THE CHINESE UNIVERSITY OF HONG KONG

Department of Mathematics
MATH1010 University Mathematics
Taylor’s theorem and L’Hôpital’s rule

1 Taylor’s polynomial, Taylor’s series and Taylor’s theorem

Exercise 1. (Level 2)
Find the Taylor polynomials of the given orders for the following functions at x = 0.

(a) ln(1 + x), order = 4

ex +e−x
(b) cosh x = 2 , order = 6

ex −e−x
(c) sinh x = 2 , order = 5

x2 x3 x4
Solution. (a) x − + −
2 3 4

x2 x4 x6
(b) 1 + + +
2 24 720

x3 x5
(c) x + +
6 120

Exercise 2. (Level 2)
Find the Taylor Series for the following functions at the given points.
1
(a) 1+x at x = 0

π
(b) sin x at x = .
2
(c) ex at x = 1.

∞
X
Solution. (a) (−1)n xn = 1 − x + x2 − x3 + · · ·
n=0

∞
X (−1)2n π 1 π 1 π
(b) (x − )2n = 1 − (x − )2 + (x − )4 − · · ·
n=0
n! 2 2! 2 4! 2

∞
X e e
(c) (x − 1)n = e + e(x − 1) + (x − 1)2 + · · ·
n=0
n! 2!

Exercise 3. (Level 2)

(a) Write down the Taylor polynomial P3 (x) of degree 3 generated by f (x) = ln(1 − x) at 0
 2

(b) Hence approximate ln 0.99.

x2 x3
Solution. (a) P3 (x) = −x − −
2 3
(b) ln 0.99 = f (0.01) ≈ P3 (0.01) = −0.01005033

Exercise 4. (Level 2)
Suppose that f : R → R is a function that can be differentiated infinitely many times and the Taylor series
expansion of f at x = 0 is
a0 + a1 x + a2 x2 + a3 x3 + · · · .

(a) If f is an odd function, show that a0 = a2 = a4 = · · · = 0.

(b) If f is an even function, show that a1 = a3 = a5 = · · · = 0.

Solution. (a) Firstly, since f is odd, −f (0) = f (−0). Therefore, f (0) = 0. Then we have,

−f (x) = f (−x)
0
−f (x) = −f 0 (−x)
f 0 (x) = f 0 (−x)
f 00 (x) = −f 00 (−x)
−f 00 (x) = f 00 (−x)

As we can see, f 0 is an even function and f 00 is an odd function again. By repeating the above, it can be
shown that f (2n) is an odd function and so f (2n) (0) = 0 for all positive integers n. The result follows as
f (2n) (0)
a2n = = 0.
(2n)!

(b) Similar to (a), what we have to show is f (2n−1) (0) = 0 for all positive integers n.

Exercise 5. (Level 2)
Find the Taylor series generated by the given functions at specified points:

(a) f (x) = sin x at x = π;

(b) f (x) = e2x at x = 1;

(c) f (x) = x3 + 2x2 − 4x + 5 at x = −1.

∞
X 1
Solution. (a) sin x = (−1)n+1 (x − π)2n+1 .
n=0
(2n + 1)!

∞
X 2 n e2
(b) e2x = (x − 1)n .
n=0
n!
 3

(c) x3 + 2x2 − 4x + 5 = 10 − 5(x + 1) − (x + 1)2 + (x + 1)3 .

(Remark: The above expression can also be obtained by performing long division.)

Exercise 6. (Level 2)
Refer to the respective pairs of functions and numbers below (separated by the semi-colon). Denote the function
concerned by f , and the number by c. Find the Taylor series Tc,f (x) of the function f about the point c, giving
the form of its general term where appropriate.

(a) i. ex ; 0. vi. eπx ; 0. xi. cosh(x); 0.

x −x xii. sin(πx); 0.
ii. e ; 1. vii. e ; 0.
iii. ex ; 2. viii. sin(x); 0. xiii. cos(π 2 x); π −1 .
iv. ex ; −1. ix. cos(x); 0. xiv. sinh(ex + π); 0.
v. e2x ; 0. x. sinh(x); 0. xv. cosh(x); 1.

(b) In this part the only difficulty is to perform differentiation cleverly.

i. 4x100 + 3x10 + 2x + 1; 0. 1 xii. ln(1 + x); 0.

vii. ; 1.
2−x
ii. 4x3 + 3x2 + 2x + 1; 1.
1 xiii. ln(1 + 2x); 0.
1 viii. x2 + 2x + 3 + ; 0.
iii. ; 0. 1−x
1+x xiv. ln(2 + x); 0.
x
1 ix. ; 0.
iv. ; 0. 1 − 2x xv. ln(x); 1.
1−x
1 x2
v. ; 1. x. ; 0. xvi. ln(x); e.
x 1 − 3x
1 2 1 1+x
vi. ; 0. xi. ; 0. xvii. ln( ); 0.
2−x 1 − x2 2 1−x

∞
X 1
Solution. (a) i. The Taylor series of the function ex about the point 0 is given by xk .
k!
k=0
∞
X e
ii. The Taylor series of the function ex about the point 1 is given by (x − 1)k .
k!
k=0
∞ 2
X e
iii. The Taylor series of the function ex about the point 2 is given by (x − 2)k .
k!
k=0
∞ −1
X e
iv. The Taylor series of the function ex about the point −1 is given by (x + 1)k .
k!
k=0
∞ k
X 2
v. The Taylor series of the function e2x about the point 0 is given by xk .
k!
k=0
∞
X πk
vi. The Taylor series of the function eπx about the point 0 is given by xk .
k!
k=0
 4

∞
X (−1)k
vii. The Taylor series of the function e−x about the point 0 is given by xk .
k!
k=0
∞
X (−1)j 2j+1
viii. The Taylor series of the function sin(x) about the point 0 is given by x .
j=0
(2j + 1)!
∞
X (−1)j
ix. The Taylor series of the function cos(x) about the point 0 is given by x2j .
j=0
(2j)!
∞
X 1
x. The Taylor series of the function sinh(x) about the point 0 is given by x2j+1 .
j=0
(2j + 1)!
∞
X 1 2j
xi. The Taylor series of the function cosh(x) about the point 0 is given by x .
j=0
(2j)!

xii. The Taylor series of the function sin(πx) about the point 0 is given by

∞
X (−1)j π 2j+1
x2j+1 .
j=0
(2j + 1)!

∞
X (−1)j+1 π 4j
xiii. The Taylor series of the function cos(π 2 x) about the point π −1 is given by (x − π −1 )2j .
j=0
(2j)!
∞
X 1 ek [eπ + (−1)k+1 e−π ] k
xiv. The Taylor series of the function sinh(ex+π) about the point 0 is given by · x .
k! 2
k=0

xv. The Taylor series of the function cosh(x) about the point 1 is given by

∞
X e + (−1)k e−1
(x − 1)k .
2(k!)
k=0

(b) i. The Taylor series of the function 4x100 + 3x10 + 2x + 1 about the point 0 is given by 1 + 2x + 3x10 + 4x100 .
ii. The Taylor series of the function 4x3 + 3x2 + 2x + 1 about the point 1 is given by 10 + 20(x − 1) + 15(x −
1)2 + 4(x − 1)3 .
∞
1 X
iii. The Taylor series of the function about the point 0 is given by (−1)k xk .
1+x
k=0
∞
1 X
iv. The Taylor series of the function about the point 0 is given by xk .
1−x
k=0
∞
1 X
v. The Taylor series of the function about the point 1 is given by (−1)k (x − 1)k .
x
k=0
∞
1 X 1 k
vi. The Taylor series of the function about the point 0 is given by x .
2−x 2k+1
k=0
∞
1 X
vii. The Taylor series of the function about the point 1 is given by (x − 1)k .
2−x
k=0
 5

∞
1 X
viii. The Taylor series of the function x2 + 2x + 3 + about the point 0 is given by 4 + 3x + 2x2 + xk .
1−x
k=3
∞
x X
ix. The Taylor series of the function about the point 0 is given by 2k−1 xk .
1 − 2x
k=1
∞
x2 X
x. The Taylor series of the function about the point 0 is given by 3k−2 xk .
1 − 3x
k=2
∞
2 X
xi. The Taylor series of the function about the point 0 is given by 2x2j .
1 − x2 j=0

∞
X (−1)k−1
xii. The Taylor series of the function ln(1 + x) about the point 0 is given by xk .
k
k=1
∞
X 2k (−1)k−1
xiii. The Taylor series of the function ln(1 + 2x) about the point 0 is given by xk .
k
k=1
∞
X (−1)k−1
xiv. The Taylor series of the function ln(2 + x) about the point 0 is given by ln(2) + xk .
2k k
k=1
∞
X (−1)k−1
xv. The Taylor series of the function ln(x) about the point 1 is given by (x − 1)k .
k
k=1
∞
X (−1)k−1
xvi. The Taylor series of the function ln(x) about the point e is given by 1 + (x − e)k .
ek k
k=1
∞
1 1+x X 1
xvii. The Taylor series of the function ln( ) about the point 0 is given by x2j+1 .
2 1−x j=0
(2j + 1)

Exercise 7. (Level 2/Level 3)

π
Find the Taylor polynomial of ecos x with order (degree) n = 2 centered at x = 2.

Solution.
π
f ( ) = 1.
2
 
f 0 (x) = sin(x) −ecos(x) .

π
f 0 ( ) = −1.
2
f 00 (x) = sin2 (x)ecos(x) − ecos(x) cos(x).
π
f 00 ( ) = 1.
2
π
The Taylor polynomial of f (x) at x = 2 is

π 1 π
1 − (x − ) + (x − )2 .
2 2 2
 6

Exercise 8. (Level 2/Level 3)

Let f (x) is a polynomial of degree n > 0 and let a ∈ R.

(a) If Pn (x) is the Taylor polynomial of degree n generated by f (x) at x = a, show that f (x) = Pn (x).

(b) Suppose that f (a) = f 0 (a) = · · · = f (r−1) (a) = 0 and f (r) (a) 6= 0, where 1 ≤ r ≤ n.
Prove that (x − a) is a factor of f (x) with multiplicity r, i.e. f (x) = (x − a)r g(x) for some polynomial g(x)
such that g(x) is not divisible by x − a.

Solution. (a) Let x ∈ R. By Taylor’s theorem, there exists c that lies on the open interval between x and a such
that
f n+1 (c)
f (x) − Pn (x) = Rn (x) = (x − a)n+1 = 0.
(n + 1)!

We have the last equality since f is only a polynomial of degree n.

(b) By (a), we have

f (x) = Pn (x)
f (r−1) (a)
= f (a) + f 0 (a)(x − a) + · · · + (x − a)r−1 +
(r − 1)!
f (r) (a) f (n) (a)
(x − a)r + · · · + (x − a)n
r! n!
f (r) (a) f (n) (a)
= (x − a)r + · · · + (x − a)n (By assumption)
r! n!
 (r)
f (n) (a)

f (a)
= (x − a)r + ··· + (x − a)n−r
r! n!
= (x − a)r g(x)

f (r) (a) f (n) (a)

where g(x) = + ··· + (x − a)n−r which is a polynomial.
r! n!
f (r) (a)
Note that g(a) = 6= 0, therefore g(x) is not divisible by x − a.
r

Exercise 9. (Level 2/Level 3)

1
Find the Taylor polynomial of x2 with order (degree) n, centered at x = −1.

Solution.
d −2
x = (−2)x−3 = −2!x−3 .
dx
d2 −2
x = (−2)(−3)x−4 = 3!x−4 .
dx2
d3 −3
x = (−2)(−3)(−4)x−5 = −4!x−5 .
dx3
 7

Generally
dn −2
x = (−1)n (n + 1)!x−n−2 .
dxn
f (n) (−1) = (n + 1)!
The Taylor Series of degree n is

n k
X (k + 1)! X
(x + 1)k = (k + 1)(x + 1)k .
k!
k=0 k=0

Exercise 10. (Level 2/Level 3)

ex +e−x
Find the Taylor polynomial of cosh x = 2 with order (degree) n centered at x = 0.

Solution. Let f (x) = cosh x. Then

f 0 (x) = sinh(x),

f 00 (x) = cosh(x),

f 000 (x) = sinh(x).

Generally
(
cosh(x) if n is even
f (n) (x) =
sinh(x) if n is odd.

So (
(n) 1 if n is even
f (0) =
0 if n is odd.

Hence The Taylor polynomial is

X xm
pn (x) = .
m!
m≥0,m even

Exercise 11. (Level 2/Level 3)

Suppose pn (x) is the Taylor series of f (x) with degree n centered at x = c. Show that pn (x + c) is the Taylor
series of g(x) = f (x + c) with degree n centered at x = 0.

Solution.
n
X f (k) (c)
pn (x) = (x − c)k .
k!
k=0

Let g(x) = f (x + c). Then g (k) (x) = f (k) (x + c). Hence g (k) (0) = f (k) (c). The Taylor series of g(x) with degree
n centered at x = 0 is
n n
X g (k) (0) k X f (k) (c) k
x = x = pn (x + c).
k! k!
k=0 k=0
 8

Exercise 12. (Level 2/Level 3)

1 1 √ 1
Show that 1 + x − x2 ≤ 1 + x ≤ 1 + x for x > 0.
2 8 2
(Hint: Use Taylor’s theorem.)

√
Solution. Let f (x) = 1 + x and let P1 (x) and P2 (x) be the Taylor polynomials of degree 1 and 2 generated by
f at the point x = 0. We have

f (0) = 1
1 1
f 0 (x) = √ ⇒ f 0 (0) =
2 1+x 2
−1 1
f 00 (x) = 3/2
⇒ f 00 (0) = −
4(1 + x) 4
3
f 000 (x) =
8(1 + x)5/2

1 1 1 1
Therefore, P1 (x) = f (0) + f 0 (0)x = 1 + x and P2 (x) = f (0) + f 0 (0)x + f 00 (0) = 1 + x − x2 .
2 2! 2 8
Let x > 0. By Taylor’s theorem, there exist b and c with 0 < b, c < x such that

√ 1 1 1
1 + x = f (x) = P1 (x) + R1 (x) = 1 + x − x2 ≤ 1 + x
2 4(1 + b)3/2 2

and
√ 1 1 3 1 1
1 + x = f (x) = P2 (x) + R2 (x) = 1 + x − x2 + x3 ≥ 1 + x − x2 .
2 8 8(1 + c)5/2 2 8
√
Therefore, 1 + 21 x − 18 x2 ≤ 1 + x ≤ 1 + 12 x for x > 0.

Exercise 13. (Level 2/Level 3)

Evaluate the limits below by considering appropriate Taylor series:

2x − sin(x) − sinh(x)
(a) lim
x→0 4 sin(x) sin(x2 )

24 cos(x) − 24 + 12 sin2 (x) − sinh2 (x2 )

(b) lim
x→0 sin2 (x) sinh(x2 )

ln(1 + x) sin(x3 )
(c) lim √
x→0 cos(x) − 1 − x2

tan(x) − x
(d) lim+
x→0 x2 tan(x)

∞
X
Solution. Whenever f (x) = an (x − c)n at and near c, we agree to write ‘f (x) = a0 + a1 (x − c) + · · · + aN (x −
n=0
N
c) + HOTN +1 at and near c’. ‘HOTN +1 ’ is the short-hand for the phrase ‘terms of order higher than or equal
to (x − c)N +1 ’.
 9

(a) At and near 0, we have

∞
X x2n+1 x3 x5
sinh(x) = =x+ + + HOT7 ,
n=0
(2n + 1)! 6 120
∞
X (−1)n x2n+1 x3 x5
sin(x) = =x− + + HOT7 ,
n=0
(2n + 1)! 6 120
∞
X (−1)n x4n+2 x6
sin(x2 ) = = x2 − + HOT10 .
n=0
(2n + 1)! 6

When x is near 0,

2x − sin(x) − sinh(x)
4 sin(x) sin(x2 )
2x − (x − x3 /6 + x5 /120 + HOT7 ) − (x + x3 /6 + x5 /120 + HOT7 )
=
4(x − x3 /6 + x5 /120 + HOT7 )(x2 − x6 /6 + HOT10 )
−x5 /60 + HOT7
=
4(x + HOT3 )(x2 + HOT6 )
−x2 /60 + HOT4
= −→ 0 as x −→ 0.
4(1 + HOT2 )(1 + HOT4 )

2x − sin(x) − sinh(x)
Then lim = 0.
x→0 4 sin(x) sin(x2 )

(b) At and near 0, we have

∞
X x2n+1 x3 x5
sinh(x) = =x+ + + HOT7 ,
n=0
(2n + 1)! 6 120
∞
X x4n+2 x6
sinh(x2 ) = = x2 + + HOT10 ,
n=0
(2n + 1)! 6
∞
X (−1)n x2n+1 x3 x5
sin(x) = =x− + + HOT7 ,
n=0
(2n + 1)! 6 120
∞
X (−1)n x2n x2 x4 x6
cos(x) = =1− + − + HOT8 .
n=0
(2n)! 2 24 720
 10

When x is near 0,

24 cos(x) − 24 + 12 sin2 (x) − sinh2 (x2 )

sin2 (x) sinh(x2 )
x2 x4 x6 3 5
x6
24(1 − 2 + 24 − 720+ HOT8 ) − 24 + 12(x − x6 + 120 x
+ HOT7 )2 − (x2 + 6 + HOT10 )2
=
(x − x3 /6 + x5 /120 + HOT7 )2 (x2 + x6 /6 + HOT10 )
x2 x4 x6 4 6
24(1 − 2 + 24 − 720 + HOT8 ) − 24 + 12(x2 − x3 + 16x 4
360 + HOT7 ) − (x + HOT8 )
=
(x2 + HOT4 )(x2 + HOT6 )
−4x4 + HOT6
=
(x2 + HOT4 )(x2 + HOT6 )
−4 + HOT2
= −→ −4 as x −→ 0.
(1 + HOT2 )(1 + HOT4 )

24 cos(x) − 24 + 12 sin2 (x) − sinh2 (x2 )

Then lim = −4.
x→0 sin2 (x) sinh(x2 )

(c) At and near 0, we have

∞
X (−1)n x2n+1 x3
sin(x) = =x− + HOT5 ,
n=0
(2n + 1)! 6
∞
X (−1)n x6n+3 x9
sin(x3 ) = = x3 − + HOT15 ,
n=0
(2n + 1)! 6
∞
X (−1)n xn+1 x2 x3
ln(1 + x) = =x− + + HOT4 ,
n=0
n+1 2 3
∞
X (−1)n x2n x2 x4
cos(x) = =1− + + HOT6 ,
n=0
(2n)! 2 24
∞
p X (1/2)(1/2 − 1) · ... · (1/2 − n + 1) 2n
1 − x2 = (−1)n · x
n=0
n!

x2 x4
= 1− − + HOT6 .
2 8

When x is near 0,

ln(1 + x) sin(x3 ) (x − x2 /2 + x3 /3 + HOT4 )(x3 − x9 /6 + HOT15 )

√ =
cos(x) − 1 − x2 (1 − x2 /2 + x4 /24 + HOT6 ) − (1 − x2 /2 − x4 /8 + HOT6 )
x4 − x5 /2 + x6 /3 + HOT7
=
x4 /6 + HOT6
1 − x/2 + x2 /3 + HOT3
= −→ 6 as x −→ 0.
1/6 + HOT2

ln(1 + x) sin(x3 )
Then lim √ = 6.
x→0 cos(x) − 1 − x2
 11

(d) At and near 0, we have

∞
X (−1)n x2n+1 x3 x5
sin(x) = =x− + + HOT7 ,
n=0
(2n + 1)! 6 120
∞
X (−1)n x2n x2 x4
cos(x) = =1− + + HOT6 .
n=0
(2n)! 2 24

When x is near 0,

tan(x) − x sin(x) − x cos(x)

=
x2 tan(x) x2 sin(x)
(x − x3 /6 + x5 /120 + HOT7 ) − x(1 − x2 /2 + x4 /24 + HOT6 )
=
x2 (x − x3 /6 + x5 /120 + HOT7 )
(x − x3 /6 + x5 /120 + HOT7 ) − (x − x3 /2 + x5 /24 + HOT7 )
=
x3 (1 − x2 /6 + x4 /120 + HOT6 )
x3 /3 + HOT5
=
x3 (1 − x2 /6
+ x4 /120 + HOT6 )
x3 /3 + HOT5
=
x3 (1 + HOT2 )
1/3 + HOT2 1
= −→ as x −→ 0+ .
1 + HOT2 3

tan(x) − x 1
Then lim = .
x→0+ x2 tan(x) 3

Exercise 14. (Level 3)

√
(a) Find the Taylor polynomial P2 (x) of degree 2 generated by the function 3 1 + x.
√
(b) Hence, approximate 3 1.3 and show that the error of your approximation is less that 0.2 × 10−2 .

f 00 (0) 2 x x2
Solution. (a) P2 (x) = f (0) + f 0 (0)x + x =1+ −
2! 3 9
√
3
(b) We can approximate 1.3 by P2 (0.3) = 1.09. The error of this approximation is E2 (0.3) which can be
estimated by

f 000 (c)
E2 (0.3) = (0.3)3 for some c ∈ (0, 0.3)
3!
 3
5 3
= (1 + c)−8/3
81 10
 3
5 3
<
81 10
1
=
600
< 2 × 10−3
 12

Exercise 15. (Level 3)

Evaluate the limits below with the help of Taylor series:
√ √
sin(3x) 3
2+x− 32−x
(a) lim (g) lim √ √
x→0+ sin(2x) x→0+
3
3+x− 33−x

x2 − 3x + 2 arcsin(x)
(b) lim− (h) lim
x→2 x2 − 5x + 6 x→0 arcsinh(x)

(ex − 1) sin(x) ex − 1 − sin(x)

(c) lim+ . (i) lim
x→0 1 − cos(x) x→0 x2

(ex − e−x )2 x2 + 2x + 2 − 2ex

(d) lim− (j) lim
x→0 1 − cos(2x) x→0 x3

sin(x) sin(x) − x + x3 /6 + x5 /60

(e) lim (k) lim
x→0 sinh(x) x→0 sinh(x) − x − x3 /6 − x5 /60

√ √
2+x− 2−x sin(x2 ) − x2
(f) lim+ √ √ (l) lim
x→0 3+x− 3−x x→0 sinh(x2 ) − x2

∞
X
Solution. Whenever f (x) = an (x − c)n at and near c, we agree to write ‘f (x) = a0 + a1 (x − c) + · · · + aN (x −
n=0

c)N + HOTN +1 at and near c’. ‘HOTN +1 ’ is the short-hand for the phrase ‘terms of order higher than or equal
to (x − c)N +1 ’.

(a) At and near 0, we have

∞
X (−1)n · 32n+1 2n+1 3 3 x3
sin(3x) = x = 3x − + HOT5 ,
n=0
(2n + 1)! 3!
∞
X (−1)n · 22n+1 2n+1 2 3 x3
sin(2x) = x = 2x − + HOT5 .
n=0
(2n + 1)! 3!

When x is ‘slightly’ greater than 0,

sin(3x) 3x − 33 x3 /3! + +HOT5 3 − 33 x2 /3! + +HOT4 3

= = −→ as x −→ 0+ .
sin(2x) 2x − 23 x3 /3! + HOT5 2 − 23 x2 /3! + +HOT4 2

sin(3x) 3
Therefore lim = .
x→0+ sin(2x) 2

(b) For any x ∈ R, we have x2 − 3x + 2 = (x − 2) + (x − 2)2 and x2 − 5x + 6 = −(x − 2) + (x − 2)2 .

When x is ‘slightly’ less than 2,

x2 − 3x + 2 (x − 2) + (x − 2)2 1 + (x − 2)
2
= 2
= −→ −1 as x −→ 2− .
x − 5x + 6 −(x − 2) + (x − 2) −1 + (x − 2)2

x2 − 3x + 2
Therefore lim− = −1.
x→2 x2 − 5x + 6
 13

(c) At and near 0, we have

∞
X (−1)n x2n x2
1 − cos(x) = 1− = + HOT4 ,
n=0
(2n)! 2
∞
X xn
ex − 1 = − 1 = x + HOT2 ,
n=0
n!
∞
X (−1)n x2n+1
sin(x) = = x + HOT3 .
n=0
(2n + 1)!

When x is ‘slightly’ greater than 0,

(ex − 1) sin(x) (x + HOT2 )(x + HOT3 )

=
1 − cos(x) x2 /2 + HOT4
x2 + HOT3
=
x2 /2 + HOT3
1 + HOT1
= −→ 2 as x −→ 0+ .
1/2 + HOT1

(ex − 1) sin(x)
Then lim+ = 2.
x→0 1 − cos(x)

(d) At and near 0, we have

∞
X (−1)n x2n x2
1 − cos(x) = 1− = + HOT4 ,
n=0
(2n)! 2
∞
X x2n+1
ex − e−x = 2 sinh(x) = 2 · = 2x + HOT3 ,
n=0
(2n + 1)!

(ex − e−x )2 = (2x + HOT3 )(2x + HOT3 ) = 4x2 + HOT4 .

When x is ‘slightly’ less than 0,

(ex − e−x )2 4x2 + HOT4

= 2 −→ 8 as x −→ 0+ .
1 − cos(x) x /2 + HOT4

(ex − e−x )2
Then lim = 8.
x→0− 1 − cos(2x)

(e) At and near 0, we have

∞
X x2n+1
sinh(x) = = x + HOT3 ,
n=0
(2n + 1)!
∞
X (−1)n x2n+1
sin(x) = = x + HOT3 .
n=0
(2n + 1)!

When x is near 0,
sin(x) x + HOT3 1 + HOT2
= = −→ 1 as x −→ 0.
sinh(x) x + HOT3 1 + HOT2
 14

sin(x)
Then lim = 1.
x→0 sinh(x)

(f) At and near 0, we have

√ √ √ p p
2+x− 2−x = 2 · ( 1 + x/2 − 1 − x/2)
√

1 x (1/2)(1/2 − 1) x 2
= 2 · [1 + · + · ( ) + HOT3 ]
2 2 2 2

1 x (1/2)(1/2 − 1) x 2
− [1 − · + · ( ) + HOT3 ]
2 2 2 2
x
= √ + HOT3 ,
2
√ √ √ p p
3+x− 3−x = 3 · ( 1 + x/3 − 1 − x/3)
√

1 x (1/2)(1/2 − 1) x 2
= 3 · [1 + · + · ( ) + HOT3 ]
2 3 2 3

1 x (1/2)(1/2 − 1) x 2
− [1 − · + · ( ) + HOT3 ]
2 3 2 3
x
= √ + HOT3 .
3

When x is ‘slightly’ greater than 0,

√ √ √ √ r
2+x− 2−x x/ 2 + HOT3 1/ 2 + HOT2 3
√ √ = √ = √ −→ as x −→ 0+ .
3+x− 3−x x/ 3 + HOT3 1/ 3 + HOT2 2

√ √ r
2+x− 2−x 3
Then lim+ √ √ = .
x→0 3+x− 3−x 2

(g) At and near 0, we have

√
3
√
3
√
3
p p
2+x− 2−x = 2 · ( 3 1 + x/2 − 3 1 − x/2)
√

3 1 x (1/3)(1/3 − 1) x 2
= 2 · [1 + · + · ( ) + HOT3 ]
3 2 2 2

1 x (1/3)(1/3 − 1) x 2
− [1 − · + · ( ) + HOT3 ]
3 2 2 2
√
3
2
= x + HOT3 ,
3
√
3
√
3
√
3
p p
3+x− 3−x = 3 · ( 3 1 + x/3 − 3 1 − x/3)
√

3 1 x (1/3)(1/3 − 1) x 2
= 3 · [1 + · + · ( ) + HOT3 ]
3 3 2 3

1 x (1/3)(1/3 − 1) x 2
− [1 − · + · ( ) + HOT3 ]
3 3 2 3
√
2· 33
= x + HOT3 .
9
 15

When x is ‘slightly’ greater than 0,

√ √ √
3
2+x− 32−x [( 3 2)/3]x + HOT3
√ √ = √
3
3+x− 33−x [(2 · 3 3)/9]x + HOT3
√ r
( 3 2)/3 + HOT2 3 9
= √ −→ as x −→ 0+ .
(2 · 3 3)/9 + HOT2 4

√
3
√ r
2+x− 32−x 3 9
Then lim √ √ = .
x→0+
2
3+x− 23−x 4

(h) At and near 0, we have

∞
12 · 32 · ... · (2n − 3)2 · (2n − 1)2 2n+1
X  
n
arcsinh(x) = x+ (−1) x = x + HOT3
n=1
(2n + 1)!
∞  2
1 · 32 · ... · (2n − 3)2 · (2n − 1)2
X 
arcsin(x) = x+ x2n+1 = x + HOT3
n=1
(2n + 1)!

When x is near 0,
arcsin(x) x + HOT3 1 + HOT2
= = −→ 1 as x −→ 0.
arcsinh(x) x + HOT3 1 + HOT2
arcsin(x)
Then lim = 1.
x→0 arcsinh(x)

(i) At and near 0, we have

∞ ∞
x
X xn X (−1)m x2m+1
e − 1 − sin(x) = −1−
n=0
n! m=0
(2m + 1)!

x2
 
= 1+x+ + HOT3 − 1 − (x + HOT3 )
2
x2
= + HOT3 .
2

When x is near 0,

ex − 1 − sin(x) x2 /2 + HOT3 1 1
2
= = + HOT1 −→ as x −→ 0.
x x2 2 2

ex − 1 − sin(x) 1
Then lim 2
= .
x→0 x 2
(j) At and near 0, we have
∞
2 x 2
X xn
x + 2x + 2 − 2e = x + 2x + 2 − 2 ·
n=0
n!

x2 x3
 
= x2 + 2x + 2 − 2 1 + x + + + HOT4
2 6
x3
= − + HOT4 .
3
 16

When x is near 0,

x2 + 2x + 2 − 2ex −x3 /3 + HOT4 1 1

= = − + HOT1 −→ − as x −→ 0.
x3 x3 3 3

x2 + 2x + 2 − 2ex 1
Then lim =− .
x→0 x3 3
(k) At and near 0, we have
∞
x3 x5 X (−1)n x2n+1 x3 x5
sin(x) − x + + = −x+ +
6 60 n=0
(2n + 1)! 6 60

x3 x5 x3 x5
 
= x− + + HOT7 − x + +
6 120 6 60
x5
= + HOT7 ,
40
∞
x3 x5 X x2n+1 x3 x5
sinh(x) − x − − = −x− −
6 60 n=0
(2n + 1)! 6 60

x3 x5 x3 x5
 
= x+ + + HOT7 − x − −
6 120 6 60
x5
= − + HOT7 .
120

When x is near 0,

sin(x) − x + x3 /6 + x5 /60 x5 /40 + HOT7

=
sinh(x) − x − x3 /6 − x5 /60 −x5 /120 + HOT7
1/40 + HOT2
= −→ −3 as x −→ 0.
−1/120 + HOT2

sin(x) − x + x3 /6 + x5 /60
Then lim = −3.
x→0 sinh(x) − x − x3 /6 − x5 /60

(l) At and near 0, we have

∞
(−1)n x4n+2 x6 x6
X  
2 2 2 2
sin(x ) − x = −x = x − + HOT10 − x2 = − + HOT10 ,
n=0
(2n + 1)! 6 6
∞
x4n+2 x6 x6
X  
2 2 2 2
sinh(x ) − x = −x = x + + HOT10 − x2 = + HOT10 .
n=0
(2n + 1)! 6 6

When x is near 0,

sin(x2 ) − x2 −x6 /6 + HOT10 −1/6 + HOT4

= = −→ −1 as x −→ 0.
sinh(x2 ) − x2 x6 /6 + HOT10 1/6 + HOT4

sin(x2 ) − x2
Then lim = −1.
x→0 sinh(x2 ) − x2
 17

Exercise 16. (Level 3)

∞  2 2
1 · 3 · ... · (2n − 3)2 · (2n − 1)2 n

1 X
(a) i. Show that √ =1+ (−1)n x at and near 0.
1 + x2 n=1
(2n)!
∞
12 · 32 · ... · (2n − 3)2 · (2n − 1)2 2n+1
X  
n
ii. Show that arcsinh(x) = x + (−1) x at and near 0.
n=1
(2n + 1)!

∞  2
1 · 32 · ... · (2n − 3)2 · (2n − 1)2
X 
(b) Show that arcsin(x) = x + x2n+1 at and near 0.
n=1
(2n + 1)!

Remark. Actually all three identities hold on the interval (−1, 1).

Solution. (a) i. According to the Generalized Binomial Formula, we have, at and near 0,

∞  
1 X 1 · 3 · ... · (2n − 3) · (2n − 1) n
√ = 1+ (−1)n x
1+x n=1
2 · 4 · ... · (2n − 2) · (2n)
∞
12 · 32 · ... · (2n − 3)2 · (2n − 1)2 n
X  
≡ 1+ (−1)n x ,
n=1
(2n)!

Then, at and near 0,

∞  2 2 2 2

1 n 1 · 3 · ... · (2n − 3) · (2n − 1)
X
√ =1+ (−1) x2n .
1+x 2
n=1
(2n)!

1
ii. Note that arcsinh0 (x) = √ for any x ∈ R. Also note that arcsinh(0) = 0.
1 + x2
Then, at and near 0, we have

∞
12 · 32 · ... · (2n − 3)2 · (2n − 1)2 2n+1
X  
n
arcsinh(x) = x + (−1) x
n=1
(2n + 1)!

1
(b) Modify the argument in the previous part. First find the Taylor series of the function √ about the
1−x
1
point 0. Then write down the Taylor series of the function √ about the point 0. Now recall that
1 − x2
1
arcsin0 (x) = √ for any x ∈ (−1, 1). Also recall that arcsin(0) = 0. Et cetera.
1 − x2

Exercise 17. (Level 3)

3
Let α be a non-zero constant, and f : R −→ R be the function defined by f (x) = eαx for any x ∈ R.

(a) Show that f 0 (x) = 3αx2 f (x) for any x ∈ R.

(b) Find an identity relating the functions f (n+3) (x), f (n+2) (x), f (n+1) (x), f (n) (x).

(c) Hence, or otherwise, determine the value of f (n) (0) for each n ∈ N.
 18

(d) Find the Taylor series of the function f about the point 0.

N
Remark. Can you imitate the above procedure to find the Taylor series of the function eαx about the point 0 for
N
each positive integer N greater than 1? Better still, can you guess the Taylor series of the function eαx without
going through all these calculations?
3 3
Solution. (a) For any x ∈ R, f (x) = eαx . Then f 0 (x) = 3αx2 eαx = 3αx2 f (x).

(b) Let n ∈ N. For any x ∈ R,

dn+2 0 dn+2
f (n+3) (x) = n+2
(f (x)) = n+2 (3αx2 f (x))
dx dx
(n + 2)(n + 1)
= 3αx2 f (n+2) + (n + 2) · 3α · (2x)f (n+1) (x) + · 3α · 2f (n) (x)
2
= 3αx2 f (n+2) + 6α(n + 2)xf (n+1) (x) + 3α(n + 2)(n + 1)f (n) (x).

(c) Let n ∈ N.
f (n+3) (0) = 3α · 02 · f (n+2) (0) + 6α(n + 2) · 0 · f (n+1) + 3α(n + 2)(n + 1)f (n) (0) = 3α(n + 2)(n + 1)f (n) (0).
Note that f (0) = 1, f 0 (0) = 0.
For any x ∈ R, f 00 (x) = 6αxf (x) + 3αx2 f 0 (x). Then f 00 (0) = 0.

• Suppose n = 3k + 1 for some k ∈ N.

Then

f (n) (0) = f (3k+1) (0) = 3α(3k)(3k − 1)f (3k−2) (0)

= ···
= (3α)k (3k)(3k − 1)(3k − 3)(3k − 4) · ... · 3 · 2 · f 0 (0) = 0.

• Suppose n = 3` + 2 for some ` ∈ N.

Then

f (n) (0) = f (3`+2) (0) = 3α(3` + 1)(3`)f (3ell−1) (0)

= ···
= (3α)` (3` + 1)(3`)(3` − 2)(3` − 3) · ... · 4 · 3 · f 00 (0) = 0.

• Suppose n = 3m for some m ∈ N.

Then

f (n) (0) = f (3m) (0)

= 3α(3m − 1)(3m − 2)f (3m−3) (0)
= ···
= (3α)m (3m − 1)(3m − 2)(3m − 4)(3m − 5) · ... · 2 · 1 · f (0)
αm [(3m)!]
= .
(m!)
 19

(d) The Taylor series of the function f about the point 0 is given by

α3 · [(3 · 1)!] 3 α6 · [(3 · 2)!] 6

1+ x + x + ···
(3!)(1!) (6!)(2!)
αm [(3m)!] 3m αm+1 [(3m + 3)!]
+ x + x3m+3 + · · · ,
[(3m)!](m!) [(3m + 3)!][(m + 1)!]

∞
X αm 3m
which we may write as x .
m=0
m!

Exercise 18. (Level 3)

(a) Let f, g : R → R be two infinitely differentiable functions and let n be a natural number.
By using mathematical induction, prove Leibniz’s rule for the n-th derivative of a product:

n
X
(n)
(f g) (x) = Ckn f (k) (x)g (n−k) (x).
k=0

(b) Find the Taylor polynomial of degree 3 generated by e2x sin x at the point x = 0.

n
X
Solution. (a) Let P (n) be the statement that ”(f g)(n) (x) = Ckn f (k) (x)g (n−k) (x)”.
k=0

• When n = 1, by the product rule LHS=(f g)0 (x) = f (x)g 0 (x) + f 0 (x)g(x)=RHS. Therefore, P (1) is true.
n
X
• Suppose P (n) is true for some natural number n, i.e. (f g)(n) (x) = Ckn f (k) (x)g (n−k) (x).
k=0
 20

Then,

(f g)(n+1) (x)
d  
= (f g)(n) (x)
dx
n
!
d X
= Ckn f (k) (x)g (n−k) (x)
dx
k=0
n
X d  (k) 
= Ckn f (x)g (n−k) (x)
dx
k=0
n
X  
= Ckn f (k+1) (x)g (n−k) (x) + f (k) (x)g (n+1−k) (x)
k=0
n
X n
X
= Ckn f (k+1) (x)g (n−k) (x) + Ckn f (k) (x)g (n+1−k) (x)
k=0 k=0

n+1
X n
X
n
= Ck−1 f (k) (x)g (n+1−k) (x) + Ckn f (k) (x)g (n+1−k) (x)
k=1 k=0

n+1
!
X
= C0n f (x)g (n+1) (x) + n
(Ck−1 + Ckn )f (k) (x)g (n+1−k) (x) + Cnn f (n+1) (x)g(x)
k=1

n+1
!
X
= C0n+1 f (x)g (n+1) (x) + Ckn+1 f (k) (x)g (n+1−k) (x) n+1 (n+1)
+ Cn+1 f (x)g(x)
k=1

n+1
X
= Ckn+1 f (k) (x)g (n+1−k) (x)
k=0

Therefore, P (n + 1) is true.

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

(b) Let h(x) = e2x sin x. By using the result in (a) with f (x) = e2x and g(x) = sin x. We have

h0 (x) = e2x cos x + 2e2x sin x

h00 (x) = −e2x sin x + C12 2e2x cos x + 4e2x sin x
h000 (x) = −e2x cos x − C13 2e2x sin x + C23 4e2x cos x + 8e2x sin x

Therefore, h(0) = 0, h0 (0) = 1, h00 (0) = 4 and h000 (0) = 11 and the required Taylor polynomial is

h00 (0) 2 h000 (0) 3 11

h(0) + h0 (0)x + x + x = x + 2x2 + x3 .
2! 3! 6

Exercise 19. (Level 3)

1
Let f : R −→ R be the function defined by f (x) = for any x ∈ R.
1 + x2

(a) Show that (1 + x2 )f 0 (x) + 2xf (x) = 0 for any x ∈ R.

 21

(b) Find an identity relating the functions f (n+1) (x), f (n) (x), f (n−1) (x).

(c) Hence, or otherwise, determine the value of f (n) (0) for each n ∈ N.

(d) Find the Taylor series of the function f about the point 0.

(e) Apply part (c), or otherwise, to find the Taylor series of the arctangent function about the point 0. (First
ask what the first derivative of arctan is.)

1
Solution. (a) For any x ∈ R, f (x) = . Then (1 + x2 )f (x) = 1. Therefore (1 + x2 )f 0 (x) + 2xf (x) = 0.
1 + x2

(b) Let n ∈ N\{0}. For any x ∈ R,

0
dn
= (0)
dxn
dn
(1 + x2 )f 0 (x) + 2xf (x)


=
dxn
n(n − 1)
= (1 + x2 )f (n+1) (x) + n · (2x)f (n) (x) + · 2f (n−1) (x)
2
+2xf (n) (x) + n · 2f (n−1) (x)
= (1 + x2 )f (n+1) (x) + 2(n + 1)xf (n) (x) + (n + 1)nf (n−1) (x)

(c) For any n ∈ N\{0}, we have 0 = f (n+1) (0) + (n + 1)nf (n−1) (0).
Let m ∈ N.
• Suppose m = 2` + 1 for some ` ∈ N.
Then f (m) (0) = f (2`+1) (0) = −(2` + 1)(2`)f (2`−1) (0) = · · · = (−1)` (2` + 1)(2`) · ... · 5 · 4 · 3 · 2 · f 0 (0) = 0.
• Suppose m = 2k for some k ∈ N.
Then f (m) (0) = f (2k) (0) = −(2k)(2k − 1)f (2k−2) (0) = · · · = (−1)k (2k)(2k − 1) · ... · 4 · 3 · 2 · 1 · f (0) =
(−1)k [(2k)!].

(d) The Taylor series of the function f about the point 0 is given by

(2 · 1)! 2 (2 · 2)! 4 (−1)k [(2k)!] 2k (−1)k+1 [(2k + 2)!] 2k+2

1− x + x + ··· + x + x + ··· ,
2! 4! (2k)! (2k + 2)!

∞
X
which we may write as (−1)k x2k .
k=0

(e) Write g(x) = arctan(x). Note that g 0 (x) = 1

1+x2 = f (x) for any x ∈ R.

Then, for any n ∈ N, g (n+1) (x) = f (n) (x) for any x ∈ R. Therefore
(
(n+1) (n) 0 if n is odd
g (0) = f (0) =
(−1)k [(2k)!] if n is even and n = 2k for some k ∈ N
 22

The Taylor series of g about 0 is given by

g 00 (0) 2 g 000 (0) 3 g (m) (0) m g (m+1) (0) m+1

g(0) + g 0 (0)x + x + x + ··· + x + x + ··· .
2! 3! m! (m + 1)!

This is

(2 · 1)! 3 (2 · 2)! 5 (−1)k [(2k)!] 2k+1 (−1)k+1 [(2k + 2)!] 2k+3

x− x + x + ··· + x + x + ··· ,
3! 5! (2k + 1)! (2k + 3)!

∞
X (−1)k 2k+1
which is x .
2k + 1
k=0

Exercise 20. (Level 3/Level 4)

Consider the exponential function ex . Write f (x) = ex . Let a 6= 0. Let N be a positive integer which satisfies
N ≥ |a|.

(a) Apply Taylor’s Theorem to show that for each n ∈ N, the inequality

n
X f (k) (0) e|a| · |a|n+1
ak − f (a) ≤ holds.
k! (n + 1)!
k=0

n n+1
f (k) (0) e|a| (N +1)N

X
k |a|
(b) Hence deduce that a − f (a) ≤ · whenever n is an integer greater than
k! N! N +1
k=0
N.
∞
X f (k) (0)
(c) Hence show that ak = ea (when we ‘sum to infinity’).
k!
k=0

∞
X ak
Remark. So we have shown that for each a ∈ R, the equality = ea holds. Therefore the Taylor series
k!
k=0
∞
X xn
of the function ex about the point 0 ‘adds up’ to some function (which assigns each a ∈ R to the number
n=0
n!
∞
X ak
), and the resultant function turns out to be identical to the function ex throughout R, and in particular
k!
k=0
at and near 0. Hence the function ex is analytic at 0.

Solution. (a) By Taylor’s Theorem, there exists some ζa strictly between 0 and a such that

n
X f (k) (0) f (n+1) (ζa ) n+1
f (x) = ak + a .
k! (n + 1)!
k=0

Note that f (n+1) (ζa ) = eζa . Since 0 < |ζa | < |a|, we have |f (n+1) (ζa )| = eζa ≤ e|ζa | < e|a| . Then

n
X f (k) (0) f (n+1) (ζa ) n+1 |f (n+1) (ζa )| · |a|n+1 e|a| · |a|n+1
ak − f (a) = − a = ≤
k! (n + 1)! (n + 1)! (n + 1)!
k=0
 23

(b) Suppose n is a positive integer greater than N . Then n > N ≥ |a|. Therefore

n
X f (k) (0)
ak − f (a)
k!
k=0

e|a| · |a|n+1
≤
(n + 1)!
e|a| · |a|N |a| |a| |a| |a|
= · · · ... · ·
N! N + 1 (N + 1) + 1 (N + 1) + (n − N − 1) (N + 1) + (n − N )
n−N +1
e|a| |a|N
 
|a|
≤ ·
N! N +1
n+1
e|a| (N + 1)N

|a|
= ·
N! N +1

|a|
(c) Note that 0 ≤ < 1.
N +1
n n+1
f (k) (0) e|a| (N + 1)N

X
k |a|
As n −→ ∞, we have a − f (a) ≤ · −→ 0.
k! N! N +1
k=0
n n
X f (k) (0) X f (k) (0)
Then ak − f (a) −→ 0 as well. Therefore ak −→ f (a) = ea .
k! k!
k=0 k=0

Exercise 21. (Level 3/Level 4)

Consider the functions sin(x), cos(x). Write f (x) = sin(x), g(x) = cos(x). Let a 6= 0. Let N be a positive integer
which satisfies N ≥ |a|.

(a) i. Apply Taylor’s Theorem to show that for each n ∈ N, the inequality

n
X f (k) (0) |a|n+1
ak − f (a) ≤ holds.
k! (n + 1)!
k=0

n n+1
f (k) (0) (N + 1)N

X |a|
ii. Hence deduce that ak − f (a) ≤ · whenever n is an integer greater than
k! N! N
k=0
N.
∞
X f (k) (0)
iii. Hence show that ak = sin(a) (when we ‘sum to infinity’).
k!
k=0

∞
X g (k) (0)
(b) Show that ak = cos(a) (when we ‘sum to infinity’).
k!
k=0

Remark. So the respective Taylor series of the functions sin(x), cos(x) about the point 0 ‘adds up’ to the functions
sin(x), cos(x) themselves throughout R, and in particular at and near 0. Hence the functions sin(x), cos(x) are
analytic at 0.
 24

Solution. (a) i. By Taylor’s Theorem, there exists some ζa strictly between 0 and a such that
n
X f (k) (0) f (n+1) (ζa ) n+1
f (x) = ak + a .
k! (n + 1)!
k=0

(
(n+1) | cos(ζa )| if n is odd
Note that |f (ζa )| = . Then we have |f (n+1) (ζa )| ≤ 1. Therefore
| sin(ζa )| if n is even

n
X f (k) (0) f (n+1) (ζa ) n+1 |f (n+1) (ζa )| · |a|n+1 |a|n+1
ak − f (a) = − a = ≤
k! (n + 1)! (n + 1)! (n + 1)!
k=0

ii. Suppose n is a positive integer greater than N . Then n > N ≥ |a|. Therefore

n
X f (k) (0)
ak − f (a)
k!
k=0

|a|n+1
≤
(n + 1)!
|a|N |a| |a| |a| |a|
= · · · ... · ·
N ! N + 1 (N + 1) + 1 (N + 1) + (n − N − 1) (N + 1) + (n − N )
n−N +1
|a|N

|a|
≤ ·
N! N +1
n+1
(N + 1)N

|a|
= ·
N! N +1

|a|
iii. Note that 0 ≤ < 1.
N +1
n n+1
f (k) (0) (N + 1)N

X
k |a|
As n −→ ∞, we have a − f (a) ≤ · −→ 0.
k! N! N +1
k=0
n n
X f (k) (0) X f (k) (0)
Then ak − f (a) −→ 0 as well. Therefore ak −→ f (a) = sin(a).
k! k!
k=0 k=0

(b) Imitate the argument in part (a), and the result follows. (Simply replace every f by g in the whole calculation
in part (a), and sin by cos.)

Exercise 22. (Level 4)

X (−1)k a2k+1 n
π a2n+3
(a) Let 0 < a < . Show that, for each n ∈ N, sin(a) − ≤ .
2 (2k + 1)! (2n + 3)!
k=0

(b) (a) For each P = 10, 20, 50, 100, find, by trial-and-error, or otherwise, one positive integer N which satisfies
0.12N +3 1
< .
(2N + 3)! 2 · 10P
(b) Hence, or otherwise, give estimates of sin(0.1) which are correct up to 10, 20, 50, 100 decimal places
respectively.
 25

(c) Give estimates of sin(0.5) which are correct up to 10, 20 decimal places respectively.

2n+2
X sin(j) (0) j sin2n+3 (ζa,n ) 2n+3
Solution. (a) For each n ∈ N, there exists some ζa,n ∈ (0, a) such that sin(a) = a + a .
j=0
j! (2n + 3)!
n
X (−1)k a2k+1 (−1)n+1 sin(ζa,n ) 2n+3
Then sin(a) = + a .
(2k + 1)! (2n + 3)!
k=0
π
Since 0 < ζa,n < , we have sin(ζa,n ) > 0.
2
n
X (−1)k a2k+1 sin(ζa,n ) 2n+3 a2n+3
Then sin(a) − = a ≤ .
(2k + 1)! (2n + 3)! (2n + 3)!
k=0

0.12N +3 1
(b) (a) Note that < is equivalent to 2 · 10P −2N −3 < (2N + 3)!
(2N + 3)! 2 · 10P
0.12·2+3 1
2 · 1010−2·2−3 = 2000 < 5040 = 7! = (2 · 2 + 3)!. Then < .
(2 · 2 + 3)! 2 · 1010
0.12·5+3 1
2 · 1020−2·5−3 = 20000000 < 13! = (2 · 5 + 3)!. Then < .
(2 · 5 + 3)! 2 · 1020
0.12·12+3 1
2 · 1050−2·12−3 = 2 · 1023 < (2 · 12 + 3)!. Then < .
(2 · 12 + 3)! 2 · 1050
0.12·21+3 1
2 · 10100−2·21−3 = 2 · 1055 < (2 · 21 + 3)!. Then < .
(2 · 21 + 3)! 2 · 10100
2
X (−1)k 0.12k+1 0.12·2+3 1
(b) sin(0.1) − ≤ < .
(2k + 1)! (2 · 2 + 3)! 2 · 1010
k=0

0.13 0.15
Therefore 0.1 − + is an estimate of sin(0.1) correct up to 10 decimal places.
3! 5!
5
X (−1)k 0.12k+1 0.12·5+3 1
sin(0.1) − ≤ < .
(2k + 1)! (2 · 5 + 3)! 2 · 1020
k=0

0.13 0.15 0.17 0.19 0.111

Therefore 0.1 − + − + − is an estimate of sin(0.1) correct up to 20 decimal
3! 5! 7! 9! 11!
places.
12
X (−1)k 0.12k+1 0.12·12+3 1
sin(0.1) − ≤ < .
(2k + 1)! (2 · 12 + 3)! 2 · 1050
k=0
12
X (−1)k 0.12k+1
Therefore is an estimate of sin(0.1) correct up to 50 decimal places.
(2k + 1)!
k=0
21
X (−1)k 0.12k+1 0.12·21+3 1
sin(0.1) − ≤ < .
(2k + 1)! (2 · 21 + 3)! 2 · 10100
k=0
21
X (−1)k 0.12k+1
Therefore is an estimate of sin(0.1) correct up to 100 decimal places.
(2k + 1)!
k=0
 26

0.52N +3 1 10P
(c) Note that < P
is equivalent to 2N +2 < (2N + 3)!.
(2N + 3)! 2 · 10 2

1010 0.52·4+3 1
• = 9765625 < 39916800 = 11! = (2 · 4 + 3)!. Then < .
22·4+2 (2 · 4 + 3)! 2 · 1010
4
X (−1)k 0.12k+1 0.52·4+3 1
sin(0.5) − ≤ < .
(2k + 1)! (2 · 4 + 3)! 2 · 1020
k=0

0.53 0.55 0.57 0.59

Therefore 0.5 − + − + is an estimate of sin(0.5) correct up to 10 decimal places.
3! 5! 7! 9!
1020 0.52·8+3 1
• 2·8+2
< (2 · 8 + 3)!. Then < .
2 (2 · 8 + 3)! 2 · 1020
8
X (−1)k 0.12k+1 0.52·8+3 1
sin(0.5) − ≤ < .
(2k + 1)! (2 · 8 + 3)! 2 · 1020
k=0
8
X (−1)k 0.12k+1
Therefore is an estimate of sin(0.5) correct up to 20 decimal places.
(2k + 1)!
k=0

Exercise 23. (Level 4)

Let α ∈ (0, 1). Consider the function f : (−1, 1) −→ R given by f (x) = cos(α arcsin(x)) for any x ∈ (−1, 1). You
may take for granted that f is analytic at 0. (f is analytic at 0 because the functions cos, α · arcsin are analytic
at 0 and f obtained by composing these functions together.)

(a) Show that (1 − x2 )f 00 (x) − xf 0 (x) + α2 f (x) = 0 for any x ∈ (−1, 1).

(b) Hence show that for any n ∈ N, for any x ∈ (−1, 1), (1−x2 )f (n+2) (x)−(2n+1)xf (n+1) (x)+(α2 −n2 )f (n) (x) =
0.
∞
X
(c) Hence at and near 0, f (x) = 1 + α2 An x2n , where {An }∞
n=1 is an infinite sequence of real numbers.
n=1
Determine the value of An for every n.

Solution. f (x) = cos(α arcsin(x)) for any x ∈ (−1, 1).

(a) For any x ∈ (−1, 1), we have

α sin(α arcsin(x))
f 0 (x) = − √
1 − x2
α2 cos(α arcsin(x)) αx sin(α arcsin(x)) α2 f (x) xf 0 (x)
f 00 (x) = − 2
− √ =− +
1−x (1 − x2 ) 1 − x2 1 − x2 1 − x2

Therefore (1 − x2 )f 00 (x) − xf 0 (x) + α2 f (x) = 0 for any x ∈ (−1, 1).

(b) Let n ∈ N. By Leibniz’s Formula, we have

n   j  n   j 
X n d 2 (n−j+2)
X n d
(1 − x ) f (x) − (x) f (n−j+1) (x) + α2 f (n) (x) = 0
j dxj j dxj
j=0 j=0
 27

for any x ∈ (−1, 1).

Simplifying the above, we obtain

(1 − x2 )f (n+2) (x) − (2n + 1)xf (n+1) (x) + (α2 − n2 )f (n) (x) = 0.

(c) For any n ∈ N, we have f (n+2) (0) = −(α2 − n2 )f (n) (0).

Note that f (0) = 1, f 0 (0) = 0. Then we deduce that for any n ∈ N,


 0 if n is odd

m
f (n) = Y


 (−1) m+1
(α2 − 4j 2 ) if n is even and n = 2m + 2 for some m ∈ N
j=0

Since f is analytic at 0, we have, at and near 0,

  
∞  n+1 n
X (−1) Y 
 (α2 − 4j 2 ) x2n+2
f (x) = 1+
 (2n + 2)! 
n=0 j=0
  
2 ∞  n+1 n
α 2 X (−1) Y 
 (α2 − 4j 2 ) x2n+2
= 1− x +
2 n=1
 (2n + 2)!
j=0

  
2 ∞  n+1 2 n
α 2 X (−1) α  Y 
= 1− x + (α2 − 4j 2 ) x2n+2
2 n=1
 (2n + 2)!
j=1

  
∞ n−1
α2 2 X  (−1)n α2  Y 2 
= 1− x + (α − 4j 2 ) x2n
2 n=2
 (2n)!
j=1


Exercise 24. (Level 4)

Let f (x) = x1/3 .

(a) Find p2 (x) centered at 8.

(b) For x > 8, show that

f 000 (2)
|R2 (x)| ≤ (x − 8)3 .
3!
Here Rn (x) denotes the Lagrange form of remainder.

(c) Approximate the error |f (8.01) − p2 (8.01)|.

1 1
Solution. (a) p2 (x) = 2 + 12 (x − 8) − 288 (x − 8)2

(b) Since the remainder is

f 000 (z)
R2 (x) = (x − 8)3 ,
3!
 28

10 − 38
where z > 8. And f 000 (z) = 27 z is decreasing over (0, ∞), thus we have

f 000 (2)
|R2 (x)| ≤ (x − 8)3 .
3!

(c) Since |f (8.01) − p2 (8.01)| = |R2 (8.01)|, by (b) we have

f 000 (z) f 000 (2)

|R2 (8.01)| = (0.01)3 ≤ (0.01)3 = 9.72 × 10−9
3! 3!

Exercise 25. (Level 4/Level 5)

Let f (x) = ex .

(a) Find f (n) (x).

(b) Find Taylor polynomial pn (x) centered at 0.

(c) Show that if x > 0

ex
|ex − pn (x)| ≤ xn+1 .
(n + 1)!

(d) Show that if x ≤ 0

1
|ex − pn (x)| ≤ |x|n+1 .
(n + 1)!

(e) Find n such that

|e−1 − pn (−1)| < 0.005.

(f) Approximate e−1 up to two decimal places.

(g) Find the Taylor series of ex .

(h) Show that for any fixed x, the Taylor series of ex is convergent to ex .
You can use the fact that
an
lim = 0.
n→∞ n!

Solution. (a) f (n) (x) = ex .

Pn 1 k
(b) pn (x) = k=0 k! x

ez ex
(c) |ex − pn (x)| = |Rn (x)| = (n+1)! x
n+1
≤ (n+1)! x
n+1
, because 0 < z < x.

ez 1
(d) |ex − pn (x)| = |Rn (x)| = (n+1)! |x|
n+1
≤ (n+1)! |x|
n+1
, because x < z < 0.

(e) By (e), we have |e−1 − pn (−1)| ≤ 1

(n+1)! | − 1|n+1 < 0.005. After calculation, we can get n = 5.

(f) Since |e−1 − p4 (−1)| < 0.01, then we calculate

1 1 1
p4 (−1) = 1 + (−1) + − + = 0.375.
2! 3! 4!
 29

(g) The Taylor series of ex is

1 2 1 1
1+x+ x + x3 + x4 + · · ·
2! 3! 4!

(h) Because ex = pn (x) + Rn (x), and

ez
lim Rn (x) = lim xn+1 = 0,
n→∞ n→∞ (n + 1)!

we know the Taylor series is convergent to ex .

Exercise 26. (Level 4/Level 5)

Let f (x) = sin(x).

(a) Find f (n) (x).

(b) Let pn (x) is the Taylor series of f (x) centered at x = c.

(c) Show that

1
|Rn (x)| ≤ |x − c|n+1 .
(n + 1)!

(d) Hence show that the Taylor series of f (x) centered at x = c is convergent to f (x).

(e) Let pn (x) be Taylor series of f (x) centered at x = 2 and qn (x) be the Taylor series centered at x = 3.
Compare the error |f (2.1) − pn (2.1)| and |f (2.1) − qn (2.1)|. Which one is a better approximation to f (2.1)?

Solution. (a) f (4n) = sin x, f (4n+1) = cos x, f (4n+2) = − sin x and f (4n+3) = − cos x, where n ∈ N.

sin(c) cos(c)
(b) pn (x) = sin(c) + cos(c)(x − c) − 2! (x − c)2 − 3! (x − c)3 + · · ·

|f (n+1) (z)| 1
(c) |Rn (x)| = (n+1)! |x − c|n+1 ≤ (n+1)! |x − c|n+1 .

(d) By (c), we know

1
0 ≤ |Rn (x)| ≤ |x − c|n+1 .
(n + 1)!

So by Sandwich theorem, we have lim Rn (x) = 0.

n→∞

|f (n+1) (z)| n+1 |f (n+1) (z)| n+1

(e) Since |f (2.1)−pn (2.1)| = (n+1)! (0.1) and |f (2.1)−qn (2.1)| = (n+1)! (0.9) , so we can see |f (2.1)−
pn (2.1)| ≤ |f (2.1) − qn (2.1)|. That is |f (2.1) − pn (2.1)| is a better approximation.

Exercise 27. (Level 4)

 
1 1+x
Note that arctanh(x) = ln for any x ∈ (−1, 1).
2 1−x

(a) Compute arctanh(m) (x) for any m ∈ N.

 30

(b) Let n ∈ N, and x ∈ (0, 1).

i. Apply Taylor’s Theorem with remainder of Lagrange form to show that there exists θx ∈ (0, x) such that

arctanh(x)
x3 x5 x2n−1
 
1 1 1
= x+ + + ··· + + + x2n+1 .
3 5 2n − 1 2(2n + 1) (1 + θx )2n+1 (1 − θx )2n+1

ii. Hence show that

x3 x5 x2n−1
 
arctanh(x) − x + + + ··· +
3 5 2n − 1
"  2n+1 #
1 x
≤ x2n+1 + .
2(2n + 1) 1−x

1
iii. Further suppose 0 < x ≤ . Show that
3

x3 x5 x2n−1
 
arctanh(x) − x + + + ··· +
3 5 2n − 1
1
≤ .
22n+1 · (2n + 1)

1 1
(c) Note that 210 = 1024 ≥ 1000, 220 = 10242 ≥ 1000000 et cetera. Hence ≤ 0.001, 20 ≤ 0.000001.
210 2

i. Apply the above result to give estimates of ln(2), ln(3/2), ln(5/4) correct up to 7 decimal places respec-
tively.
ii. Hence give estimates of ln(3), ln(5) correct up to 6 decimal places respectively.

 
1 1+x 1
Solution. (a) arctanh(x) = ln = (ln(1 + x) − ln(1 − x)).
2 1−x 2
For any positive integer m,

(−1)m−1 · [(m − 1)!] (−1)m−1 (−1)m · [(m − 1)!]

 
(m) 1
arctanh (x) = −
2 (1 + x)m (1 − x)m
(m − 1)! (−1)m−1
 
1
= · m
+ .
2 (1 + x) (1 − x)m

(b) Let n ∈ N, and x ∈ (0, 1).

i. By Taylor’s Theorem, there exists θx ∈ (0, x) such that

2n
X arctanh(m) (0) m arctanh(2n+1) (θx ) 2n+1
arctanh(x) = x + x .
m=0
m! ((2n + 1)!)

Note that arctanh(0) = 0.

 31

For each k = 1, 2, · · · , n, we have

(2k − 1)! (−1)2k−1

 
1 (2k − 1)!
arctanh(2k) (0) = · 2k
+ 2k
= · (−1 + 1) = 0.
2 (1 + 0) (1 − 0) 2

For each k = 0, 1, 2, · · · , n − 1, we have

(−1)2k
 
(2k)! 1 (2k)!
arctanh(2k+1) (0) = · + = · (1 + 1) = (2k)!.
2 (1 + 0)2k+1 (1 − 0)2k+1 2

(−1)2n
 
(2n)! 1
arctanh(2n+1) (θx ) = · +
2 (1 + θx )2n+1 (1 − θx )2n+1
 
(2n)! 1 1
= · + .
2 (1 + θx )2n+1 (1 − θx )2n+1

Then

arctanh(x)
n−1
X arctanh(2k+1) (0) 2k+1 arctanh(2n+1) (θx ) 2n+1
= x + x
(2k + 1)! (2n + 1)!
k=0

n−1  
X (2k)! [(2n)!]/2 1 1
= x2k+1 + · + x2n+1
(2k + 1)! (2n + 1)! (1 + θx )2n+1 (1 − θx )2n+1
k=0

x3 x5 x2n−1
 
1 1 1
= x+ + + ··· + + + x2n+1 .
3 5 2n − 1 2(2n + 1) (1 + θx )2n+1 (1 − θx )2n+1

1 1 1 1
ii. Note that 0 < θx < x < 1. Then 0 < < = 1. Also, 0 < < . Then
1 + θx 1+0 1 − θx 1−x

x3 x5 x2n−1
 
arctanh(x) − x + + + ··· +
3 5 2n − 1
 
1 1 1
= + x2n+1
2(2n + 1) (1 + θx )2n+1 (1 − θx )2n+1
 
1 1 1
= + x2n+1
2(2n + 1) (1 + θx )2n+1 (1 − θx )2n+1
 
1 1
≤ 1+ x2n+1
2(2n + 1) (1 − x)2n+1
"  2n+1 #
1 x
= x2n+1 +
2(2n + 1) 1−x

.
1 2
iii. Further suppose 0 < x ≤ . Then 1 − x ≥ > 0. Therefore
3 3

x 1/3 1
≤ = .
1−x 2/3 2
 32

1
Also, x ≤ .
2
Hence

x3 x5 x2n−1
 
arctanh(x) − x + + + ··· +
3 5 2n − 1
"  2n+1 #
1 x
≤ x2n+1 +
2(2n + 1) 1−x
"   2n+1 #
2n+1
1 1 1
≤ +
2(2n + 1) 2 2
1
=
22n+1 · (2n + 1)

1
(c) i. Whenever 0 < x ≤ ,
3

x3 x5 x2·10−1
 
1
arctanh(x) − x + + + ··· + ≤
3 5 2 · 10 − 1 22·10+1 · (2 · 10 + 1)
1
≤
106 · 2 · (2 · 10 + 1)
1
≤ .
4 · 107

Then
x3 x5 x2·10−1
   
1+x 1 1
ln −2 x+ + + ··· + ≤2· ≤ .
1−x 3 5 2 · 10 − 1 4 · 107 2 · 107

1 3 1 5 1
Note that ln(2) = 2arctanh( ), ln( ) = 2arctanh( ), ln( ) = 2arctanh( ).
3 2 5 4 9

9
X 1 1 1
ln(2) − 2 · 2k+1 ≤
2k + 1 3 2 · 107
k=0

9
3 X 1 1 1
ln( ) − 2 · ≤
2 2k + 1 52k+1 2 · 107
k=0

9
5 X 1 1 1
ln( ) − 2 · ≤
4 2k + 1 92k+1 2 · 107
k=0

ln(2) = 0.693147181... ≈ 0.6931471, correct up to 7 decimal places.

3
ln( ) = 0.405465108... ≈ 0.4054651, correct up to 7 decimal places.
2
5
ln( ) ≈ 0.2231435, correct up to 7 decimal places.
4
 33

ii. We have
9 9
X 1 1 X 1 1
ln(3) − 2 · 2k+1 − 2 · 2k+1
2k + 1 3 2k + 1 5
k=0 k=0

9
! 9
!
X 1 1 3 X 1 1
= ln(2) − 2 · + ln( ) − 2 ·
2k + 1 32k+1 2 2k + 1 52k+1
k=0 k=0

9 9
X 1 1 3 X 1 1
≤ ln(2) − 2 · + ln( ) − 2 ·
2k + 1 32k+1 2 2k + 1 52k+1
k=0 k=0

2 1
≤ ≤
2 · 107 2 · 106

We also have
9 9
X 1 1 X 1 1
ln(5) − 4 · 2k+1 − 2 ·
2k + 1 3 2k + 1 92k+1
k=0 k=0

9
! 9
!
X 1 1 5 X 1 1
= 2 ln(2) − 4 · + ln( ) − 2 ·
2k + 1 32k+1 4 2k + 1 92k+1
k=0 k=0

9 9
X 1 1 5 X 1 1
≤ 2 ln(2) − 2 · 2k+1 + ln( ) − 2 ·
2k + 1 3 4 2k + 1 92k+1
k=0 k=0

3 1
≤ 7
≤
2 · 10 2 · 106

ln(3) = 0.40546510... + 0.69314718... ≈ 1.098612, correct up to 6 decimal places.

ln(5) = 0.22314355... + 2 × 0.69314718... ≈ 1.609437, correct up to 6 decimal places.

Exercise 28. (Level 5)

(a) Let A0 , A1 , · · · , Am be real numbers, B be a non-negative real number, and δ be a positive real number.
Suppose the polynomial function A0 + A1 x + · · · + Am xm satisfies

|A0 + A1 x + · · · + Am xm | ≤ Bxm+1 for any x ∈ (0, δ].

Show that A0 = A1 = · · · = Am = 0. (Hint: Take limit at 0.)

(b) Let f be a function defined on an open interval I which contains the point 0. Suppose f is differentiable on
I for as many times as we like.
Let g be the function defined by g(x) = f (x2 ). You may take for granted that g is also differentiable on its
domain for as many times as we like.

i. Let a be positive real number satisfying a2 ∈ I. Let n be a positive integer.

Show that for every x ∈ (0, a], there exist some ζx ∈ (0, x2 ), ωx ∈ (0, x) such that

n 2n+1
X f (k) (0) f (n+1) (ζx ) 2n+2 X g j (0) g (2n+2) (ωx ) 2n+2
x2k + x = xj + x
k! (n + 1)! j=0
j! (2n + 2)!
k=0
 34

ii. Here take for granted that every continuous function defined on a closed and bounded interval attains
absolute maximum in that interval.
Apply the previous part to show that there is some non-negative real number B, whose value is indepen-
dent of that of x, such that for every x ∈ (0, a],

n 2n+1
X f (k) (0) X g j (0) j
x2k − x = Bx2n+2
k! j=0
j!
k=0

(Hints: You may start by asking whether each of |f n+1 |, |g 2n+2 | attains absolute maximum on [0, a]. At
some stage you may also recall the Triangle Inequality, which reads: |s + t| ≤ |s| + |t| for any x, t ∈ R.)
iii. Apply the previous part to show that


 0 if m is odd

(m)
g (0) =
 (2k)! f (k) (0)

if m is even and m = 2k for some integer k

k!

Solution. (a) Let A0 , A1 , · · · , Am be real numbers, B be a non-negative real number, and δ be a positive real
number. Suppose the polynomial function A0 + A1 x + · · · + Am xm satisfies

|A0 + A1 x + · · · + Am xm | ≤ Bxm+1 for any x ∈ (0, δ].

• We have −Bxm+1 ≤ A0 + A1 x + · · · + Am xm ≤ Bxm+1 for any x ∈ (0, δ].

Note that lim+ −Bxm+1 = 0 and lim+ −Bxm+1 = 0.
x→0 x→0

Then, by Sandwich Rule, we have lim+ (A0 + A1 x + · · · + Am xm ) = 0.

x→0

Therefore A0 = lim+ (A0 + A1 x + · · · + Am xm ) = 0.

x→0

• Now, for any x ∈ (0, δ], −Bxm+1 ≤ A1 x + A2 x2 + · · · + Am xm ≤ Bxm+1 .

Then −Bxm ≤ A1 + A2 x + · · · + Am xm−1 ≤ Bxm .
Note that lim+ −Bxm = 0 and lim+ −Bxm = 0.
x→0 x→0

Then, by Sandwich Rule, we have lim+ (A1 + A2 x · · · + Am xm−1 ) = 0.

x→0

Therefore A1 = lim+ (A1 + A2 x + · · · + Am xm−1 ) = 0.

x→0

• Now, for any x ∈ (0, δ], −Bxm ≤ A2 x + A3 x2 + · · · + Am xm ≤ Bxm .

Repeating the previous argument, we obtain A2 = 0.
• Further repeating the argument, we obtain A3 = 0, A4 = 0, ..., Am = 0.

(b) Let f be a function defined on an open interval I which contains the point 0. Suppose f is differentiable on
I for as many times as we like.
Let g be the function defined by g(x) = f (x2 ). You may take for granted that g is also differentiable on its
domain for as many times as we like.
 35

i. Let a be positive real number satisfying a2 ∈ I. Let n be a positive integer.

Suppose x ∈ (0, a].
2n+1
X g j (0) j g (2n+2) (ωx ) 2n+2
By Taylor’s Theorem, there exists some ωx ∈ (0, x) such that g(x) = x + x .
j=0
j! (2n + 2)!

Since x ∈ (0, a], we have x ∈ (0, a ]. By Taylor’s Theorem, there exists some ζx ∈ (0, x2 ) such that
2 2

n
X f (k) (0) 2k f (n+1) (ζx ) 2n+2
f (x2 ) = x + x .
k! (n + 1)!
k=0

Therefore

n 2n+1
X f (k) (0) f (n+1) (ζx ) 2n+2 X g j (0) g (2n+2) (ωx ) 2n+2
x2k + x = f (x2 ) = g(x) = xj + x .
k! (n + 1)! j=0
j! (2n + 2)!
k=0

ii. The function f (n+1) is continuous on [0, a2 ]. Then the function |f (n+1) | is continuous on [0, a2 ], and hence
attains absolue maximum in [0, a2 ]. Denote the absolute maximum value of |f (n+1) | on [0, a2 ] by M1 .
The function g (2n+2) is continuous on [0, a]. Then the function |g (2n+2) | is continuous on [0, a], and hence
attains absolute maximum in [0, a]. Denote the absolute maximum value of |g (2n+2) | on [0, a] by M2 .
For every x ∈ (0, a], there exist some ζx ∈ (0, x2 ), ωx ∈ (0, x) such that

n 2n+1
X f (k) (0) f (n+1) (ζx ) 2n+2 X g j (0) g (2n+2) (ωx ) 2n+2
x2k + x = xj + x .
k! (n + 1)! j=0
j! (2n + 2)!
k=0

Then

n 2n+1
X f (k) (0) X g j (0) j f (n+1) (ζx ) 2n+2 g (2n+2) (ωx ) 2n+2
x2k − x = − x + x
k! j=0
j! (n + 1)! (2n + 2)!
k=0

|f (n+1) (ζx )| 2n+2 |g (2n+2) (ωx )| 2n+2

= x + x
(n + 1)! (2n + 2)!
M1 M2
≤ x2n+2 + x2n+2
(n + 1)! (2n + 2)!
 
M1 M2
= + x2n+2 = Bx2n+2 ,
(n + 1)! (2n + 2)!

M1 M2
where B = + is a non-negative real number whose value is independent of that of x.
(n + 1)! (2n + 2)!
g (2j) (0) f (j) (0) g (2j+1) (0)
iii. For each j = 0, 1, 2, · · · , n, define A2j = − , A2j+1 = .
(2j)! j! (2j + 1)!
Then, for every x ∈ (0, a], we have

2n+1 2n+1 n
X X g j (0) j X f (k) (0) 2k
Aj x j = x − x ≤ Bx2n+1
j=0 j=0
j! k!
k=0

Then A0 = A1 = A2 = A3 = · · · = A2n = A2n+1 = 0.

 36

It follows that


 0 if m is odd

g (m) (0) =
 (2k)! f (k) (0)

if m is even and m = 2k for some integer k

k!

Exercise 29. (Level 5)

Here you may take for granted that lim e−1/x = 0.
x→0+
(
0 if x ≤ 0
Let f : R −→ R be the function defined by f (x) = .
e−1/x if x > 0

(a) Apply L’Hôpital’s Rule, or otherwise, to show that lim x−n e−1/x = 0 for any positive integer n.
x→0+

(b) i. Find the ‘explicit’ formula for the function f 0 on R\{0}.

ii. Show that f is differentiable at 0.
iii. Show that f is continuously differentiable at 0.
(
(n) 0 if x < 0
(c) i. Show that for each positive integer n, f (x) = , where Pn (t) is some
Pn (1/x)e−1/x if x > 0
polynomial expression in t. (You may need mathematical induction here.)
ii. Show that f is smooth at 0, and find the value of f (n) (0) for each n ∈ N.

(d) i. Write down the Taylor series Tf,0 (x) of the function f about the point 0.
ii. Does Tf,0 (x) define a function at and near the point 0? If yes, what is it explicitly? Is there any δ > 0
so that this function is equal to f on the interval (0, δ)? Why?

Remark. This exercise tells you that the definition of analyticity is highly non-trivial: a function may be smooth
at a certain point without being analytic at that point.

Solution. (a) Let n be a positive integer. We mark by ‘(H)’ beneath the equality at which we apply L’Hôpital’s
Rule,

x−n −nx−n−1
lim+ x−n e−1/x = lim+ = lim
x→0 x→0 e1/x (H) x→0+ −e1/x /x2
nx−n+1 n(−n + 1)x−n
= lim+ 1/x
= lim+
x→0 e (H) x→0 −e1/x /x2
n(n − 1)x−n+2
= lim+
x→0 e1/x
= ···
[n(n − 1) · ... · 2]x−1 [n(n − 1) · ... · 2](−1)x−2
= lim+ = lim
x→0 e1/x (H) x→0+ −e1/x /x2
n!
= lim = lim+ (n!)e−1/x = 0
x→0+ e1/x x→0
 37

(
0 if x ≤ 0
(b) f (x) = .
e−1/x if x > 0
(
0 0 if x < 0
i. f (x) = .
x−2 e−1/x if x > 0

ii. Provided that t is ‘slightly’ greater than 0,

f (0 + t) − f (0) t−2 e−1/t

= = t−3 e−1/t −→ 0 as t −→ 0+ .
t t

f (0 + t) − f (0)
Provided that t is ‘slightly’ less than 0, = 0 −→ 0 as t −→ 0− .
t
f (0 + t) − f (0) f (0 + t) − f (0)
Then lim , lim exist and are equal to 0. Therefore f is differentiable
t→0+ t t→0 − t
at 0, and f 0 (0) = 0.
iii. Provided that x is ‘slightly’ greater than 0, f 0 (x) = x−2 e−1/x −→ 0 = f 0 (0) as x −→ 0+ .
Provided that x is ‘slightly’ less than 0, f 0 (x) = 0 −→ 0 = f 0 (0) as x −→ 0− .
Then lim+ f 0 (x), lim− f 0 (x) exist and are equal to f 0 (0). Therfore f is continuously differentiable at 0.
x→0 x→0

(
0 if x < 0
(c) i. Denote by S(n) the proposition that for each positive integer n, f (n) (x) = ,
Pn (1/x)e−1/x if x > 0
where Pn (t) is some polynomial expression in t.
• According to the result in part (i), S(1) is true. (Here P1 (t) = t2 .)
(
(k) 0 if x < 0
• Assume S(k) is true. Then f (x) = −1/x
, where Pk (t) is some polyno-
Pk (1/x)e if x > 0
mial expression in t.
We are going to deduce that S(k + 1) is true:
∗ For any x ∈ (−∞, 0), we have f (k+1) (x) = (f (k) )0 (x) = 0.
For any x ∈ (0, +∞), we have f (k+1) (x) = (f (k) )0 (x) = −x−2 Pk0 (1/x)e−1/x + Pk (1/x)x−2 e−1/x =
x−2 (Pk (1/x) − Pk0 (1/x))e−1/x = Pk+1 (1/x)e−1/x , where Pk+1 (t) = t2 (Pk (t) − Pk0 (t)).
Since Pk (t) is a polynomial expression in t, Pk0 (t) is also a polynomial expression in t. Then
Pk (t) − Pk0 (t) is a polynomial expression in t. Therefore Pk+1 (t) is a polynomial expression in t.
(
0 if x < 0
To summarize, we have f (k+1) (x) = −1/x
, where Pk+1 (t) is the
Pk+1 (1/x)e if x > 0
polynomial expression given above.
Hence S(k + 1) is true.
By the Principle of Mathematical Induction, S(n) is true for any positive integer n.
ii. Denote by T (n) the proposition that f is n-times differentiable at 0 and f (n) (0) = 0.
• By the results in parts (ii), (iii) together, T (1) is true.
• Assume T (k) is true. Then f is k-times differentiable at 0 and f (k) (0) = 0.
 38

f (k) (0 + t) − f (k) (0)

∗ Provided that t is ‘slightly’ less than 0, = 0 −→ 0 as t −→ 0− .
t
Whenever t > 0, f (k) (t) = Pk (1/t)e−1/t , where Pk (s) is a polynomial expression in s. Now
Pk (s) = am sm + am−1 sm−1 + · · · + a1 s + a0 for some real constants a0 , a1 , · · · , am−1 , am .
Then, provided t is ‘slightly’ greater than 0,

f (k) (0 + t) − f (k) (0)

t
= t−1 f (k) (0)
= t−1 (am t−m + am−1 t−m+1 + · · · + a1 t−1 + a0 )e−1/t
= am t−m−1 e−1/t + am−1 t−m e−1/t + · · · + a1 t−2 e−1/t + a0 t−1 e−1/t
−→ 0 + 0 + ···0 + 0 = 0 as t −→ 0+

f (k) (0 + t) − f (k) (0) f (k) (0 + t) − f (k) (0)

Now lim− , lim− exist and are equal to 0. Therefore
t→0 t t→0 t
f (k) is differentiable at 0, f is (k + 1)-times differentiable at 0, and f (k+1) (0) = 0.
Hence T (k + 1) is true.
By the Principle of Mathematical Induction, T (n) is true for any positive integer n.
It follows that f is smooth at 0. Moreover, for any n ∈ N, f (n) (0) = 0.
∞
X
(d) i. The Taylor series Tf,0 (x) of the function f about the point 0 is 0 · xn .
n=0

ii. Tf,0 (x) adds up to the constant zero function, as we sum to infinity.
Note that f (x) > 0 for any x ∈ (0, +∞). For any δ > 0, we have f (δ/2) 6= 0. Therefore f fails to be
equal to the constant zero function on (0, δ),

Exercise 30. (Level 5)

(1 − τ )|v| 1 + |v|
(a) Let v, τ ∈ R. Suppose 0 < |v| < 1 and 0 < τ < 1. Show that 0 < ≤ < 1.
1 + τv 2

(b) Consider the function ln(1 + x). Write f (x) = ln(1 + x). Let a ∈ (−1, 1)\{0}.

i. Let n ∈ N. Apply Taylor’s Theorem with remainder of Cauchy form to show that there exists some
θn ∈ (0, 1) such that
n n
f (k) (0) k |a|n+1

X 1 − θn
a − f (a) = · .
k! 1 + θn a 1 + θn a
k=0

∞
X f (k) (0)
ii. Hence show that ak = ln(1 + a) (when we ‘sum to infinity’).
k!
k=0

∞
X (−1)m−1 m
Remark. Hence the Taylor series x of the function ln(1 + x) about the point 0 ‘adds up’ to
m=1
m
some function, which turns out to be identical to the function ln(1 + x) on (−1, 1).
 39

(c) Let α ∈ (0, 1). Consider the function (1 + x)α . Write f (x) = (1 + x)α . Let a ∈ (−1, 1)\{0}.

i. Let n ∈ N. Apply Taylor’s Theorem with remainder of Cauchy form to show that there exists some
θn ∈ (0, 1) such that

n
X f (k) (0)
ak − f (a)
k!
k=0
 n
h α α i 1 − θn
= α(1 − α)(1 − ) · ... · (1 − ) · (1 + θn a)α−1 · · |a|n+1 .
2 n 1 + θn a

∞
X f (k) (0)
ii. Show that ak = (1 + a)α (when we ‘sum to infinity’).
k!
k=0

Remark. Therefore, whenever 0 < α < 1 and −1 < x < 1,

(1 + x)α
α(α − 1) 2 α(α − 1)(α − 2) 3
= 1 + αx + x + x + ···
2! 3!
α(α − 1)(α − 2) · ... · (α − n + 1) n
+ x + ··· .
n!

We have established a particular case of the Generalized Binomial Theorem.

Solution. (a) • Note that 1 − τ > 0.

Also note that 0 < τ |v| = |τ v| < 1. Then τ v > −1. Therefore 1 + τ v > 0.
(1 − τ )|v|
Hence > 0.
1 + τv
1 + |v| 1+1
• Since |v| < 1, we have < = 1.
2 2
• Note that

(1 + τ v)(1 + |v|) − 2(1 − τ )|v| = 1 + τ v + |v| + τ v · |v| − 2|v| + 2τ |v|

= 1 + τ v − |v| + τ v · |v| + 2τ |v|.

∗ When 0 < v < 1, we have

(1 + τ v)(1 + |v|) − 2(1 − τ )|v| = (1 − v) + 3τ v + τ v 2 > 0.

∗ When −1 < v < 0, we have 0 < −v = |v| < 1. Then

(1 + τ v)(1 + |v|) − 2(1 − τ )|v| = 1 + v − τ v − τ v2

= (1 + v)(1 − τ v)
= (1 − |v|)(1 + τ |v|) > 0.

(1 − τ )|v| 1 + |v|
Then ≤ .
1 + τv 2
(1 − τ )|v| 1 + |v|
Hence 0 < ≤ < 1.
1 + τv 2
 40

(b) f (x) = ln(1 + x) for any x ∈ (−1, +∞). Let a ∈ (−1, 1)\{0}.

i. Let n ∈ N. There exists some ωn strictly between 0, a such that

n
X f (k) (0) f (n+1) (ωn )
ak − f (a) = − (a − ωn )n a
k! n!
k=0

ωn
Define θn = . Note that θn ∈ (0, 1) and ωn = θn a.
a
(−1)n (n!)
Also note that f (n+1) (ωn ) = . Now
(1 + ωn )n+1

n
X f (k) (0) f (n+1) (ωn )
ak − f (a) = (a − ωn )n a
k! n!
k=0

1 (−1)n (n!)
= · · (a − ωn )n a
n! (1 + ωn )n+1
1
= · |a − ωn |n · |a|
(1 + ωn )n+1
1
= · |a − θn a|n · |a|
(1 + θn a)n+1
1
= · |1 − θn |n · |a|n+1
(1 + θn a)n+1
n
|a|n+1

1 − θn
= · .
1 + θn a 1 + θn a

(1 − θn )|a| 1 + |a| 1 1
ii. For the same a, n, θn in part (i), we have 0 < < < 1. Also, 0 < ≤ .
1 + θn a 2 1 + θn a 1 − |a|
Then
n
X f (k) (0)
ak − f (a)
k!
k=0
n
|a|n+1

1 − θn
= ·
1 + θn a 1 + θn a
 n
1 + |a| |a|
≤ · −→ 0 as n −→ ∞
2 1 − |a|

n
X f (k) (0)
Therefore ak −→ f (a) = ln(1 + a) as n −→ ∞.
k!
k=0

(c) Here α ∈ (0, 1), and f (x) = (1 + x)α for any x ∈ (−1, +∞). Let a ∈ (−1, 1)\{0}.

i. Let n ∈ N.
There exists some ωn strictly between 0, a such that

n
X f (k) (0) f (n+1) (ωn )
ak − f (a) = − (a − ωn )n a.
k! n!
k=0
 41

ωn
Define θn = . Note that θn ∈ (0, 1) and ωn = θn a.
a
Also note that f (n+1) (ωn ) = [α(α − 1)(α − 2) · ... · (α − n)](1 + ωn )α−n−1 . Now

n
X f (k) (0)
ak − f (a)
k!
k=0

f (n+1) (ωn )
= (a − ωn )n a
n!
[α(α − 1)(α − 2) · ... · (α − n)](1 + ωn )α−n−1
= · (a − ωn )n a
n!
(1 − α)(2 − α) · ... · (n − α)
= α· · (1 + ωn )α−n−1 · |a − ωn |n · |a|
n!
h α α i
= α(1 − α)(1 − ) · ... · (1 − ) (1 + θn a)α−n−1 · (1 − θn )n · |a|n+1
2 n
 n
h α α i 1 − θn
= α(1 − α)(1 − ) · ... · (1 − ) · (1 + θn a)α−1 · · |a|n+1 .
2 n 1 + θn a

(1 − θn )|a| 1 − θn |a|
ii. For the same a, n, θn in part (i), we have 0 < < < 1.
1 + θn a 1 − θn |a|
1 1
Also, since 0 < α < 1 and 0 < |θn a| < |a| < 1, we have 0 < (1 + θn a)α−1 = ≤ .
(1 + θn a)1−α (1 − |a|)1−α
α
Moreover, 0 < 1 − < 1 for each j = 1, 2, · · · , n. Then
j

n
X f (k) (0)
ak − f (a)
k!
k=0
 n
h α α i 1 − θn
= α(1 − α)(1 − ) · ... · (1 − ) · (1 + θn a)α−1 · · |a|n+1
2 n 1 + θn a
α
≤ · |a|n+1 −→ 0 as n −→ ∞
(1 − |a|)1−α

n
X f (k) (0)
Therefore ak −→ f (a) = (1 + a)α as n −→ ∞.
k!
k=0

2 Arithmetics of series
Exercise 1. (Level 1)
∞
x X
Show that = 22n x2n+1 at and near 0.
1 − 4x2 n=0
 
1 1
Remark. Actually the identity holds on − , .
2 2
 42

∞ ∞
x X
2 n
X
Solution. At and near 0, we have = x (4x ) = 22n x2n+1 . (The first step requires the expansion
1 − 4x2 n=0 n=0
of the geometric series.)

Exercise 2. (Level 1/Level 2)

Find the Taylor series of the following function, centered at 0.
ex −e−x
(a) sinh(x) = 2 .
2
(b) ex .
Recall the Taylor series of ex centered at 0 is

x2 x3 x4
ex = 1 + x + + + + ···
2! 3! 4!

Solution. (a) The Taylor series of e−x centered at 0 is

(−x)2 (−x)3 (−x)4

1 + (−x) + + + + ···
2! 3! 4!
x2 x3 x4
=1−x+ − + + ···
2! 3! 4!
They Taylor series of sinh(x) centered at 0 is therefore

1−1 1+1 1−1 2 1+1 3 1−1 4

+ x+ x + x + x + ···
2 2 2 × 2! 2 × 3! 2 × 4!

x3 x5
x+ + + ···
3! 5!
2
(b) The Taylor series of ex centered at 0 is

x4 x6 x8
1 + (x2 ) + + + + ···
2! 3! 4!

Exercise 3. (Level 1/Level 2)

1
(a) Find the Taylor series of 2 sin 2x centered at x = 0.

(b) Find the Taylor series of cos x sin x centered at x = 0 with order 5 by using the the product of the above
Taylor series.
Check: Because sin 2x = 2 sin x cos x, your answer should be same as the previous part.

Recall the Taylor series of cos x and sin x centered at 0 is

x2 x4 x6
cos x = 1 − + − + ···
2! 4! 6!
x3 x5 x7
sin x = x − + − + ···
3! 5! 7!
 43

1
Solution. (a) The Taylor series of 2 sin 2x is

2x (2x)3 (2x)5 (2x)7

− + − + ···
2 2 × 3! 2 × 5! 2 × 7!

2 2 x3 24 x5 2 6 x7
=x− + − + ··· .
3! 5! 7!
2 2 4
= x − x3 + x5 − x7 + · · ·
3 15 45

(b) The Taylor series of cos x sin x is

x2 x4 x3 x5
  
1− + − ··· x− + − ...
2! 4! 3! 5!
   
1 1 1 1 1
=x− + x3 + + + x5 + · · ·
2! 3! 4! 2!3! 5!
2 2
= x − x3 + x5 − · · ·
3 15

Exercise 4. (Level 1/Level 2)

4 − 3x K L
(a) Find some constants K, L so that = + for any x ∈ R\{1, 2}.
(1 − x)(2 − x) 1−x 2−x
Remark. This process is called partial fraction decomposition.
∞
4 − 3x X
(b) Hence show that = (1 + An )xn at and near 0. Here {An }∞
n=0 is an infinite sequence of real
(1 − x)(2 − x) n=0
numbers. You have to determine the value of An for any n ∈ N.

4 − 3x 1 2
Solution. (a) For any x ∈ R\{1, 2}, we have = + .
(1 − x)(2 − x) 1−x 2−x

(b) At and near 0, we have

∞ ∞  n
4 − 3x 1 2 1 1 X X x
= + = + = xn +
(1 − x)(2 − x) 1−x 2−x 1 − x 1 − x/2 n=0 n=0
2
∞ h
X  x n i
= xn +
n=0
2
∞  
X 1
= 1+ xn .
n=0
2n

Exercise 5. (Level 2)
 44

∞
X (−1)n
(a) Show that x ln(x) = (x − 1) + (x − 1)n at and near 1.
n=2
n(n − 1)

Remark. Actually the identity holds on the interval (0, 2).

∞
X 4(−1)k k
(b) Show that (1 − x2 ) arctan(x) = x + x2k+1 at and near 0.
(2k + 1)(2k − 1)
k=1

Remark. Actually the identity holds on the interval (−1, 1).

∞
X (−1)n−1 n
Solution. (a) At and near 0, we have ln(1 + x) = x at and near 0. Then, at and near 1, we have
n=1
n
∞
X (−1)n−1
ln(x) = ln(1 + (x − 1)) = (x − 1)n . Therefore, at and near 1, we have
n=1
n

x ln(x) = [(x − 1) + 1] ln(x) = (x − 1) ln(x) + ln(x)

∞ ∞
X (−1)n−1 X (−1)n−1
= (x − 1) (x − 1)n + (x − 1)n
n=1
n n=1
n
∞ ∞
X (−1)n−1 X (−1)n−1
= (x − 1)n+1 + (x − 1)n
n=1
n n=1
n
∞ ∞
X (−1)n X (−1)n−1
= (x − 1)n + (x − 1) + (x − 1)n
n=2
n − 1 n=2
n
∞  
X
n 1 1
= (x − 1) + (−1) − (x − 1)n
n=2
n−1 n
∞
X (−1)n
= (x − 1) + (x − 1)n .
n=2
n(n − 1)

(b) At and near 0, we have

(1 − x2 ) arctan(x) = arctan(x) − x2 arctan(x)

∞ ∞
X (−1)k 2k+1 X (−1)k 2k+1
= x − x2 · x
2k + 1 2k + 1
k=0 k=0
∞ k ∞
X (−1) 2k+1 X (−1)k 2k+3
= x − x
2k + 1 2k + 1
k=0 k=0
∞ ∞
" #
X (−1)k 2k+1 X (−1)k−1 2k+1
= x+ x − x
2k + 1 2k − 1
k=1 k=1
∞ 
(−1)k (−1)k−1 2k+1
X 
= x+ − x
2k + 1 2k − 1
k=1
∞
X 4(−1)k k
= x+ x2k+1 .
(2k + 1)(2k − 1)
k=1
 45

Exercise 6. (Level 2)
∞ ∞
X x2m+1 X x2m
(a) Show that sinh(x) = and cosh(x) = at and near 0.
m=0
(2m + 1)! m=0
(2m)!

∞
3 X (−1)n (1 − 32n )x2n+1
(b) i. Show that sin3 (x) = at and near 0.
4 n=1 (2n + 1)!
∞
x2 X (−1)n (32n+2 − 1)x2n
ii. Hence, or otherwise, show that sin2 (x) cos(x) = at and near 0.
4 n=0 (2n + 2)!

∞
X A2m+1
(c) Show that cosh2 (x2 ) − cos2 (x2 ) = xN (2m+1) at and near 0. Here A, B, C are rational numbers
m=0
(Bm + C)!
and N is an integer. You have to determine their respective values.

Remark. Actually all these identities hold on the whole of R, and not only at and near 0.
∞ ∞
X xn X (−1)n xn
Solution. (a) At and near 0, we have ex = and e−x = . Then
n=0
n! n=0
n!

sinh(x)
ex − e−x
=
2
"∞ ∞
#
1 X xn X (−1)n xn
= −
2 n=0 n! n=0 n!
∞
X 1 − (−1)n xn
= ·
n=0
2 n!
∞
X x2m+1
= ,
m=0
(2m + 1)!

cosh(x)
ex + e−x
=
2
"∞ ∞
#
1 X xn X (−1)n xn
= +
2 n=0 n! n=0 n!
∞
X 1 + (−1)n xn
= ·
n=0
2 n!
∞
X x2k
= .
(2k)!
k=0

(b) i. For any x ∈ R, we have

sin(3x) = sin(2x) cos(x) + cos(2x) sin(x)

= 2 sin(x) cos2 (x) + (1 − 2 sin2 (x)) sin(x)
= 3 sin(x) − 4 sin3 (x).
 46

Therefore, at and near 0, we have

" ∞ ∞
#
1 1 X (−1)n x2n+1 X (−1)n (3x)2n+1
sin3 (x) = (3 sin(x) − sin(3x)) = 3 −
4 4 n=0
(2n + 1)! n=0
(2n + 1)!
∞
3 X (−1)n (1 − 32n )x2n+1
=
4 n=0 (2n + 1)!
∞
3 X (−1)n (1 − 32n )x2n+1
= .
4 n=1 (2n + 1)!

ii. At and near 0, we have

∞
!
2 1 d 1 d 3 X (−1)n (1 − 32n )x2n+1
sin3 (x) =


sin (x) cos(x) =
3 dx 3 dx 4 n=1 (2n + 1)!
∞
1 3 X d (−1)n (1 − 32n )x2n+1
 
= ·
3 4 n=1 dx (2n + 1)!
∞
1 X (−1)n (1 − 32n )x2n
=
4 n=1 (2n)!
∞
x2 X (−1)n−1 (32n − 1)x2n−2
=
4 n=1 (2n)!
∞
x2 X (−1)n (32n+2 − 1)x2n
=
4 n=0 (2n + 2)!

(c) For any x ∈ R, we have

∞ ∞
cosh(2x) + 1 1 1 1 1 X (2x)2n 1 X 22n−1 2n
cosh2 (x) = = + cosh(2x) = + = + x
2 2 2 2 2 n=0 (2n)! 2 n=0 (2n)!
∞
X 22n−1 2n
= 1+ x
n=1
(2n)!

and

1 + cos(2x) 1 1
cos2 (x) = = + cos(2x)
2 2 2
∞
1 1 X (−1)n (2x)2n
= +
2 2 n=0 (2n)!
∞
1 X (−1)n 22n−1 2n
= + x
2 n=0 (2n)!
∞
X (−1)n 22n−1 2n
= 1+ x .
n=1
(2n)!
 47

Therefore, at and near 0, we have

∞ ∞
X 22n−1 2n X (−1)n 22n−1 2n
cosh2 (x) − cos2 (x) = x − x
n=1
(2n)! n=1
(2n)!
∞  2n−1
(−1)n 22n−1 2n

X 2 2n
= x − x
n=1
(2n)! (2n)!
∞ 
1 − (−1)n 22n
X 
= · x2n
n=1
2 (2n)!
∞ 
1 − (−1)n+1 22(n+1)
X 
= · x2(n+1)
n=0
2 [2(n + 1)]!
∞ 
1 + (−1)n 22n+2
X 
= · x2n+2
n=0
2 (2n + 2)!
∞
X 24n+2 4n+2
= x
n=0
(4n + 2)!
∞ ∞
X 24n+2 X 42n+1 4(2n+1)
cosh2 (x2 ) − cos2 (x2 ) = (x2 )4n+2 = x .
n=0
(4n + 2)! n=0
(4n + 2)!

Exercise 7. (Level 2/Level 3)

∞
1 X
Here you may take for granted that = (−1)n xn whenever |x| < 1, as you have in school mathematics.
1 + x n=0

∞ ∞
X (−1)m−1 m X 1
(a) Show that ln(1 + x) = x and ln(1 − x) = − xm at and near 0.
m=1
m m=1
m

∞
1 1+x X x2n+1 1 1+x
(b) Show that ln( ) = at and near 0. ( ln( ) is the hyperbolic arctangent function in
2 1−x n=0
2n + 1 2 1−x
disguise.)

Remark. Actually all three identities hold on the interval (−1, 1), and not only at and near 0.

d 1
Solution. (a) Note that (ln(1 + x)) = for any x ∈ (−1, +∞).
dx 1+x
∞ ∞ ∞
1 X X (−1)n n+1 X (−1)m−1 m
At and near 0, we have = (−1)n xn . Then ln(1 + x) = ln(1 + 0) + x = x
1 + x n=0 n=0
n+1 m=1
m
.
∞ ∞
X (−1)m−1 m m
X 1
Therefore, we also have ln(1 − x) = · (−1) x = − xm .
m=1
m m=1
m
 48

(b) At and near 0, we have

1 1+x 1
ln( ) = (ln(1 + x) − ln(1 − x))
2 1−x 2
" ∞ ∞
#
1 X (−1)m−1 m X 1 m
= x − − x
2 m=1 m m=1
m
∞
X (−1)m−1 + 1 m
= x
m=1
2m
∞
X x2n+1
= .
n=0
2n + 1

Exercise 8. (Level 3)
√ √
1 + x2 − 1 − x2
Let f : (−1, 1) −→ R be the function defined by f (x) = √ for any x ∈ (−1, 1).
2 1 − x4
∞  
X (Ak + B)!
(a) Show that f (x) = xP k+Q at and near 0.
C Dk+E [(F k + G)!]H
k=0
Here A, B, C, D, E, F, G, H, P, Q are positive integers whose values you have to determine explicitly.
Remark. Actually the identity holds on (−1, 1).

(b) Hence, or otherwise, find f (n) (0) for any n ∈ N.

√ √
1+x− 1−x
Solution. (a) Define the function g : (−1, 1) −→ R by g(x) = √ for any x ∈ (−1, 1). Note that
2 1 − x2
f (x) = g(x2 ) for any x ∈ (−1, 1).
At and near 0, we have
 49

g(x)
√ √ √ √
1+x− 1−x 1+x− 1−x 1 1
= √ = √ √ = √ − √
2 1−x 2 2 1+x 1−x 2 1−x 2 1+x
∞ ∞
1 X − 21 1 X − 21
   
= (−1)n xn − xn
2 n 2 n
n=0 n=0
∞ 
− 12 (−1)n − 1 n
X  
= x
n 2
n=0
∞
− 12
X  
= − x2k+1
2k + 1
k=0
∞
X (− 12 )(− 12 −1)(− 21 −2)···[− 12 −(2k−2)][− 12 −(2k−1)](− 12 −2k)
= − (2k+1)! x2k+1
k=0
∞
X (−1)2k+1 [1 · 3 · 5 · ... · (4k − 3)(4k − 1)(4k + 1)] 2k+1
= − x
22k+1 [(2k + 1)!]
k=0
∞  
X 1 · 3 · 5 · ... · (4k − 3)(4k − 1)(4k + 1)
= x2k+1
22k+1 [(2k + 1)!]
k=0
∞  
X (4k + 2)!
= x2k+1
22k+1 [(2k + 1)!][2 · 4 · 6 · ... · (4k − 2)(4k)(4k + 2)]
k=0
∞  
X (4k + 2)!
= 4k+2
x2k+1 .
2 [(2k + 1)!]2
k=0

Then, at and near 0, we have

∞  
2
X (4k + 2)!
f (x) = g(x ) = x4k+2
24k+2 [(2k + 1)!]2
k=0

(b) Note that

∞ 
[(4k + 2)!]2
X 
f (x) = g(x2 ) = 4k+2
x4k+2 .
2 [(2k + 1)!]2 [(4k + 2)!]
k=0

For any n ∈ N, we have

[(4k + 2)!]2


 if n = 4k + 2 for some k ∈ N
24k+2 [(2k + 1)!]2

f (n) (0) =


0 otherwise


Exercise 9. (Level 3)

(a) Denote by f (x) the function sinh2 (x2 ) − sin2 (x2 ) on R. Compute f (n) (0) for any n ∈ N.
 50

(b) Denote by g(x) the function cos(9x4 ) cos(5x4 ) on R. Compute g (n) (0) for any n ∈ N.

Solution. (a) At and near 0, we have

cosh(2x) − 1
sinh2 (x) =
2
1 1
= − + cosh(2x)
2 2
∞
1 1 X (2x)2n
= − +
2 2 n=0 (2n)!
∞
1 X 22n−1 2n
= − + x
2 n=0 (2n)!
∞
X 22n−1 2n
= x
n=1
(2n)!

and

1 − cos(2x)
sin2 (x) =
2
1 1
= − cos(2x)
2 2
∞
1 1 X (−1)n (2x)2n
= −
2 2 n=0 (2n)!
∞
1 X (−1)n−1 22n−1 2n
= + x
2 n=0 (2n)!
∞
X (−1)n−1 22n−1 2n
= x
n=1
(2n)!

Therefore, at and near 0, we have

∞ ∞
X 22n−1 2 2n X 22n−1 4n
sinh2 (x2 ) = (x ) = x
n=1
(2n)! n=0
(2n)!
∞ ∞
X (−1)n−1 22n−1 2 2n X (−1)n−1 22n−1 4n
sin2 (x2 ) = (x ) = x
n=1
(2n)! n=1
(2n)!
 51

Hence
∞ ∞
X 22n−1 4n X (−1)n−1 22n−1 4n
sinh2 (x2 ) − sin2 (x2 ) = x − x
n=1
(2n)! n=1
(2n)!
∞  2n−1
(−1)n−1 22n−1 4n

X 2
= x4n − x
n=1
(2n)! (2n)!
∞ 
1 + (−1)n 22n
X 
= · x4n
n=1
2 (2n)!
∞
X 24n 8n
= x
n=1
(4n)!
∞
24n [(8n)!]
 
X 1
= x8n ,
n=1
(8n)! (4n)!

∞
24n [(8n)!]
 
X 1
x8n is the Taylor series of the (analytic) function f (x) about the point 0. Therefore
n=1
(8n)! (4n)!

 4k
2 [(8k)!]
if n = 8k for some k ∈ N\{0}


(4k)!

f (n) (0) = .


0 otherwise


1
cos(14x4 ) + cos(4x4 ) . Therefore, at and near 0, we


(b) For any x ∈ R, we have g(x) = cos(9x4 ) cos(5x4 ) =
2
have
"∞ ∞
#
1 X (−1)n (14x4 )2n X (−1)n (4x4 )2n
g(x) = +
2 n=0 (2n)! n=0
(2n)!
∞
X (−1)n (142n + 42n ) 8n
= x
n=0
2[(2n)!]
∞
22n−1 (−1)n (72n + 22n )[(8n)!]
 
X 1
= x8n
n=0
(8n)! [(2n)!]

Hence
 2m−1
2 (−1)m (72m + 22m )[(8m)!]

 if n = 8m for some m ∈ N
[(2m)!]

g (n) (0) = .


0 otherwise


Exercise 10. (Level 3)

∞
1 X
(a) Show that = (−1)n x2n at and near 0.
1 + x2 n=0
 52

∞
X (−1)n x2n+1
(b) i. Show that arctan(x) = at and near 0.
n=0
2n + 1
∞
X (−1)n x2n+2
ii. Show that ln(1 + x2 ) = at and near 0.
n=0
n+1
∞
1 X F n xGn
iii. Hence show that ln(1 + x2 ) − x arctan(x) = at and near 0.
2 n=1
Hn(Hn − 1)
Here F, G, H are integers whose values you have to determine.

Remark. Actually all four identities hold on the interval (−1, 1).
∞ ∞
1 X
n 2 n
X
Solution. (a) At and near 0, = (−1) (x ) = (−1)n x2n .
1 + x2 n=0 n=0

1
(b) (a) Note that arctan0 (x) = for any x ∈ R.
1 + x2
∞ ∞
X (−1)n x2n+1 X (−1)n x2n+1
Then, at and near 0, we have arctan(x) = arctan(0) + = .
n=0
2n + 1 n=0
2n + 1

d 2x
ln(1 + x2 ) =


(b) Note that for any x ∈ R.
dx 1 + x2
∞ ∞
2x X
n 2 n
X
At and near 0, we have = 2x · (−1) (x ) = 2(−1)n x2n+1 .
1 + x2 n=0 n=0
∞ ∞
X 2(−1)n x2n+2 X (−1)n x2n+2
Then, ln(1 + x2 ) = ln(1 + 02 ) + = .
n=0
2n + 2 n=0
n+1

(c) At and near 0, we have

∞ ∞
1 1 X (−1)n x2n+2 X (−1)n x2n+1
ln(1 + x2 ) − x arctan(x) = · −x·
2 2 n=0 n+1 n=0
2n + 1
∞ ∞
X (−1)n x2n+2 X (−1)n x2n+2
= −
n=0
2n + 2 n=0
2n + 1
∞ 
(−1)n (−1)n
X 
= − x2n+2
n=0
2n + 2 2n + 1
∞
X (−1)n+1
= x2n+2
n=0
(2n + 2)(2n + 1)
∞
X (−1)n x2n
= .
n=1
2n(2n − 1)

Exercise 11. (Level 3)

x2 + x 1
Let f, h : R\{1} −→ R be the functions defined by f (x) = , h(x) = for any x ∈ R\{1}.
(1 − x)3 1−x
 53

(a) Verify that f (x) = h(x) − 3h0 (x) + h00 (x) for any x ∈ R\{1}.
∞
X
(b) Show that f (x) = n2 xn at and near 0.
n=0

∞
X n(n − 3)
(c) Show that f (x) = n+3
(x + 1)n at and near −1.
n=0
2

1 3 2 x2 + x
Solution. (a) For any x ∈ R\{1}, h(x) − 3h0 (x) + h00 (x) = − + = · · · = = f (x).
1 − x (1 − x)2 (1 − x)3 (1 − x)3
∞
X
(b) At and near 0, we have h(x) = xn . Hence
n=0

f (x) = h(x) − 3h0 (x) + h00 (x)

∞
X ∞
X ∞
X
= xn − 3 · nxn−1 + n(n − 1)xn−2
n=0 n=1 n=2
∞
X ∞
X ∞
X
= xn − 3 · (n + 1)xn + (n + 2)(n + 1)xn
n=0 n=0 n=0
∞
X
= [1 − 3(n + 1) + (n + 1)(n + 2)]xn
n=0
∞
X
= n2 xn .
n=0

1 1/2
(c) Note that h(x) = = for any x ∈ R\{1}.
1−x 1 − (x + 1)/2
∞ n X ∞
(x + 1)n

X 1 x+1
At and near −1, we have h(x) = = . Hence
n=0
2 2 n=0
2n+1

f (x) = h(x) − 3h0 (x) + h00 (x)

∞ ∞ ∞
X (x + 1)n X n n−1
X n(n − 1)
= n+1
− 3 · n+1
(x + 1) + n+1
(x + 1)n−2
n=0
2 n=1
2 n=2
2
∞ ∞ ∞
X (x + 1)n X n+1 n
X (n + 2)(n + 1)
= n+1
− 3 · n+2
(x + 1) + n+3
(x + 1)n
n=0
2 n=0
2 n=0
2
∞
X 4 − 6(n + 1) + (n + 1)(n + 2)
= n+3
(x + 1)n
n=0
2
∞
X n(n − 3)
= n+3
(x + 1)n .
n=0
2
 54

Exercise 12. (Level 3)

1 1 1
Consider the rational functions f (x) = , g(x) = , h(x) = .
(x − 1)2 (x − 2) 1−x 2−x

(a) Verify that f (x) = g(x) − g 0 (x) − h(x) for any x ∈ R\{1, 2}.

(b) Find the Taylor series of f about the point 0. Hence, or otherwise, find f (n) (0) for any n ∈ N.

(c) Find the Taylor series of f about the point 3. Hence, or otherwise, find f (n) (3) for any n ∈ N.

1 1 1 1
Solution. (a) For any x ∈ R\{1, 2}, g(x) − g 0 (x) − h(x) = − 2
− = ··· = =
1 − x (1 − x) 2−x (x − 1)2 (x − 2)
f (x).
∞ ∞ ∞
X X 1/2 1 X xn
(b) At and near 0, we have g(x) = xn , g 0 (x) = nxn−1 , h(x) = = · . Then
n=0 n=1
1 − x/2 2 n=0 2n

f (x) = g(x) − g 0 (x) − h(x)

∞ ∞ ∞
X X 1 X xn
= xn − nxn−1 − ·
n=0 n=1
2 n=0 2n
∞ ∞ ∞
X X X xn
= xn − (n + 1)xn −
n=0 n=0 n=0
2n+1
∞  
X 1
= 1 − (n + 1) − xn
n=0
2n+1
∞  
X 1
= −n − xn
n=0
2n+1

 
(n) 1
Therefore f (0) = −n − · (n!) for any n ∈ N.
2n+1

(c) At and near 3, we have

∞
1 −1/2 1 X (−1)n (x − 3)n
g(x) = = =− · ,
−2 − (x − 3) 1 + (x − 3)/2 2 n=0 2n
∞ ∞
0 1 X (−1)n n(x − 3)n−1 1 X (−1)n+1 (n + 1)(x − 3)n
g (x) = − · =− · ,
2 n=1 2n 2 n=0 2n+1
∞
1 −1 X
h(x) = = =− (−1)n (x − 3)n .
−1 − (x − 3) 1 + (x − 3) n=0
 55

Then

f (x)
= g(x) − g 0 (x) − h(x)
∞ ∞
" #
1 X (−1)n (x − 3)n 1 X (−1)n+1 (n + 1)(x − 3)n
= − · − − ·
2 n=0 2n 2 n=0 2n+1
" ∞ #
X
n n
− − (−1) (x − 3)
n=0
∞ 
1 (−1)n 1 (−1)n+1 (n + 1)
X 
= − · n
+ · n+1
+ (−1) (x − 3)n
n

n=0
2 2 2 2
∞  
X
n n+3
= (−1) 1 − n+2 (x − 3)n .
n=0
2

 
n+3
Therefore f (n) (3) = (−1)n 1 − n+2 · (n!) for any n ∈ N.
2

Exercise 13. (Level 3) Let a be a non-zero constant, and N be a positive integer greater than 2. Denote by
N
f (x) the function eax on R.
∞
X ak
(a) Show that f (x) = xN k at and near 0.
k!
k=0

Remark. Actually this identity holds on the whole of R.

(b) Compute f (n) (0) for any n ∈ N.

∞ ∞
N
X (axN )k X ak
Solution. (a) At and near 0, we have f (x) = eax = = xkN .
k! k!
k=0 k=0

∞
X 1 ak [(kN )!] kN
(b) Note that f (x) = · x at and near 0.
(kN )! (k!)
k=0
 k
a [(kN )!]
if n = N k for some k ∈ N



Then f (n) (0) = k! .


 0 otherwise

Exercise 14. (Level 3)

(a) Find the Taylor series of (1 − x)−1/2 centered at 0.

(b) Find d
dx sin−1 x.

(c) Find the Taylor series of sin−1 x centered at 0 with order 6.

 56

Solution. (a) The Taylor series of (1 − x)−1/2 is

∞
X − 12 (− 12 − 1) · · · (− 21 − (n − 1))
(−x)n
n=0
n!

(b) d
dx sin−1 x = √ 1
1−x2

(c) First we find the Taylor series for d

dx sin−1 x with order 5, which is

1 3
1 + x2 + x4 .
2 8

Since sin−1 x is the antiderivative of d

dx sin−1 x, we have the Taylor series of sin−1 x with order 6 is

1 3
x + x3 + x5 .
6 40

Exercise 15. (Level 3/Level 4)

1
Find the Taylor series of 1+ln x centered at x = 1 with order 3.
Recall the Taylor series of 1 + ln x centered at x = 1 is

(x − 1)2 (x − 1)3 (x − 1)4

1 + (x − 1) − + − + ···
2 3 4

1
Solution. Suppose the Taylor series of ln x centered at x = 1 with order 3 is

a0 + a1 (x − 1) + a2 (x − 1)2 + a3 (x − 1)3 + · · ·

Then
a0 + a1 (x − 1) + a2 (x − 1)2 + a3 (x − 1)3 + · · ·

(x − 1)2 (x − 1)3
 
× 1 + (x − 1) − + − · · · = 1.
2 3

Hence
a0 = 1
a1 + a0 = 0
a0
a2 + a1 − =0
2
a1 a0
a3 + a2 − + = 0.
2 3
So
3 7
a0 = 1, a1 = −1, a2 = , a3 = − .
2 3
 57

3 Intermediate forms and L’Hôpital’s rule

Exercise 1. (Level 2)
Evaluate the following limits

tan x − x
(a) lim+
x→0 x3

x + ln x
(b) lim
x→+∞ x ln x
 
1 1
(c) lim+ −
x→0 x tan−1 x

1
Solution. (a)
3

(b) 0

(c) 0

Exercise 2. (Level 2)
Use L’Hopital’s rule to evaluate the following limits.

1 − x cot x e − (1 + x) x
1
(a) lim (g) lim
x→0 x sin x x→0 x
sinh x − sin x
(b) lim
x→0 x(cosh x − cos x) 1
(h) lim x 1−x
x→1
ln cos 2x
(c) lim
x→0ln cos x
ln(2x3 − 5x2 + 3)
  (i) lim
1 1 x→+∞ ln(4x2 + x − 7)
(d) lim − x
x→0 x e −1
  π 
1 1 (j) lim x − tan−1 x
(e) lim − x→+∞ 2
x→1 ln x x−1
ex − x − 1 1
(f) lim (k) lim (ex + x) x
x→0 cosh x − 1 x→+∞

1 − x cot x sin x − x cos x

Solution. (a) lim = lim
x→0 x sin x x→0 x sin2 x
cos x − (cos x − x sin x)
= lim
x→0 sin2 x + 2x sin x cos x
x
= lim
x→0 sin x + 2x cos x
x 1
= lim =
1→0 cos x + 2(cos x − sin x) 3
 58

sinh x − sin x cosh x − cos x

(b) lim = lim
x→0 x(cosh x − cos x) x→0 cosh x − cos x + x(sinh x + sin x)

sinh x + sin x
= lim
x→0 x(cosh x + cos x) + 2(sinh x + sin x)

cosh x + cos x
= lim
x→0 3(cosh x + cos x) + x(sinh x − sin x)
1
= 3

ln cos 2x −2 tan 2x
(c) lim = lim
x→0 ln cos x x→0 − tan x
4 cos2 x
= lim =4
x→0 cos2 2x

ex − 1 − x
 
1 1
(d) lim − x = lim
x→0 x e −1 x→0 x(ex − 1)
x
e −1
= lim x
x→0 xe + ex − 1
ex
= lim x
x→0 xe + 2ex

= 12
 
1 1 x − 1 − ln x
(e) lim − = lim
x→1 ln x x−1 x→1 ln x(x − 1)

x−1
= lim
x→1 x − 1 + x ln x
1
= lim
x→1 2 + ln x

= 12

ex − x − 1 ex − 1
(f) lim = lim
x→0 cosh x − 1 x→0 sinh x
x
e
= lim
x→0 cosh x
=1
1 1
e − (1 + x) x e − e x ln(1+x)
(g) lim = lim
x→0 x x→0 x
1 1 ln(x+1)
= lim −e x ln(1+x) ( x(x+1) − x2 )
x→0
x − (x + 1) ln(x + 1)
= −e lim
x→0 x2 (x + 1)
− ln(x + 1)
= −e lim
x→0 3x2 + 2x
−1
= −e lim
x→0 (x + 1)(6x + 2)

= 2e
1 ln x
(h) lim x 1−x = lim e 1−x
x→1 x→1
ln x
lim
=e x→1 1−x
 59

1
lim − x
= ex→1
= 1e

ln(2x3 − 5x2 + 3) (6x2 − 10x)(4x2 + x − 7)

(i) lim = lim
x→+∞ ln(4x2 + x − 7) x→+∞ (2x3 − 5x2 + 3)(8x + 1)

(6 − 10/x)(4 + 1/x − 7/x2 )

= lim
x→+∞ (2 − 5/x + 3/x3 )(8 + 1/x)
3
= 2

π
π  − tan−1 x
(j) lim x − tan −1
x = lim 2
x→+∞ 2 x→+∞ 1
x
x2
= lim
x→+∞ 1 + x2
=1
ln(ex +x)
1 lim
lim (ex + x) x = ex→+∞
x
(k)
x→+∞
ex +1
lim x +x
= ex→+∞ e
=e

Exercise 3. (Level 2)
Use L’Hopital’s rule to evaluate the following limits.

sin 3x 1
 
1 1 (m) lim x 1−x
(a) lim (g) lim − x x→1
x→0 sin 5x x→0 x e −1

sin2 x
  ln(2x3 − 5x2 + 3)
(b) lim 1 1 (n) lim
x→0 1 − cos x
(h) lim − x→+∞ ln(4x2 + x − 7)
x→1 ln x x − 1
 
2 sin x − sin 2x ex − x − 1 1
(c) lim (i) lim (o) lim x sin
x→0 x − sin x x→0 cosh x − 1 x→+∞ x
1 − x cot x 1
π 
(d) lim e − (1 + x) x (p) lim x − tan−1 x
x→0 x sin x (j) lim x→+∞ 2
x→0 x
sinh x − sin x 
3

(e) lim 2x − 1 (q) lim x ln 1 +
x→0 x(cosh x − cos x) (k) lim x→+∞ x
x→0 x
ln cos 2x 1 1
(f) lim (l) lim+ x 1+ln x (r) lim (ex + x) x
x→0 ln cos x x→0 x→+∞

3
Solution. (a) 5 (f) 4 (k) ln 2
1 (l) e
(b) 2 (g) 2

1
(c) 6 (h) 1 (m) e
2
1 3
(d) 3 (i) 1 (n) 2

1 e
(e) 3 (j) 2 (o) 1
 60

(p) 1 (q) 3 (r) e

Exercise 4. (Level 2)
Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule. (Do not apply L’Hôpital’s
Rule blindly. In some situations, even though it is appropriate to apply L’Hôpital’s Rule, there may be a better
and faster road to the answer.)

ex−1 − x sinh(x) − x 1 + ln(x) − xx

(a) i. lim ii. lim iii. lim
x→1 (x − 1)2 x→0 x − sin(x) x→1 1 + ln(x) − x

ln(sin(3x)) ln(1 + sin(3x)) ln(cos(x3 ))

(b) i. lim+ iii. lim+ v. lim+
x→0 ln(sin(2x)) x→0 ln(1 + sin(2x)) x→0 ln(cos(x2 ))
ln(cos(3x)) ln(sin(x3 )) ln(1 + sin(x3 ))
ii. lim+ iv. lim+ vi. lim+
x→0 ln(cos(2x)) x→0 ln(sin(x2 )) x→0 ln(1 + sin(x2 ))

(c) lim+ sin(sin(x)) ln(arctan(e−1/x ))

x→0
   
1 1 1 cos(x)
(d) i. lim+ − ii. lim+ −
x→0 sin(x) x − x2 x→0 x2 x sin(x)
ex sin(x)
i. lim+ (x + sin(x))x

(e) ii. lim+ x 1−x 1
x→0 x→1 iii. lim+
x→0 sinh(x)

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.

ex−1 − x ex−1 − 1 ex−1 1

(a) i. lim 2
= lim = lim = .
x→1 (x − 1) (H) x→1 2(x − 1) (H) x→1 2 2
sinh(x) − x cosh(x) − 1 sinh(x) cosh(x)
ii. lim = lim = lim = lim = 1.
x→0 x − sin(x) (H) x→0 1 − cos(x) (H) x→0 sin(x) (H) x→0 cos(x)

iii.

1 + ln(x) − xx 1/x − xx (ln(x) + 1)

lim = lim
x→1 1 + ln(x) − x (H) x→1 1/x − 1
−1/x2 − xx (ln(x) + 1)2 − xx /x
= lim
(H) x→1 −1/x2
= lim [1 + xx (ln(x) + 1)2 x2 + xx · x] = 3.
(H) x→1

(b) i.

ln(sin(3x)) (3 cos(3x)/ sin(3x))

lim+ = lim+
x→0 ln(sin(2x)) (H) x→0 (2 cos(2x)/ sin(2x))
 
cos(3x) 3x sin(2x)
= lim · ·
x→0+ cos(2x) sin(3x) 2x
= 1.
 61

ii.

ln(cos(3x)) (−3 sin(3x)/ cos(3x))

lim = lim
x→0+ ln(cos(2x)) (H) x→0+ (−2 sin(2x)/ cos(2x))
 
9 cos(2x) 2x sin(3x)
= lim+ · · ·
x→0 4 cos(3x) sin(2x) 3x
9
= .
4

iii.

ln(1 + sin(3x)) [3 cos(3x)/(1 + sin(3x))]

lim = lim
x→0+ ln(1 + sin(2x)) (H) x→0+ [2 cos(2x)/(1 + sin(2x)]
 
3 cos(3x) 1 + sin(2x)
= lim+ · ·
x→0 2 cos(2x) 1 + sin(3x)
3
= .
2

iv.

ln(sin(x3 )) (3x2 cos(x3 )/ sin(x3 ))

lim+ = lim+
x→0 ln(sin(x2 )) (H) x→0 (2x cos(x2 )/ sin(x2 ))
3 cos(x3 ) sin(x2 ) x3
 
= lim · · ·
x→0+ 2 cos(x2 ) x2 sin(x3 )
3
= .
2

v.

ln(cos(x3 )) (−3x2 sin(x3 )/ cos(x3 ))

lim+ = lim+
x→0 ln(cos(x2 )) (H) x→0 (−2x sin(x2 )/ cos(x2 ))
 2
3x cos(x2 ) sin(x3 ) x2

= lim · · ·
x→0+ 2 cos(x3 ) x3 sin(x2 )
= 0.

vi.

ln(1 + sin(x3 )) [3x2 cos(x3 )/(1 + sin(x3 ))]

lim+ = lim
x→0 ln(1 + sin(x2 )) (H) x→0+ [2x cos(x2 )/(1 + sin(x2 )]

cos(x3 ) 1 + sin(x2 )
 
3
= lim x· ·
x→0+ 2 cos(x2 ) 1 + sin(x3 )
= 0.
 62

(c)

lim sin(sin(x)) ln(arctan(e−1/x ))

x→0+

ln(arctan(e−1/x ))
= lim+
x→0 csc(sin(x))
(1/ arctan(e−1/x )) · [1/(1 + e−2/x )] · e−1/x · (1/x2 )
= lim
(H) x→0+ − csc(sin(x)) cot(sin(x)) · cos(x)
e−1/x sin2 (sin(x))
= lim+ −
x→0 x2 (1 + e−2/x ) arctan(e−1/x ) cos(sin(x)) cos(x)
" 2
e−1/x

sin(sin(x))
= lim − ·
x→0+ arctan(e−1/x ) sin(x)
 2 #
sin(x) 1
· ·
x (1 + e−2/x ) cos(sin(x)) cos(x)
= −1

(d) i.

x − x2 − sin(x)
 
1 1
lim+ − = lim+
x→0 sin(x) x − x2 x→0 (x − x2 ) sin(x)
1 − 2x − cos(x)
= lim+
(H) x→0 (1 − 2x) sin(x) + (x − x2 ) cos(x)
−2 + sin(x)
= lim+ = −1
(H) x→0 −(2 + x − x2 ) sin(x) + 2(1 − 2x) cos(x)

ii.
 
1 cos(x) sin(x) − x cos(x)
lim+ 2
− = lim+
x→0 x x sin(x) x→0 x2 sin(x)
x sin(x)
= lim+
(H) x→0 2x sin(x) + x2 cos(x)
sin(x)
= lim
x→0+ 2 sin(x) + x cos(x)
cos(x)
= lim
(H) x→0+ 3 cos(x) − x sin(x)
1
=
3

(e) i.

ln(x + sin(x))
lim x ln(x + sin(x)) = lim
x→0+ x→0+ x−1
[(1 + cos(x))/(x + sin(x))]
= lim+
(H) x→0 −x−2
x(1 + cos(x))
= lim − = 0.
x→0+ 1 + x−1 sin(x)
 63

Then lim (x + sin(x))x = lim ex ln(x+sin(x)) = e0 = 1.

x→0+ x→0+

ex ex ln(x) ex ln(x) + ex (1/x)

 
ii. lim · ln(x) = lim − = lim − = −e
x→1+ 1−x x→1+ x − 1 (H) x→1+ 1
ex ex
Then lim+ x 1−x = lim+ e 1−x ·ln(x) = e−e .
x→1 x→1

iii.
 
1 ln (sinh(x))
lim+ sin(x) ln = lim+ −
x→0 sinh(x) x→0 csc(x)
(cosh(x)/ sinh(x))
= lim −
(H) x→0+ − csc(x) cot(x)
 
cosh(x) sin(x) x
= lim+ · · · sin(x) = 0
x→0 cos(x) x sinh(x)

 sin(x)
1
Then lim+ = lim+ esin(x) ln(1/ sinh(x)) = e0 = 1.
x→0 sinh(x) x→0

Exercise 5. (Level 2/level 3)

1
Let f (x) = x2 sin( ) for x 6= 0 and let g(x) = sin x.
x
f (x) f 0 (x)
Show that lim = 0 but lim 0 does not exist.
x→0 g(x) x→0 g (x)

f (x)
(Remark: L’Hôpital’s rule cannot be applied to find lim = 0 in this case.)
x→0 g(x)

x2 sin( x1 )  x  
f (x) 1
Solution. lim = lim = lim x sin( ) = (1)(0) = 0.
x→0 g(x) x→0 sin x x→0 sin x x
1
( Remark: lim x sin( ) = 0 can be proved by using the sandwich theorem.)
x→0 x
1 1
However, for x 6= 0, we have f 0 (x) = 2x sin( ) − cos( ) and g 0 (x) = cos x. Therefore,
x x

f 0 (x) 2x sin( x1 ) − cos( x1 )

= .
g 0 (x) cos x

1 1
Consider two sequences {an } and {bn } defined by an = and bn = for all natural numbers n. Then
2nπ (2n + 1)π
we have lim an = lim bn = 0. However,
n→∞ n→∞

− cos(2nπ)
lim f (an ) = lim 1 = −1,
n→∞ n→∞ cos( 2nπ )

and
− cos((2n + 1)π)
lim f (bn ) = lim 1 = 1.
n→∞ n→∞ cos( (2n+1)π )
 64

f 0 (x)
Therefore, lim does not exist.
x→0 g 0 (x)

Exercise 6. (Level 2/level 3)

xn
Prove that lim x = 0 for all natural numbers n.
x→+∞ e
p(x)
Hence show that lim = 0 for any polynomial p(x).
x→+∞ ex

xn
Solution. Let P (n) be the statement that ” lim = 0”.
x→+∞ ex

x (x)0 1
• When n = 1, By L’Hôpital’s rule, lim = lim = lim x = 0.
x→+∞ ex x→+∞ (ex )0 x→+∞ e

Therefore, P (1) is true.

xn
• Suppose P (n) is true for some natural number n, i.e. lim = 0.
x→+∞ ex

Then,

xn+1 ∞
lim ( )
x→+∞ ex ∞
(n + 1)xn
= lim
x→+∞ ex
= 0 (By assumption)

Therefore, P (n + 1) is true.

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

Let P (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 . By using the above result,

p(x) an xn + an−1 xn−1 + · · · + a1 x + a0

lim = lim
x→+∞ ex x→+∞ ex
xn
     
x 1
= an lim + · · · + a1 lim + a0 lim
x→+∞ ex x→+∞ ex x→+∞ ex

= 0 + ··· + 0 + 0
= 0

Exercise 7. (Level 2/level 3)

ln x
Show that for any real number α > 0, lim = 0.
x→+∞ xα

Solution. Let f (x) = ln x and g(x) = xα . Because limx→+∞ ln x = +∞ and limx→+∞ xα = +∞, g 0 (x) = αxα−1 .
The function g 0 (x) is not equal to 0 for x very large. Then

ln x f 0 (x)
lim = lim
x→+∞ xα x→+∞ g 0 (x)
 65

provided the limit of the right hand side exists. The RHS is

x−1 1
lim = = 0.
x→+∞ αxα−1 αxα

Exercise 8. (Level 2/level 3)

Apply L’Hôpital’s Rule to evaluate the limits below:
√ √
sin(3x) 2+x− 2−x ex − 1 − sin(x)
(a) lim+ (f) lim+ √ √ (k) lim
x→0 sin(2x) x→0 3+x− 3−x x→0 x2
√ √ x2 + 2x + 2 − 2ex
x2 − 3x + 2 3
2+x− 32−x
(b) lim− 2 (g) lim+ √ √ (l) lim
x→2 x − 5x + 6 x→0
3
3+x− 33−x x→0 x3

(ex − 1) sin(x) arcsin(x) sin(x) − x + x3 /6 + x5 /60

(c) lim+ . (h) lim (m) lim
x→0 1 − cos(x) x→0 arcsinh(x) x→0 sinh(x) − x − x3 /6 − x5 /60

(ex − e−x )2 xx − x sin(e−x )

(d) lim (i) lim (n) lim √
x→0− 1 − cos(2x) x→1 ln(x) − x + 1 x→+∞ sinh(1/ x)

√ √
sin(x) sin( x − 4 x) sin(x2 ) − x2
(e) lim (j) lim+ √ √ (o) lim
x→0 sinh(x) x→0 x+ 4x x→0 sinh(x2 ) − x2

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.

sin(3x) 3 cos(3x) 3
(a) lim+ = lim = .
x→0 sin(2x) (H) x→0+ 2 cos(2x) 2

x2 − 3x + 2 2x − 3
(b) lim− = lim = −1.
x→2 x2 − 5x + 6 (H) x→2− 2x − 5

(c)

(ex − 1) sin(x) ex sin(x) + (ex − 1) cos(x)

lim+ = lim+
x→0 1 − cos(x) (H) x→0 sin(x)
ex sin(x) + 2ex cos(x) − (ex − 1) sin(x)
= lim+
(H) x→0 cos(x)
= 2.

(d)

(ex − e−x )2 2(ex − e−x )(ex + e−x )

lim− = lim−
x→0 1 − cos(2x) (H) x→0 2 sin(2x)
e2x − e−2x
= lim−
x→0 sin(2x)
2e2x + 2e−2x
= lim
(H) x→0− 2 cos(2x)
= 2.
 66

sin(x) cos(x)
(e) lim = lim = 1.
x→0 sinh(x) (H) x→0 cosh(x)

1 1
√ √ √ + √ r
2+x− 2−x 2 2+x 2 2−x 3
(f) lim+ √ √ = lim = .
x→0 3 + x − 3 − x (H) x→0+ √ 1 + √
1 2
2 3+x 2 3−x

1 1
√
3
√
3
√ + √ r
2+x− 2−x 3
3( 2 + x) 2 3( 2 − x)2
3
3 9
(g) lim √ √ = lim = .
x→0+
3
3 + x − 3 3 − x (H) x→0+ √ 1 + √
1 4
3( 3 3 + x)2 3( 3 3 − x)2
√
arcsin(x) 1/ 1 − x2
(h) lim = lim √ = 1.
x→0 arcsinh(x) (H) x→0 1/ 1 + x2

xx − x xx (ln(x) + 1) − 1 xx (ln(x) + 1)2 + xx (1/x)

(i) lim = lim = lim = −2.
x→1 ln(x) − x + 1 (H) x→1 1/x − 1 (H) x→1 −1/x2

(j)

1 1 √ √
√ √ [ √ − √ ] cos( x − 4 x)
sin( x − 4 x) 2 x 4( x) 4 3
lim+ √ √ = lim
x→0 x+ 4x (H) x→0+ 1 1
√ + √
2 x 4( x)3 4

√ √ √
(2 4 x − 1) cos( x − 4 x)
= lim+ √
x→0 24x+1
= −1.

ex − 1 − sin(x) ex − cos(x) ex + sin(x) 1

(k) lim 2
= lim = lim = .
x→0 x (H) x→0 2x (H) x→0 2 2

x2 + 2x + 2 − 2ex 2x + 2 − 2ex 2 − 2ex −(ex − 1) 1

(l) lim = lim = lim = lim =− .
x→0 x3 (H) x→0 3x2 (H) x→0 6x x→0 3x 3

(m)

sin(x) − x + x3 /6 + x5 /60 cos(x) − 1 + x2 /2 + x4 /12

lim = lim
x→0 sinh(x) − x − x3 /6 − x5 /60 (H) x→0 cosh(x) − 1 − x2 /2 − x4 /12

− sin(x) + x + x3 /3
= lim
(H) x→0 sinh(x) − x − x3 /3

− cos(x) + 1 + x2
= lim
(H) x→0 cosh(x) − 1 − x2

sin(x) + 2x
= lim
(H) x→0 sinh(x) − 2x

cos(x) + 2
= lim = −3.
(H) x→0 cosh(x) − 2

3
sin(e−x ) −e−x cos(e−x ) 2x 2 e−x cos(e−x )
(n) lim √ = lim √ 3 √ = lim √ = 0.
x→+∞ sinh(1/ x) (H) x→+∞ [−1/(2( x) )] cosh(1/ x) x→+∞ cosh(1/ x)
 67

(o)

sin(x2 ) − x2 2x cos(x2 ) − 2x
lim = lim
x→0 sinh(x2 ) − x2 (H) x→0 2x cosh(x2 ) − 2x

cos(x2 ) − 1
= lim
x→0 cosh(x2 ) − 1

−2x sin(x2 )
= lim
(H) x→0 2x sinh(x2 )

− sin(x2 )
= lim
x→0 sinh(x2 )

−2x cos(x2 )
= lim
(H) x→0 2x cosh(x2 )

− cos(x2 )
= lim = −1
x→0 cosh(x2 )

Exercise 9. (Level 2/Level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.

x5 + x4 + x2 + x + 1 ln(ln(ln(x))) 2
ln(e2x + x3 )
(a) lim (c) lim (e) lim
x→+∞ x5 − x4 − x − 1 x→+∞ ln(ln(x)) x→+∞ x2 + x

ln(x + 1) (ln(x))4 ln(x + 1) ln(x − 1)

(b) lim (d) lim (f) lim
x→+∞ ln(x) x→+∞ x3 x→+∞ (ln(x))2

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.

(a)

x5 + x4 + x2 + x + 1 5x4 + 4x3 + 2x + 1
lim = lim
x→+∞ x5 − x4 − x − 1 (H) x→+∞ 5x4 − 4x3 − 1
20x3 + 12x2 + 2
= lim
(H) x→+∞ 20x3 − 12x2
60x2 + 24x
= lim
(H) x→+∞ 60x2 − 24x

5x + 2
= lim
x→+∞ 5x − 2
5
= lim
(H) x→+∞ 5
= 1.

ln(x + 1) 1/(x + 1) x 1
(b) lim = lim = lim = lim = 1.
x→+∞ ln(x) (H) x→+∞ 1/x x→+∞ x + 1 (H) x→+∞ 1

ln(ln(ln(x))) (1/x)(1/ ln(x))(1/ ln(ln(x))) 1

(c) lim = lim = lim = 0.
x→+∞ ln(ln(x)) (H) x→+∞ (1/x)(1/ ln(x)) x→+∞ ln(ln(x)))
 68

(d)

(ln(x))4 4x−1 · (ln(x))3

lim = lim
x→+∞ x3 (H) x→+∞ 3x2
4(ln(x))3
= lim
x→+∞ 3x3
12x−1 · (ln(x))2
= lim
(H) x→+∞ 9x2
4(ln(x))2
= lim
x→+∞ 3x3
8x−1 ln(x)
= lim
(H) x→+∞ 9x2
8 ln(x)
= lim
x→+∞ 9x3

8x−1
= lim
(H) x→+∞ 27x2

8
= lim
x→+∞ 27x3
= 0

2
(e) Note that lim e−2x = 0, and
x→+∞

2
2 x 1 e−2x
lim xe−2x = lim = lim 2 = lim =0
x→+∞ x→+∞ e2x2 (H) x→+∞ 4xe 2x x→+∞ 4x

2
2 x 2x 1 −2x2
lim x2 e−2x = lim = lim 2 = lim e =0
x→+∞ x→+∞ e2x2 (H) x→+∞ 4xe 2x x→+∞ 2
2 x3 3x2 3 −x2
lim x3 e−2x = lim 2 = lim 2 = lim xe =0
x→+∞ x→+∞ e2x (H) x→+∞ 4xe2x x→+∞ 4

2
Inductively, we deduce that lim xn e−2x = 0 for each n ∈ N.
x→+∞

2 2 2
ln(e2x + x3 ) (4xe2x + 3x2 )/(e2x + x3 )
lim = lim
x→+∞ x2 + x (H) x→+∞ 2x + 1
2
4xe2x + 3x2
= lim 2
x→+∞ (2x + 1)e2x + 2x4 + x3

2
4x + 3x2 e−2x
= lim 2
x→+∞ (2x + 1) + (2x4 + x3 )e−2x

2 2
4 + 6xe−2x + 3x2 · (−4x)e−2x
= lim 2 2
(H) x→+∞ 2 + (8x3 + 3x2 )e−2x + (2x4 + x3 ) · (−4x)e−2x

2
4 + (−12x3 + 6x)e−2x
= lim 2
x→+∞ 2 + (−8x5 − 4x4 + 8x3 + 3x2 )e−2x

= 2
 69

ln(x + 1) 1/(x + 1) x 1
(f) lim = lim = lim = lim = 1.
x→+∞ ln(x) (H) x→+∞ 1/x x→+∞ x + 1 (H) x→+∞ 1
ln(x − 1) 1/(x − 1) x 1
lim = lim = lim = lim = 1.
x→+∞ ln(x) (H) 1/x
x→+∞ x→+∞ x − 1 (H) x→+∞ 1
  
ln(x + 1) ln(x − 1) ln(x + 1) ln(x − 1)
Then lim = lim = 1.
x→+∞ (ln(x))2 x→+∞ ln(x) ln(x)

Exercise 10. (Level 2/Level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.

(a) lim+ x csc(2x) (c) lim x5 e−2x (e) lim+ sin(x) ln(x)
x→0 x→+∞ x→0

 x 
−10 (−x−2 )
1
3
(b) lim x (ln(x)) 4
(d) lim x e (f) lim x 1+ −e
x→0+ x→0+ x→+∞ x

x 1 1
Solution. (a) lim+ x csc(2x) = lim+ = lim = .
x→0 x→0 sin(2x) (H) x→0+ 2 cos(2x) 2

(b)

(ln(x))4
lim+ x3 (ln(x))4 = lim+
x→0 x→0 x−3
4x−1 (ln(x))3
= lim
(H) x→0+ −3x−4
4(ln(x))3
= lim
x→0+ −3x−3
12x−1 (ln(x))2
= lim+
(H) x→0 9x−4
4(ln(x))2
= lim+
x→0 3x−3
8x−1 ln(x)
= lim+
(H) x→0 −9x−4
8 ln(x)
= lim
x→0+ −9x−3
8x−1
= lim+
x→0 27x−4
8x3
= lim+
(H) x→0 27
= 0

x5 5x4 20x3 5!
(c) lim x5 e−2x = lim = lim = lim = ··· = lim = 0.
x→+∞ x→+∞ e2x (H) x→+∞ 2e2x (H) x→+∞ 4e2x (H) (H) x→+∞ 25 e2x
 70

(d)

−2 x−10
lim+ x−10 e(−x )
= lim+
x→0 x→0 e(x−2 )
−10x−11
= lim+
(H) x→0 −2x−3 e(x−2 )
5x−8
= lim
x→0+ e(x−2 )
5 · (−8x−9 )
= lim
(H) x→0+ (−2x−3 )e(x−2 )
5 · 4x−6
= lim+
x→0 e(x−2 )
= ···
(H)

−2
= lim 5! · e(−x )
x→0+

= 0

ln(x) 1/x sin2 (x)

(e) lim sin(x) ln(x) = lim = lim = lim − = 0.
x→0+ x→0+ csc(x) (H) x→0+ − csc(x) cot(x) x→0+ x cos(x)

(f)

− ln(1 + t) + 1 − 1/(1 + t) −1/(1 + t) + 1/(1 + t)2

lim+ = lim+
t→0 t2 (H) t→0 2t
1/(1 + t)2 − 2/(1 + t)3
= lim
(H) t→0+ 2
1
= − .
2

x
ex ln(1+1/x) − e
 
1
lim x 1+ −e = lim
x→+∞ x x→+∞ 1/x
−1
et ln(1+t)
−e
= lim+
t=1/x t→0 t
−1
[−t−2 ln(1 + t) + t−1 /(1 + t)]et ln(1+t)
= lim+
(H) t→0 1
− ln(1 + t) + 1 − 1/(1 + t) 1
= lim 2
· (1 + t) t
t→0+ t
e
= −
2

Exercise 11. (Level 2/Level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.
 71

 
(a) lim (csc(x) − cot(x)) 2 1
x→0 (d) lim+ +
x→1 1 − x2 ln(x)

(b) limπ (sec(x) − tan(x))

 
x 2
x→ 2 (e) lim+ −
x→0 ex − 1 − x x + x2
   
1 1 1 1
(c) lim − (f) lim+ −
x→0 arctanh(x) tanh(x) x→0 sin(x) sinh(x) x2

1 − cos(x) sin(x)
Solution. (a) lim (csc(x) − cot(x)) = lim = lim = 0.
x→0 x→0 sin(x) (H) x→0 cos(x)

1 − sin(x) − cos(x)
(b) limπ (sec(x) − tan(x)) = limπ = limπ = 0.
x→ 2 x→ 2 cos(x) (H) x→ 2 − sin(x)

(c)
 
1 1 tanh(x) − arctanh(x)
lim − = lim
x→0 arctanh(x) tanh(x) x→0 arctanh(x) tanh(x)
sech2 (x) − 1/(1 − x2 )
= lim
(H) x→0 arctanh(x)sech2 (x) + tanh(x)/(1 − x2 )

1 − x2 − cosh2 (x)
= lim
x→0 (1 − x2 )arctanh(x) + sinh(x) cosh(x)

−2x − 2 sinh(x) cosh(x)

= lim
(H) x→0 1 − 2xarctanh(x) + sinh2 (x) + cosh2 (x)
= 0.

(d)

2 ln(x) + 1 − x2
 
2 1
lim + = lim
x→1+ 1 − x2 ln(x) x→1+ (1 − x2 ) ln(x)
2/x − 2x
= lim
(H) x→1+ −2x ln(x) + (1 − x2 )/x
2/x2 − 2
= lim+
x→1 −2 ln(x) + 1/x2 − 1
−4/x3
= lim+
(H) x→1 −2/x − 2/x3
= 1.
 72

(e)
 
x 2
lim+ −
x→0 e − 1 − x x + x2
x

x(x + x2 ) − 2(ex − 1 − x)
= lim+
x→0 (ex − 1 − x)(x + x2 )
x3 + x2 + 2x + 2 − 2ex
= lim+
x→0 (ex − 1 − x)(x + x2 )
3x2 + 2x + 2 − 2ex
= lim+
(H) x→0 (ex − 1)(x + x2 ) + (ex − 1 − x)(1 + 2x)
6x + 2 − 2ex
= lim+
(H) x→0 ex (x + x2 ) + 2(ex− 1)(1 + 2x) + 2(ex − 1 − x)
6 − 2ex
= lim+
(H) x→0 ex (x + x2 ) + 3ex (1 + 2x) + 6(ex − 1)
4
=
3

(f)

1 − cos(x) cosh(x) sin(x) cosh(x) − cos(x) sinh(x)

lim = lim
x→0+ x2 (H) x→0+ 2x
2 sin(x) sinh(x)
= lim = 0.
(H) x→0+ 2

Then
 
1 1
lim −
x→0+ sin(x) sinh(x) x2
x2 − sin(x) sinh(x)
= lim
x→0+ x2 sin(x) sinh(x)
2x − sin(x) cosh(x) − cos(x) sinh(x)
= lim
(H) x→0+ 2x sin(x) sinh(x) + x2 cos(x) sinh(x) + x2 sin(x) cosh(x)
2 − 2 cos(x) cosh(x)
= lim+
(H) x→0 2 sin(x) sinh(x) + 4x cos(x) sinh(x) + 4x sin(x) cosh(x) + 2x2 cos(x) cosh(x)
1 − cos(x) cosh(x)
= lim+
(H) x→0 x2
 −1
sin(x) sinh(x) sinh(x) sin(x)
· · + 2 cos(x) · + 2 cosh(x) · + cos(x) cosh(x)
x x x x
= 0.

Exercise 12. (Level 2/Level 3)

3ex + 2
 
1
(a) Apply L’Hôpital’s Rule, or otherwise, to evaluate the limit lim ln .
x→0+ sin(x) 5
 73

1
3ex + 2
  sin(x)
(b) Hence, or otherwise, evaluate the limit lim .
x→0+ 5

Solution. (a)

3ex + 2 (3ex /5) · [5/(3ex + 2)]

 
1
lim+ ln = lim+ by L’Hôpital’s Rule,
x→0 sin(x) 5 x→0 cos(x)
3ex
= lim+
x→0 (3ex + 2) cos(x)
3
=
5

1
3ex + 2
  sin(x)
3ex +2
= lim+ e sin(x) ln( ) = e 35 .
1
(b) lim+ 5
x→0 5 x→0

Exercise 13. (Level 2/Level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.

(a) lim+ xx (c) lim+ (sin(x))x (e) lim+ (ln(1 + x))x

x→0 x→0 x→0

√ 2 1
(b) lim+ (x3 + 3x + x)x +2x
(d) lim+ xsin(x) (f) lim+ (sinh(x)) ln(x)
x→0 x→0 x→0

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.

ln(x) 1/x
(a) lim+ x ln(x) = lim+ = lim = lim+ −x = 0.
x→0 x→0 1/x (H) x→0+ −1/x2 x→0

ln(x)
Then lim xx = lim e 1/x = e0 = 1.
x→0+ x→0+

(b)
√
2 3
√ ln(x3 + 3x + x)
lim (x + 2x) ln(x + 3x + x) = lim
x→0+ x→0+ 1/(x2 + 2x)
1 √
(3x2 + 3 + x− 2 /2)/(x3 + 3x + x)
= lim+
(H) x→0 −(2x + 2)/(x2 + 2x)2
5 1
x(3x 2 + 3x 2 + 1/2)(x + 2)2
= lim 5 1
x→0+ −(2x + 2)(x 2 + 3x 2 + 1)
= 0.
√ 2 √
(x2 +2x) ln(x3 +3x+ x)
lim (x3 + 3x + x)x +2x
= lim+ e = e0 = 1.
x→0+ x→0
 74

(c)

ln(sin(x))
lim x ln(sin(x)) = lim
x→0+ x→0+ 1/x
cos(x)/ sin(x)
= lim+
(H) x→0 −1/x2
 
x
= lim+ −x cos(x) ·
x→0 sin(x)
= 0.

Then lim+ (sin(x))x = lim+ ex ln(sin(x)) = e0 = 1.

x→0 x→0

(d) Similar arguments as in (c) gives

lim sin(x) ln(x) = · · · = 0.
x→0+

Then lim xsin(x) = lim esin(x) ln(x) = e0 = 1.

x→0+ x→0+

(e)

ln(ln(1 + x))
lim x ln(ln(1 + x)) = lim
x→0+ x→0+ 1/x
[1/(1 + x)][1/ ln(1 + x)]
= lim
(H) x→0+ −1/x2
−x2
= lim
x→0+ (1 + x) ln(1 + x)
−2x
= lim+
(H) x→0 1 + ln(1 + x)
= 0.

Then lim+ (ln(1 + x))x = lim+ ex ln(ln(1+x)) = e0 = 1.

x→0 x→0

(f)

ln(sinh(x)) cosh(x)/ sinh(x)

lim = lim
x→0+ ln(x) (H) x→0+ 1/x
x cosh(x)
= lim
x→0+ sinh(x)
cosh(x) + x sinh(x)
= lim
(H) x→0+ cosh(x)
= 1.

1 ln(sinh(x))
Then lim (sinh(x)) ln(x) = lim e ln(x) = e1 = e.
x→0+ x→0+

Exercise 14. (Level 2/Level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.
 75

(a) lim x x
1
1 (e) lim (− ln(x))x
x→+∞ (c) lim+ ( )sin(x) x→0+
x→0 x
1
(b) lim+ (csc(x))x (d) lim+ (csc(x))sinh(x) (f) lim (ln(x)) ln(x)
x→0 x→0 x→+∞

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.

ln(x) 1/x
(a) lim = lim = 0.
x→+∞ x (H) x→+∞ 1
1 ln(x)
Then lim x x = lim e x = e0 = 1.
x→+∞ x→+∞

(b)

− ln(sin(x))
lim+ x ln(csc(x)) = lim+
x→0 x→0 1/x
− cos(x)/ sin(x)
= lim
(H) x→0+ −1/x2
x2 cos(x)
= lim+
x→0 sin(x)
 
x
= lim · x cos(x)
x→0+ sin(x)
= 0.

Then lim+ (csc(x))x = lim ex ln(csc(x)) = e0 = 1.

x→0 x→+∞

1
(c) lim sin(x) ln( ) = lim − sin(x) ln(x) = · · · = 0.
x→0 + x x→0+

1 1
Then lim ( )sin(x) = lim esin(x) ln( x ) = e0 = 1.
x→0+ x x→+∞

sinh(x) cosh(x)
(d) lim = lim = 1.
x→0+ sin(x) (H) x→0+ cos(x)

ln(sin(x))
lim sinh(x) ln(csc(x)) = lim −
x→0+ x→0+ 1/ sinh(x)
cos(x)/ sin(x)
= lim −
(H) x→0+ − cosh(x)/ sinh2 (x)
 
cos(x) sinh(x)
= lim · · sinh(x)
x→0+ cosh(x) sin(x)
= 0

Then lim+ (csc(x))sinh(x) = lim esinh(x) ln(csc(x)) = e0 = 1.

x→0 x→+∞

ln(− ln(x)) (−1/x) · [1/(− ln(x))] x

(e) lim+ x ln(− ln(x)) = lim+ = lim = lim+ − = 0.
x→0 x→0 1/x (H) x→0+ −1/x2 x→0 ln(x)

Then lim+ (− ln(x))x = lim ex ln(− ln(x)) = e0 = 1.

x→0 x→+∞
 76

ln(ln(x)) (1/x) · (1/ ln(x)) 1

(f) lim = lim = lim = 0.
x→+∞ ln(x) (H) x→+∞ 1/x x→+∞ ln(x)

1 ln(ln(x))
Then lim (ln(x)) ln(x) = lim e ln(x) = e0 = 1.
x→+∞ x→+∞

Exercise 15. (Level 2/level 3)

Evaluate the limits below. Where appropriate and convenient, apply L’Hôpital’s Rule.
√1 1
(a) lim (1 + 2x) x (c) lim (1 + sin(x)) sinh(x) (e) lim (cos(x))ln(sin(x))
x→0 x→0+ x→ π
2
−

 x2 +2x   1
3 1 arcsin(x) x2
(b) lim 1+ 2 (d) lim+ (tan(3x) + cos(4x)) sin(x) (f) lim
x→+∞ x +x+1 x→0 x→0 x

Solution. An equality at which L’Hôpital’s Rule is applied is indicated with ‘(H)’ marked beneath the equality
symbol.
√
ln(1 + 2x) 2/(1 + 2x) 4 x
(a) lim √ = lim √ = lim = 0.
x→0 x (H) x→0 1/(2 x) x→0 1 + 2x

ln(1+2x)
√1 √
Then lim (1 + 2x) x = lim e x = e0 = 1.
x→0 x→+∞

(b)
 
2 3
lim (x + 2x) ln 1 + 2
x→+∞ x +x+1
ln(1 + 3/(x2 + x + 1))
= lim
x→+∞ (x2 + 2x)−1
[−3(x2 + x + 1)−2 (2x + 1)]/[1 + 3/(x2 + x + 1)]
= lim
(H) x→+∞ −(x2 + 2x)−2 (2x + 2)
3(2x + 1)(x2 + 2x)2
= lim
x→+∞ (2x + 2)(x2 + x + 1)2 (1 + 3/(x2 + x + 1))

= 3
 x2 +2x
3
 
(x2 +2x) ln 1+ x2 +x+1
3
Then lim 1+ 2 = lim e = e3
x→+∞ x +x+1 x→+∞

ln(1 + sin(x)) cos(x)/(1 + sin(x))

(c) lim+ = lim = 1.
x→0 sinh(x) (H) x→0+ cosh(x)
1 ln(1+sin(x))
Then lim (1 + sin(x)) sinh(x) = lim e sinh(x) = e1 = e.
x→0+ x→+∞

ln(tan(3x) + cos(4x)) (3 sec2 (3x) − 4 sin(4x))/(tan(3x) + cos(4x))

(d) lim+ = lim+ = 3.
x→0 sin(x) (H) x→0 cos(x)
1 ln(tan(3x)+cos(4x))
Then lim (tan(3x) + cos(4x)) sin(x) = lim e sin(x) = e3 .
x→0+ x→+∞
 77

ln(cos(x)) − sin(x)/ cos(x)

(e) lim cos(x) ln(cos(x)) = lim = lim = lim − cos(x) = 0. Then
x→ π
2
− π−
x→ 2 sec(x) (H) x→ π
2
− sec(x) tan(x) x→ π
2
−

ln(sin(x))
lim ln(sin(x)) ln(cos(x)) = lim
x→ 2π−
x→ 2 π− 1/ ln(cos(x))
cos(x)/ sin(x)
= lim
(H) x→ π
2
− (− sin(x)/ cos(x)) · [−1/(ln(cos(x)))2 ]
(cos(x) ln(cos(x)))2
= lim
x→ 2 π−
sin2 (x)
= 0

Therefore lim (cos(x))ln(sin(x)) = lim eln(sin(x)) ln(cos(x)) = e0 = 1.

π−
x→ 2 x→+∞

(f)
√
ln (arcsin(x)/x) {[x/ 1 − x2 − arcsin(x)]/x2 }[x/ arcsin(x)]
lim = lim
x→0 x2 (H) x→0 2x
1
x(1 − x2 )− 2 − arcsin(x)
= lim
x→0 2x2 arcsin(x)
1 3 √
[(1 − x2 )− 2 + x2 (1 − x2 )− 2 ] − 1/ 1 − x2
= lim √
(H) x→0 4x arcsin(x) + 2x2 / 1 − x2
3
x(1 − x2 )− 2
= lim 1
x→0 4 arcsin(x) + 2x(1 − x2 )− 2
3 5
(1 − x2 )− 2 + 3x2 (1 − x2 )− 2
= lim √ 1 3
(H) x→0 4/ 1 − x2 + 2(1 − x2 )− 2 + 2x2 (1 − x2 )− 2

(1 − x2 )−1 + 3x2 (1 − x2 )−2

= lim
x→0 6 + 2x2 (1 − x2 )−1
1
=
6
  1
arcsin(x) x2 ln(arcsin(x)/x) 1
Then lim = lim e x2 = e6 .
x→0 x x→+∞

Exercise 16. (Level 3)

Evaluate the each of the limits below. Think carefully whether to apply L’Hôpital’s Rule or not.

x2 + cos(ex ) ex + x sin(x) + cos(x)

(a) lim (c) lim
x→+∞ x2 + cos(e2x ) x→+∞ ex + cos(x)

x + sin(x) x2 + sin(2x)
(b) lim (d) lim
x→+∞ x − sin(x) x→+∞ (2x3 + x + sin(x))esin(x)
 78

Solution. (a) Applying the Sandwich Rule, we obtain

lim x−2 cos(ex ) = 0, and lim x−2 cos(e2x ) = 0.

x→+∞ x→+∞

x2 + cos(ex ) 1 + x−2 cos(ex )

Then lim = lim = 1.
x→+∞ x2 + cos(e2x ) x→+∞ 1 + x−2 cos(e2x )

(b) Applying the Sandwich Rule, we obtain lim x−1 sin(x) = 0.

x→+∞

x + sin(x) 1 + x−1 sin(x)

Then lim = lim = 1.
x→+∞ x − sin(x) x→+∞ 1 − x−1 sin(x)

(c) Applying the Sandwich Rule, we obtain lim e−x cos(x) = 0.

x→+∞

Applying L’Hôpital’s Rule, we obtain lim xe−x = 0. Then lim xe−x sin(x) = 0 by the Sandwich Rule.
x→+∞ x→+∞

x
e + x sin(x) + cos(x) 1 + xe sin(x) + e−x cos(x)
−x
Then lim = lim = 1.
x→+∞ ex + cos(x) x→+∞ 1 + e−x cos(x)

(d) Applying the Sandwich Rule, we obtain lim x−2 sin(2x) = 0, lim x−3 sin(x) = 0, and lim x−1 e− sin(x) =
x→+∞ x→+∞ x→+∞
0.
x2 + sin(2x) (1 + x−2 sin(2x)) · x−1 e− sin(x)
Then lim = lim = 0.
x→+∞ (2x3 + x + sin(x))e sin(x) x→+∞ 2 + x−2 + x−3 sin(x)

Exercise 17. (Level 3)

√
Let an = n n. Show that lim an = 1. (Hint: use L’Hopital’s rule).
n→∞

ln x
Solution. Let f (x) = x1/x = e x . Then by the L’Hoptial’s rule

ln x 1
lim = lim = 0.
x→+∞ x x→+∞ x

Therefore
lim f (x) = e0 = 1.
x→+∞

This implies lim an = lim f (n) = 1

n→∞ n→∞

Exercise 18. (Level 3)

Suppose the second derivative of f (x) exists, then

f (a + h) + f (a − h) − 2f (a)
f 00 (a) = lim .
h→0 h2
 79

Solution. Let F (x) = f (a + x) + f (a − x) − 2f (a). Let G(x) = x2 . Then G0 (x) = 2x. It is not equal to 0 when x
is close to 0. Also limx→0 F (x) = f (a) + f (a) − 2f (a) = 0 and limx→0 G(x) = 0. Hence by L’Hopital’s rule

F (x) F 0 (x)
lim = lim 0
x→0 G(x) x→0 G (x)

provided the RHS exists.

F 0 (x) = f 0 (a + x) − f 0 (a − x) and G0 (x) = 2x. Then G00 (x) = 2. It is not equal to 0 when x is close to to 0. Also
limx→0 F 0 (x) = f 0 (a) − f 0 (a) = 0 and limx→0 G0 (x) = 0. Hence by L’Hopital’s rule again

F 0 (x) F 00 (x)
lim = lim
x→0 G0 (x) x→0 G00 (x)

provided the RHS exists.

The RHS is
f 00 (a + x) + f 00 (a − x)
lim = f 00 (a).
x→0 2

Exercise 19. (Level 4)

Suppose f (x) is a differentiable function defined on (−a, a) satisfying

• f (0) = 0

• There exists a constant k such that −kx ≤ f (x) ≤ kx for x ∈ (−a, a).

• f (x) is not a constant function.

(a) Show that |f 0 (0)| ≤ k.

(b) Show that lim+ xf (x) = 1.

x→0

Solution. (a)
f (h) − f (0) kh
f 0 (0) = lim ≤ lim = k.
h→0 h h→0 h

So f 0 (0) ≤ k. Similarly, f 0 (0) ≥ −k. So |f 0 (0)| ≤ k.

(b)
lim xf (x) = lim+ ef (x) ln x .
x→0+ x→0

Then
−kx ≤ f (x) ≤ kx

−kx ln x ≤ f (x) ln x ≤ kx ln x.
By L’Hopital’s rule
ln x
lim x ln x = lim+
x→0+ x→0 1/x

1/x
= lim+ = lim+ (−x) = 0.
x→0 −1/x2 x→0
 80

By the Sandwich theorem,

lim f (x) ln x = 0.
x→0+

Hence
lim xf (x) = e0 = 1.
x→0+

Exercise 20. (Level 4)

x sin x
(a) Evaluate lim .
x→0 1 − cos x
1
(b) Using (a), evaluate lim+ (1 − cos x) ln x .
x→0

Solution. (a)

x sin x 2x sin x2 cos x2

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

1 ln(1−cos x)
(b) Let y = (1 − cos x) ln x . Then ln y = ln x . By L’Hopital’s rule and (a), one gets

ln(1 − cos x) x sin x

lim = lim = 2.
x→0 ln x x→0 1 − cos x

Hence we have lim ln y = 2, so

x→0

1 lim ln y
lim (1 − cos x) ln x = lim eln y = ex→0 = e2 ,
x→0+ x→0+

by the continuity of exponential function.

Exercise 21. (Level 5)

Let α, β ∈ R. Suppose α > β > 0. Let f : [0, +∞) −→ R be the function defined by
 √
 αβ
 if x=0
1
f (x) =
 x
α + βx x
 if x>0
2


(a) i. Show that f is continuous at 0.

ii. Does lim f (x) exist? If yes, what is its value?
x→+∞
 81

(b) Define the function h : [0, 1) −→ R by h(x) = (1 + x) ln(1 + x) + (1 − x) ln(1 − x) for any x ∈ [0, 1).

i. Show that h(x) > h(0) for any x ∈ (0, 1).

αx − β x
ii. Suppose x > 0. Write τx = .
αx + β x
 x
α ln(αx ) + β x ln(β x )
 
2
Show that 0 < τx < 1 and h(τx ) = 2 + ln .
αx + β x αx + β x

(c) Define the function g : [0, +∞) −→ R by g(x) = ln(f (x)) for any x ∈ [0, +∞).

αx ln(αx ) + β x ln(β x )
 
2 0 2
i. Suppose x > 0. Show that x g (x) = + ln .
αx + β x αx + β x
ii. Show that f is strictly increasing on [0, +∞).

Solution. (a) i. When x > 0, we have

 x1 !
αx + β x αx + β x ln (αx + β x ) − ln(2)
  
1
ln(f (x)) = ln = ln = .
2 x 2 x

Now apply L’Hôpital’s Rule:

ln(αx + β x ) − ln(2)
lim+
x→0 x
(αx ln(α) + β x ln(β))/(αx + β x )
= lim
x→0+ 1
αx ln(α) + β x ln(β) ln(α) + ln(β) p
= lim+ x x
= = ln( αβ).
x→0 α +β 2
√ p
By the continuity of the exponential function, lim+ f (x) = lim+ eln(f (x)) = lim+ eln( αβ)
= αβ = f (0).
x→0 x→0 x→0

Hence f is continuous at 0.
 t
β β
ii. Note that 0 < < 1. Then lim = 0.
α t→+∞ α
When x > 0 and x is of ‘large magnitude’, we have

ln(f (x))
 x 1 !
α + βx x
= ln
2
 x
α + βx

1
= ln
x 2
ln (αx + β x ) − ln(2)
=
x
x ln(α) + ln (1 + (β/α)x ) − ln(2)
=
x
ln (1 + (β/α)x ) − ln(2)
= ln(α) + −→ ln(α) as x −→ +∞.
x
 82

Then lim ln(f (x)) exists and is of value ln(α).

x→+∞

Therefore, by the continuity of the exponential function, lim f (x) exists and is of value α.
x→+∞

(b) i. h is continuous on [0, 1) and is differentiable on (0, 1).

1+x
For any x ∈ (0, 1), we have h0 (x) = ln(1 + x) + 1 − ln(1 − x) − 1 = ln( ).
1−x
1+x 1+x
Note that 0 < 1 − x < 1 + x. Then > 1. Therefore h0 (x) = ln( ) > 0.
1−x 1−x
Then h is strictly increasing on [0, 1). Hence h(x) > h(0) for any x ∈ (0, 1).
αx − β x
ii. Suppose x > 0. Write τx = .
αx + β x
αx − β x αx − 0
Since β > 0, we have τx = < = 1.
αx + β x αx + 0
αx − β x 0
Since α > β > 0, we have αx > β x > 0 and hence τx = x x
> x = 0.
α +β α + βx
Therefore 0 < τx < 1.

h(τx )
= (1 + τx ) ln(1 + τx ) + (1 − τx ) ln(1 − τx )
αx − β x αx − β x αx − β x αx − β x
       
= 1+ x ln 1 + + 1 − ln 1 −
α + βx αx + β x αx + β x αx + β x
2αx 2αx 2β x 2β x
   
= ln + ln
αx + β x αx + β x αx + β x αx + β x
2αx 2β x
     
x 2 x 2
= ln(α ) + ln + x ln(β ) + ln
αx + β x αx + β x α + βx αx + β x
 x
α ln(αx ) + β x ln(β x )
 
2
= 2 x x
+ ln
α +β α + βx
x

(c) i. Suppose x > 0.

αx + β x ln (αx + β x ) − ln(2)
 
1
Note that g(x) = ln(f (x)) = ln = .
x 2 x
ln (αx + β x ) − ln(2)
We have xg(x) = x ln(f (x)) = x · = ln(αx + β x ) − ln(2).
x
αx ln(α) + β x ln(β)
Then xg 0 (x) + g(x) = (xg(x))0 = .
αx + β x
Therefore

αx ln(α) + β x ln(β)
x2 g 0 (x) = x· − xg(x)
αx + β x
αx · x ln(α) + β x · x ln(β)
 x
α + βx

1
= − x · ln
αx + β x x 2
αx ln(αx ) + β x ln(β x )
 
2
= + ln .
αx + β x αx + β x
 83

h(τx ) h(0) 0 f 0 (x)

ii. Suppose x > 0. We have x2 g 0 (x) = > = = 0. Then = g 0 (x) > 0.
2 2 2 f (x)
Note that f (x) > 0. Then f 0 (x) > 0.
Recall that f is continuous at 0. Then f is continuous on [0, +∞). Therefore f is strictly increasing on
[0, +∞).