1

THE CHINESE UNIVERSITY OF HONG KONG

Department of Mathematics
Exercise for MATH1010 University Mathematics
September 2016

Contents
1 Preliminaries 2
1.1 Trigonometric identities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Mathematical Induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Sequences 6
2.1 Intuitive definition of limits of sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Monotone Convergence theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Sandwich theorem for sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3 Limits of functions 15
3.1 Intuitive definition of limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Sandwich theorem for functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Two important limits: lim (1 + x1 )x = e and lim sinx x = 1 . . . . . . . . . . . . . . . . . . . . . . 17
x→+∞ x→0

4 Continuity 21
4.1 Definition of continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Intermediate value theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5 Differentiation 24
5.1 Definition of differentiability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.2 Derivatives and Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.3 Higher derivatives (Leibniz’s Rule) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

6 Application of differentiation 31
6.1 Rolle’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.2 Lagrange’s mean value theorem and Cauchy’s mean value theorem . . . . . . . . . . . . . . . . . . 31
6.3 First and second derivative check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.4 Curve sketching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

7 Indefinite integration 42
7.1 Primitive functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.2 Integration by substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7.3 Integration by parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
7.4 Trigonometric substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
7.5 Integration of rational functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.6 Reduction formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

8 Definite Integration 57
8.1 Mean value theorem for integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
8.2 Fundamental theorem of Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
8.3 Reduction formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
8.4 Improper integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.5 Inequalities Involving Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
 2

1 Preliminaries
1.1 Trigonometric identities

Exercise 1. (Level 1)
Let a, b, θ be three real numbers. Prove that
p
|a sin θ + b cos θ| ≤ a2 + b2 .

Solution.
p
|a sin θ + b cos θ| ≤ |a sin θ + b cos θ|2 + |a cos θ − b sin θ|2
p
= a2 + b2

Exercise 2. (Level 1/Level 2)

Using the well-known formula cos α + cos β = 2 cos α+β α−β
2 cos 2 , prove that

5π π π
cos + cos = cos .
12 4 12

Solution. Using the well-known formula cos α + cos β = 2 cos α+β α−β
2 cos 2 , we obtain that

5π π 5π π
5π π 12 + 4 12 − 4 π π π
cos + cos = 2 cos cos = 2 cos cos = cos .
12 4 2 2 3 12 12

Exercise 3. (Level 3)
tan A+tan B
Using tan(A + B) = 1−tan A tan B , show that

tan−1 1 + tan−1 2 + tan−1 3 = π.

tan A+tan B
Solution. Using tan(A + B) = 1−tan A tan B , one gets

tan(tan−1 1 + tan−1 2 + tan−1 3)

tan(tan−1 1 + tan−1 2) + tan(tan−1 3)
=
1 − tan−1 1 + tan−1 2) tan(tan−1 3)
tan(tan−1 1) + tan(tan−1 2) + tan(tan−1 3) − tan(tan−1 1) tan(tan−1 2) tan(tan−1 3)
=
1 − tan(tan−1 1) tan(tan−1 3) − tan(tan−1 2) tan(tan−1 3) − tan(tan−1 1) tan(tan−1 2)
1+2+3−1·2·3
=
1 − tan(tan−1 1) tan(tan−1 3) − tan(tan−1 2) tan(tan−1 3) − tan(tan−1 1) tan(tan−1 2)
= 0.

The result follows.

 3

Exercise 4. (Level 3/Level 4)

tan A+tan B
Using tan(A + B) = 1−tan A tan B , show that

5π π 7π 4π
tan + tan + tan = tan .
18 3 18 9

5π π 7π π tan A+tan B
√
Solution. Write α = 18 , β = 3,γ = 18 and δ = 18 . Using tan(A + B) = 1−tan A tan B and tan π3 = 3, we have

tan α tan γ
= tan(β − δ) tan(β + δ)
√ √
3 − tan δ 3 + tan δ
= √ √
1 + 3 tan δ 1 − 3 tan2 δ
2

3 − tan2 δ
= .
1 − 3 tan2 δ

tan A+tan B tan A+tan B+tan C−tan A tan B tan C

Iterate tan(A + B) = 1−tan A tan B to get tan(A + B + C) = 1−tan A tan B−tan B tan C−tan A tan C . So

3 tan δ − tan3 δ 3 − tan2 δ

tan(3δ) = 2 = tan δ = tan α tan γ tan δ
1 − 3 tan δ 1 − 3 tan2 δ

which implies the result by tan A = 1/ tan( π2 − A).

1.2 Mathematical Induction

Exercise 5. (Level 2)
A sequence {an } is defined as follows:

1 1 1
a1 = and − = 2n + 5 for n = 1, 2, 3, · · · .
5 an+1 an

1
Show that an = n2 +4n for n = 1, 2, 3, · · · .

1
Solution. Let P (n) be the statement that an = n2 +4n .

1 1
• When n = 1, a1 = 5 = (1)2 +4(1) , hence P (1) is true.

1
• Suppose P (n) is true for some natural number n, i.e. an = n2 +4n for some natural number n.
Then,

1 1
= + 2n + 5
an+1 an
. = n2 + 4n + 2n + 5
= (n + 1)2 + 4(n + 1).

1
Therefore, an+1 = (n+1)2 +4(n+1) , hence P (n + 1) is true.
 4

By mathematical induction, P (n) is true for all natural numbers n.

Exercise 6. (Level 2/Level 3)

(a) Prove that, for n ∈ N,

xn − y n = (x − y)(xn−1 + xn−2 y + · · · + xy n−2 + y n−1 ).

(b) Factorize a4 + a2 + 1.

Solution. (a) Let P (n) be the statement that xn − y n = (x − y)(xn−1 + xn−2 y + · · · + xy n−2 + y n−1 ).

• When n = 1, P (1) is true by direct checking.

• Suppose P (n) is true for some natural number n, i.e. xn −y n = (x−y)(xn−1 +xn−2 y+· · ·+xy n−2 +y n−1 )
for some natural number n.
Then,

xn+1 − y n+1 = xn+1 − xy n + xy n − y n+1

. = x(xn − y n ) + y n (x − y)
= x(x − y)(xn−1 + xn−2 y + · · · + xy n−2 + y n−1 ) + y n (x − y)
= (x − y)(xn + xn−1 y + · · · + xy n−1 + y n ).

Therefore, P (n + 1) is true.

By mathematical induction, P (n) is true for all natural numbers n.

a6 −1 (a3 −1)(a3 +1)

(b) Take x = a2 , y = 1 and n = 3 in the previous part. Then we get a4 + a2 + 1 = a2 −1 = a2 −1 =
3 3
a −1 a +1
a−1 a−1 = (a2 − a + 1)(a2 + a + 1).

Exercise 7. (Level 4)
Show that for every positive integer n, there are r, cr , · · · , c0 such that rN ∪ {0}, cr = 1, cj ∈ {0, 1} for 0 ≤ j < r
and
n = cr 2r + cr−1 2r−1 + · · · + c1 2 + c0 .

Remark: Such a representation is unique.

Solution. Let P (n) be the statement that there are r, cr , · · · , c0 such that r ∈ N ∪ {0}, cr = 1, cj ∈ {0, 1} for
0 ≤ j < r and
n = cr 2r + cr−1 2r−1 + · · · + c1 2 + c0 .

• When n = 1, take r = 0 and c0 = 1. Then P (1) is true as 1 = 020 + 1.

 5

• Suppose P (m) is true for 1 ≤ m < k, i.e. there are r, cr , · · · , c0 such that r ∈ N ∪ {0}, cr = 1, cj ∈ {0, 1}
for 0 ≤ j < r and
m = cr 2r + cr−1 2r−1 + · · · + c1 2 + c0 .

We want to show that there are r0 , c0r0 , · · · , c00 such that r0 ∈ N ∪ {0}, c0r0 = 1, c0j ∈ {0, 1} for 0 ≤ j < r0 and

k = c0r0 2r + c0r0 −1 2r−1 + · · · + c01 2 + c00 .

Consider two cases according to the parity of k. First, suppose k is even. By induction hypothesis, there
are r, cr , · · · , c0 such that r ∈ N ∪ {0}, cr = 1, cj ∈ {0, 1} for 0 ≤ j < r and

k
= cr 2r + cr−1 2r−1 + · · · + c1 2 + c0 ,
2

then
k = cr 2r+1 + cr−1 2r + · · · + c1 22 + c0 2.

We set r0 = r + 1, c00 = 0 and c0j+1 = cj for 0 ≤ j ≤ r. Thus, there are r0 , c0r0 , · · · , c00 such that r0 ∈ N ∪ {0},
c0r0 = 1, c0j ∈ {0, 1} for 0 ≤ j < r0 and

k = c0r0 2r + c0r0 −1 2r−1 + · · · + c01 2 + c00 .

Now suppose that k is odd, that is, k = 2m + 1 for some m < k. Apply the induction hypothesis on m.
There are r, cr , · · · , c0 such that r ∈ N ∪ {0}, cr = 1, cj ∈ {0, 1} for 0 ≤ j < r and

m = cr 2r + cr−1 2r−1 + · · · + c1 2 + c0 ,

then
k = cr 2r+1 + cr−1 2r + · · · + c1 22 + c0 2 + 1.

We set r0 = r + 1, c00 = 1 and c0j+1 = cj for 0 ≤ j ≤ r. Thus, there are r0 , c0r0 , · · · , c00 such that r0 ∈ N ∪ {0},
c0r0 = 1, c0j ∈ {0, 1} for 0 ≤ j < r0 and

k = c0r0 2r + c0r0 −1 2r−1 + · · · + c01 2 + c00 .

Therefore, P (k) is true.

By mathematical induction, P (n) is true for all natural numbers n.

 6

2 Sequences
2.1 Intuitive definition of limits of sequences

Exercise 1. (Level 2)
Evaluate the following limits.

n5 +n2 −n+1
(a) lim 5 4 2
n→∞ 3n +4(n+1) +2(2n−1)

√ √
3
(b) lim ( 3 n2 + 1 − n2 )
n→∞
h 3 3
i √ √
(c) lim { (n + 1) 2 − n 2 ( n + 1 − n)}
n→∞

1 1 1




(d) lim 1− 22 1− 32 ··· 1 − n2
n→∞

Solution. (a)

n5 + n2 − n + 1 1 + n13 − n14 + n15

lim = lim 4
n→∞ 3n5 + 4(n + 1)4 + 2(2n − 1)2 n→∞ 3 + + n1 )4 + n23 (2 − n1 )2
n (1

1
=
3

(b)
p
3
√
3
lim ( n2 + 1 − n2 )
n→∞
√
√
p p 3
p
3 3
3
(n2 + 1)2 + 3 n2 (n2 + 1) + n4
= 2 2
lim ( n + 1 − n p p √
3
)
n→∞ 3
(n2 + 1)2 + 3 n2 (n2 + 1) + n4
1
= lim p p √
3
n→∞ 3
(n2 + 1)2 + n2 (n2 + 1) + n4
3

= 0

(c)
h 3 3
i √ √
lim { (n + 1) 2 − n 2 ( n + 1 − n)}
n→∞
i (n + 1) 32 + n 23 √ √ √
h 3 3 √ n+1+ n
= lim { (n + 1) − n
2 2
3 3 ( n + 1 − n) √ √ }
n→∞ (n + 1) 2 + n 2 n+1+ n
3n2 + 3n + 1
= lim 3 3√ √
n→∞ ((n + 1) + n 2 )( n + 1 +
2 n)
3
+ n12
3+ n
= lim q
n→∞ 3
((1 + n1 ) 2 + 1)( 1 + 1
n + 1)

3
=
4
 7

(d)
    
1 1 1
lim 1− 2 1 − 2 ··· 1 − 2
n→∞ 2 3 n
       
1 1 1 1 1 1
= lim 1− 1+ 1− 1+ ··· 1 − 1+
n→∞ 2 2 3 3 n n
          
1 3 2 4 n−1 n+1
= lim ···
n→∞ 2 2 3 3 n n
   
1 n+1
= lim
n→∞ 2 n
1
=
2

Exercise 2. (Level 2)
Let a ∈ R. A sequence {un } is defined by
an
un = .
1 + an+1
h i
1 1
(a) Show that un = a 1− 1+an+1 .

(b) Discuss the behaviour, as n → ∞, of the sequence {un } for the cases

(i) |a| > 1,

(ii) |a| < 1,
(iii) a = 1,
(iv) a = −1.

(Hint: You may use lim an = ±∞ for |a| > 1 and lim an = 0 for |a| < 1.)
n→∞ n→∞

Solution. (a) Straightforward.

(b) (i) For |a| > 1, using the well-known limits lim an = ±∞, one gets
n→∞
 
1 1 1
lim un = lim 1− n+1
= .
n→∞ n→∞ a 1+a a

(ii) For |a| < 1, using the well-known limit lim an = 0, we have
n→∞
 
1 1
lim un = lim 1− = 0.
n→∞ n→∞ a 1 + an+1
1
(iii) For a = 1, un = 2 for all positive integers n, thus, lim un = 12 .
n→∞

(iv) For a = −1, the sequence becomes

(
− 12 if n is odd
un =
undefined if n is even.

The limit does not exist.

 8

2.2 Monotone Convergence theorem

Exercise 3. (Level 2)
For each of the following sequence,

(i) Determine whether the sequence is monotonic;

(ii) Determine whether the sequence is bounded.

1 1 1
(a) an = n + n+1 + ··· + 2n
√
n2 +1−n
(b) an = n2

Solution. (a) (i) For all n ≥ 1,

 
1 1 1 1 1 1
an+1 − an = + + ··· + − + + ··· +
n+1 n+1 2n + 2 n n+1 2n
1 1 1
= + −
2n + 1 2n + 2 n
1 1 1
< + −
2n 2n n
= 0
implying that {an } is monotonic (strictly) decreasing.
(ii) Noting that
1 1 1 1 1 1 n+1
0< + + ··· + < + + ··· + < ,
n n+1 2n n n n n
this shows that {an } is bounded.
√
n2 +1−n
(b) an = n2

(i) For all n ≥ 1,

an+1
an
p
(n + 1)2 + 1 − (n + 1) n2
= √
(n + 1)2 n2 + 1 − n
√ 2
√ 2 √
n2 (n+1) +1−(n+1) (n+1) +1+(n+1) n2 +1+n
= (n+1)2
√ √ √
n2 +1−n n2 +1+n (n+1)2 +1+(n+1)
√
n2 n2 + 1 + n
= 2
p
(n + 1) (n + 1)2 + 1 + (n + 1)
< 1
implying that {an } is monotonic (strictly) decreasing.
(ii) Noting that for n ≥ 1
√ √ √ √
n2 − n n2 + 1 − n n2 + n2 2
0= < < < < 3,
n2 n2 n2 n
this shows that {an } is bounded.
 9

Exercise 4. (Level 3)
A sequence is defined by x1 = 1, xn+1 = 23 xn + 9
x2n for n ≥ 1.

(a) (i) Show that xn > 3 for n ≥ 2.

(ii) Prove that xn+1 < xn for n ≥ 2.

(b) Hence show that {xn } converges and find lim xn .

n→∞

Solution. (a) (i) Consider xn+1 − 3. Then,

2 9
xn+1 − 3 = xn + 2 − 3
3 xn
2x3n − 9x2n + 27
=
3x2n
(xn − 3)2 (2xn + 3)
= .
3x2n

Note that x2 > 3. Thus xn+1 − 3 > 0 for n ≥ 2. Hence xn > 3 for n ≥ 2.
(ii) Noting that

2 9
xn+1 − xn = xn + 2 − xn
3 xn
27 − x3n
=
3x2n
(3 − xn )(x2n + xn + 9)
= ,
3x2n

one gets xn+1 < xn for n ≥ 2 provided that xn > 3 for n ≥ 2.

(b) From parts (i) and (ii) of (a), {xn } converges by monotone convergence theorem. Let lim xn = L. Then, by
n→∞
the convergence of sub-sequence,
 
2 9
lim xn+1 = lim xn + 2
n→∞ n→∞ 3 xn
2 9
L = L+ 2
3 L

gives L3 = 27. On solving, it gives a real solution L = 3. Hence lim xn = 3.

n→∞

Exercise 5. (Level 4)
Let a1 and b1 be any two positive numbers such that a1 > b1 . Let {an } and {bn } be two sequences satisfying the
equations
an + bn p
an+1 = and bn+1 = an bn
2
for all positive integers n.
 10

(a) (i) Show that an > bn for for all positive integers n.
(ii) Deduce that {an } is monotonic decreasing and {bn } is monotonic increasing.

(b) Show that both lim an and lim bn exist. Hence prove that lim an = lim bn .
n→∞ n→∞ n→∞ n→∞

Solution. (a) (i) We note that

√ √
a n + bn p ( an + bn )2
an+1 − bn+1 = − an bn = >0
2 2
for n ≥ 1. The results follows.
(ii) For n ≥ 1, by previous part, we get

an + bn an + an
an+1 = < = an ,
2 2
that is, {an } is monotonic decreasing. Also, by similar arguments, one obtains, for n ≥ 1,
p p
bn+1 = an bn > bn · bn = bn

which says that {bn } is monotonic increasing.

(b) From (a), we note that for n ≥ 1

a1 > an+1 > bn+1 > b1 .
We see that {an } is monotonic decreasing and bounded below while {an } is monotonic increasing and bounded
above. By monotone convergence theorem, both lim an and lim bn exist. Let lim an = A and lim bn = B,
n→∞ n→∞ n→∞ n→∞
then, by the recurrence relation, we have
a n + bn
lim an+1 = lim
n→∞ n→∞ 2
A+B
so A = 2 , hence A = B. Therefore, lim an = lim bn is obtained.
n→∞ n→∞

Exercise 6. (Level 4)
The two sequences of positive integers {a1 , a2 , · · · , an , · · · }, {b1 , b2 , · · · , bn , · · · } satisfy the following conditions:

a1 = b1 = 1 and an+1 = an + 2bn , bn+1 = an + bn

for all positive integers n.

(a) Prove that for each positive integer n,

an ≥ n, bn ≥ n and a2n − b2n = (−1)n .

(b) (i) Deduce from (a) that

an √ an √
< 2 if n is odd, > 2 if n is even,
bn bn
and

an+2 an 2(−1)n+1
− = .
bn+2 bn bn (2an + 3bn )
 11

(ii) Hence show that the two sequences { ab2k−1

2k−1
: k = 1, 2, 3 · · · } and { ab2k
2k
: k = 1, 2, 3 · · · } converge to the
√
same limit 2.

Solution. (a) For n = 1, a1 = b1 = 1.

Suppose that an ≥ n, bn ≥ n. Then

an+1 = an + 2bn ≥ n + 2n ≥ n + 1 and bn+1 = an + bn ≥ n + n ≥ n + 1.

By the principle of induction, the results are true for all positive integers n. Next, noting that a2n+1 − 2b2n+1 =
(an + 2bn )2 − 2(an + bn )2 = −(a2n − 2b2n ), one gives that

a2n − 2b2n = −(a2n−1 − 2b2n−1 )

..
.
= (−1)n−1 (a21 − 2b21 )
= (−1)n .

an
√
(b) (i) Using (a), if n is odd, one gets a2n − 2b2n = −1 < 0 implying that bn < 2 and if n is even, one gets
√
a2n − 2b2n = 1 > 0 implying that (by (a), abnn > 0 ) abnn > 2.
Again, by (a), we see that

an+2 an an+1 + 2bn+1 an

− = −
bn+2 bn an+1 + bn+1 bn
an + bn + 2an + 4bn an
= −
an + bn + an + 2bn bn
2(a2n − 2b2n )
= −
bn (2an + 3bn )
2(−1)n+1
= .
bn (2an + 3bn )

an a2k+1 a2k−1 2 a2k a2k−2

(ii) From above results, we get (by (a), bn > 0) b2k+1 − b2k−1 = b2k−1 (2a2k−1 +3b2k−1 ) > 0 and b2k − b2k−2 =
−2
b2k−2 (2a2k−2 +3b2k−2 ) < 0. Thus, { ab2k−1
2k−1
: k = 1, 2, 3 · · · } is monotonic increasing and { ab2k
2k
: k = 1, 2, 3 · · · }
√
is monotonic decreasing. By (b)(i), { ab2k−1
2k−1
: k = 1, 2, 3 · · · } is bounded above by 2 and { ab2k
2k
:k =
√
1, 2, 3 · · · } is bounded below by 2, therefore, by monotone convergence theorem, both { ab2k−1
2k−1
} and
a2k−2
{ ab2k
2k
} converge. Let lim = L1 and lim a2k
= L2 then
k→∞ b2k−2 k→∞ b2k

a2k−2
a2k−1 a2k−2 + 2b2k−2 b2k−2 +2
lim = lim = lim a2k−2
k→∞ b2k−1 k→∞ a2k−2 + b2k−2 k→∞ +1
b2k−2

and

a2k−1
a2k a2k−1 + 2b2k−1 b2k−1 +2
lim = lim = lim a2k−1 .
k→∞ b2k k→∞ a2k−1 + b2k−1 k→∞ +1
b2k−1
 12

These set up two equations

L2 + 2 L1 + 2
L1 = and L2 = .
L2 + 1 L1 + 1
an
√
On solving, one obtains (by (a), bn > 0 implies that L1 and L2 both are non-negative) L1 = L2 = 2,
hence the results follow.

2.3 Sandwich theorem for sequences

Exercise 7. (Level 3)
Evaluate the following limits
√
n+1
(a) lim n
n→∞

(−1)n n2 −1
(b) lim n3
n→∞

sin2 (n!)
(c) lim n!
n→∞

sinh2 (n−2 )
(d) lim n
n→∞

Solution. (a) Note that for any positive integer n,

√ √ √
n+1 2n 2
0< 2n
< 2n < .
n n n
√
1 n+1
Together with lim = 0, by sandwich theorem, we get lim n2n = 0.
n→∞ n n→∞

(b) Observe that for any positive integer n,

−n2 − 1 (−1)n n2 − 1 n2 − 1
< < .
n3 n3 n3

−n2 −1 n2 −1 (−1)n n2 −1
With this fact lim n3 = lim n3 = 0, by sandwich theorem, one has lim n3 = 0.
n→∞ n→∞ n→∞

sin2 (n!)
(c) Since 0 < sin2 (n!) < 1, then we have that 0 < n! < 1
n! ≤ 1
n for all positive integer n. Hence lim 1
=0
n→∞ n
sin2 (n!)
gives lim n! = 0 by the sandwich theorem.
n→∞

ex −e−x
(d) Recall that sinh x = 2 , then 0 < sinh2 (n−2 ) < e−1
2 < 1 for all positive integer n. Thus, it implies that

sinh2 (n−2 ) 1
0< < .
n n

1 sinh2 (n−2 )
Therefore, because lim = 0, lim = 0 by sandwich theorem.
n→∞ n n→∞ n
 13

Exercise 8. (Level 3)

(a) For 1 ≤ k ≤ n, show that (n + 1 − k)k ≥ n. Hence, deduce that (n!)2 ≥ nn .

1
(b) Find lim √ , if exists.
n→∞ n n!

Solution. (a) Observe that, for 1 ≤ k ≤ n, (n + 1 − k)k − n = (n − k)(k − 1) ≥ 0, that is (n + 1 − k)k ≥ n.

Multiplying the inequalities (n + 1 − k)k ≥ n from k = 1 to k = n, we then get

(n!)2 ≥ nn .

1 1 1 1
(b) From the previous part, one has 0 < √
n
≤ √ . Since lim √ = 0, by the sandwich theorem, lim √ =
n! n n→∞ n n→∞ n n!

0.

Exercise 9. (Level 3)
1 1 1
Let sn = 2
+ 2
+ ··· + , show that lim sn = 0.
(2n) (2n + 1) (3n)2 n→∞

Solution. Observe that

2n + 1 1 1 1 1 1 1
≤ + + ··· + ≤ + + ··· +
(3n)2 (3n)2 (3n)2 (3n)2 (2n)2 (2n + 1)2 (3n)2

and

1 1 1 1 1 1 2n + 1
+ + ··· + ≤ + + ··· + ≤ .
(2n)2 (2n + 1)2 (3n)2 (2n)2 (2n)2 (2n)2 (3n)2
2n+1 2n+1 2n+1 2n+1
Then (3n)2 ≤ sn ≤ (2n)2 , together with lim 2 = lim 2 = 0, by the sandwich theorem, lim sn = 0 as
n→∞ (2n) n→∞ (3n) n→∞
desired.

Exercise 10. (Level 4/Level 5)

(a) Let x be a positive real number. Prove that for any positive integer n greater than 1,

(1 + x)n > 1 + nx.

(b) Let t be any fixed positive number. Consider the sequence

a1 , a2 , · · · , an , · · ·
√
n
where an = t.

(i) Let t > 1.

A. Show that an > 1.
 14

B. Putting an = 1 + xn , show that, for n ≥ 2,

t−1
1 < an < 1 + .
n

C. Show that for t > 1, lim an = 1.

t→∞

(ii) Show that for t > 0, lim an = 1.

t→∞

(c) For any k fixed positive numbers a1 , a2 , · · · , ak , prove that the sequence
p
vn = n
an1 + an2 + · · · + ank

converges and find its limit.

Solution. (a) For any positive integer n greater than 1, using binomial theorem, one gets

(1 + x)n = 1 + nx + C2n x2 + · · · + xn > 1 + nx.

√
n
(b) (i) A. Since t > 1, an > t > 1.
B. By the previous result, an > 1 for n ≥ 2. For the other inequality, consider (1 + xn )n . Using part (a),
t−1 t−1
we have t = (1 + xn )n > 1 + nxn , therefore, xn < n for n ≥ 2. Hence 1 < an = 1 + xn < 1 + n
for n ≥ 2.
t−1
C. Since lim n = 0, by sandwich theorem, we get lim an = 1.
n→+∞ t→∞

(ii) We consider three cases:

• For t > 1, by (b)(i), we get lim an = 1.
t→∞

• For t = 1, an = 1 for all positive integers n. Then, lim an = 1.

t→∞
q
• For 0 < t < 1, Consider bn := a1n = n 1t for all positive integers n. Since 1
t > 1, lim bn = 1, by
t→∞
(b)(i), and hence lim an = 1.
t→∞

(c) Let a be the largest of the numbers a1 , a2 , · · · , ak , then

p √
n
a< n
an1 + an2 + · · · + ank ≤ a k.

√
n
By above result, lim k = 1, therefore, by sandwich theorem, we get
t→∞

p
lim n
an1 + an2 + · · · + ank = a.
t→∞
 15

3 Limits of functions
3.1 Intuitive definition of limits of functions
Exercise 1. (Level 2)
Evaluate the following limits.
x3 −2x−1
(a) lim 5
x→−1 x −2x−1

x+x2 +x3 +···+xn −n

(b) lim x−1
x→1

x2 +3
(c) lim √
x→+∞ x x2 +1

x2 +3
(d) lim x
√
x2 +1
x→−∞
h 3 √ √ √
i
(e) lim x2 x+2−2 x+1+ x
x→+∞

Solution. (a)

x3 − 2x − 1 (x + 1)(x2 − x − 1)
lim = lim
x→−1 x5 − 2x − 1 x→−1 (x + 1)(x4 − x3 + x2 − x − 1)

(x2 − x − 1)
= lim
x→−1 (x4 − x3 + x2 − x − 1)
1
=
3

xk −1 (x−1)(xk−1 +xk−2 +···+1)

(b) Note that for any positive integer k, lim = lim = k. Then,
x→1 x−1 x→1 x−1

x + x2 + x3 + · · · + xn − n
lim
x→1 x−1
(x − 1) + (x2 − 1) + (x3 − 1) + · · · + (xn − 1)
= lim
x→1 x−1
x−1 x2 − 1 x3 − 1 xn − 1
= lim + lim + lim + · · · + lim
x→1 x − 1 x→1 x − 1 x→1 x − 1 x→1 x − 1

= 1 + 2 + 3 + ··· + n
n(n + 1)
= .
2

(c)
x2 + 3 1 + 32
lim √ = lim q x = 1
x→+∞ x x2 + 1 x→+∞
1 + 12 x

(d) We have, by the substitution, y = −x, y → +∞ as x → −∞,

x2 + 3 y2 + 3 1 + y32
lim √ = lim − p = lim − q = −1.
x→−∞ x x2 + 1 y→+∞ y y 2 + 1 y→+∞ 1 + 12 y
 16

(e)
h 3 √ √ √
i
lim x2 x+2−2 x+1+ x
x→+∞
h 3 √ √ √ √
i
= lim x2 x+2− x+1+ x− x+1
x→+∞
h 3 √ √ √ √ √ √ √ √ i
= lim x 2 ( x+2−√x+1)(√
x+2+ x+1)
x+2+ x+1
+ ( x− x+1)( x+ x+1)
√ √
x+ x+1
x→+∞
  
3 1 1
= lim x2 √ √ −√ √
x→+∞ x+2+ x+1 x+ x+1
  √ √ 
3 ( x − x + 2)
= lim x 2 √ √ √ √
x→+∞ ( x + 2 + x + 1)( x + x + 1)
  √ √ √ √ 
3 ( x − x + 2)( x + x + 2)
= lim x 2 √ √ √ √ √ √
x→+∞ ( x + 2 + x + 1)( x + x + 1)( x + x + 2)
  
3 −2
= lim x 2 √ √ √ √ √ √
x→+∞ ( x + 2 + x + 1)( x + x + 1)( x + x + 2)
 
−2
= lim  q q q q 
x→+∞
( 1 + x2 + 1 + x1 )(1 + 1 + x1 )(1 + 1 + x2 )

1
= −
4

3.2 Sandwich theorem for functions

Exercise 2. (Level 2)
Evaluate the following limits.

(a) lim− x| sin x1 |

x→0

sin tan x+tan sin x

(b) lim x+1
x→+∞

x5 +x3 sin 1
x2
(c) lim x5 +7
x→+∞

3x2 +sin 5x+cos 7x

(d) lim 2
x→+∞ 11x +sin 13x+cos 17x

Solution. (a) Since for x < 0, 0 < | sin x1 | < 1, we get x < x| sin x1 | < 0 for x < 0. Because lim x = 0, by the
x→0−
1
sandwich theorem, lim− x| sin x| = 0.
x→0

(b) Provided that x is large enough, we have −1 < sin tan x < 1 and −tan1 < tan sin x < tan1. Then
one has − 1+tan1
x+1 < sin tan x+tan sin x
x+1 < 1+tan1
x+1 . Because lim 1+tan1
x+1 = 0, by the sandwich theorem,
x→+∞
sin tan x+tan sin x
lim x+1 =0
x→+∞
 17

(c) When x is very large,

1
x5 − x3 ≤ x5 + x3 sin ≤ x5 + x3 .
x2
Hence we get
x5 − x3 x5 + x3 sin x12 x5 + x3
≤ ≤ .
x5 + 7 x5 + 7 x5 + 7
x5 −x3 x5 +x3
Let g(x) = x5 +7 and x5 +7 .
lim g(x) = lim h(x) = 1,
x→+∞ x→+∞

5 3 1
x +x sin x2
thus, by the sandwich theorem, we have lim x5 +7 = 1.
x→+∞

sin nx cos mx 3x2 + sin 5x + cos 7x

(d) By sandwich theorem, we have lim 2 = 0 and lim x2 = 0 for any n, m. Hence lim =
x→+∞ x x→+∞ x→+∞ 11x2 + sin 13x + cos 17x

3 + sinx25x + cosx27x 3
lim = 11 .
x→+∞ 11 + sin 13x + cosx17x
x2 2

3.3 Two important limits: lim (1 + x1 )x = e and lim sin x

x
=1
x→+∞ x→0

Exercise 3. (Level 3)
Evaluate the following limits.

1+sin x−cos x
(a) lim
x→0 1+sin px−cos px

cot(a+2x)−2 cot(a+x)+cot a
(b) lim x2
x→0

x2 √
(c) lim √
1+x sin x− cos x
x→0

1−cos x cos 2x cos 3x

(d) lim 1−cos x
x→0

(x+a)x+a (x+b)x+b
(e) lim 2x+a+b
x→+∞ (x+a+b)

√
1+x sin x−1
(f) lim ex2 −1
x→0

Solution. (a)

1 + sin x − cos x 2 sin2 x2 + sin x

lim = lim
x→0 1 + sin px − cos px x→0 2 sin2 px + sin px
2
sin2 x sin x
2 2
+
= lim sin2xpx x
x→0 sin px
2 x2 + x

1
=
p
 18

(b) Note that, by trigonometry identities,

cot(a + 2x) − 2 cot(a + x) + cot a

= cot(a + 2x) − cot(a + x) + cot a − cot(a + x)
cos(a+2x) sin(a+x)−sin(a+2x) cos(a+x) cos(a) sin(a+x)−sin(a) cos(a+x)
= sin(a+2x) sin(a+x) + sin(a) sin(a+x)

− sin x sin x
= +
sin(a + 2x) sin(a + x) sin(a) sin(a + x)
sin x sin(a + 2x) − sin(a)
= ·
sin(a + x) sin(a + 2x) sin(a)
sin x 2 cos(a + x) sin x
= ·
sin(a + x) sin(a + 2x) sin(a)
2 cos(a + x) sin2 x
= .
sin(a + 2x) sin(a + x) sin(a)

Hence, it implies that

cot(a + 2x) − 2 cot(a + x) + cot a

lim
x→0 x2
2 cos(a + x) sin2 x
= lim
x→0 x2 sin(a + 2x) sin(a + x) sin(a)

2
2 cos(a + x) sinx2 x
= lim
x→0 sin(a + 2x) sin(a + x) sin(a)

2 cos a
= .
sin3 a

(c)

x2
lim √ √
x→0 1 + x sin x − cos x
√ √
x2 1 + x sin x + cos x
= lim √ √ · √ √
x→0 1 + x sin x − cos x 1 + x sin x + cos x
√ √
x2 ( 1 + x sin x + cos x)
= lim
x→0 1 + x sin x − cos x
2
√ √
x ( 1 + x sin x + cos x)
= lim
x→0 2 sin2 x2 + x sin x
√ √
1 + x sin x + cos x
= lim sin2 x
x→0
2 x2 2 + sinx x
4
=
3
 19

(d) First note that

1 − cos x cos 2x cos 3x

1 − cos x
1 − cos x + cos x − cos x cos 2x cos 3x
=
1 − cos x
1 − cos 2x cos 3x
= 1 + cos x
1 − cos x
1 − cos 2x + cos 2x − cos 2x cos 3x 1 − cos 2x
= 1 + cos x · ·
1 − cos 2x 1 − cos x
 
1 − cos 3x 1 − cos 2x
= 1 + cos x · 1 + cos 2x · ·
1 − cos 2x 1 − cos x
1 − cos 2x 1 − cos 3x
= 1 + cos x · + cos x · cos 2x ·
1 − cos x 1 − cos x
and
1 − cos nx
lim
x→0 1 − cos x
sin2 nx
2
= lim
x→0 sin2 x
2
n2 x2
sin2 nx
2 22
= lim ·
n x2 sin2 x2
2
x→0
22

= n2 ,

one gets

1 − cos x cos 2x cos 3x

lim
x→0 1 − cos x
 
1 − cos 2x 1 − cos 3x
= lim 1 + cos x · + cos x · cos 2x ·
x→0 1 − cos x 1 − cos x
= 1 + 2 2 + 32
= 14.
 20

(e)

(x + a)x+a (x + b)x+b
lim
x→+∞ (x + a + b)2x+a+b

(1 + xa )x+a (1 + xb )x+b
= lim a+b 2x+a+b
x→+∞ (1 + x )
x a  x b
(1 + xa )1+ a (1 + xb )1+ b

= lim ia+b
x→+∞
h 2x
1+ a+b
(1 + a+b
x )
x a  x b
(1 + xa ) a (1 + xa ) (1 + xb ) b (1 + xb )

= lim ia+b
x→+∞
h 2x
(1 + a+b x )
a+b (1 + a+b )
x

ea · ea
=
e2(a+b)
= e−a−b

(f)
√ √
1 + x sin x − 1 1 + x sin x − 1 x2
lim = lim
x→0 ex2 − 1 x→0 x2 ex2 − 1
√ √
1 + x sin x − 1 1 + x sin x + 1 x2
= lim √ 2
x→0 x2 1 + x sin x + 1 ex − 1
x sin x x2
= lim √ x 2
x→0 x2 ( 1 + x sin x + 1) e − 1
sin x
x x2
= lim √ x2
x→0 1 + x sin x + 1 e − 1
1
=
2
 21

4 Continuity
4.1 Definition of continuity

Exercise 1. (Level 2)
The function f is continuous at x = 0 and is defined for −1 < x < 1 by
 2a
 x ln(1 + x) if − 1 < x < 0

f (x) = b if x = 0
 x2√cos x

if 0 < x < 1.
1− 1−x2

Determine the values of the constants a and b.

Solution. Consider two-sided limits:

1
lim f (x) = lim− 2a ln(1 + x) x = 2a
x→0− x→0

and
p
lim+ f (x) = lim+ cos x(1 + 1 − x2 ) = 2.
x→0 x→0

f is continuous at 0, then 2a = b = 2, hence we get a = 1 and b = 2.

Exercise 2. (Level 3)
Evaluate the following limits.
√ √

(a) lim sin x + 1 − sin x
x→+∞

1
cos x


x2
(b) lim cos 2x
x→0

ax +bx +cx

x1
(c) lim 3 (a, b, c > 0)
x→0

cos(xex )−cos(xe−x )
 
(d) lim x3
x→0

Solution. (a) We have

√ √

lim sin x + 1 − sin x
x→+∞
 √ √
√  √
x+1− x
x+1+ x
= lim 2 cos sin
x→+∞ 22
 √ √ √ √ √ √ 
x+1+ x x+1− x x+1+ x
= lim 2 cos sin √ √
x→+∞ 2 2 x+1+ x
 √ √ 
x+1+ x 1
= lim 2 cos sin √ √
x→+∞ 2 2( x + 1 + x)
= 0,

The last equality is justified by the sandwich theorem.

 22

(b) Note that

1
 nx  x12
lim (cos nx) x2 = lim 1 − 2 sin2
x→0 x→0 2
 nx  x12
lim ln 1−2 sin2
= e x→0 2
−2 sin2 nx
2
 nx  2 sin−1
2 nx
lim ln 1−2 sin2 2

= ex→0 x2 2
n2
= e− 2 .

1 12 22 3
cos x
= e−


x2
We get lim cos 2x
2 e 2 = e2 .
x→0

dx −1
(c) By using lim x = ln d for d > 0, one has
x→0

 x1
ax + bx + cx

lim
x→0 3
x +bx +cx 1
lim ln( a 3 )x
= ex→0
3
ax − 1 + bx − 1 + cx − 1 ax − 1 + bx − 1 + cx − 1 ax −1+bx −1+cx −1
 

lim ln1+ 
x→0 3x 3
= e
ln a+ln b+ln c
= e 3

√3
= abc

(d)

cos(xex ) − cos(xe−x )
 
lim
x→0 x3
ex + e−x ex − e−x
 
−2
= lim sin( x) sin( x)
x→0 x3 2 2
+e−x
x x −x
!
sin( e 2 x) ex + e−x sin( e −e 2 x) ex − e−x
= lim −2 ex +e−x e x −e−x
x→0
2 x 2 2 x 2x
x
+e−x x −x
!
−x sin(
e
2 x) ex + e−x sin( e −e
2 x) e2x − 1
= lim −2e ex +e−x x
e −e −x
x→0
2 x 2 2 x 2x
= −2

4.2 Intermediate value theorem

Exercise 3. (Level 3)
Suppose f : [0, 1] → [0, 1] is a continuous function on [0, 1]. Show that there exists a point x ∈ [0, 1] such that
f (x) = x.
 23

Solution. Since 0 ≤ f (x) ≤ 1 for x ∈ [0, 1], we may assume that f (0) > 0 and f (1) < 1. Otherwise f (0) = 0 or
f (1) = 1, then we are done. Let g : [0, 1] → [0, 1] is defined by g(x) = f (x) − x for x ∈ [0, 1]. Note that g is a
continuous function. Since g(0) = f (0) > 0 and g(1) = f (1) − 1 < 0, by intermediate value theorem, there exists
a point x ∈ (0, 1) such that g(x) = 0 i.e. f (x) = x.

Exercise 4. (Level 4)
Let n be a positive integer greater than 2. Let f : [0, 1] → R is a continuous function. Suppose f (0) = f (1).
Define a function g : [0, 1 − n1 ] → R by g(x) = f (x + n1 ) − f (x) for all x ∈ [0, 1 − n1 ].

n−1
g( nk ) = 0.
P
(a) Show that
k=0

(b) Deduce that there exists a point x ∈ [0, 1 − n1 ] such that f (x + n1 ) = f (x).

n−1 n−1
g( nk ) = f ( k+1 k
P P

Solution. (a) n ) − f ( n ) = f (1) − f (0) = 0.
k=0 k=0

(b) If there exists a point x ∈ {0, n1 , n2 , · · · , n−1 1

n } such that f (x + n ) = f (x). Then we are done.
Now suppose that for all k ∈ {0, 1, 2, · · · , n − 1}, g( nk ) 6= 0. Then there exist k0 , k1 ∈ {0, 1, 2, · · · , n − 1}
such that g( kn0 ) > 0 and g( kn1 ) < 0. By intermediate value theorem with the continuity of g, we get a point
x ∈ [0, 1 − n1 ] such that g(x) = 0, that is f (x + n1 ) = f (x).
 24

5 Differentiation
5.1 Definition of differentiability

Exercise 1. (Level 2)
Investigate the differentiability of the following functions at x = 0.
|x|
(a) y = 1+x2
(
tan−1 1
x x 6= 0
(b) y =
0 x=0
(
x tan−1 1
x x 6= 0
(c) y =
0 x=0
(
x2 tan−1 1
x x 6= 0
(d) y =
0 x=0

|x| f (0+h)−f (0) f (0+h)−f (0) 1 |h|

Solution. (a) Let f (x) = 1+x2 . Consider h . Note that lim h = lim h 1+h2 = −1 and
h→0− h→0−
f (0+h)−f (0) 1 |h|
lim h = lim h 1+h2 = 1, hence f (x) is not differentiable at x = 0.
h→0+ h→0+

(b) The function is not continuous at x = 0, so it is not differentiable.

(
x tan−1 x x 6= 0
(c) Let g(x) = .
0 x=0
g(0+h)−g(0) g(0+h)−g(0) g(0+h)−g(0)
Consider h . Note that lim h = lim tan−1 1
h = − π2 and lim h = lim tan−1 1
h =
h→0− h→0− h→0+ h→0+
π
2, hence g(x) is not differentiable at x = 0.
(
x2 tan−1 1
x x 6= 0
(d) Let k(x) = .
0 x=0
k(0+h)−k(0) k(0+h)−k(0)
Consider h . Note that, by the sandwich theorem lim− h = lim− h tan−1 1
h = 0 and
h→0 h→0
k(0+h)−k(0)
lim h = lim h tan−1 1
h = 0, hence k(x) is differentiable at x = 0.
h→0+ h→0+

Exercise 2. (Level 2)
dy
Find dx , from the first principles, of the following functions:

1
(a) y = x2 +x+1

(b) y = 2x sin 2x
√
√4
3
(c) y = 6 x4 + x
 25

Solution. (a)
 
dy 1 1 1
= lim −
dx h→0 h (x + h)2 + (x + h) + 1 x2 + x + 1
2x + h + 1
= lim −
h→0 [(x + h)2 + (x + h) + 1](x2 + x + 1)
2x + 1
= −
(x2 + x + 1)2

(b)

dy 2(x + h) sin(2x + 2h) − 2x sin 2x

= lim
dx h→0 h
2x[sin(2x + 2h) − sin 2x] + 2h sin(2x + 2h)
= lim
h→0 h
4x cos(2x + h) sin h + 2h sin(2x + 2h)
= lim
h→0 h
 
4x cos(2x + h) sin h
= lim + 2 sin(2x + 2h)
h→0 h
= 4x cos 2x + 2 sin 2x

(c)
 
dy 1 4 4 4 4
= lim 6(x + h) 3 + 1 − 6x 3 −
1
dx h→0 h (x + h) 2 x2
 
1 4 4 4 4
= lim 6(x + h) 3 − 6x 3 + 1 − 1
h→0 h (x + h) 2 x2
6(x + h)4 − 6x4
 
1 4x − 4(x + h)
= lim 8 4 4 8 + 1 1 1 1
h→0 h (x + h) 3 + (x + h) 3 x 3 + x 3 x 2 (x + h) 2 [x 2 + (x + h) 2 ]
1 6(4x3 + 6x62h + 4xh2 + h3 )
 
4
= lim 8 4 4 8 − 1 1 1 1
h→0 h (x + h) 3 + (x + h) 3 x 3 + x 3 x 2 (x + h) 2 [x 2 + (x + h) 2 ]
1 3
= 8x 3 − 2x− 2

5.2 Derivatives and Chain Rule

Exercise 3. (Level 1)
3
2x2 −7x
Let f : (−1, 1) → R be a function defined by f (x) = xesec x
sin cos x sin−1 tan−1 9 . Find f 0 (0).

Solution. We start by the first principle:

2
 
0 f (h) − f (0) sec3 h −1 −1 2h − 7h
f (0) = lim = lim e sin cos h sin tan = 0.
h→0 h h→0 9
 26

Exercise 4. (Level 2)
Find the derivative of the following:
sin x
(a) y = √
a2 cos2 +b2 sin2 x

a a x a x x x a
(b) y = xa + ax + aa + xx + xa + ax + xx + aa

(c) y = x10x

(d) y = xsin x

(e) y = x(sin x)cos x

3
(f) y = (x2 + 2x − 1) 2 (x4 − 3)3

Solution. (a)

dy
dx
p −2a2 cos x sin x + 2b2 sin x cos x
cos x a2 cos2 +b2 sin2 x − sin x p
2 a2 cos2 +b2 sin2 x
=
a2 cos2 +b2 sin2 x
cos x(a2 cos2 +b2 sin2 x) + (a2 − b2 ) cos x sin2 x
= 3
(a2 cos2 +b2 sin2 x) 2
a2 cos x(cos2 + sin2 x)
= 3
(a2 cos2 +b2 sin2 x) 2
a2 cos x
= 3
(a2 cos2 +b2 sin2 x) 2

(b) Note that

(xx )0 = (ex ln x )0 = (1 + ln x)xx ,
a a a a
(xx )0 = (ex ln x 0
) = (xa−1 + axa−1 ln x)ex ln x
= xa−1 (1 + a ln x)xx ,
x x ax x 1 x
(xa )0 = (ea ln x 0
) =( + ln aax ln x)ea ln x = ax ( + ln a ln x)xa ,
x x
x x x x
(ax )0 = (ex ln a 0
) = xx ln a(1 + ln x)ex ln a
= xx ln a(1 + ln x)ax
and
x x 1 xx ln x 1 x
(xx )0 = (ex ln x 0
) = xx (ln x + (ln x)2 + )e = xx (ln x + (ln x)2 + )xx ,
x x
we have

dy
dx
a a x a
−1
= aa xa + a ln axa−1 ax + (ln a)2 ax aa + xa−1 (1 + a ln x)xx
1 x x 1 x
+ax ( + ln a ln x)xa + xx ln a(1 + ln x)ax + xx (ln x + (ln x)2 + )xx .
x x
 27

(c)
dy
= 10x + x10x ln 10
dx

(d)
dy desin x ln x sin x sin x
= =( + cos x ln x)esin x ln x = ( + cos x ln x)xsin x
dx dx x x

(e) Note that

0
0
((sin x)cos x ) = ecos x ln sin x

= (cot x cos x − sin x ln sin x)ecos x ln sin x

= (cot x cos x − sin x ln sin x)(sin x)cos x ,

we have
dy
= x(cot x cos x − sin x ln sin x)(sin x)cos x + (sin x)cos x .
dx

(f) Taking the natural logarithmic

3
ln y = ln(x2 + 2x − 1) + 3 ln(x4 − 3),
2

and differentiation give

4x3
 
dy 3 2x + 2
= y · +3· 4
dx 2 x2 + 2x − 1 x −3
12x3
 
2 3
4 3 3x + 3
= (x + 2x − 1) (x − 3)
2 + .
x2 + 2x − 1 x4 − 3

Exercise 5. (Level 2)
dy xy ln y−y 2
If xy = y x , prove that dx = xy ln x−x2 .

xy ln y−y 2
Solution. Differentiation both sides gives xy ( xy + ln x dx
dy
) = y x (ln y + x dy
y dx ), and hence dy
dx = xy ln x−x2 .

Exercise 6. (Level 3)
Using the identity

x x x sin x
cos cos · · · cos n = n ,
2 4 2 2 sin 2xn

derive the sum

1 x 1 x 1 x
Sn = tan + tan + · · · + n tan n .
2 2 4 4 2 2
 28

Solution. Taking the natural logarithmic

x x x x
ln cos + ln cos + · · · + ln cos n = ln sin x − ln 2n − ln sin n ,
2 4 2 2

and differentiation give

1 x 1 x 1 x x
tan + tan + · · · + n tan n = cot x − cot n .
2 2 4 4 2 2 2

Exercise 7. (Level 3)
Find a polynomial P (x) such that
P 0 (x) − 2P (x) = xn ,

n being a given positive integer.

m
ar xr , then from P 0 (x) − 2P (x) = xn , we have
P
Solution. Let P (x) =
r=0

m
X m
X
rar xr−1 − 2 a r xr = xn
r=1 r=0

m−1
X m
X
(r + 1)ar+1 xr − 2 a r xr = xn
r=0 r=0

n−1
X n−1
X m−1
X m
X
(r + 1)ar+1 xr − 2 ar xr + (r + 1)ar+1 xr − 2 a r xr = xn
r=0 r=0 r=n r=n

Comparing coefficients, one gives when m ≥ r ≥ n + 1, then ar = 0. Hence we have

n−1
X n−1
X
(r + 1)ar+1 xr − 2 ar xr − 2an xn = xn
r=0 r=0

Again by comparing coefficients, it gives

−2an = 1 and (r + 1)ar+1 − 2ar = 0 for r = 1, 2, 3, · · · , n − 1,

then
1 r+1
an = − and ar = ar+1 for r = 1, 2, 3, · · · , n − 1.
2 2
Thus, for r = 1, 2, 3, · · · , n,
1 n(n − 1) · · · (n − r)
an−r = − ,
2 2r
implies that
 
1 n n n−1 n!
P (x) = − x + x + ··· + n .
2 2 2
 29

5.3 Higher derivatives (Leibniz’s Rule)

Exercise 8. (Level 3)
Let f (x) = xn ex where n is a positive integer.

(a) Prove that

n
X n! xn−r
e−x f (n) (x) = { }2 .
r=0
(n − r)! r!

(b) Evaluate f (2n) (0).

Solution. (a) By Leibniz theorem,

n
dn n x X
(x e ) = Crn (xn )(r) (ex )(n−r)
dxn r=0
n
X n!
= n(n − 1) · · · (n − r + 1)xn−r ex
r=0
r!(n − r)!
n
X n! n!
= xn−r ex
r=0
r!(n − r)! (n − r)!

which gives the result.

(b) By Leibniz theorem,

2n
d2n n x X
(x e ) = Cr2n (xn )(r) (ex )(2n−r)
dx2n r=0
n
X (2n)!
= n(n − 1) · · · (n − r + 1)xn−r ex
r=0
r!(2n − r)!

(2n)!
which gives f (2n) (0) = n! .

Exercise 9. (Level 4)
Let y = (sin−1 x)2

(a) Prove that

(i) (1 − x2 )y 00 − xy 0 = 2,
(ii) (1 − x2 )y (n+2) − (2n + 1)xy (n+1) − n2 y (n) = 0.

(b) Deduce that y (n+2) (0) = n2 y (n) (0) and that

y (2n) (0) = 2[(2n − 2)(2n − 4) · · · 4 · 2]2 and y (2n+1) (0) = 0.

 30

Solution. (a) Prove that

sin−1 x
√ √
(i) y 0 = 2√
1−x2
gives 1 − x2 y 0 = 2 sin−1 x. Differentiation gives 1 − x2 y 00 − √ x
1−x2
y0 = √ 2
1−x2
, that is,
2 00 0
(1 − x )y − xy = 2.
(ii) Differentiate n times by Leibniz theorem, we have that

0 = ((1 − x2 )y 00 − xy 0 )(n)
n(n − 1)
= (1 − x2 )y (n+2) + n(−2x)y (n+1) + (−2)y (n) − xy (n+1) − ny (n)
2
= (1 − x2 )y (n+2) − (2n + 1)xy (n+1) − n2 y (n)

(b) Putting x = 0 in (a)(ii), we get that y (n+2) (0) = n2 y (n) (0).

sin−1 0
By the preceding equation and y 0 (0) = 2√
1−02
= 0, we have

y (2n+1) (0) = y (2n−1) (0) = · · · = y 0 (0) = 0.

Again, by y 00 (0) = 2, one obtains

y (2n) (0) = (2n − 2)2 y (2n−2) (0)

..
.
= [(2n − 2)(2n − 4) · · · 4 · 2]2 y 00 (0)
= 2[(2n − 2)(2n − 4) · · · 4 · 2]2 .
 31

6 Application of differentiation
6.1 Rolle’s Theorem
Exercise 1. (Level 3)
Prove that the equation
a1 a2 an
a1 x + a2 x2 + · · · + an xn = + + ··· +
2 3 n+1
is solvable.

Solution. Let f ; R → R defined by

a1 x2 a2 x3 an xn
 
a1 a2 an
f (x) = + + ··· + − + + ··· + x
2 3 n+1 2 3 n+1
 
a1 a2 an
for all x ∈ R. Note that f is differentiable on R and that f 0 (x) = a1 x+a2 x2 +· · ·+an xn − 2 + 3 + ··· + n+1 .
For f (0) = f (1) = 0 therefore, by Rolle’s theorem, there is an c ∈ (0, 1) such that f 0 (c) = 0 which shows the
result.

Exercise 2. (Level 3)
Let f be a smooth function. Prove that, if a < c < b, there is an ζ ∈ (a, b) such that

f (a) f (b) f (c) f 00 (ζ)

+ + = .
(a − b)(a − c) (b − c)(b − a) (c − b)(c − a) 2

(x−b)(x−c) (x−c)(x−a) (x−b)(x−a)

(Hint: Consider F (x) = f (x) − (a−b)(a−c) f (a) + (b−c)(b−a) f (b) + (c−b)(c−a) f (c).)

(x−b)(x−c) (x−c)(x−a) (x−b)(x−a)

Solution. Let F (x) = f (x) − (a−b)(a−c) f (a) + (b−c)(b−a) f (b) + (c−b)(c−a) f (c). Also F is smooth. Since F (a) =
F (b) = F (c) = 0, by Rolle’s theorem, there exist ξ ∈ (a, c) and η ∈ (b, c) such that F (ξ) = F (η) = 0. Again, by
Rolle’s theorem, there is an ζ ∈ (ξ, η) ⊂ (a, b) such that F 00 (ζ) = 0. This shows the result.

6.2 Lagrange’s mean value theorem and Cauchy’s mean value theorem

Exercise 3. (Level 1)
Suppose a function f is continuous on [a, b] and differentiable on (a, b). Decide whether it is possible to find for
each point ξ of (a, b), there are two points c, d, a ≤ c < ξ < d ≤ b, such that

f (c) − f (d)
= f 0 (ξ).
c−d

Solution. It is impossible. Suppose not, we consider f (x) = x3 and ξ = 0. Then c must be negative while d must
be positive, and
f (c) − f (d) c3 − d3
0 = 3(0)2 = = >0
c−d c−d
which is absurd.
 32

Exercise 4. (Level 2)
Show that for x ∈ (0, π2 ), f (x) = sin x tan x − 2 ln sec x is always positive.

Solution. Differentiation gives

2 sec x tan x
f 0 (x) = cos x tan x + sin x sec2 x − = sin x(sec x − 1)2 > 0,
sec x
therefore f (x) > f (0) = 0 for all x ∈ (0, π2 ).

Exercise 5. (Level 3)
Let I be an open interval and a, b ∈ I with a < b. If f is differentiable on I and if λ is a number between f 0 (a)
and f 0 (b), show that there is at least one point c ∈ (a, b) such that f 0 (c) = λ.
(
f 0 (a) if t = a
(Hint: You may start with defining a function fa (t) = f (a)−f (t) .
t−a if t 6= a

Solution. Define functions fa , fb : [a, b] → R by

(
f 0 (a) if t = a
fa (t) = f (a)−f (t)
t−a 6 a
if t =

and (
f 0 (b) if t = b
fb (t) = f (b)−f (t) .
t−b 6 b
if t =
Since f is differentiable at x = a, x = b, fa and fb are continuous. From the definitions, it follows that
fa (a) = f 0 (a), fa (b) = fb (a) and fb (b) = f 0 (b). Hence λ lies between fa (a) and fa (b) or λ lies between fb (a) and
fb (b). Without loss of generality, we assume λ lies between fa (a) and fa (b). By the intermediate value theorem,
there exists s in (a, b) such that
f (a) − f (s)
λ = fa (s) = .
a−s
Thanks the mean value theorem, there is an c ∈ (a, s) such that

f (a) − f (s)
f 0 (c) = = λ.
a−s

Exercise 6. (Level 3/Level 4)

θ2 θ4 θ2
Show that, when 0 < θ < 2π, 1 − 2 + 24 > cos θ > 1 − 2 .

θ2 θ4
Solution. Let f (θ) = 1 − 2 + 24 − cos θ. Note that for 0 < θ < 2π,

θ3
f 0 (θ) = −θ + + sin θ,
6
θ2
f 00 (θ) = −1 + + cos θ,
3
f (3) (θ) = θ − sin θ,
f (4) (θ) = 1 − cos θ > 0,
 33

and
f (0) = f 0 (0) = f 00 (0) = f (3) (0) = f (4) (0) = 0.

It follows that f (3) is increasing on [0, 2π], and hence f (3) (θ) > f (3) (0) = 0 for 0 < θ < 2π. Apply the argument
00 0
in the same manner, we get f (θ) > 0, f (θ) > 0 and f (θ) > 0 for 0 < θ < 2π. Then one gets when 0 < θ < 2π,
θ2 θ4
1− 2 + 24 > cos θ.
θ2
Similarly, let g(θ) = cos θ − 1 + 2 . Note that for 0 < θ < 2π,

g 0 (θ) = − sin θ + θ,
00
g (θ) = 1 − cos θ > 0,

and
g(0) = g 0 (0) = 0.
θ2 θ2 θ4 θ2
We get when 0 < θ < 2π, cos θ > 1 − 2 , so 1 − 2 + 24 > cos θ > 1 − 2 is established.

Exercise 7. (Level 5)
π
(a) Prove that for 0 < θ < 2, tan θ > θ > sin θ.

(b) (i) Show that C32n−1 − 2C12n−1 − 2C32n+1 − 2C12n+1 = −C32n+3 .

2n+1 2n+1 2n−1 2n−1 2n−1 2n+3
(ii) Show that 2C2r+1 + 2C2r−1 − C2r+1 + 2C2r−1 − C2r−3 = C2r+1 for r ∈ {2, 3, · · · , n − 1}.
(iii) Prove that for n ∈ N,
n
X
2n+1
sin(2n + 1)θ = sin2n+1 θ (−1)r C2r+1 (cot2 θ)n−r ,
r=0

π
where 0 < θ < 2.

(c) (i) Given that α1 , · · · , αn are the roots of an xn + an−1 xn−1 + · · · + a0 = 0 where an 6= 0.
A. Show that α1 + · · · + αn = − an−1
an .

B. Furthermore, α1 , · · · , αn are pairwise distinct, does there exist a β 6∈ {α1 , · · · , αn } such that an β n +
an−1 β n−1 + · · · + a0 = 0?
(ii) Using (b), show that
n
X kπ n(2n − 1)
cot2 = ,
2n + 1 3
k=1

and deduce that

n
X kπ n(2n + 2)
csc2 = .
2n + 1 3
k=1

n
1
P
(d) Let An = k2 . Show that
k=1

π2 π2
     
2n 2n − 1 2n 2n + 2
< An < .
6 2n + 1 2n + 1 6 2n + 1 2n + 1

Hence evaluate lim An .

n→+∞
 34

Solution. (a) Let f (θ) = tan θ − θ and f (θ) = sin θ − θ for 0 < θ < π
2. Differentiation gives f 0 (θ) = sec2 θ − 1 > 0
and g 0 (θ) = cos θ − 1 < 0. Then f is strictly increasing and g is strictly decreasing. Thus, for 0 < θ < π
2,
f (θ) > f (0) and g(θ) < g(0), i.e. tan θ > θ > sin θ.

(b) (i) Compute in a straightforward way.

(ii) Using the fact that Crn + Cr−1
n
= Crn+1 repeatedly, one obtains for r ∈ {2, 3, · · · , n − 1}

2n+3 2n+2 2n+2

C2r+1 = C2r+1 + C2r
2n+1 2n+1 2n+1
= C2r+1 + 2C2r + C2r−1

and
2n+1 2n+1 2n+1
C2r+1 − 2C2r + C2r−1
2n+1 2n+1 2n+1 2n+1
= C2r+1 − C2r + C2r−1 − C2r
2n 2n 2n 2n 2n 2n 2n 2n
= C2r+1 + C2r − C2r − C2r−1 + C2r−1 + C2r−2 − C2r − C2r−1
2n 2n 2n 2n
= C2r+1 − C2r−1 + C2r−2 − C2r
2n−1 2n−1 2n−1 2n−1 2n−1 2n−1 2n−1 2n−1
= C2r+1 + C2r − C2r−1 − C2r−2 + C2r−2 + C2r−3 − C2r − C2r−1
2n−1 2n−1 2n−1
= C2r+1 − 2C2r−1 + C2r−3 .

Combining, we have for r ∈ {2, 3, · · · , n − 1}

2n+1 2n+1 2n−1 2n−1 2n−1

2C2r+1 + 2C2r−1 − C2r+1 + 2C2r−1 − C2r−3
2n+1 2n+1 2n+1
= C2r+1 + 2C2r + C2r−1
2n+3
= C2r+1 .

(iii) For n = 1, we have

sin 3θ = sin(2θ + θ)
= sin 2θ cos θ + sin θ cos 2θ
= 2 sin θ cos2 θ + sin θ(cos2 θ − sin2 θ)
= 3 sin θ cos2 θ − sin3 θ
= sin3 θ(3 cot2 θ − 1)
n
X
= sin3 θ (−1)r C2r+1
3
(cot2 θ)1−r .
r=0

Suppose that
k
X
2k+1 2k+1
sin(2k + 1)θ = sin θ (−1)r C2r+1 (cot2 θ)k−r ,
r=0
 35

for all k ≤ n. Noting that sin(2n + 1)θ cos 2θ = 21 (sin(2n + 3)θ + sin(2n − 1)θ). It gives that

sin(2n + 3)θ
= − sin(2n − 1)θ + 2 sin(2n + 1)θ cos 2θ
n−1
X
2n−1
= sin2n−1 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r
r=0
n
X
2n+1
+ sin2n+1 θ (−1)r C2r+1 2 cos 2θ(cot2 θ)n−r
r=0

n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r csc4 θ
r=0
n
X
2n+1 2 cos 2θ
+ sin2n+3 θ (−1)r C2r+1 (cot2 θ)n−r
r=0
sin2 θ
n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r (cot4 θ + 2 cot2 θ + 1)
r=0
n
X
2n+3 2n+1
+ sin θ (−1)r C2r+1 (2 cot2 θ − 2)(cot2 θ)n−r .
r=0

The first and second summations would be treated as follows:

n−1
X
2n−1
sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r (cot4 θ + 2 cot2 θ + 1)
r=0

n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n+1−r
r=0

n−1
X
2n−1
+ sin2n+3 θ (−1)r+1 2C2r+1 (cot2 θ)n−r
r=0

n−1
X
2n−1
+ sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r
r=0

n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n+1−r
r=0
n
X
2n−1
+ sin2n+3 θ (−1)r 2C2r−1 (cot2 θ)n+1−r
r=1

n+1
X
2n−1
+ sin2n+3 θ (−1)r+1 C2r−3 (cot2 θ)n+1−r ,
r=2
 36

n
X
2n+1
sin2n+3 θ (−1)r C2r+1 (2 cot2 θ − 2)(cot2 θ)n−r
r=0
n
X
2n+1
= sin2n+3 θ (−1)r 2C2r+1 (cot2 θ)n+1−r
r=0
n
X
2n+1
+ sin2n+3 θ (−1)r 2C2r+1 (cot2 θ)n−r
r=0
n
X
2n+1
= sin2n+3 θ (−1)r 2C2r+1 (cot2 θ)n+1−r
r=0

n+1
X
2n+1
+ sin2n+3 θ (−1)r+1 2C2r−1 (cot2 θ)n+1−r .
r=1
 37

Add up, by the above results, one arrives

sin(2n + 3)θ
n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n−1−r (cot4 θ + 2 cot2 θ + 1)
r=0
n
X
2n+1
+ sin2n+3 θ (−1)r C2r+1 (2 cot2 θ − 2)(cot2 θ)n−r
r=0

n−1
X
2n−1
= sin2n+3 θ (−1)r+1 C2r+1 (cot2 θ)n+1−r
r=0
n
X
2n−1
+ sin2n+3 θ (−1)r 2C2r−1 (cot2 θ)n+1−r
r=1

n+1
X
2n−1
+ sin2n+3 θ (−1)r+1 C2r−3 (cot2 θ)n+1−r
r=2
n
X
2n+1
+ sin2n+3 θ (−1)r 2C2r+1 (cot2 θ)n+1−r
r=0

n+1
X
2n+1
+ sin2n+3 θ (−1)r+1 2C2r−1 (cot2 θ)n+1−r
r=1

= sin 2n+3
θ(−C12n−1 + 2C12n+1 )(cot2 θ)n+1
+ sin2n+3 θ(C32n−1 − 2C12n−1 − 2C32n+1 − 2C12n+1 )(cot2 θ)n
n−1
X
2n+1 2n+1 2n−1 2n−1 2n−1
+ sin2n+3 θ (−1)r (2C2r+1 + 2C2r−1 − C2r+1 + 2C2r−1 − C2r−3 )(cot2 θ)n+1−r
r=2
2n+3 2n−1 2n−1 2n+1 2n+1
+ sin θ(−1)n (2C2n−1 − C2n−3 + 2C2n+1 + 2C2n−1 ) cot2 θ
2n−1 2n+1
+ sin2n+3 θ(−1)n+1 (−C2n−1 + 2C2n+1 )
= sin2n+3 θ(cot2 θ)n+1 + sin2n+3 θ(−C32n+3 )(cot2 θ)n
n−1
X
2n+3 2n+3
+ sin2n+3 θ (−1)r (C2r+1 )(cot2 θ)n+1−r + sin2n+3 θ(−1)n (C2n+1 ) cot2 θ
r=2

+ sin2n+3 θ(−1)n+1
n+1
X
2n+3
= sin2n+3 θ (−1)r C2r+1 (cot2 θ)n+1−r .
r=0

Hence, we are done by mathematical induction.

(c) (i) A. Note that

an−1 n−1 a0
xn + x + ··· + = (x − α1 )(x − α2 ) · · · (x − αn )
an an
= xn − (α1 + · · · + αn )xn−1 + · · · + α1 · · · αn

The result follows by comparing the coefficients of xn−1

 38

B. No. Otherwise 0 = an β n + an−1 β n−1 + · · · + a0 = an (β − α1 ) · · · (β − αn ) which implies that an = 0.

It contradicts the given condition.
kπ
Pn 2n+1 n−r
(ii) From (b), we may observe that cot2 2n+1 for k = 1, 2, · · · , n are distinct roots of r=0 (−1)r C2r+1 x =
0. By A. of (i), we get
n
X kπ C 2n+1 n(2n − 1)
cot2 = 32n+1 = ,
2n + 1 C1 3
k=1

and hence
n n  
X kπ X kπ n(2n − 1) n(2n + 2)
csc2 = 1 + cot2 =n+ = .
2n + 1 2n + 1 3 3
k=1 k=1

(d) By (a), for k ∈ {1, 2, · · · , n}, we get

 2
kπ kπ kπ
cot2 < < csc2 .
2n + 1 2n + 1 2n + 1

Summing k from 1 to n, one obtains

n n  2 n
X kπ X 2n + 1 X kπ
cot2 < < csc2 .
2n + 1 kπ 2n + 1
k=1 k=1 k=1

π2
Applying the results of (c)(ii) and multiplying the constant (2n+1)2 , we have

π2 π2
     
2n 2n − 1 2n 2n + 2
< An < .
6 2n + 1 2n + 1 6 2n + 1 2n + 1
   
2n 2n−1 2n 2n+2
Since lim 2n+1 · 2n+1 = 1 and lim 2n+1 · 2n+1 = 1, by the sandwich theorem, one gets lim An =
n→+∞ n→+∞ n→+∞
π2
6 .

6.3 First and second derivative check

Exercise 8. (Level 3)
Prove that the function
9 1
f (x) = +
x 4−x
has a maximum point and a minimum point.

Solution. Note that

(x − 6)(x − 3) 18 2
f 0 (x) = −8 and f 00 (x) = + .
x2 (4 − x)2 x3 (4 − x)3

The critical values are x = 0, 3, 4, 6. f 0 does not change sign through x = 0 and x = 4, so x = 0 and x = 4 do not
serve as an extremum point. (Or, you may say x = 0 and x = 4 are not any candidate for extremum point, since
the domain of function exclude those two points.) Because f 00 (6) < 0 and f 00 (3) > 0, by the second derivative
test, the maximum point is (6, f (6)) and the minimum point is (3, f (3)).
 39

Exercise 9. (Level 5)
1 5


Let I = [2, 3] and g(x) = 2 x+ x , where x ∈ I. Let x0 ∈ I and define xn+1 = g(xn ) for n = 1, 2, 3, · · · .

(a) Show that the equation x = g(x) has exactly one root in I.

(b) Show that xn ∈ I for n = 1, 2, 3, · · · .

(c) Show that |g 0 (x)| ≤ 2

9 for all x ∈ I.

(d) Let α be the root of g(x) = x mentioned in (a).

(i) Show that |xn − α| ≤ 29 |xn−1 − α|.

(ii) Show that {xn } converges and find the limit.

Solution. (a) Let f (x) = x − g(x). Since f (2)f (3) < 0, by intermediate value theorem, f (x) = 0 i.e. x = g(x)
has at least one root in I. Note that f 0 (x) = 1
2 + 5
2x2 > 0 for all x ∈ I. It says that f is a strictly increasing,
and the root in I is unique.

(b) We shall investigate the absolute maximum and minimum of g on I. The derivative g 0 (x) = 1 5


2 1− x2 . From
√ √ √
g 0 (x) = 0, we have x = 5 or − 5. We look at x = 5. Note that

√ 5
g 00 ( 5) = √ > 0,
( 5)3
√
by the second derivative test, g(x) attains the local minimum at x = 5. Comparing the following three
candidates:
9 √ √ 7
g(2) = , g( 5) = 5 and g(3) = ,
4 3
√ √
one obtains 5 = g( 5) ≤ g(x) ≤ g(3) = 37 , and hence 2 < g(x) < 3 for all x ∈ I. From xn = g(xn−1 ), we
have 2 < xn < 3.

(c) For g 0 (x) = 21 1 − x52 and g 00 (x) = x53 > 0 for all x ∈ I, so g 0 is increasing, therefore,

1 2
− = g 0 (2) ≤ g 0 (x) ≤ g 0 (3) = .
8 9

Then, we get |g 0 (x)| ≤ 2

9 for all x ∈ I.

(d) (i) By the mean value theorem, there exists ξ ∈ (2, 3) such that

g(xn−1 ) − g(α)
= g 0 (ξ).
xn−1 − α

Then, by (c), we have

2
|xn − α| = |g 0 (ξ)||xn−1 − α| ≤ |xn−1 − α|.
9
 40

(ii) Inductively, we get

 n
2 2
|xn − α| ≤ |xn−1 − α| · · · ≤ |x0 − α|.
9 9
2 n


Because lim 9 |x0 − α| = 0, the sandwich theorem tells us that lim |xn − α| = 0. It implies, by
n→∞ n→∞
the sandwich theorem with the inequalities −|xn − α| ≤ xn − α ≤ |xn − α|, that lim xn − α = 0, i.e.
n→∞
 
lim xn = α. The sequence {xn } converges. From xn+1 = g(xn ) = 21 xn + x5n , we have
n→∞

 
1 5
lim xn+1 = lim xn + .
n→∞ n→∞ 2 xn

1 5

√
One arrives α = 2 α+ α which gives α2 = 5. Since α ∈ [2, 3], then we have α = 5, in other words,
√
lim xn = 5.
n→∞

6.4 Curve sketching

Exercise 10. (Level 4)

Let f (x) = (1 + x)e−2x be a real values function defined on R.

(a) Find f 0 (x) and f 00 (x).

(b) Find, if any, the relative maximum and minimum points and the point of inflexion of f (x).

(c) Find, if any, the equation of asymptote.

(d) Sketch the graph of y = f (x).

(e) Sketch the graph of y 2 = f (x).

(f) Sketch the graph of y = |f (x)|.

(g) Sketch the graph of y = f (|x|).

(h) Sketch the graph of y = f (−|x|).

(i) Sketch the graph of y = f (3x + 5).

Solution. (a) Differentiation gives f 0 (x) = −(2x + 1)e−2x and f 00 (x) = 4xe−2x .

(b) f (x) = 0 gives x = − 12 . Because f 00 (− 12 ) < 0, by the second derivative test, the maximum point is
(− 21 , f (− 12 )) = (− 12 , 2e ).
Note that

< 0 if x < 0

00
f (x) = 0 if x = 0 .

> 0 if x > 0


Therefore the point of inflexion is (0, f (0)) = (0, 1).

 41

(c) First notice that lim f (x) = ∞. Making use of the mean value theorem, we get ex > 1 + x for x > 0,
x→−∞

and hence 0 < (1 + x)e−2x < e−x for x > 0. Then, by the sandwich theorem, we have lim f (x) =
x→+∞

lim (1 + x)e−2x = 0. Therefore, y = 0 is the equation of asymptote.

 
x→+∞

(d)

(e)

(f)

(g)

(h)

(i)
 42

7 Indefinite integration
7.1 Primitive functions

Exercise 1. (Level 2)
R
Find f (x)dx for the following functions f (x).

(a) f (x) = πx

(b) f (x) = xπ

(c) f (x) = x−π

√
(d) f (x) = π
x
1
(e) f (x) = x− π

(f) f (x) = eπ

(g) f (x) = π x

(h) f (x) = ln x logx π

(i) f (x) = logπ x

π 2
R
Solution. (a) f (x)dx = 2x +C

xπ+1
R
(b) f (x)dx = π+1 +C

x−π+1
R
(c) f (x)dx = −π+1 +C

π+1
π
R
(d) f (x)dx = π+1 x
π +C

π−1
π
R
(e) f (x)dx = 1−π x
π +C

f (x)dx = eπ x + C
R
(f)

ex ln π πx
ex ln π dx =
R R
(g) f (x)dx = ln π +C = ln π +C

ln x ln π
R R
(h) f (x)dx = ln x dx = x ln π + C

ln x 1
R R
(i) f (x)dx = ln π dx = x ln π +C

Exercise 2. (Level 2)
q
Let |x| < 1. Show that the following three functions cos−1 x, − sin−1 x and 2 cos−1 1
2 (x + 1) are the primitive
functions of a same function.
 43

Solution. Differentiation gives

−1
d
dx cos−1 x = − √1−x
1
2
, d
dx (− sin
1
x) = − √1−x 2
and

r ! 1
√ √
d −1 1 2 2 x+1 1
2 cos (x + 1) = −2 q = −√
dx 2 1− 1 1 − x2
2 (x + 1)

. Thus they are the primitive functions of a same function.

Exercise 3. (Level 3)
R
Find f (x)dx for the following functions f (x).

2
(x + 2)
 for x > 0
(a) f (x) = 2 for x = 0

1 for x < 0


(b) f (x) = |x2 + x + 6|

(c) f (x) = |x2 + x − 6|

(d) f (x) = |x + 3| + |x − 2|
1
(e) f (x) = |x+3|+|x−2|

(x+2)3
(
R
3 for x ≥ 0
Solution. (a) f (x)dx = F (x) + C where F (x) =
x + 83 for x < 0

x3 x2
R
(b) f (x)dx = 3 + 2 + 6x + C
 3
x x2
3 + 2 − 6x for x ≥ 2

R 3 2
(c) f (x)dx = F (x) + C where F (x) = − x3 − x2 + 6x + 44 3 for −3 ≤ x < 2

x 3 2
x 37
3 + 2 − 6x − 3 for x < −3

2
x + x
 for x ≥ 2
R
(d) f (x)dx = F (x) + C where F (x) = 5x − 4 for −3 ≤ x < 2

 2
−x − x − 13 for x < −3
 ln |2x+1|
 2
 for x ≥ 2
R
(e) f (x)dx = F (x) + C where F (x) = x5 + ln25 − 52 for −3 ≤ x < 2
 ln |2x+1|

− 2 + ln 5 − 1 for x < −3

Exercise 4. (Level 3)
Find

(a) sin2 x1cos2 x dx and

R

1√
R
(b) √
ax+b− ax+c
dx provided that b 6= c.
 44

Solution. (a)

sin2 x + cos2 x
Z Z
1
dx = dx
sin2 x cos2 x sin2 x cos2 x
Z
1 1
= 2 + 2x
dx
sin x cos
= tan x − cot x + C

(b)
Z Z √ √
1 ax + b + ax + c
√ √ dx = dx
ax + b − ax + c (ax + b) − (ax + c)
2 h 3 3
i
= (ax + b) 2 + (ax + c) 2 + C
3a(b − c)

7.2 Integration by substitution

Exercise 5. (Level 2)
Evaluate the integrals.
x
R
(a) √1+x 2
dx
R√
(b) x3 + 1x2 dx

ex cos ex dx
R
(c)

ln3 x
R
(d) x dx

√
√ x 1 √ 1 + x2 ) =
R R
Solution. (a) 1+x2
dx = 2 1+x2
d(1 1 + x2 + C
R√ 1
R√ 3
(b) x3 + 1x2 dx = 3 x3 + 1d(1 + x3 ) = 92 (x3 + 1) 2 + C

ex cos ex dx = cos ex dex = sin ex + C

R R
(c)

ln3 x ln3 x ln4 x

R R
(d) x dx = x d ln x = 4 +C

Exercise 6. (Level 3)
Evaluate the following integrals.

(a) 1 + tan6 xdx

R

x2n−1
R
(b) xn +1 dx

√1
R
(c) x+ x2 +1
dx

√ln x
R
(d) x 1+ln x
dx
 45

Solution. (a)
Z Z
6
1 + tan xdx = (tan4 x − tan2 x + 1)(1 + tan2 x)dx
Z
= (tan4 x − tan2 x + 1)(sec2 x)dx
Z
= (tan4 x − tan2 x + 1)d tan x

tan5 x tan3 x
= − + tan x + C
5 3

(b)

x2n−1 xn−1 (xn + 1) − xn−1

Z Z
dx = dx
xn + 1 xn + 1
xn−1
Z
= xn−1 − dx
xn + 1
xn
Z
1 1
= − d(xn + 1)
n n xn +1
xn ln(xn + 1)
= − + C.
n n

√ t2 −1 t2 +1
(c) Let t = x + x2 + 1, then x = 2t and dx = 2t2 dt. We have

t2 + 1
Z Z
1
√ dx = dt
x + x2 + 1 2t3
Z
1 1 1
= + 3 dt
2 t t
 
1 1
= ln |t| − 2 + C
2 2t
 
1 p 1
= 2
ln |x + x + 1| − √ + C.
2 2(x + x2 + 1)2

ln x
(d) Let t = 1 + ln x, then x dx = (t − 1)dt. We have
Z Z
ln x 1 2 1 2 1
√ dx = t− 2 (t − 1)dt = (t − 3)t 2 + C = (ln x − 2)(1 + ln x) 2 + C.
x 1 + ln x 3 3

Exercise 7. (Level 3/Level 4)

Using cos(A − B) = cos A cos B + sin A sin B, find the following integrals.
dx
R
(a) sin(x+a) cos(x+b) if cos(a − b) 6= 0.

x+a
R cosn−1
(b) sinn+1
2
x−a dx if cos a 6= 0.
2
 46

Solution. (a)

cos(a − b)
Z Z
dx 1
= dx
sin(x + a) cos(x + b) cos(a − b) sin(x + a) cos(x + b)
cos[(x + a) − (x + b)]
Z
1
= dx
cos(a − b) sin(x + a) cos(x + b)
Z
1 cos(x + a) sin(x + b)
= + dx
cos(a − b) sin(x + a) cos(x + b)
Z Z
1 d sin(x + a) d cos(x + b)
= −
cos(a − b) sin(x + a) cos(x + b)
 
1  sin(x + a) 
= ln  +C
cos(a − b)  cos(x + b) 

cos x+a sin x+a

sin x−a
2 +cos 2
x+a
cos x−a
cos a
(b) Let t = 2
sin x−a
, then dt = − 2
2 sin2 x+a
2
dx = − 2 sin 2 x+a dx. One obtains
2 2 2

cosn−1 x+a
Z Z
1
2
dx = − 2tn−1 dt
sinn+1 x−a
2
cos a
tn
= − +C
n cos a
1 cosn x+a 2
= − + C.
n cos a sinn x−a
2

7.3 Integration by parts

Exercise 8. (Level 2)
Evaluate the following integrals:

(a) x sec2 xdx

R

R
(b) ln xdx

(c) sec2 x ln tan xdx

R

Solution. (a)
Z Z Z
2
x sec xdx = xd tan x = x tan x − tan xdx = x tan x − ln | sec x| + C

(b)
Z Z Z
ln xdx = x ln x − xd ln x = x ln x − 1dx = x ln x − x + C

(c)
Z Z
2
sec x ln tan xdx = ln tan xd tan x = tan x ln tan x − tan x + C
 47

Exercise 9. (Level 3/Level 4)

Evaluate the following integrals:

sin2 xdx
R
(a)

x5
R
(b) (1+x3 )3 dx

1+x
x2 ln 1−x
R
(c) dx
 
1+sin x
ex
R
(d) 1+cos x dx
R
(e) sin ln xdx

Solution. (a)
Z Z
sin2 xdx = − sin xd cos x
Z
= − sin x cos x + cos2 xdx
Z
= − sin x cos x + 1 − sin2 xdx
Z
= − sin x cos x + x − sin2 xdx

gives that
Z
sin x cos x x
sin2 xdx = − + + C.
2 2

(b)

x5 −x3
Z Z
dx = d(1 + x3 )−2
(1 + x3 )3 6
x3 x2
Z
= − + dx
6(1 + x3 )2 2(1 + x3 )2
x3
Z
1
= − + d(1 + x3 )
6(1 + x3 )2 6(1 + x3 )2
x3 1
= − − +C
6(1 + x3 )2 6(1 + x3 )
 48

(c)
Z Z
2 1+x 1 1+x 3
x ln dx = ln dx
1−x 3 1−x
x3
Z
1+x 1 1+x
= ln − x3 d ln
3 1−x 3 1−x
x3 x3
Z
1+x 2
= ln − dx
3 1−x 3 1 − x2
x3 x(x2 − 1) + x
Z
1+x 2
= ln − dx
3 1−x 3 1 − x2
x3
Z Z
1+x 2 2 x
= ln + xdx − dx
3 1−x 3 3 1 − x2
x3 1 + x x2
Z
1 1
= ln + + d(1 − x2 )
3 1−x 3 3 1 − x2
x3 1 + x x2 1
= ln + + ln(1 − x2 ) + C
3 1−x 3 3

(d)
Z  
x 1 + sin x
e dx
1 + cos x
Z  
1 + sin x
= dex
1 + cos x
  Z  
1 + sin x 1 + sin x
= ex − ex d
1 + cos x 1 + cos x
  Z  
1 + sin x 1 + sin x + cos x
= ex − ex dx
1 + cos x (1 + cos x)2
  Z   Z  
1 + sin x 1 sin x
= ex − ex dx − ex dx
1 + cos x 1 + cos x 1 + cos x
  Z   Z  
x 1 + sin x x 1 x 1
= e − e dx − e d
1 + cos x 1 + cos x 1 + cos x
  Z     Z  
x 1 + sin x x 1 x 1 x 1
= e − e dx − e + e dx
1 + cos x 1 + cos x 1 + cos x 1 + cos x
 
x sin x
= e +C
1 + cos x
 49

(e) Note that

Z
sin ln xdx
Z
= x sin ln x − xd sin ln x
Z
= x sin ln x − cos ln xdx
Z
= x sin ln x − x cos ln x + xd cos ln x
Z
= x sin ln x − x cos ln x − x sin ln xdx,

then, by rearrangement, we get

Z
1
sin ln xdx = (x sin ln x − x cos ln x) + C.
2

7.4 Trigonometric substitution

Exercise 10. (Level 2)

Evaluate the following integrals:

(a) x21+9 dx
R

√ 1
R
(b) x2 +9
dx
R√
(c) x2 + 9dx

√ 1
R
(d) 25−x2
dx
R√
(e) 25 − x2 dx

√ 1
R
(f) x2 −16
dx
R√
(g) x2 − 16dx

Solution. (a) Let x = 3 tan z. Then dx = 3 sec2 zdz, we get

Z Z
1 1 x
dx = · 3 sec2 zdz = 3z + C = 3 tan−1 + C.
x2 + 9 9 sec2 z 3
 50

(b) Let x = 3 tan z. Then dx = 3 sec2 zdz, we get

Z Z
1 1
√ dx = · 3 sec2 zdz
x2 +9 3 sec z
Z
= sec zdz

= ln | tan z + sec z| + C 0
√
x x2 + 9
= ln | + | + C0
3 3
p
= ln |x + x2 + 9| + C.

(c)
Z p p Z p
x2 + 9dx = x x2 + 9 − xd x2 + 9

x2
p Z
= x x2 +9− √ dx
x2 + 9
x2 + 9
Z
Z
p 9
= x +9− x2√ dx + √ dx
2
x +9 2
x +9
p Z p p
= x x2 + 9 − x2 + 9dx + 9 ln |x + x2 + 9|

gives that
√
x x2 + 9 9
Z p p
x2 + 9dx = + ln |x + x2 + 9| + C
2 2

(d) Let x = 5 sin z. Then dx = 5 cos zdz, we get

Z Z
1 1 x
√ dx = · 5 cos zdz = z + C = sin−1 + C.
25 − x2 5 cos z 5

(e)
Z p p Z p
25 − x2 dx = x 25 − x2 − xd 25 − x2

x2
p Z
= x 25 − x2 + √ dx
25 − x2
x2 − 25
Z
Z
p 25
= x 25 − + √ x2 dx + √ dx
25 − x 2 25 − x2
Z p Z
p 25
= x 25 − x2 − 25 − x2 dx + √ dx
25 − x2

gives that
√
x 25 − x2
Z p
25 x
25 − x2 dx = + sin−1 + C.
2 2 5
 51

(f) Let x = 4 sec z. Then dx = 4 sec z tan zdz, we get

Z Z
1 1
√ dx = · 4 sec z tan zdz
2
x − 16 4 tan z
Z
= sec zdz

= ln | sec z + tan z| + C 0
√
x x2 − 16
= ln | + | + C0
4 4
p
= ln |x + x2 − 16| + C.

(g)
Z p p Z p
x2 − 16dx = x x2 − 16 − xd x2 − 16

x2
p Z
= 2
x x − 16 − √ dx
x2 − 16
x2 − 16
Z Z
p 1
= 2
x x − 16 − √ dx − 16 √ dx
2
x − 16 2
x − 16
Z p Z
p 1
= x x2 − 16 − x2 − 16dx − 16 √ dx
x2 − 16

gives that
√
x x2 − 16
Z p p
x2 − 16dx = − 8 ln |x + x2 − 16| + C.
2

Exercise 11. (Level 3)

Evaluate the following integrals:
√
x2 −a2
R
(a) x4 dx

1
R
(b) x2 +x+1 dx

2 sin x
R
(c) 3 sin2 x+4 cos2 x
dx

1
R
(d) x2 −2x cos α+1 dx (0 < α < π)

R (sin−1 x)2
(e) √
1−x2
dx

1√
R
(f) (1+x2 ) 1−x2
dx
 52

Solution. (a) Let x = a sec z, then dx = a sec x tan zdz

Z √ 2
x − a2
Z
a tan z
dx = a sec x tan zdz
x4 a4 sec4 z
Z
1
= sin2 z cos zdz
a2
Z
1
= sin2 zd sin z
a2
1 sin3 z
= +C
a2 3
3
(x2 − a2 ) 2
= +C
3a2 x3

(b)
Z Z
1 1 2 2x + 1
dx = 1 2 dx = √ tan−1 √
3 + C.
x2 + x + 1 (x + 2) + 4 3 3

(c)
Z Z
2 sin x d cos x 2 cos x
dx = −2 = − √ tan−1 √ + C.
3 sin2 x + 4 cos2 x 3 + cos2 x 3 3
(d)
Z Z
1 1
dx = dx
x2 − 2x cos α + 1 x2 − 2x cos α + cos2 α + sin2 α
Z
1
= dx
(x − cos α)2 + sin2 α
1 x − cos α
= tan−1 +C
sin α sin α

(e) Let x = sin z, then dx = cos zdz. We get

(sin−1 x)2 z 2 cos zdz z3 (sin−1 x)3

Z Z Z
√ dx = p = z 2 dz = +C = + C.
1 − x2 1 − sin2 z 3 3

(f) Let x = sin z, then dx = cos zdz. We get

Z Z
1 1
√ dx = dz
(1 + x2 ) 1 − x2 (1 + sin2 z)
csc2 z
Z
= dz
(1 + csc2 z)
Z
1
= − d cot z
(2 + cot2 z)
1 cot z
= − √ tan−1 √ + C
2 2
r
1 1 − x2
= − √ tan−1 + C.
2 2x2
 53

7.5 Integration of rational functions

Exercise 12. (Level 2)

R dx
Find x2 −100

1 A B 1 1
Solution. Put x2 −100 = x−10 + x+10 . Then we have 1 = A(x + 10) + B(x − 10), hence A = 20 and B = − 20 .
Thus,
Z Z
dx 1 1
= − dx
x2 − 100 20(x − 10) 20(x + 10)
1 1
= ln |x − 10| − ln |x + 10| + C
20 20
 
1  x − 10 
= ln   + C.
20 x + 10 

Exercise 13. (Level 3/Level 4)

(a) (i) Show that x31+1 = 1

3(1+x) + 2−x
3(1−x+x2 ) .

(ii) Find x31+1 dx.

R

√ √
(b) (i) Show that 1
= √ x+ √2 − √ x− √2 .
x4 +1 2
2 2(x + 2x+1) 2
2 2(x − 2x+1)

1
R
(ii) Find x4 +1 dx.

Solution. [(Level 3/Level 4)]

(a) (i) The solution is omitted since it is straightforward.

(ii) Noting that

2−x −1 2x − 1
Z Z
1 1
dx = + dx
3(1 − x + x2 ) 6 1−x+x 2 2 1 − x + x2
d(1 − x + x2 ) 1
Z Z
1 dx
= − +
6 1−x+x 2 2 (x − 21 )2 + 3
2

1 1 2x − 1
= − ln(1 − x + x2 ) + √ tan−1 √ + C,
6 3 3

and using (a)(i), we get

2−x
Z Z
1 1
dx = + dx
x3 + 1 3(1 + x) 3(1 − x + x2 )
1 1 1 2x − 1
= ln |x + 1| − ln(1 − x + x2 ) + √ tan−1 √ + C.
3 6 3 3

(b) (i) The solution is omitted since it is straightforward.

 54

(ii) Noting that

Z √
x+ 2
√ √ dx
2 2(x2 + 2x + 1)
Z √ √
1 2x + 2 2
= √ √ + √ dx
4 2 x2 + 2x + 1 x2 + 2x + 1
√
d(x2 + 2x + 1) 1
Z Z
1 1
= √ √ + 1 2 dx
4 2 2
x + 2x + 1 4 (x + √
2
) + 12
1 √ 1
= √ ln(x2 + 2x + 1) + √ tan−1 (2x + 1) + C1
4 2 2 2

and
√
−x + 2
Z
√ √ dx
2 2(x2 − 2x + 1)
Z √
y+ 2
= − √ √ dy
2 2(y 2 + 2y + 1)
1 √ 1
= − √ ln(y 2 + 2y + 1) − √ tan−1 (2y + 1) + C2
4 2 2 2
1 √ 1
= − √ ln(x2 − 2x + 1) + √ tan−1 (2x − 1) + C2 ,
4 2 2 2

one obtains
Z
1
dx
x4 +1
√ √
x− 2
Z
x+ 2
= √ √ − √ √ dx
2 2(x2 + 2x + 1) 2 2(x2 − 2x + 1)
1 √ 1
= √ ln(x2 + 2x + 1) + √ tan−1 (2x + 1)
4 2 2 2
1 √ 1
− √ ln(x2 − 2x + 1) + √ tan−1 (2x − 1) + C
4 2 2 2
√
1 x2 + 2x + 1 1 1
= √ ln √ + √ tan−1 (2x + 1) + √ tan−1 (2x − 1) + C.
4 2 x2 − 2x + 1 2 2 2 2

7.6 Reduction formulae

Exercise 14. (Level 3)
Let Im,n = sinm x cosn xdx.
R

(a) Derive the following reduction formulae for m + n 6= 0

(i)
sinm+1 x cosn−1 x n−1
Im,n = + Im,n−2
m+n m+n
 55

(ii)
sinm+1 x cosn+1 x m−1
Im,n = − + Im,n−2
m+n m+n
sin4 x cos4 xdx.
R
(b) Find

Solution. (a) (i) Using integration by parts,

Z
Im,n = sinm x cosn xdx
Z
1
= cosn−1 xd sinm+1 x
m+1
cosn−1 x sinm+1 x n−1
Z
= + sinm+2 x cosn−2 xdx
m+1 m+1
cosn−1 x sinm+1 x n−1
Z
= + sinm x(1 − cos2 x) cosn−2 xdx
m+1 m+1
cosn−1 x sinm+1 x n−1 n−1
= + Im,n−2 − Im,n
m+1 m+1 m+1
gives the result by a simple rearrangement.
(ii) Using integration by parts,
Z
Im,n = sinm x cosn xdx
Z
1
= − sinm−1 xd cosn+1 x
n+1
sinm−1 x cosn+1 x m − 1
Z
= − + sinm−2 x cosn+2 xdx
n+1 n+1
sinm−1 x cosn+1 x m − 1
Z
= − + sinm−2 x cosn x(1 − sin2 x)dx
n+1 n+1
sinm−1 x cosn+1 x m − 1 m−1
= − + Im−2,n − Im,n
n+1 n+1 n+1
gives the result by a simple rearrangement.

(b) Using the previous results, one obtains,

I4,4
sin5 x cos3 x 3
= + I4,2
8 8
5 3
sin3 x cos3 x 3
 
sin x cos x 3
= + − + I2,2
8 8 8 6
5 3 3 3
 3 
sin x cos x sin x cos x 3 sin x cos x 1
= − + + I2,0
8 16 16 4 4
5 3 3
sin x cos3 x sin x cos3 x 3 sin x cos x
 
3 sin x cos x 1
= − + + − + I0,0
8 16 64 64 2 2
sin5 x cos3 x sin3 x cos3 x 3 sin3 x cos x 3 sin x cos x 3
= − + − + x + C.
8 16 64 128 128
56
 57

8 Definite Integration
8.1 Mean value theorem for integrals

Exercise 1. (Level 2)
R1 R1
Given that 0 x21+1 dx = π
4 and 0
sin πxdx = 2
π. Show that there exist two points a and b in the closed interval
[0, 1] such that
Z 1
sin πx 2 π
= = sin πb.
0 x2 + 1 π(a2 + 1) 4

1
Solution. Both functions x2 +1 and sin πx are continuous and positive for over (0, 1). Applying the mean value
theorem twice, we get that
there exist two points a and b in the closed interval [0, 1] such that
Z 1 Z 1
sin πx 1 2
= 2 sin πxdx = ,
0 x2 + 1 a +1 0 π(a2 + 1)

and
Z 1 Z 1
sin πx 1 π
= sin πb dx = sin πb.
0 x2 + 1 0 x2 + 1 4

8.2 Fundamental theorem of Calculus

Exercise 2. (Level 2)
Evaluate the following integrals:

y2
R 20
(a) −4
y+5− 16 dy

R2 1
(b) 1
2y + 3 − 4(2x+3) dy

R2
(c) 0
|1 − y|dy
π
y
sec3 tan y2 dy
R
(d) 2
0 2

R 2017 1
(e) −2017
y 2017 (y 7102 + 7201)y 2017 sin2017 y ln |y 7102 + 2017|dy

Solution. (a)
20 20
y2
 2
y3
Z
y
y+5− dy = + 5y − = 144
−4 16 2 48 −4

(b)
Z 2  2
1 1 1 7
2y + 3 − dy = y 2 + 3y + ln |2x + 3| = 6 + ln | |
1 4(2x + 3) 8 1 8 5
 58

(c)
Z 2 Z 1 Z 2  1  2
1 1
|1 − y|dy = 1 − ydy + y − 1dy = y − y 2 + y 2 − y = 1
0 0 1 2 0 2 1

(d)
π π √
2 h 3 y i π2 2 8−2
Z Z
2
3 y y 2 y 2y
sec tan dy = 2 sec d sec = sec =
0 2 2 0 2 2 3 2 0 3

(e) Since the integrand is an odd function, the integral is 0.

Exercise 3. (Level 3)
Evaluate the following integrals:
√
3
1√
R
(a) 1
2
sin−1 x 1−x2
dx
2

R ln 2 √
(b) 0
ex − 1dx
R2
(c) 1
x2 ln xdx
R2
(d) 1
x2 ln2 xdx

Solution. (a)
√ √
3 3
d sin−1 x
Z Z
2 1 2
√ dx = dx
1
2
sin−1 x 1 − x2 1
2
sin−1 x
√
 3
ln sin−1 x 12

=
2

= ln 2

π
(b) Let sec2 y = ex , then dx = 2 tan ydy. When x = 0 or x = ln 2, y = 0 or y = 4 accordingly. We have that

π
Z ln 2 √
Z 4
q
ex − 1dx = tan2 y · 2 tan ydy
0 0
Z π
4
= 2 tan2 ydy
0
Z π
4
= 2 sec2 y − 1dy
0
π
= 2 [tan y − y]04
π
= 2−
2
 59

(c)

2 2
x3
Z Z
x2 ln xdx = ln xd
1 1 3
 3 2 Z 2 2
x x
= ln x − dx
3 1 1 3
 3 2
8 x
= ln 2 −
3 9 1
8 7
= ln 2 −
3 9

(d)

2 2
x3
Z Z
x2 ln2 xdx = ln2 xd
1 1 3
 3 2 Z 2 2
x x
= ln2 x − d ln2 x
3 1 1 3
Z 2 2
8 2 2x
= ln 2 − ln xdx
3 1 3
 
8 2 2 8 7
= ln 2 − ln 2 −
3 3 3 9
8 2 16 14
= ln 2 − ln 2 +
3 9 27

Exercise 4. (Level 3/Level 4)

Using some suitable substitution, show that
π
sinm x π
R
(a) 2
0 sinm x+cosm x dx = 4;

π
1+sin x
R
(b) 2
0
ln 1+cos x dx = 0;

π
1 π
R
(c) 2
0 1+tanλ x
dx = 4;

R1
(d) √1 dx = π
0 x+ 1−x2 4.

π
Solution. (a) By means of the substitution x = 2 − y,

π π
sinm x cosm x
Z 2
Z 2

m dx = m dx.
0 sin x + cosm x 0 sin x + cosm x
 60

Thus,
π
sinm x
Z 2

m dx
0 sin x + cosm x
Z π2 Z π2 !
1 sinm x cosm x
= dx + dx
2 0 sinm x + cosm x 0 sinm x + cosm x
Z π
1 2
= 1dx
2 0
π
= .
4

π
(b) Making use of the substitution x = 2 − y,

Z π Z π
2 2
ln(1 + sin x)dx = ln(1 + cos x)dx.
0 0

We then have
Z π Z π Z π
2 1 + sin x 2 2
ln dx = ln(1 + sin x)dx − ln(1 + cos x)dx = 0.
0 1 + cos x 0 0

π
(c) By substitution x = 2 − y, it suggests us that
π π π
tanλ x
Z Z Z
2 1 2 1 2
dx = dx = dx.
0 1 + tanλ x 0 1 + cotλ x 0 1 + tanλ x

We hence get
π π π
!
tanλ x
Z Z Z
2 1 1 2 1 2
dx = dx + dx
0 1 + tanλ x 2 0 1 + tanλ x 0 1 + tanλ x
Z π
1 2
= 1dx
2 0
π
= .
4

π
(d) Let x = sin z, then dx = cos zdz and when x = 0, z = 0; when x = 1, z = 2. One arrives that
π
Z 1 Z
1 2 sin x π
√ dx = dx = ,
0 x + 1 − x2 0 sin x + cos x 4

the last equality is justified by (a).

Exercise 5. (Level 2)
Find the derivative of the functions below.
R 2x
(a) a sin t ln(1 + t)dt
R 2x
(b) b
sin t ln(1 + x)dt
 61

Rc
(c) 2x
sin t ln(1 + t)dt
R 2x
(d) 2x
sin t ln(1 + x)dt

Solution. (a) The derivative is 2 sin(2x) ln(1 + 2x).

R 2x
sin tdt cos b−cos(2x)dt
(b) The derivative is 2 ln(1 + x) sin(2x) + b
1+x = 2 ln(1 + x) sin(2x) + 1+x .

(c) The derivative is −2 sin(2x) ln(1 + 2x).

(d) The derivative is 0.

Exercise 6. (Level 2/Level 3)

Define for − π2 < x < π2 ,
Z sec x √ 2
F (x) = ( 2)t dt.
tan x

0
Solve F (x) = 0.

R sec x √ t2 R sec x √ 2 R tan x √ t2

Solution. Write F (x) = tan x
( 2) dt = 0 ( 2)t dt − 0 ( 2) dt, then we have
√ 2 √ 2
F 0 (x) = ( 2)sec x sec x tan x − ( 2)tan x sec2 x.

Then one has

F 0 (x) = 0
√ 2 √ 2
( 2)sec x sec x tan x − ( 2)tan x sec2 x = 0
tan x √ 2 2
= ( 2)tan x−sec x
sec x
1
sin x = √
2
π
x = .
4

Exercise 7. (Level 3)
Evaluate the following limits

1
R x2 sin2 t

(a) lim x4 0 t2 ln(1 + t)dt
x→0


1
R x2 sin2 t

(b) lim x3 0 t2 ln(1 + x)dt
x→0
 
(c) lim Rx 1 − 3
x→0 x 0
sin t2 dt x4
 62

Solution. (a) Noting

sin2 x2
lim =1
x→0 x4
and
2x
ln(1 + x2 ) 2
lim = lim 1+x = 1,
x→0 x2 x→0 2x

one gets
2 2
x2
!
sin2 t ( sinx4x ln(1 + x2 ))(2x)
Z
1
lim ln(1 + t)dt = lim
x→0 x4 0 t2 x→0 4x3

1 sin2 x2 ln(1 + x2 )
= lim · lim
2 x→0 x4 x→0 x2
1
= .
2

(b) Noting
2 2
x2
!
sin2 t 2x( sinx4x )
Z
1
lim dt = lim =1
x→0 x2 0 t2 x→0 2x

and
ln(1 + x) 1
lim = lim = 1,
x→0 x x→0 1 + x

one gets

x2 x2
! !
sin2 t sin2 t
Z Z
1 1 ln(1 + x)
lim ln(1 + x)dt = lim dt · lim
x→0 x3 0 t2 x→0 x2 0 t2 x→0 x
= 1.

(c) Noting
x3 3x2
 
lim R x = lim =3
x→0
0
sin t2 dt x→0 sin x2

and
Rx
x3 − 3 0
sin t2 dt 3x2 − 3 sin x2
lim = lim
x→0 x7 x→0 7x6
6x − 6x cos x2
= lim
x→0 42x5
1 − cos x2
= lim
x→0 7x4
2x sin x2
= lim
x→0 28x3
1
= ,
14
 63

it gives
! Rx
1 3 x3 − 3 0 sin t2 dt
lim Rx
2
− 4 = lim Rx
x→0 x 0 sin t dt x x→0 x4 0 sin t2 dt
Rx
x3 − 3 0 sin t2 dt x3
= lim · lim R x
x→0 x7 x→0
0
sin t2 dt
3
= .
14

Exercise 8. (Level 3)
Rx Rx
Let f is a continuous function on R. Suppose lim ( 0 f (t)dt + f (x)) exists and that lim (ex 0 f (t)dt) = ∞.
x→+∞ x→+∞
Show that lim f (x) = 0.
x→+∞

Solution. By L’Hopital’s rule and the fundamental theorem of calculus, then we get

x
Rx
ex f (t)dt
Z
0
lim f (t)dt = lim
x→+∞ 0 x→+∞ ex
Rx
ex ( 0 f (t)dt + f (x))
= lim
x→+∞ ex
Z x
= lim ( f (t)dt + f (x)).
x→+∞ 0

Rx
Hence lim f (t)dt exists and lim f (x) = 0.
x→+∞ 0 x→+∞

Exercise 9. (Level 3/Level 4)

Rx 1
(a) For x ∈ (−1, 1), by considering 0 1−t
dt, show that

∞
X xn
− ln(1 − x) = .
n=1
n

(b) Can you write down a function whose Taylor’s series is

∞
X xn
?
n=1
n2

Solution. (a) Making use of geometric series, for x ∈ (−1, 1),

x ∞
xX ∞ Z x ∞ ∞
xn+1 xn
Z Z
1 X X X
dt = tn dt = tn dt = = .
0 1−t 0 n=0 n=0 0 n=0
n + 1 n=1 n
 64

Rx 1
On the other hand, we have 0 1−t
dt = − ln(1 − x). Hence

∞
X xn
− ln(1 − x) = .
n=1
n

(b) A direction calculation yields that

x ∞ n−1
xX ∞ Z x n−1 ∞
− ln(1 − t) xn
Z Z
t X t X
dt = dt = dt =
0 t 0 n=1
n n=1 0 n n=1
n2

for x ∈ (−1, 1).

Exercise 10. (Level 5)

π 1
(a) For 0 < x < 2, show that ln sec x < 2 sin x tan x.
Rx 8 ln sec x
(b) Define g(x) = 0
(1 + sec t)dt and ρ(x) = [g(x)]2 for x ∈ (0, π2 ).

(i) Show that g(x) > 0 for x ∈ (0, π2 ), and find g(x).
(ii) Show that ρ(x) is decreasing.
(Hint: Consider [g(x)]3 ρ0 (x) cot x.)
(iii) Find lim ρ(x).
x→0+

(iv) Find lim

π
ρ(x).
x→ 2 −

Rx f (x)
(c) Define f (x) = 0
(1 + sec t) ln sec tdt and σ(x) = g(x) ln sec x for x ∈ (0, π2 ).

(i) Find lim+ σ(x).

x→0

(ii) Find lim

π−
σ(x).
x→ 2

(iii) It is given the following fact: Let u, v : [0, c) → R be two continuous functions and continuously
u0
differentiable on (0, c), in which satisfied that u, v and v0 are positive and increasing. Furthermore if
u
u(0) = v(0) = 0, then v is increasing.
A. Show that 1 − csc x cot x ln sec x is increasing.
(1+sec
R x) ln sec x
B. Show that tan x 0x 1+sec tdt
is increasing.
(1+sec
R x) ln sec x .
C. Find lim+ tan x 0x 1+sec tdt
x→0

D. Show that σ(x) is increasing.

2
Solution. (a) Let h(x) = ln sec x− 12 sin x tan x. Differentiation gives h0 (x) = tan x 1 − cos x+sec x
= − tan x (coscos
x−1)


2 x <
π 1
0 for 0 < x < 2. Hence h(x) < h(0), that is, ln sec x < 2 sin x tan x.
Rx
(b) (i) Since the integrand 1 + sec t > 0 for t ∈ (0, π2 ), g(x) > 0 for x ∈ (0, π2 ). And g(x) = 0
1 + sec tdt =
x + ln(sec x + tan x).
 65

(ii) Note that, by the fundamental theorem of calculus, g 0 (x) = 1 + sec x. Then

[g(x)]2 tan x − 2g 0 (x)g(x) ln sec x

ρ0 (x) = 8
[g(x)]4
tan x
= 8 (g(x) − 2(csc x + cot x) ln sec x)
[g(x)]3

which gives
[g(x)]3 ρ0 (x) cot x = 8(g(x) − 2(csc x + cot x) ln sec x).
ln sec x tan x
Note that lim+ sin x = lim+ cos x = 0, and hence
x→0 x→0

lim ([g(x)]3 ρ0 (x) cot x) = lim+ 8(g(x) − 2(csc x + cot x) ln sec x) = 0.

x→0+ x→0

Consider [g(x)]3 ρ0 (x) cot x. Then, by (a),

d[g(x)]3 ρ0 (x) cot x

dx
= 8(1 + sec x − 2(csc x + cot x) tan x + 2 csc x(csc x + cot x) ln sec x)
1
< 8(1 + sec x − 2(csc x + cot x) tan x + 2 csc x(csc x + cot x) sin x tan x)
2
= 0.

One arrives that for x ∈ (0, π2 )

[g(x)]3 ρ0 (x) cot x < lim+ ([g(x)]3 ρ0 (x) cot x) = 0

x→0

which implies that ρ0 (x) < 0 provided that g(x) and cot x both are positive on the interval. Thus ρ(x)
is decreasing.
sin x cos x
(iii) We first compute lim g(x) = lim 1+sec x = 21 . Then,
x→0+ x→0+

8 ln sec x
lim ρ(x) = lim
x→0+ x→0+ [g(x)]2
4 tan x
= lim+
x→0 g(x)(1 + sec x)
 
4 sin x
= lim+
x→0 1 + cos x g(x)
= 1.

(iv) Using lim

π−
g(x) = lim
π−
(x + ln(sec x + tan x)) = ∞ we get
x→ 2 x→ 2

8 ln sec x 4 sin x
lim ρ(x) = lim = lim = 0.
x→ π
2
− π
x→ 2 − [g(x)]2 x→ π
2
− g(x)(1 + cos x)

Rx
(1+sec t) ln sec tdt
(c) Define σ(x) = 0
g(x) ln sec x for x ∈ (0, π2 ).
 66

(i) Note that, by the fundamental theorem of calculus,

Rx
0
(1 + sec t) ln sec tdt ln sec x
lim = lim
x→0+ [g(x)]3 x→0+ 3[g(x)]2
tan x
= lim+
x→0 6g(x)(1 + sec x)
sin x
= lim
x→0+ 6g(x)(1 + cos x)
cos x
= lim
x→0+ 6((1 + sec x)(1 + cos x) − g(x) sin x))
1
= ,
24

and
Rx
8 0
(1 + sec t) ln sec tdt
σ(x) =
ρ(x)[g(x)]3

then we get
Rx
8
(1 + sec t) ln sec tdt
0
lim+ σ(x) = lim+
x→0 x→0 ρ(x)[g(x)]3
Rx
1 8 0 (1 + sec t) ln sec tdt
= lim · lim
x→0+ ρ(x) x→0+ [g(x)]3
1
= .
3

π− 49π π 49π
(ii) When x → 2 , x would be viewed as any number greater than 100 , then ln sec t > 1 for 2 >t> 100 .
π
For 2 >t > 49π
100 ,

Z x
(1 + sec t) ln sec tdt
0
Z x Z cos−1 1
e
= (1 + sec t) ln sec tdt + (1 + sec t) ln sec tdt
1
cos−1 e 0
Z x
> (1 + sec t)dt
1
cos−1 e

1 1 1
= x + ln(sec x + tan x) − cos−1 + ln(sec(cos−1 ) + tan cos−1 ( ))
e e e
π−
→ ∞ as x → .
2
 67

ln sec x tan x
Using lim
π g(x) = lim
π 1+sec x = 1, we get
x→ 2 − x→ 2 −

Rx
0
(1 + sec t) ln sec tdt
lim σ(x) = lim
x→ π
2
− x→ π
2
− g(x) ln sec x
(1 + sec x) ln sec x
= lim
x→ π
2
− g(x) tan x + (1 + sec x) ln sec x
1
= lim g(x)
x→ π − tan x
2
ln sec x 1+sec x +1
1
= .
2

(iii) A. We need to show that the derivative is non-negative:

(1 − csc x cot x ln sec x)0

= csc x cot2 x ln sec x + csc3 x ln sec x − csc x cot x tan x
= csc x(cot2 x ln sec x + csc2 x ln sec x − 1).

To achieve that, we observe that for x > 0,

cot2 x ln sec x + csc2 x ln sec x − 1 is increasing,

as, by (a),

(cot2 x ln sec x + csc2 x ln sec x − 1)0

= ((csc2 −1) ln sec x + csc2 x ln sec x − 1)0
= (2 csc2 x ln sec x − ln sec x − 1)0
= −4 csc2 x cot x ln sec x + 2 csc2 x cot x − tan x
1
≥ −4 csc2 x cot x( sin x tan x) + 2 csc2 x cot x − tan x
2
−2 cos x + 2 − sin2 x
=
sin x cos x
(cos x − 1)2
=
sin x cos x
≥ 0,

ln sec x tan x 1
then, lim+ sin2 x
= lim+ 2 sin x cos x = 2 gives us that
x→0 x→0

cot2 x ln sec x + csc2 x ln sec x − 1

≥ lim (cot2 x ln sec x + csc2 x ln sec x − 1)
x→0+

= 0.

Hence
1 − csc x cot x ln sec x is increasing

because (1 − csc x cot x ln sec x)0 ≥ 0.

 68

B. Note that
(1 + sec x) ln sec x (cot x + csc x) ln sec x
Rx = Rx ,
tan x 0 1 + sec tdt 0
1 + sec tdt

((cot x + csc x) ln sec x)0

Rx
( 0 1 + sec tdt)0
(cot x + csc x) tan x − csc x(cot x + csc x) ln sec x
=
1 + sec x
= 1 − csc x cot x ln sec x
≥ lim (1 − csc x cot x ln sec x)
x→0+

1
= ,
2
ln sec x tan x
and lim+ sin x = lim+ cos x = 0. Since
x→0 x→0

Z x
lim+ ((cot x + csc x) ln sec x) = 0 and lim+ 1 + sec tdt = 0,
x→0 x→0 0

and by the previous result, 1 − csc x cot x ln sec x is positive and increasing, by the given fact we get
(1+sec
R x) ln sec x is increasing.
tan x 0x 1+sec tdt
Rx
1+sec tdt
C. lim+ 0 sin x = lim+ 1+sec
cos x
x
= 2 gives, by (b)(iii),
x→0 x→0

(1 + sec x) ln sec x
lim
Rx
tan x 0 1 + sec tdt
x→0+
Rx !
1 + sec tdt ρ(x)
= lim+ (1 + cos x) · 0 ·
x→0 sin x 8
1
= .
2

D. Note that
Z x Z x
lim (1 + sec t) ln sec tdt = 0 and lim (ln sec x 1 + sec tdt) = 0,
x→0+ 0 x→0+ 0

and
Rx
( 0 (1 + sec t) ln sec tdt)0
Rx
(ln sec x 0 1 + sec tdt)0
(1 + sec x) ln sec x
= Rx
(1 + sec x) ln sec x + tan x 0 1 + sec tdt
1
= tan x 0x 1+sec tdt
R
1+ (1+sec x) ln sec x
 69

which implies that it is increasing by the previous result. And

Rx
( 0 (1 + sec t) ln sec tdt)0 1
Rx = tan x 0x 1+sec tdt
R
(ln sec x 0 1 + sec tdt)0 1+ (1+sec x) ln sec x

1
≥ lim tan x 0x 1+sec tdt
R
x→0+ 1+ (1+sec x) ln sec x

1
=
3

which says that it is positive. Using the given fact, one obtains that σ(x) is increasing.

8.3 Reduction formulae

Exercise 11. (Level 4)

For any non-negative integer n, let
π
Z2
In = ex sin x cosn xdx.
−π
2

(a) Evaluate I0 and I1 .

(b) Show that for n ≥ 2

(n + 1)2 + 1 In = n(n − 1)In−2 .

 

(c) Show that when n is odd

n! π π
In = (e 2 − e− 2 )
[(n + 1)2 + 1] · · · [22 + 1]

and when n is even

n! π π
In = (e 2 + e− 2 ).
[(n + 1)2 2
+ 1] · · · [1 + 1]
 70

Solution. (a) We first note that

π
Z2
I0 = ex sin xdx
−π
2

π
Z2
= sin xdex
−π
2

π
Z2
π
= [ex sin x]− π −
2
ex d sin x
2
−π
2

π
Z2
π
−π
= e +e
2 2 − ex cos xdx
−π
2

π
Z2
π
−π
= e +e
2 2 − cos xdex
−π
2

π
Z2
π π π
= e 2 + e− 2 − [ex cos x]− π − 2
ex sin xdx
2
−π
2

π π
= e 2 + e− 2 − I0 .

It gives
π π
e 2 + e− 2
I0 = .
12 + 1
 71

By several times of integration by parts, we get

π
Z2
I1 = ex sin x cos xdx
−π
2

π
Z2
= sin x cos xdex
−π
2

π
Z2
π
= [ex sin x cos x]− π − 2
ex d(sin x cos x)
2
−π
2

π
Z2
= − ex (cos2 x − sin2 x)dx
−π
2

π
Z2
= − (cos2 x − sin2 x)dex
−π
2

π
Z2
π
= − ex (cos2 x − sin2 x) −2 π + ex d(cos2 x − sin2 x)

2
−π
2

π
Z2
π π
= e 2 − e− 2 − 4 ex sin x cos xdx
−π
2

π π
= e 2 − e− 2 − 4I1 .

Hence, one obtains that

π π
e 2 − e− 2
I0 = .
22 + 1
(Remark: The calculation of I1 will be much simpler, if double angle formulae are used.)

(b) Integration by parts, we get

π
Z2
In = ex sin x cosn xdx
−π
2

π
Z2
1
= − ex sin xd cosn+1 x
n+1
−π
2

π
Z2
1
= ex sin x cosn+1 xdx
n+1
−π
2

In+1
= .
n+1
 72

On the other hand, again, integration by parts tells us that

π
Z2
In = ex sin x cosn xdx
−π
2

π
Z2
= sin x cosn xdex
−π
2

π
Z2
= − ex d(sin x cosn x)
−π
2

π
Z2
= − ex (cosn+1 x − n sin2 x cosn−1 x)dx
−π
2

π
Z2
= − ex ((n + 1) cosn+1 x − n cosn−1 x)dx
−π
2

= −(n + 1)In+1 + nIn−1 .

Using the preceeding two equations, we have, for n ≥ 2,

In = −(n + 1)In+1 + nIn−1 = −(n + 1)2 In + n(n − 1)In−2 ,

that is,
(n + 1)2 + 1 In = n(n − 1)In−2 .
 

(c) By (b), we have, for n ≥ 2,

n(n − 1)
In = In−2
(n + 1)2 + 1
n(n − 1) (n − 2)(n − 3)
= · In−4
(n + 1)2 + 1 (n − 1)2 + 1
..
.
( π π
n!
[(n+1)2 +1]···[22 +1] (e
2 − e− 2 ) when n is odd
= n! π
−π
[(n+1)2 +1]···[12 +1] (e +e ) when n is even.
2 2

8.4 Improper integrals

Exercise 12. (Level 2/Level 3)

Discuss the convergence of the improper integrals below.
R∞ dx
(a) 0 (1+x)(2+3x)
 73

R∞ dx
(b) 2 x(ln x)k

R∞ 1−ln x
(c) 2 x2 dx

R1
(d) 0
3x2 ln xdx

1 −1 3
Solution. (a) Note that (1+x)(2+3x) = 1+x + 2+3x . Since

Z ∞ Z X
dx dx
= lim
0 (1 + x)(2 + 3x) X→∞ 0 (1 + x)(2 + 3x)
X
−1
Z
3
= lim + dx
X→∞ 0 1 + x 2 + 3x
 X
2 + 3x
= lim ln
X→∞ 1+x 0
 
2 + 3X
= lim ln − ln 2
X→∞ 1+X
3
= ln ,
2

the integral is convergent.

(b) We examine the convergence for the following three cases:

k = 1: Since
Z ∞ Z X
dx dx X
= lim = lim [ln ln x]2 = lim [ln ln X − ln ln 2] = ∞,
2 x ln x X→∞ 2 x ln x X→∞ X→∞

the integral is divergent.

k < 1: Since
Z ∞ Z X
dx d ln x 1 X
lim (ln x)1−k 2 = ∞,

= lim =
2 x(ln x)k X→∞ 2 (ln x) k 1 − k X→∞

the integral is divergent.

k > 1: Since
∞ X
(ln 2)1−k
Z Z
dx d ln x 1  1−k X

= lim = lim (ln x) = ,
2 x(ln x)k X→∞ 2 (ln x)k 1 − k X→∞ 2 k−1

the integral is convergent.

(c) Because

X X X
1 − ln x
Z Z Z
1 1
dx = dx + ln xd
2 x2 2 x2 2 x
Z X  X Z X
1 ln x 1
= dx + − dx
2 x2 x 2 2 x2
ln X ln 2
= −
X 2
 74

ln X 1
and by l’Hopital’s rule, lim = lim = 0, therefore,
X→∞ X X→∞ X

∞
1 − ln x
Z
ln 2
dx = − .
2 x2 2

The integral is convergent.

3x2 ln xdx:
R
(d) Apply integration by parts on

x3
Z Z Z
2 3 3
3x ln xdx = ln xdx = x ln x − x2 dx = x3 ln x − .
3

Consequently,
Z 1 Z 1
3x2 ln xdx = lim 3x2 ln xdx
0 ε→0+ ε
1
x3

3
= lim x ln x −
ε→0+ 3 ε
ε3
 
1 3
= lim − + ε ln ε −
ε→0+ 3 3
1
= − .
3

The last equality is justified by the limit (obtained by L’hopital’s rule),

ln ε 1
lim+ ε3 ln ε = lim+ −3
= lim+ = 0.
ε→0 ε→0 ε ε→0 −3ε−3

Exercise 13. (Level 3/level 4)

2 R 3 f 0 (x)
Let f (x) = (x+1) (x−1)
x3 (x−2) . Find −1 1+f 2 (x) dx.

(x+1)2 (x−1)
Solution. The integrand f (x) = x3 (x−2) is unbounded at x = 0 and x = 2. We first treat the following four
improper integrals:

0 1 2 3
f 0 (x) f 0 (x) f 0 (x) f 0 (x)
Z Z Z Z
dx, dx, dx and dx.
−1 1 + f 2 (x) 0 1 + f 2 (x) 1 1 + f 2 (x) 2 1 + f 2 (x)
 75

f 0 (x)
= tan−1 f (x) + C, we get
R
Using 1+f 2 (x) dx

0 a
f 0 (x) f 0 (x)
Z Z
dx = lim− dx
−1 1 + f 2 (x) a→0 2
−1 1 + f (x)
a
lim− tan−1 f (x) −1

=
a→0

= lim tan−1 f (a)

a→0−
π
= − ,
2
1 1
f 0 (x) f 0 (x)
Z Z
dx = lim+ dx
0 1 + f 2 (x) b→0 2
b 1 + f (x)
1
lim+ tan−1 f (x) b

=
b→0

= lim − tan−1 f (b)

b→0+
π
= − ,
2
2 c
f 0 (x) f 0 (x)
Z Z
dx = lim dx
1 1 + f 2 (x) c→2− 2
1 1 + f (x)
c
lim− tan−1 f (x) 1

=
c→2

= lim tan−1 f (c)

c→2−
π
= − , and
2
3 3
f 0 (x) f 0 (x)
Z Z
dx = lim+ dx
2 1 + f 2 (x) d→2 2
d 1 + f (x)
3
lim+ tan−1 f (x) d

=
d→2
 
32
= lim tan−1 − tan−1 f (d)
d→2+ 27
32 π
= tan−1 − .
27 2

Therefore, we arrive at

3
f 0 (x)
Z
dx
−1 1 + f 2 (x)
0 1 2 3
f 0 (x) f 0 (x) f 0 (x) f 0 (x)
Z Z Z Z
= dx + dx + dx + dx
−1 1 + f 2 (x) 0 1 + f 2 (x) 1 1 + f 2 (x) 2 1 + f 2 (x)
π π π 32 π
= − − − + tan−1 −
2 2 2 27 2
32
= tan−1 − 2π.
27
 76

8.5 Inequalities Involving Integrals

Exercise 14. (Level 2)

By comparing 1 + 2x, 1 + x and 1 + x2 , show that, for x ≥ 1

√ x
ln 1 + 2x ≤ ln(1 + x) ≤ ln(1 + )2 .
2

Solution. Observe that for t ≥ 0,

1 1 1
≤ ≤ .
1 + 2t 1+t 1 + 2t

Integrating from 0 to x, it gives

Z x Z x Z x
1 1 1
dt ≤ dt ≤ dt
0 1 + 2t 0 1+t 0 1 + 2t

which says that

√ x
ln 1 + 2x ≤ ln(1 + x) ≤ ln(1 + )2 .
2

Exercise 15. (Level 3)

Suppose that h is positive and continuous on [0, ∞) and that h(x) ≥ H(x), for x > 0, where
Z x
H(x) = 1 + h(t)dt.
0

Prove that, for any x > 0,

h(x) ≥ ex .

Solution. Since h is positive, H is positive. By the fundamental theorem of calculus, H 0 (t) = h(t) ≥ H(t), we
then integrate from 0 to x,
Z x 0 Z x
H (t)
dt ≥ dt.
0 H(t) 0

With H(0) = 1, we have ln H(x) ≥ x, i.e.

h(x) ≥ H(x) ≥ ex .

Exercise 16. (Level 3)

Let x be a real number. Express ln(1 + x2 ) in integral form and, considering this integral, show that, when n > 0,

nx2
< ln(1 + x2 )n < nx2 .
1 = x2
 77

R 1+x2 1
Solution. Observe that ln(1 + x2 ) = 1 t dt and for 1 ≤ t ≤ 1 + x2 ,

1 1
≤ ≤ 1.
1 + x2 t

Therefore,
1+x2 1+x2
nx2
Z Z
2 n 1 1
ln(1 + x ) = n dt ≥ n dt =
1 t 1 1 + x2 1 + x2
and
Z 1+x2 Z 1+x2
2 n 1
ln(1 + x ) = n dt ≤ n 1dt = nx2 .
1 t 1

Combining, we get
nx2
< ln(1 + x2 )n < nx2 ,
1 = x2
as desired.

Exercise 17. (Level 5)

(a) Show that, for any real numbers a and x,

(1 − |a|)2 ≤ 1 − 2a cos x + a2 ≤ (1 + |a|)2 .

(b) For any real number a such that |a| =

6 1, let
Z π
I(a) = ln(1 − 2a cos x + a2 )dx.
0

(i) By using (a), show that

π(1 − |a|)2 ≤ I(a) ≤ π(1 + |a|)2 ,

and deduce that lim I(a) = 0.

a→0

(ii) Using cos(2x) = cos2 x − sin2 x = 2 cos2 x − 1 = 1 − 2 sin2 x, show that

I(a) + I(−a) = I(a2 ) and I(a) = I(−a).

Hence show that

1 n
I(a) = n
I(a2 )
2
for all positive integer n.
(iii) Show that
1
I( ) = I(a) − 2π ln |a|.
a

(iv) Deduce from above results that

(
0 if |a| < 1
I(a) =
2π ln |a| if |a| > 1
 78

(c) Show that I(1) = I(−1) = 0.

Solution. (a) We begin with

−2|a| ≤ −2a cos x ≤ 2|a|,

then we get
(1 − |a|)2 ≤ 1 − 2a cos x + a2 ≤ (1 + |a|)2 .

(b) (i) Using the inequalities derived in (a),

Z π Z π Z π
(1 − |a|)2 dx ≤ ln(1 − 2a cos x + a2 )dx ≤ (1 + |a|)2 dx,
0 0 0

i.e.
π(1 − |a|)2 ≤ I(a) ≤ π(1 + |a|)2 .

As lim (1 ± |a|)2 = 0, by the sandwich theorem, we get lim I(a) = 0.

a→0 a→0

(ii) We first have that

I(a) + I(−a)
Z π Z π
= ln(1 − 2a cos x + a2 )dx + ln(1 + 2a cos x + a2 )dx
0 0
Z π
= ln(1 − 2a2 cos(2x) + a4 )dx
0
Z 2π
1
= ln(1 − 2a2 cos t + a4 )dt
2 0
Z π Z 2π
1 1
= ln(1 − 2a2 cos t + a4 )dt + ln(1 − 2a2 cos t + a4 )dt
2 0 2 π
Z 2π
1 1
= I(a2 ) − ln(1 − 2a2 cos(2π − t) + a4 )d(2π − t)
2 2 π
Z π
1 1
= I(a2 ) + ln(1 − 2a2 cos s + a4 )ds
2 2 0

1 1
= I(a2 ) + I(a2 )
2 2
= I(a2 )

and
Z π
I(−a) = ln(1 − 2a cos x + a2 )dx
0
Z π
= − ln(1 + 2a cos(π − x) + a2 )d(π − x)
0
Z π
= ln(1 + 2a cos t + a2 )dt
0

= I(a).
 79

Hence, inductively, we get

1 1 n
I(a) = I(a2 ) = · · · = n I(a2 ).
2 2
(iii)
Z π
1 2 1
I( ) = ln(1 − cos x + 2 )dx
a 0 a a
Z π Z π
= ln(1 − 2a cos x + a2 )dx − ln a2 dx
0 0

= I(a) − 2π ln |a|.
n
(iv) When |a| < 1, lim a2 = 0, by (ii) and (i)
n→∞

1 n
I(a) = lim n
I(a2 ) = 0.
n→∞ 2

When |a| > 1, we use the result in (iii), we then have

1
I(a) = I( ) + 2π ln |a| = 2π ln |a|.
a

The last equability follows from I( a1 ) = 0 as 1

|a| < 1.

(c) The derivations of I(a) + I(−a) = I(a2 ) and I(a) = I(−a) haven’t used the assumption |a| 6= 1, the results
remain true when a = ±1. Hence solving I(1) + I(−1) = I(1) and I(1) = I(−1), we have I(1) = I(−1) = 0,
as desired.

Exercise 18. (Level 5)

(a) Suppose f (x), g(x) are continuously differentiable functions such that f 0 (x) ≥ 0 for a ≤ x ≤ b.
Rx
(i) Let w(x) = a
g(t)dt. Show that
Z b Z b
f (x)g(x)dx = f (b)w(b) − f 0 (x)w(x)dx.
a a

(ii) Using the mean value theorem for integral, show that
Z b Z b Z c
f (x)g(x)dx = f (b) g(x)dx + f (a) g(x)dx
a c a

for c ∈ [a, b].

(b) Let F (x) be a function with a continuous second derivative such that F 00 (x) ≥ 0 and F 0 (x) ≥ m > 0 for
a ≤ x ≤ b. Using (a) with f (x) = − F 01(x) and g(x) = −F 0 (x) cos F (x). Show that
Z 
 b  4
cos F (x)dx ≤ .
 
 m

 a

(Remark: This lemma plays an important role in the theory of exponential sums.)
 80

(c) (i) Show that

Z 1 Z 1
n+1
cos x dx ≤ cos xn dx.
0 0
R1
Hence show that lim cos xn dx exists.
n→∞ 0
R 2π
(ii) Using (b), or otherwise, show that lim cos xn dx exists.
n→∞ 0

Solution. (a) (i) Note that w(a) = 0, we get

Z b Z b Z b
f (x)g(x)dx = f (x)dw(x) = f (b)w(b) − f 0 (x)w(x)dx.
a a a

(ii) By the mean value theorem for integral, there is a c ∈ [a, b] such that
Z b Z b
f 0 (x)w(x)dx = w(c) f 0 (x)dx = f (b)w(c) − f (a)w(c).
a a

Plugging in the equation of (i), one obtains

Z b Z b
f (x)g(x)dx = f (b)w(b) − f 0 (x)w(x)dx
a a
Z b
= f (b)w(b) − w(c) f 0 (x)dx
a

= f (b)w(b) − f (b)w(c) + f (a)w(c)

Z b Z c Z c
= f (b) g(x)dx − f (b) g(x)dx + f (a) g(x)dx
a a a
Z b Z c
= f (b) g(x)dx + f (a) g(x)dx.
c a

(b) Note that

1 1
|f (a)| = | |≤ ,
F 0 (a) m
1
|f (b)| ≤ ,
m
Z b Z b Z b
| g(x)dx| = | F 0 (x) cos F (x)dx| = | cos F (x)dF (x)| = | sin F (b) − sin F (c)| ≤ 2,
c c c

and
Z c
| g(x)dx| ≤ 2.
a
 81

F 00 (x)
F 00 (x) ≥ 0 ensures that f 0 (x) = (F 0 (x))2 ≥ 0. We are ready to use (a) and the estimations above. Then
Z 
 b  Z b Z c
cos F (x)dx = |f (b) g(x)dx + f (a) g(x)dx|
 

 a  c a
Z b Z c
≤ |f (b) g(x)dx| + |f (a) g(x)dx|
c a

2 2
≤ +
m m
4
≤ .
m

(c) (i) For x ∈ [0, 1], xn ≥ xn+1 , then cos xn ≤ cos xn+1 ≤ 1 so, by integrating from 0 to 1,
Z 1 Z 1
n
cos x dx ≤ cos xn+1 dx ≤ 1.
0 0

R1
Hence the sequence { 0 cos xn dx} is increasing and bounded by 1, by monotone convergence theorem,
R1
lim 0 cos xn dx exists.
n→∞

(ii) For 1 ≤ x ≤ 2π, let F (x) = xn . Then F 0 (x) = nxn−1 ≥ n and F 00 (x) = n(n−1)xn−2 ≥ 0 for 1 ≤ x ≤ 2π.
We are in position to use the result of (b), we get
Z 2π
4
| cos xn dx| ≤ .
1 n

By the sandwich theorem tells us that

Z 2π
lim cos xn dx = 0.
n→∞ 1

R 2π R1 R 2π R1
Hence, by (c)(i), lim cos xn dx = lim cos xn dx + lim cos xn dx = lim cos xn dx exists.
n→∞ 0 n→∞ 0 n→∞ 1 n→∞ 0