Department of Mathematics

Indian Institute of Technology Guwahati

MA 101: Mathematics I
Sequence of real numbers
July-December 2019

√
We know that
√ 2 is not a rational number, but we can find rational numbers as close as
we wish to 2. For instance, the sequence of rationals

1.4, 1.41, 1.414, 1.4142, 1.41421, . . .

√
seem to get closer and closer to 2, as their squares indicate:

1.96, 1.9881, 1.99396, 1.99996164, 1.9999899241, . . . .

Thus it seems that we can create a square root of 2 by taking a “limit” of a sequence of
rational numbers. This is how we shall construct the reals.
Definition 1 (Sequence). A sequence of real numbers or a sequence in R is a mapping
f : N → R. We write xn for f (n), n ∈ N and it is customary to denote a sequence as hxn i
or (xn ) or {xn }.
Example 1. There are different ways of expressing a sequence. For example:
(1) Constant sequence: (a, a, a, . . .), where a ∈ R

(2) Sequence defined by listing: (1, 4, 8, 11, 52, . . .)

(3) Sequence defined by rule: (xn ), where xn = 3n2 for all n ∈ N

(4) Sequence defined recursively: (xn ), where x1 = 4 and xn+1 = 2xn − 5 for all n ∈ N
Convergence: What does it mean?

Think of the examples:

1. (2, 2, 2, . . .)

2. ( n1 )

3. ((−1)n n1 )

4. (1, 2, 1, 2, . . .)
√
5. ( n)

6. ((−1)n (1 − n1 ))

7. (n2 − 1)
Definition 2 (Convergent sequence). A sequence (xn ) is said to be convergent if there
exists ` ∈ R such that for every ε > 0, there exists n0 ∈ N satisfying |xn − `| < ε for all
n ≥ n0 . We say that ` is a limit of (xn ).
Notation: We write lim xn = ` or xn → `.
n→∞

Theorem 1. Limit of a convergent sequence is unique.

Proof. Let (xn ) be a convergent sequence. Assume that lim xn = `1 and lim xn = `2 .
n→∞ n→∞
We claim that `1 = `2 . To see this, by way of contradiction assume that `1 6= `2 . Then
ε = |`1 −`
3
2|
> 0. By definition of convergence of a sequence, there are positive integers n1
and n2 such that

|xn − `1 | < ε for all n ≥ n1 and |xn − `2 | < ε for all n ≥ n2 .

Let n0 = max{n1 , n2 }. Then

|xn − `1 | < ε and |xn − `2 | < ε for all n ≥ n0 .

Using triangle inequality, we have

3ε = |`1 − `2 | = |`1 − xn0 + xn0 − `2 | ≤ |`1 − xn0 | + |xn0 − `2 | < 2ε,

which is a contradiction. Hence, we must have `1 = `2 .

1
Example 2. Using the definition of convergence of a sequence, show that lim = 0.
n→∞ n

Solution. Let ε > 0. By using the Archimedean property, we can find a positive integer
n0 such that n0 · ε > 1, that is, ε > n10 . Now, for all n ≥ n0 , we have

1 1 1
| − 0| = ≤ < ε.
n n n0
1
This proves that lim = 0.
n→∞ n

Example 3. Consider the sequence (xn ) where xn = (−1)n . The terms of the sequence
are −1, 1, −1, 1, −1, 1, . . .. It is intuitively clear that this sequence does not approach to
any real number. Therefore, the sequence does not converge. We now establish this fact
by using the definition.

Solution. By way of contradiction assume that the given sequence converges to `. Then
for ε = 21 , there exists a natural number n0 such that

1
|(−1)n − `| < for all n ≥ n0 .
2
1
This gives |1 − `| < 2
and |1 + `| < 12 . Now, using triangle inequality we have

2 = |1 + 1| = |1 − ` + ` + 1| ≤ |1 − `| + |1 + `| < 1,

which is a contradiction. This proves that ((−1)n ) does not converge.

Example 4. If |α| < 1, then the sequence (αn ) converges to 0.

Solution. If α = 0, then αn = 0 for all n ∈ N and so (αn ) converges to 0. Now we assume
1 1
that α 6= 0. Since |α| < 1, |α| > 1 and so |α| = 1 + h for some h > 0. For all n ∈ N, we
have (1 + h)n = 1 + nh + n(n−1)
2!
h2 + · · · + hn > nh ⇒ |α|n = (1+h)1 1
n < nh for all n ∈ N.
1
Given ε > 0, we choose n0 ∈ N satisfying n0 > hε . Then |αn − 0| = |α|n < n10 h < ε for all
n ≥ n0 and hence (αn ) converges to 0.
log ε
Alternative proof: Given ε > 0, we choose n0 ∈ N satisfying n0 > log |α|
. Then for all
n n n0 n
n ≥ n0 , we have |α − 0| = |α| ≤ |α| < ε and hence (α ) converges to 0.
Bounded sequence: Given a sequence (xn ), we can ask whether the set {x1 , x2 , x3 , . . .}
is bounded or not. If this set is bounded then we call that the sequence (xn ) is bounded.
Equivalently, the sequence (xn ) is bounded if there is a positive number M such that
|xn | ≤ M for all n ∈ N. If (xn ) is not bounded then it is said to be √ unbounded. For
example, (a), ((−1)n ), ( n1 ) are bounded sequences; whereas (n2 ) and (2 n) are unbounded
sequences.
Theorem 2. Every convergent sequence is bounded.
Proof. Let (xn ) be a convergent sequence and let lim xn = `. By taking ε = 1, we find a
n→∞
positive integer n0 such that |xn − `| < 1 for all n ≥ n0 . Equivalently,

` − 1 < xn < ` + 1 for all n ≥ n0 .

Thus, the set {xn0 , xn0 +1 , xn0 +2 , . . .} is bounded from below by ` − 1 and bounded from
above by ` + 1. Using the triangle inequality, we have

|xn | = |xn − ` + `| ≤ |xn − `| + |`| < 1 + |`| for all n ≥ n0 .

Now, there are only finitely many elements left, namely x1 , x2 , . . . , xn0 −1 . Let

M = max{|x1 |, |x2 |, . . . , |xn0 −1 |, |`| + 1}.

Then we have |xn | ≤ M for all n ≥ 1. Hence, (xn ) is a bounded sequence.

Remark 1. From the above theorem, it follows that √ if a sequence is not bounded then
it is not convergent. For example, the sequence ( n) is unbounded and hence is not
convergent. However, every bounded sequence is not convergent. For example, ((−1)n ) is
a bounded sequence but it does not converge.
Limit rules for convergent sequences:
Theorem 3. Let xn → x and yn → y. Then
(a) xn + yn → x + y.

(b) αxn → αx for all α ∈ R.

(c) |xn | → |x|.

(d) xn yn → xy.
xn x
(e) yn
→ y
if yn 6= 0 for all n ∈ N and y 6= 0.
 −3n 2
Example 5. The sequence ( 3n2n2 +5n+3 ) is convergent with limit 32 .
3
2n2 −3n 2− n 1
Solution. We have 3n2 +5n+3
= 5
3+ n + 3 for all n ∈ N. Since n
→ 0, the limit rules for
n2
algebraic operations on sequences imply that the given sequence is convergent with limit
2−0
3+0+0
= 23 .
√ √
Example 6. The sequence ( n + 1 − n) is convergent with limit 0.
√ √ √1
Solution. For all n ∈ N, n + 1 − n = √n+1+ 1 √
n
= √ n
1
. Since n1 → 0, the limit
1+ n +1
rules for algebraic operations on sequences imply that the given sequence is convergent
0
with limit √1+0+1 = 0.
Theorem 4 (Sandwich theorem). Let (xn ), (yn ), (zn ) be sequences such that xn ≤ yn ≤ zn
for all n ∈ N. If both (xn ) and (zn ) converge to the same limit `, then (yn ) also converges
to `.
Proof. Let ε > 0. Since xn → `, so there exists a positive integer n1 such that |xn − `| < ε
for all n ≥ n1 . Similarly, as zn → `, so there exists a positive integer n2 such that
|zn − `| < ε for all n ≥ n2 . Let n0 = max{n1 , n2 }. Then,
` − ε < xn < ` + ε and ` − ε < zn < ` + ε for all n ≥ n0 .
Using the given fact that xn ≤ yn ≤ zn for all n ∈ N, we obtain
` − ε < xn ≤ yn ≤ zn < ` + ε for all n ≥ n0 .
This proves that |yn − `| < ε for all n ≥ n0 , and hence yn → `.
 cos n  −1 cos n 1
Example 7. Consider the sequence . Since ≤ ≤ for all n ∈ N, so
n n n n
cos n
by applying Sandwich theorem we find that lim = 0.
n→∞ n
1
Example 8. If α > 0, then the sequence (α n ) converges to 1.
1
Solution. We first assume that α ≥ 1 and let xn = α n − 1 for all n ∈ N. Then xn ≥ 0
and α = (1 + xn )n = 1 + nxn + n(n−1)
2!
x2n + · · · + xnn > nxn for all n ∈ N. So 0 ≤ xn < αn
α
for all n ∈ N. Since n → 0, by Sandwich theorem, it follows that xn → 0. Consequently
1 1 1
α n → 1. If α < 1, then α1 > 1 and as proved above, ( α1 ) n → 1. It follows that α n → 1.

Alternative proof: We first assume that α ≥ 1. For each n ∈ N, applying the A.M. ≥ G.M.
1
inequality for the numbers 1, ..., 1, α (1 is repeated n − 1 times), we get 1 ≤ α n ≤ 1 + α−1
n
.
α−1 1
Since n → 0, by Sandwich theorem, it follows that α → 1. The case for α < 1 is same
n

as given in the above proof.

1
Example 9. The sequence (n n ) converges to 1.
1
Solution. For all n ∈ N, let an = n n − 1. Then for all n ∈ N,
n(n − 1) 2 n(n − 1) 2
n = (1 + an )n = 1 + nan + an + · · · + ann > an .
2! 2!
2 2
This implies 0 ≤ a2n < n−1 for all n ≥ 2. Since n−1 → 0, by Sandwich theorem, it follows
1
that a2n → 0 and so an → 0. Consequently n n → 1.
Theorem 5. Let r ∈ R. Then there exists a sequence (xn ) of rational numbers such that
lim xn = r.
n→∞

Proof. We know that between two real numbers, there is a rational number. For each
n ∈ N, consider the real numbers r − n1 and r + n1 . Let xn be a rational number such
that r − n1 < xn < r + n1 . Then (xn ) is a sequence of rational numbers, and by Sandwich
theorem (xn ) converges to r.
Divergent sequence:

Definition 3. A sequence (xn ) is said to be divergent if it has no limit.

√
Example 10. If (xn ) is unbounded then it is divergent. For example, ( n), (3n2 ),
((−1)n n3 ) are all divergent. We have seen that the sequence ((−1)n ) is not convergent,
and so it is a divergent sequence although it is bounded.

Definition 4. A sequence (xn ) is said to approach infinity or diverges to infinity if for

any real number M > 0, there is a positive integer n0 such that an ≥ M for all n ≥ n0 .
Similarly, (xn ) is said to approach −∞ or diverges to −∞ if for any real number M > 0,
there is a positive integer n0 such that an ≤ −M for all n ≥ n0 .

Remark 2. Let (xn ) and (yn ) be two sequences of real numbers.

• If (xn ) and (yn ) both diverge to ∞, then the sequences (xn + yn ) and (xn yn ) also
diverge to ∞.

• If (xn ) diverges to ∞ and (yn ) converges, then (xn + yn ) diverges to ∞.

Monotone sequence:

Definition 5. A sequence (xn ) is said to be increasing if xn+1 ≥ xn for all n ∈ N.

Similarly, (xn ) is said to be decreasing if xn+1 ≤ xn for all n ∈ N. We say that (xn ) is
monotonic if it is either increasing or decreasing.

Example 11. The sequence (1 − n1 ) is increasing.

1 1 1 1
Solution. For all n ∈ N, n+1 < n
and so 1 − n+1
> 1− n
for all n ∈ N. Therefore the
given sequence is increasing.

Example 12. The sequence (n + n1 ) is increasing.

1
Solution. For all n ∈ N, (n + 1 + n+1 ) − (n + n1 ) = 1 − n(n+1)
1 1
> 0 ⇒ n + 1 + n+1 > n + n1
for all n ∈ N. Therefore the given sequence is increasing.

Example 13. The sequence (cos nπ

3
) is not monotonic.

Solution. Since cos π3 = 12 , cos 3π

3
= −1 and cos 6π
3
= 1, we have cos π3 > cos 3π
3
< cos 6π
3
and hence the given sequence is neither increasing nor decreasing. Consequently the given
sequence is not monotonic.

Theorem 6. If (xn ) is increasing and not bounded above then (xn ) diverges to ∞. If (xn )
is decreasing and not bounded below then (xn ) diverges to −∞.
Theorem 7 (Monotone convergence theorem). Let (xn ) be a sequence of real numbers.
(a) If (xn ) is increasing and bounded above then (xn ) converges to sup{xn : n ∈ N}.
(b) If (xn ) is decreasing and bounded below then (xn ) converges to inf{xn : n ∈ N}.
(c) A monotonic sequence converges if and only if it is bounded.
Proof. We only prove (a). Since (xn ) is increasing, so it is bounded below by x1 . Also,
(xn ) is bounded above and hence (xn ) is bounded. Let s = sup{xn : n ∈ N}. We claim
that xn → s. To prove this, let ε > 0. Then s − ε is not an upper bound and so there
exists some n0 such that s − ε < xn0 . Since (xn ) is increasing so xn ≥ xn0 for all n ≥ n0 .
Therefore
s − ε < xn0 ≤ xn ≤ s < s + ε for all n ≥ n0 .
This proves that xn → s.
Example 14. Let x1 = 1 and xn+1 = 13 (xn + 1) for all n ∈ N. Then the sequence (xn ) is
convergent and lim xn = 21 .
n→∞

Solution. For all n ∈ N, we have xn+1 − xn = 13 (1 − 2xn ). Also, x1 > 21 and if we assume
that xk > 21 for some k ∈ N, then xk+1 = 13 (xk + 1) > 13 ( 12 + 1) = 12 . Hence by the
principle of mathematical induction, xn > 12 for all n ∈ N. So (xn ) is bounded below.
Again, from above, we get xn+1 − xn < 0 for all n ∈ N ⇒ xn+1 < xn for all n ∈ N ⇒ (xn )
is decreasing. Therefore (xn ) is convergent. Let ` = lim xn . Then lim xn+1 = ` and
n→∞ n→∞
since xn+1 = 31 (xn + 1) for all n ∈ N, we get ` = 13 (` + 1) ⇒ ` = 12 .
Example 15. The sequence ((1 + 1/n)n ) is convergent.
Solution. Let an = (1 + 1/n)n . Then
n    k
X n 1 n 1 n(n − 1) 1 n(n − 1) · · · 2 · 1 1
an = =1+ · + · 2 + ··· + · n
k=0
k n 1 n 2! n n! n
      
1 1 1 1 2 n−1
=1+1+ 1− + ··· + 1− 1− ··· 1 − .
2! n n! n n n
Similarly, we have
      
1 1 1 1 2 n−1
an+1 =1+1+ 1− + ··· + 1− 1− ··· 1 −
2! n+1 n! n+1 n+1 n+1
    
1 1 2 n
+ 1− 1− ··· 1 − .
(n + 1)! n+1 n+1 n+1
Note that the expression for an contains n + 1 terms, while that for an+1 contains n + 2
terms. Moreover, each term appearing in an is less than or equal to the corresponding
term in an+1 , and an+1 has one more positive term. Therefore, we have
2 ≤ a1 < a2 < · · · < an < an+1 < · · · ,
so that the sequence (an ) is increasing. For n > 1, we have
   
1 1 1 1
2 < an < 1 + 1 + + + · · · + n−1 = 2 + 1 − n−1 < 3.
2 22 2 2
By Monotone convergence theorem, the sequence (an ) converges to a real number that
lies between 2 and 3. We define the number e to be the limit of this sequence.
 xn+1
Theorem 8. Let xn 6= 0 for all n ∈ N and let L = lim xn
exist.
n→∞

(a) If L < 1, then xn → 0.

(b) If L > 1, then (xn ) is divergent.

Proof. Note that L ≥ 0. Let r be such that L < r < 1. Let ε = r − L. Then ε > 0, and
hence there is some n0 ∈ N such that
|xn+1 |
0≤ < L + ε = r for all n ≥ n0 .
|xn |

Using the above inequality for n0 , n0 + 1, · · · , n, we have

|xn0 +1 | |xn0 +2 | |xn+1 |

0≤ ··· < rn−n0 +1
|xn0 | |xn0 +1 | |xn |
 
|xn0 |
⇒ 0 ≤ |xn+1 | < · rn
rn0 −1

Since 0 < r < 1, so rn → 0. Now using Sandwich theorem, we have xn → 0. This

completes the proof of (a).
Suppose that L > 1. Let r be such that 1 < r < L. Let ε = L − r. Since ε > 0, there
is some n0 ∈ N such that
|xn+1 |
r =L−ε< for all n ≥ n0 .
|xn |

This yields (as shown before)

 
|xn0 |
|xn+1 | > · rn
rn0 −1

Since r > 1, so rn diverges to infinity and hence (xn ) also diverges.

xn+1
Remark 3. If L = lim xn
= 1, then (xn ) may converge or diverge. For example,
n→∞
n
the sequence ((−1) ) diverges and L = 1. For any nonzero constant sequence, L = 1 and
constant sequences are convergent.
n
Example 16. If α ∈ R, then the sequence ( αn! ) is convergent.
n
Solution. Let xn = αn! for all n ∈ N. If α = 0, then xn = 0 for all n ∈ N and so (xn )
|α|
converges to 0. If α 6= 0, then lim | xxn+1
n
| = lim n+1 = 0 < 1 and so (xn ) converges to
n→∞ n→∞
0.
n
Example 17. The sequence ( 2n4 ) is not convergent.
2n
Solution. If xn = n4
for all n ∈ N, then lim | xxn+1 | = lim 2
1 4 = 2 > 1. Therefore the
n→∞ n n→∞ (1+ n )
sequence (xn ) is not convergent.
Subsequence:
Definition 6 (Subsequence). Let (xn ) be a sequence in R. If (nk ) is a sequence of positive
integers such that n1 < n2 < n3 < · · · , then (xnk ) is called a subsequence of (xn ).

Example 18. Think of some divergent sequences and their convergent subsequences.

Theorem 9. If a sequence (xn ) converges to `, then every subsequence of (xn ) must

converge to `.

Remark 4. From the above theorem, we have the following:

• If (xn ) has a subsequence (xnk ) such that xnk 6→ `, then xn 6→ `.

• If (xn ) has two subsequences converging to two different limits, then (xn ) cannot be
convergent.

Example 19. If xn = (−1)n (1 − n1 ) for all n ∈ N, then (xn ) is not convergent.

1 1
Solution. We have x2n−1 = (−1)2n−1 (1 − 2n−1 ) = 2n−1 − 1 → −1.
2n 1 1
Also x2n = (−1) (1 − 2n ) = 1 − 2n → 1 6= −1. Hence, (xn ) is not convergent.

Example 20. Let (xn ) be a sequence in R. Then (x2n ) and (x2n−1 ) are two subsequences
of (xn ). Suppose that x2n → ` ∈ R and x2n−1 → `. Then xn → `.

Solution. Let ε > 0. Since x2n → ` and x2n−1 → `, there exist n1 , n2 ∈ N such that
|x2n − `| < ε for all n ≥ n1 and |x2n−1 − `| < ε for all n ≥ n2 . Taking n0 = max{2n1 , 2n2 −
1} ∈ N, we find that |xn − `| < ε for all n ≥ n0 . Hence xn → `.

Example 21. The sequence (1, 12 , 1, 23 , 1, 34 , ...) converges to 1.

n 1
Solution. If (xn ) denotes the given sequence, then x2n = n+1
= 1
1+ n
→ 1 and x2n−1 =
1 → 1. Therefore (xn ) converges to 1.

Theorem 10. Every sequence of real numbers has a monotone subsequence.

Proof. Let (xn ) be a sequence of real numbers. A term xp is called a peak in (xn ) if
xp > xm for all m > p. That is, a peak in (xn ) is a term which is greater than all the
succeeding terms. Let P be the set of all the peaks of (xn ). We now consider the following
two cases:

• P is finite: Note that in this case P can be empty also. Let p1 < p2 < · · · < p` so
that xp1 , xp2 , . . . , xp` are the only peaks of (xn ). Let n1 > p` . Then xn1 is not a peak.
Hence there is some n2 ∈ N such that n2 > n1 and xn1 ≤ xn2 . Again, since n2 > p`
so xn2 is not a peak. Hence, there is some n3 > n2 such that xn2 ≤ xn3 . In this
way, using the principle of mathematical induction, we have an increasing sequence
n1 < n2 < n3 < · · · < nk < · · · in N such that xn1 ≤ xn2 ≤ xn3 ≤ · · · < xnk < · · · .
This proves that (xnk ) is an increasing subsequence of (xn ).

• P is infinite: In this case, we have p1 < p2 < · · · < pk < · · · so that xp1 , xp2 , . . . , xpk , . . .
are the peaks of (xn ). Clearly, xp1 > xp2 > . . . > xpk > . . .. Hence, (xpk ) is a de-
creasing sequence of (xn ).
Theorem 11 (Bolzano-Weierstrass Theorem). Every bounded sequence in R has a con-
vergent subsequence.
Proof. Let (xn ) be a sequence of real numbers. By the previous theorem, (xn ) has a
monotone subsequence, say (xnk ). Since (xn ) is bounded, so (xnk ) is also bounded. By
the Monotone convergence theorem, (xnk ) is convergent.
Cauchy sequence:
Definition 7 (Cauchy sequence). A sequence (xn ) is called a Cauchy sequence if for each
ε > 0, there exists n0 ∈ N such that |xm − xn | < ε for all m, n ≥ n0 .
Theorem 12. Every Cauchy sequence is bounded.
Theorem 13 (Cauchy’s criterion for convergence). A sequence in R is convergent if and
only if it is a Cauchy sequence.
Proof. Let (xn ) be convergent and xn → `. Let ε > 0. Then there is some n0 ∈ N such
that |xn − `| < ε/2 for all n, m ≥ n0 . Now,

|xn − xm | = |xn − ` + ` − xm | ≤ |xn − `| + |xm − `| < ε/2 + ε/2 = ε for all m, n ≥ n0 .

Hence, (xn ) is a Cauchy sequence. Conversely, suppose that (xn ) is a Cauchy sequence.
Then (xn ) is bounded, and by Bolzano Weierstrass theorem, (xn ) has a convergent sub-
sequence, say (xnk ). Suppose that xnk → `. We claim that xn → `. To prove this, let
ε > 0. Since (xn ) is Cauchy, so there is some n0 ∈ N such that |xn − xm | < ε/2 for all
n, m ≥ n0 . Also, since xnk → `, there is k0 ∈ N such that |xnk − `| < ε/2 for all k ≥ k0 .
Let j = max{k0 , n0 }. Since nj ≥ j, so nj ≥ n0 . Also, j ≥ k0 . Therefore

|xn − `| ≤ |xn − xnj | + |xnj − `| < ε/2 + ε/2 = ε for all n ≥ n0 .

Hence the sequence (xn ) is convergent.

Example 22. Let (xn ) satisfy either of the following conditions:
(a) |xn+1 − xn | ≤ αn for all n ∈ N

(b) |xn+2 − xn+1 | ≤ α|xn+1 − xn | for all n ∈ N,

where 0 < α < 1. Then (xn ) is a Cauchy sequence.
Solution. (a) For all m, n ∈ N with m > n, we have

|xm − xn | ≤ |xn − xn+1 | + |xn+1 − xn+2 | + · · · + |xm−1 − xm |

≤ αn + αn+1 + · · · + αm−1
αn
= (1 − αm−n )
1−α
αn
<
1−α
n
α 0
Since 0 < α < 1, αn → 0 and so given any ε > 0, we can choose n0 ∈ N such that 1−α < ε.
α n0
Hence for all m, n ≥ n0 , we have |xm −xn | < 1−α < ε. Therefore (xn ) is a Cauchy sequence.
(b) For all m, n ∈ N with m > n, we have

|xm − xn | ≤ |xn − xn+1 | + |xn+1 − xn+2 | + · · · + |xm−1 − xm |

≤ (αn−1 + αn + · · · + αm−2 )|x2 − x1 |
αn−1
= (1 − αm−n )|x2 − x1 |
1−α
αn−1
≤ |x2 − x1 |
1−α
Since 0 < α < 1, αn−1 → 0 and so given any ε > 0, we can choose n0 ∈ N such that
αn0 −1 n0 −1
1−α
|x2 − x1 | < ε. Hence for all m, n ≥ n0 , we have |xm − xn | ≤ α1−α |x2 − x1 | < ε.
Therefore (xn ) is a Cauchy sequence.
1
Example 23. Let (xn ) be a sequence defined as x1 = 1 and xn+1 = 1 + xn
for n ∈ N.
Then xn+1 xn = 1 + xn > 2. Now,
xn+1 − xn 1
|xn+2 − xn+1 | = | | < |xn+1 − xn |.
xn+1 xn 2
Hence, (xn ) is a Cauchy sequence.
Limit superior and limit inferior:

Let (xn ) be a bounded sequence. Let y1 = sup{x1 , x2 , . . .}, y2 = sup{x2 , x3 , . . .}, and
so on. That is, for n ∈ N,

yn = sup{xn , xn+1 , . . .} = sup xk .

k≥n

Let A and B be two nonempty subsets of R such that A ⊆ B. Then clearly, sup(A) ≤
sup(B) and inf(B) ≤ inf(A). Hence,

y1 ≥ y2 ≥ y3 ≥ · · · .

Since (xn ) is bounded so the sequence (yn ) is bounded below. By Monotone convergence
theorem, (yn ) is convergent and converges to the infimum of {y1 , y2 , . . .}. The limit of
the sequence (yn ) is called the limit superior of the sequence (xn ), and is denoted by
lim sup xn . Thus,
lim sup xn := lim yn = inf sup xk .
n→∞ n k≥n

Similarly, let z1 = inf{x1 , x2 , . . .}, z2 = inf{x2 , x3 , . . .}, and so on. That is, for n ∈ N,

zn = inf{xn , xn+1 , . . .} = inf xk .

k≥n

We have z1 ≤ z2 ≤ z3 ≤ · · · . Since (xn ) is bounded so the sequence (zn ) is bounded above.

By Monotone convergence theorem, (zn ) is convergent and converges to the supremum
of {z1 , z2 , . . .}. The limit of the sequence (zn ) is called the limit inferior of the sequence
(xn ), and is denoted by lim inf xn . Thus,

lim inf xn := lim zn = sup inf xk .

n→∞ n k≥n
Example 24. Consider the sequence (xn ), where xn = (−1)n . Clearly, for any n, yn =
sup{xn , xn+1 , . . .} = 1 and zn = inf{xn , xn+1 , . . .} = −1. Hence, lim sup xn = 1 and
lim inf xn = −1.

Example 25. Consider the sequence (xn ), where xn = n1 . Clearly, for any n, yn = sup{ k1 :
k ≥ n} = n1 and zn = inf{ k1 : k ≥ n} = 0. Hence, lim sup xn = lim yn = lim n1 = 0 and
n→∞ n→∞
lim inf xn = 0.

Remark 5. Suppose that |xn | < M for n ∈ N. Then −M ≤ zn ≤ yn ≤ M for all n.

Hence,
−M ≤ lim inf xn ≤ lim sup xn ≤ M.

Theorem 14. Let (an ) and (bn ) be two bounded sequences.

(1) lim inf an ≤ lim sup an .

(2) If an ≤ bn for all n ∈ N, then lim sup an ≤ lim sup bn and lim inf an ≤ lim inf bn .

(3) lim sup(an +bn ) ≤ lim sup an +lim sup bn and lim inf(an +bn ) ≥ lim inf an +lim inf bn .

Theorem 15. If (an ) is a convergent sequence, then

lim inf an = lim an = lim sup an .

n→∞

Proof. Let ` = lim an . Let ε > 0. Then there is some n0 ∈ N such that

` − ε/2 < an < ` + ε/2 for all n ≥ n0 .

Let yn = sup{ak : k ≥ n} and zn = inf{ak : k ≥ n}. Then, for all n ≥ n0 , we have

` − ε < ` − ε/2 ≤ zn ≤ an ≤ yn ≤ ` + ε/2 < ` + ε.

Hence, lim sup an = lim yn = ` and lim inf an = lim zn = `.

Theorem 16. Let (an ) be a bounded sequence. If lim sup an = lim inf an , then (an ) is
convergent and lim an = lim sup an .
n→∞

Proof. Let lim sup an = lim inf an = `. Let yn = sup{ak : k ≥ n} and zn = inf{ak : k ≥
n}. Then we have lim sup an = lim yn = ` and lim inf an = lim zn = `. Let ε > 0. Then
there are positive integers n1 and n2 such that

` − ε < yn < ` + ε for all n ≥ n1 and ` − ε < zn < ` + ε for all n ≥ n2 .

Let n0 = max{n1 , n2 }. Then,

` − ε < zn ≤ an ≤ yn < ` + ε for all n ≥ n0 .

Hence, lim an = `.
Alternative proof: Equivalently, we can directly apply Sandwich Theorem. We have
zn ≤ an ≤ yn for all n. Since zn → ` and yn → `, by using Sandwich Theorem, we have
an → `.