓(𝑥) = 𝑥 2 +1 is positive and continuous for 𝑥 ≥ 1. We find 𝑓 ′ (𝑥) = (𝑥 2 +1)2 < 0 for 𝑥 > 1 and so 𝑓
is decreasing. 𝑓 satisfies the conditions for the integral test.
∞ 𝑥 1 ∞ 2𝑥 1 𝑏 2𝑥 1
So ∫1 𝑑𝑥 = 2 ∫1 𝑑𝑥 = 2 𝑙𝑖𝑚 ∫1 𝑑𝑥 = 2 𝑙𝑖𝑚 [𝑙𝑛(𝑏 2 + 1) + 𝑙𝑛 2] = ∞ so the series
𝑥 2 +1 𝑥 2 +1 𝑏→∞ 𝑥 2 +1 𝑏→∞
diverge.
38
REAL ANALYSIS 1 Issa Ndungo
Exercise 4.6
1) Use the integral test to determine the divergence and convergence of the following series
1 2 1 1 1 1 1
a) ∑∞
𝑛=1 𝑛+3 b) ∑∞ 𝑛=1 3𝑛+5 c) 3 + 5 + 7 + 9 + ⋯ d) ∑∞ 𝑛=1 𝑛2 +1
2) Explain why the integral test does not apply to the following series
−1𝑛 𝑠𝑖𝑛 𝑛 (𝑠𝑖𝑛 𝑛)2
a) ∑∞
𝑛=1 𝑛 b) ∑∞𝑛=1 2 + 𝑛 c) ∑∞𝑛=1 𝑛
3) use the p-series test to determine the convergence of divergence of the following series.
1 1 1
a) ∑∞
𝑛=1 5 b) 1 + + + ⋯
√𝑛 √2 √3
39
REAL ANALYSIS 1 Issa Ndungo
Example 4.7
1
1) Determine the convergent and divergent of the series ∑∞
𝑛=1 1+3𝑛
Solution
1 1
The series ∑∞ ∞
𝑛=1 1+3𝑛 resembles ∑𝑛=1 3𝑛 converging geometric series. Term by term comparison
1 1 1 1
yields: 𝑎𝑛 = 2+3𝑛 < 3𝑛 = 𝑏𝑛 so, since ∑∞ ∞
𝑛=1 3𝑛 converges then ∑𝑛=1 1+3𝑛 also converges.
1
2) Determine the convergent and divergent of the series ∑∞
𝑛=1 2+ .
√𝑛
Solution
1 1 1 1
The series ∑∞
𝑛=1 2+ resembles ∑∞
𝑛=1 divergent p-series. And 2+ ≤ 𝑛 ≥ 2 which does
√𝑛 √𝑛 √𝑛 √𝑛
∞ 1
not meet the requirement for divergence. We also compare with ∑𝑛=1 divergent harmonic series.
𝑛
1 1
𝑎𝑛 = ≤ 2+ = 𝑏𝑛 𝑛 ≥ 4 and by the direct comparison test, the given series converge.
𝑛 √𝑛
Limit comparison
Proposition 4.6
𝑎𝑛
Suppose that 𝑎𝑛 > 0, 𝑏𝑛 > 0, and 𝑙𝑖𝑚 = ℓ where ℓ is finite and positive. Then the two series
𝑛→∞ 𝑏𝑛
∑ 𝑎𝑛 and ∑ 𝑏𝑛 either both diverge of converge.
Proof
𝑎𝑛 𝑎𝑛
Because 𝑎𝑛 > 0, 𝑏𝑛 > 0 and 𝑙𝑖𝑚 = ℓ there exists 𝑁 > 0 such that 0 < = ℓ + 1 for 𝑛 ≥ 𝑁.
𝑛→∞ 𝑏𝑛 𝑏𝑛
This implies that 0 < 𝑎𝑛 < (ℓ + 1)𝑏𝑛 . So, by the direct comparison test, the convergence of
𝑏
∑ 𝑏𝑛 implies the convergence of ∑ 𝑎𝑛 . Similarly, the fact that 𝑙𝑖𝑚 𝑛 = 1/ℓ can be used to show
𝑛→∞ 𝑎𝑛
that the convergence of ∑ 𝑎𝑛 implies the convergence of ∑ 𝑏𝑛 . ∎
NB: Some examples of p-series to use in comparison tests for given series are in the table below
Given series p-series Conclusion
𝟏 1 Both series converge
∑ 𝟐 ∑ 2
𝟑𝒏 − 𝟒𝒏 + 𝟓 𝑛
𝟏 1 Both series diverge
∑ ∑
√𝟑𝒏 − 𝟐 √𝑛
𝒏𝟐 − 𝟏𝟎 1 Both series converge
∑ ∑ 3
𝟒𝒏𝟓+𝒏
𝟑 𝑛
The table above suggests that when choosing a series for comparison one disregards all but the
highest powers of 𝑛 in both the numerators and the denominator.
40
REAL ANALYSIS 1 Issa Ndungo
Example 4.8
1
1) Show that the general harmonic series ∑∞
𝑛=1 𝑎𝑛+𝑏 diverge.
Solution
1 1 1 1
By comparing with ∑∞
𝑛=1 𝑛 divergent harmonic series, we have 𝑙𝑖𝑚 𝑎𝑛+𝑏 ÷ 𝑛 = 𝑎 because this limit
𝑛→∞
is finite and positive then, the given series diverges.
√𝑛
2) Determine the convergence of divergence of ∑ 𝑛2 +1
Solution
√𝑛 1
Compare the series with ∑ 𝑛2 = ∑ 3 convergent p-series
𝑛2
√𝑛 1 𝑛2
Because 𝑙𝑖𝑚 ÷ 3 = 𝑙𝑖𝑚 𝑛2 +1 = 1 then by limit comparison test the given series converge.
𝑛→∞ 𝑛2 𝑛2 𝑛→∞
𝑛2𝑛
3) Determine the convergence of divergence of ∑ 4𝑛3 +1 .
Solution
2𝑛
Compare with ∑ 𝑛2 divergent series. Now that the series diverge, by 𝑛𝑡ℎ term test from the limit
𝑛2𝑛 2𝑛 1 1
𝑙𝑖𝑚 ÷ = 𝑙𝑖𝑚 1 = so, the given series diverge.
𝑛→∞ 4𝑛3 +1 𝑛2 𝑛→∞ 4+( 3 ) 4
𝑛
Proof
Consider the alternating series ∑(−1)𝑛+1 𝑎𝑛 . For this series, the partial sum (where 2𝑛)
𝑠2𝑛 = (𝑎1 − 𝑎2 ) + (𝑎3 − 𝑎4 ) + (𝑎5 − 𝑎6 ) + ⋯ (𝑎2𝑛−1 − 𝑎2𝑛 ) has all nonnegative terms and
therefore the sequence (𝑠2𝑛 ) is a nondecreasing sequence, we can also write:
𝑠2𝑛 = 𝑎1 − (𝑎2 − 𝑎3 ) − (𝑎4 − a5 ) − ⋯ − (a2n−2 − a2n ) − a2n which implies that s2n ≤ a1 for
every integer n. So, (s2n ) is bounded, nondecreasing and converges to some value L. Because
s2n−1 − a2n = s2n and a2n → 0 we have lim s2n−1 = lim s2n = lim a2n = L + lim a2n = L
n→∞ n→∞ n→∞ n→∞
Because both s2n and s2n−1 converges to the same limit L, it follows that (sn ) also converges to L.
Consequently, the given alternating series converge. ∎
41
REAL ANALYSIS 1 Issa Ndungo
42