CHAPTER 2

Functions of one variable

In this chapter we dene the derivative of a function of one variable
and expose the properties of dierentiable functions. We suppose that the
notions of limit and continuity for functions of one variable are known.
The dierential and integral calculus appeared in XVII century in the
papers of Fermat, Toricelli, Barrow under its geometric and cinematic
aspect. Leibniz and Newton introduced the notions of derivative and
integral at the end of XVII century. A rigorous denition of the derivative,
based on the notion of limit was given by Cauchy. The next step in
the foundamentation of the notions of dierential calculus was made by
Weierstrass.

2.1. The derivative of a function

Let f : A R, A R, be a function and a A .
Definition 2.1.1. The function f is said to be dierentiable at the
point a if
f (x) f (a)
xa
xa
lim

exists and is nite. The limit is denoted by f  (a) or

the derivative of f at the point a.
53

df
(a) and is called
dx

54

Chapter 2
If we consider the left and the right limit in Denition 2.1.1 at the

point a we obtain the left and the right derivative of f at a denoted by

fs (a) and fd (a).
The function f is called dierentiable on the set B, B A, if f is
dierentiable at every point of B. In this case we can dene the function
f  : B R called the derivative of f on B. Now, if f  is dierentiable on
B one can dene the second derivative of f by f  = (f  ) .
Generally, the nth derivative of f , denoted by f (n) is dened by
recurrence as follows:
f (n) = (f (n1) ) ,

n 1,

f (0) = f.

We recall some of the basic dierentiation rules.

(f + g) = f  + g 
(f ) = f  ,

(f g) = f  g + f g 
 
f
f g f g
=
g
g2
(f g) = (f  g)g 
(f 1 ) =

1
.
f  f 1

For more details concerning the conditions of the previous relations

see [1].
We recall Leibniz formula for the evaluation of the nth derivative
of a product. If f, g : A R are n-times dierentiable on A then
(f g)(n) = Cn0 f (n) g + Cn1 f (n1) g  + + Cnn f g (n) .

Functions of one variable

55

Application 2.1.1. Find the nth derivative of the function

f : R R,

f (x) = (x2 x + 1)e2x .

Solution. Denote u(x) = e2x , v(x) = x2 x + 1, x R. We have

u(k) (x) = 2k e2x ,

v  (x) = 2x 1,

v  (x) = 2,

v (k) (x) = 0,

k>2

f (n) (x) = Cn0 2n e2x (x2 x + 1) + Cn1 2n1 e2x (2x 1) + Cn2 2n2 e2n 2
= 2n2e2x [4x2 + (4n 4)x + n2 n + 4], n 2.

In what follows for an interval I, I R, we dene

C k (I) = {f | f : I R, f (k) and is continuous on I}
C 0 (I) = {f | f : I R, f is continuous on I}
C (I) = {f | f : I R, f (k) for all k N on I}
C (I) is called the class of indenite dierentiable functions on I.

2.2. Mean value theorems

Definition 2.2.1. Let f : A R, A R, be a function. A point
a A is called a point of local minimum (maximum) of f if there
exists V V(a) such that f (a) f (x) (f (a) f (x)) for every x V A.
A point of local minimum (maximum) of f is called a point of local
extremum of f .
Theorem 2.2.1. (Fermat) Let f : A R, A R, be a function
possessing a local extremum at a intA. If f is dierentiable at a, then
f  (a) = 0.

56

Chapter 2
Proof. Suppose that a is a point of local minimum. Then V

V(a):
f (a) f (x), x V A.
Then for all x V A, x < a, we have
f (x) f (a)
< 0.
xa
Letting x a in the previous relation we get fs (a) 0.
For x V A, x > a, we have
f (x) f (a)
>0
xa
and letting x a it follows fd (a) 0. Since f is dierentiable at a,
fs (a) = fd (a), hence f  (a) = 0.
A root of the derivative of a function is called a stationary point.
Remark 2.2.1. A stationary point is not necessary a point of local
extremum. For the function f : R R, f (x) = x3 , x = 0 is a stationary
point but it is not a local extremum point.
Theorem 2.2.2. (Rolle) Let f : [a, b] R be a function with the
properties:
(1) f is continuous on [a, b];
(2) f is dierentiable on (a, b);
(3) f (a) = f (b).
Then there exists c (a, b) such that f  (c) = 0.
Proof. By the continuity of f on [a, b] it follows that f attains
its minimum and its maximum on [a, b]. Denote m = min f (x), M =
max f (x) on [a, b]. If m = M, then f is a constant function on [a, b],
f  (x) = 0, x [a, b], and c can be every point of (a, b). If m = M,

Functions of one variable

57

since f (a) = f (b), it follows that at least an extremum point lies on

(a, b). Denoting by c one of this points, f  (c) = 0 in view of Fermats
theorem.
Theorem 2.2.3. (Cauchy) Let f, g : [a, b] R be two functions
with the following properties:
(1) f , g are continuous on [a, b];
(2) f , g are dierentiable on (a, b);
(3) g (x) = 0 for every x (a, b).
Then there exists c (a, b) such that
f  (c)
f (b) f (a)
=  .
ba
g (c)
g(b)-g(a)
Proof. First we remark that g(a) = g(b). If we suppose that g(a) =
g(b), then c (a, b) such that g  (c) = 0, according to Rolles theorem.
Now dene the function h : [a, b] R,
h(x) = (f (b) f (a))g(x) (g(b) g(a))f (x).
The function f satises the condition of Rolles theorem, hence c
(a, b) with the property h (c) = 0, which is equivalent with
f (b) f (a)
f  (c)
=  .
ba
g (c)

Theorem 2.2.4. (Lagrange) Let f : [a, b] R be a function with

the properties:
(1) f is continuous on [a, b];
(2) f is dierentiable on (a, b).
Then there exists c (a, b) such that
f (b) f (a)
= f  (c).
ba

58

Chapter 2
Proof. Take g(x) = x in Cauchys theorem.
Corollary 2.2.1. Let f : I R be dierentiable on the interval I,

I R.
(1) If f  (x) 0 on I, then f is increasing on I;
(2) If f  (x) 0 on I, then f is decreasing on I;
(3) If f  (x) = 0 on I, then f is constant on I.
Proof. (1) Let x1 , x2 I1 , x1 < x2 . By Lagranges theorem follows
that there exists c, c (x1 , x2 ) such that:
f (x2 ) f (x1 )
= f  (c).
x2 x1
Then f (x1 ) f (x2 ), since x2 x1 > 0 and f  (c) 0.
(2) Analogous with the proof of (1).
(3) Let a I be a xed point and x I be an arbitrary point. Then
there exists c between a and x such that
f (x) f (a)
= f  (c) = 0 f (x) = f (a).
xa

2.3. Taylors formula

One of the basic problems in mathematics is to approximate a function belonging to a given class by another function which is more simple
or by a sequence of functions. By Taylors formula a function of class C n
is approximated by a polynomial function.
Theorem 2.3.1. (Taylor) Let I be an open interval in R and f :
I R be a function possessing a derivative of order (n + 1) on I. Then
for every x, x0 I, x = x0 , there exists a point between x0 and x such

Functions of one variable

59

that:
f (x) = f (x0 )+

f  (x0 )
f (n) (x0 )
f (n+1) ()
(xx0 )+ +
(xx0 )n +
(xx0 )n+1 .
1!
n!
(n + 1)!

Proof. Suppose that x0 < x. Let F : [x0 , x] R be dened by

F (t) = f (t) +

f  (t)
f (n) (t)
(x t) + +
(x t)n + A(x t)n+1
1!
n!

where A is chosen such that F satises the conditions of Rolles theorem

on [x0 , x]. Then F must satises the condition
F (x) = F (x0 )
or
f (x) = f (x0 ) +

f  (x0 )
f (n) (x0 )
(x x0 ) + +
(x x0 )n + A(x x0 )n+1 .
1!
n!

Now we have to prove that there exist (x0 , x) such that

A=

f (n) ()
.
(n + 1)!

By Rolles theorem follows that there exists (x0 , x) satisfying

F  () = 0.
We have
f (t)
f  (t)
f  (t) f  (t)
+
(x t) 2
(x t)+
(x t)2 +. . .
1!
1!
2!
2!
f (n+1) (t)
f (n) (t)
(x t)n1 +
(x t)n (n + 1)A(x t)n
n
n!
n!
f (n+1) (t)
=
(x t)n A(n + 1)(x t)n .
n!

F  (t) = f  (t)

Now from the relation F  () = 0 follows A =

f (n+1) ()
.
(n + 1)!

60

Chapter 2
Remark 2.3.1. 1) The polynomial
Tn (x) = f (x0 ) +

f  (x0 )
f (n) (x0 )
(x x0 ) + +
(x x0 )n
1!
n!

is called the Taylor polynomial of nth degree of f at x0 .

Tn (x) satises the following relations:
Tn (x0 ) = f (x0 )
Tn (x0 ) = f  (x0 )
...
Tn(n) (x0 ) = f (n) (x0 ).
2) The last term from Taylors formula
Rn (x) =

f (n+1) ()
(x x0 )n+1
(n + 1)!

is called the remainder of nth order.

3) The point is called the intermediary point and
= x0 + (x x0 ),

0 < < 1.

Corollary 2.3.1. (Maclaurin) If f : I R satises the conditions

of Theorem 2.3.1 and 0 I, then there exists between 0 and x such
that:
f (x) = f (0) +

f  (0)
f (n) (0) n f (n+1) () n+1
x ++
x +
x .
1!
n!
(n + 1)!

Proof. We take x0 = 0 in Taylors formula.

We give some expansions of elementary functions using Maclaurin
formula.

Functions of one variable

61

Corollary 2.3.2. The following relations hold:

(1)

(2)

ex = 1 +

x
xn
xn+1 x
++
+
e ,
1!
n! (n + 1)!

sin x = x

+(1)n

x3 x5
x2n1
+
+ (1)n1
3!
5!
(2n 1)!

x2n
sin x,
(2n)!

cos x = 1

(3)

+(1)n+1

x2n+1
sin x,
(n + 1)!

0 < < 1, x R.

xn
x2 x3
+
+ (1)n1
2
3
n

(1)n xn+1
,
(n + 1)(1 + x)n+1

(5) (1 + x) = 1 +

0 < < 1, x R.

x2n
x2 x4
+
+ (1)n
2!
4!
(2n)!

ln(1 + x) = x

(4)

0 < < 1, x R

0 < < 1, x (1, ).

( 1) 2

( 1) . . . ( n + 1) n
x+
x ++
x
1!
2!
n!

( 1) . . . ( n)
(1 + x)n1 xn+1 ,
(n + 1)!
R, 0 < < 1, x (1, ).

Proof. The relations (1)-(5) follows from Maclaurin formula.

We give now an application of Taylors formula to nd the local extrema of a function.

62

Chapter 2
Theorem 2.3.2. Let I R be an open interval, f C n (I), n 2,

and x0 I such that

f  (x0 ) = f  (x0 ) = = f (n1) (x0 ) = 0,

f (n) (x0 ) = 0.

Then:
(1) If n is an even number, x0 is a point of local extremum of f as
follows:
ii) f (n) (x0 ) > 0 x0 is a point of local minimum;
ii) f (n) (x0 ) < 0 x0 is a point of local maximum.
(2) If n is an odd number, x0 is not a point of local extremum of f .
Proof. By the continuity of f n and f (n) (x0 ) = 0 it follows that there
exists V = (x0 , x0 + ), V I, such that f (x) > 0 or f (x) < 0 for
every x V .
Now for x V we have
f (x) = f (x0 ) +

f  (x0 )
f (n1) (x0 )
(x x0 ) + +
(x x0 )n1
1!
(n 1)!

f (n) ()
f (n) ()
(x x0 )n = f (x)+ =
(x x0 )n ,
n!
n!

= x0 + (x x0 ), 0 < < 1.
(1) i) f (n) (x0 ) > 0 f (n) () > 0 and (xx0 )n 0 f (x) f (x0 ),
x V , hence x0 is a point of local minimum of f .
ii) f (n) (x0 ) < 0 f (n) () > 0 and (x x0 )n 0 f (x) f (x0 ),
x V , hence x0 is a point of local maximum of f .
(2) If n is an odd number (x x0 )n change the sign at x0 , f (n) () has
constant sign on V , thus x0 is not a point of local extremum of f .

Functions of one variable

63

2.4. Problems
1. Find the derivative of the nth order of the functions:

b
1
, a, b R, a = 0, x R \ ;
a) f (x) =
ax + b
a


1
1
, x R \ ,1 ;
b) f (x) = 2
2x x 1
2
c) f (x) = x3 e2x , x R;
d) f (x) = x3 ln x, x (0, );
e) f (x) = ex sin x, x R;
f) f (x) = ex cos x, x R.
2. Find f (n) (0), n N, for the functions:
2

a) f : R R, f (x) = ex ;
b) f : R R, f (x) = arctan x;
arcsin x
.
c) f : (1, 1) R, f (x) =
1 x2
3. Prove that f : R R, given by the relation

f (x) =

e x , x > 0
0,

x0

is indenite dierentiable on R.
4. Prove that the equation
1+

x
x2
x2n
+
++
=0
1! 2!
(2n)!

has not real roots.

5. Find n with the property that Taylors polynomial of n degree of
the function f (x) = ex at x0 = 0, approximate f on [1, 1] with an error
smaller than 103 .
6. Let I be an open interval in R, f C 2 (I) and a, b I such
that f  (a) = f  (b) = 0. Prove that there exists c (a, b) satisfying the

64

Chapter 2

relation:

(b a)2 
|f (c)|.
4
7. Let fij : I R, 1 i, j n, be dierentiable functions on the
|f (b) f (a)|

interval I R. Prove that:



 f11 (x) f12 (x)

 f (x) f (x)
22
 21

 ...
...


 fn1 (x) fn2 (x)

...
...
...
...



 f (x) f (x)
12

 11
f1n (x) 


 ...
...
n 

f2n (x) 
 

 =
 fk1 (x) fk2
(x)

. . . 
k=1 
...

 ...

fnn (x) 
 fn1 (x) fn2 (x)

...
...
...
...
...


f1n (x) 

. . . 


fkn
(x)  .

. . . 

fnn (x) 

CHAPTER 3

Sequences and series of functions

This chapter contains notions and results concerning sequences and
series of functions. They play an important role in function approximation. As a particular case we study power series. The notion of uniform
convergence was considered rst by Stokes and Seidel (1847) who evidentiate its role preserving continuity and then by Cauchy (1953). Power
series appeared rst in the Newtons paper (1665) who believed that every function can be expanded in power series. The idea of representation
of a function by a functional series of sinus and cosinus appeared around
1750 in the papers of D. Bernoulli. The relation between the sum and
the coecients of a trigonometric series was given by J. Fourier in 1805
but he could not prove the convergence of the series. This problem was
solved by Dirichlet in 1829.

3.1. Sequences of functions

In what follows we denote by K one of the elds R or C.
Definition 3.1.1. A sequence of functions (fn )n1 , fn : A K
is said to be convergent or pointwise convergent to the function
f : A K on A if
f (x) = lim fn (x).
n

65

66

Chapter 3
We denote the pointwise convergence by
fn f

PC

or fn f.

Example 3.1.1. 1) Let (fn )n1 , fn : [0, 1] R, fn (x) = xn . We have


0, x [0, 1)
lim fn (x) =
n
1, x = 1

0, x [0, 1)
Hence fn f , f (x) =
1, x = 1
2) Let (fn )n1 , fn : R R, be given by the relation
fn (x) =

1 + x2 enx
,
1 + enx

Then

x R, n N .

x,
lim fn (x) =

x<0

0, x = 0

x2 , x > 0

Definition 3.1.2. A sequence of functions (fn )n1 , fn : A K, is

said to be uniformly convergent to the function f : A K on A if for
every > 0, there exists n N such that for all n n and all x A
|fn (x) f (x)| < .
We denote the uniform convergence by
fn f

UC

or fn f.

Remark 3.1.1. 1) Pointwise convergence does not preserve generally

the properties of the terms of a functional sequence. For instance the
terms of the sequence from Example 3.1.1,1) are continuous functions
but its limit is not a continuous function.

Sequences and series of functions

67

The functions fn from Example 3.1.1,2) are dierentiable but the

limit of the sequence is not dierentiable at x = 0.
2) Every uniformly convergent sequence of functions is pointwise convergent, but the converse of this armation is not generally true.
The sequence from Example 3.1.1 is not uniformly convergent on
[0, 1].
1
there exist n0 N such that
4
|fn (x) f (x)| < , for n n0 and for all x [0, 1]. It follows
Suppose contrary. Then for =

1
xn < , x [0, 1), n n0 .
4
1
For x = 1 , n n0 , one gets
n

n
1
1
1
< , n N .
n
4
Letting n in the previous relation we obtain

1
1
, contradice
4

tion.
We give in what follows some properties of uniformly convergent sequences of functions.
Theorem 3.1.1. (A test of uniform convergence) Let (fn )n1 be
a sequence of functions, fn : A K, n 1 and f : A K. If there
exist a sequence of positive numbers (an )n1 such that:
(1) lim an = 0;
(2) |fn (x) f (x)| an , for all n n0 and all x A
then fn f on A.
Proof. Let > 0 be given. By (1) follows that there exists n0 =
n N such that an < , n n0 . Then |fn (x) f (x)| < for all
n n0 and all x A, hence fn f on A.

68

Chapter 3
Theorem 3.1.2. (Cauchys criterion) A sequence of functions

(fn )n1 , fn : A K is uniformly convergent on A if and only if for

every > 0 there exists n N such that for all m, n n and all x A
|fn (x) fm (x)| < .
Proof. Necessity. Suppose that fn f on A and take > 0. Then
there exists n N such that

|fn (x) f (x)| < , n n , x A.

2
Then for all m, n N, m, n n and all x A we get
|fm (x) fn (x)| |fm (x) f (x)| + |f (x) fn (x)| <

+ = .
2 2

Suciency. Suppose that for every > 0, there exists n N such

that
(1)

|fm (x) fn (x)| < , m, n n , x A,

2

and prove that there exists f : A A, fn f on A.

Fix x A. Then the sequence of numbers (fn (x))n1 is fundamental
on R, hence it is convergent. Denote
lim fn (x) = f (x).

Letting m in the relation (1) we obtain

|f (x) fn (x)|

< , n n ,
2

therefore fn f on A.
Remark 3.1.2. The Theorem 3.1.2 can be reformulated as follows.
Let (fn )n1 be a functional sequence, fn : A K. Then:
(fn )n1 uniformly convergent (fn )n1 uniformly fundamental.

Sequences and series of functions

69

Theorem 3.1.3. Take (fn )n1 , fn : [a, b] R, be a sequence of

continuous functions, fn f on [a, b]. Then f is continuous on [a, b].
Proof. Let x0 [a, b] and > 0. By the uniform convergence of
(fn )n1 it follows that there exists n N such that

|fn (x) f (x)| < , n n , x [a, b].

3
The continuity of fn0 at x0 implies the existence of > 0 such that
for |x x0 | < :

|fn0 (x) fn0 (x)| < .

3
Now for all x [a, b], |x x0 | < , we have:

|f (x) f (x0 )| |f (x) fn0 (x)| + |fn0 (x) fn0 (x0 )| + |fn0 (x0 ) f (x0 )|
<

+ + = ,
3 3 3

hence f is continuous at x0 .
Theorem 3.1.4. Let (fn )n1 , fn : [a, b] R, be a sequence of continuous functions on [a, b], fn f on [a, b]. Then:
b
b
fn (x)dx =
f (x)dx.
lim
n

Proof. First we remark that f is continuous on [a, b], in virtue of

Theorem 3.1.3, therefore f is integrable on [a, b].
Let be a positive number. By fn f follows that there exists
n N such that
|fn (x) f (x)| <

, n n , x [a, b].
ba

We have:
 b
 b
b





f
(x)dx

f
(x)dx
|f
(x)

f
(x)|dx

n
n


a

dx = ,
ba

70

Chapter 3

for all n n . The theorem is proved.

Theorem 3.1.5. Let (fn )n1 , fn : [a, b] R, be a sequence of dierentiable functions such that:
PC

1) fn f on [a, b];
2) fn g on [a, b].
Then f is dierentiable on [a, b] and f  = g.
Proof. Fix x0 [a, b] and > 0. By the condition 2) follows that
there exists n N such that
(1)


(x)| < , m, n n , x [a, b].
|fn (x) fm
3


g, follows fn 0 fm
fn 0 g, and letting
Take n0 > n . Since fm

m , n = n0 in (1) we obtain
(2)

|fn 0 (x) g(x0 )| , x [a, b].

3

Since fn0 is dierentiable at x0 , there exists U V(x0 ) satisfying



 fn0 (x) fn0 (x0 )



 < , x U [a, b].
(3)

f
(x
)
0
n
0

 3
x x0
Now for m, n n one gets


 fn (x) fn (x0 ) fm (x) fm (x0 ) 





x x0
x x0


 (fn (x) fm (x)) (fn (x0 ) fm (x0 )) 


=

x x0

(c)| <
= |fn (c) fm

where c is the intermediary point from Lagrange theorem applied to the

function fn fm on [a, x].

Sequences and series of functions

71

Letting m , n = n0 in the previous inequality it follows:



 fn0 (x) fn0 (x0 ) f (x) f (x0 ) 
 , x [a, b].

(4)


x x0
x x0  3
For every x U [a, b] we get

 

 f (x) f (x0 )
  f (x) f (x0 ) fn0 (x) f (n0 )(x0 ) 




g(x0 ) 

 x x0

x x0
x x0


 fn0 (x) fn0 (x0 )



+ 
fn0 (x0 ) + |fn 0 (x0 ) g(x0 )| < + + = .
x x0
3 3 3
Since is an arbitrary positive number we have f  (x0 ) = g(x0 ).

3.2. Series of functions

The results obtained for sequences of functions may be adapted to
series of functions using the sequence of partial sums.
Let (fn )n1 , fn : A K, (K is R or C) be a sequence of func

tion. The series of function
fn is said to be pointwise (uniformly)
n=1

convergent if the sequence of its partial sums (sn )n1 ,

sn = f1 + f2 + + fn ,

n 1,

is pointwise (uniformly) convergent.

If there exists lim sn = f , then f is called the sum of
n

We denote

fn .

n=1
PC

fn = f

n=1

and

UC

fn = f

n=1

for pointwise, respectively uniform convergence.

For the evaluation of uniform convergence of a series of functions we
frequently use the next result.

72

Chapter 3
Theorem 3.2.1. (Weierstrass) Let

fn , fn : A K, be a series

n=1

of function and (an )n1 , sequence of positive numbers such that:

1) |fn (x)| an , n n0 , x A;


2)
an is convergent.
n=1

Then

fn is uniformly and absolutely convergent on A.

n=1

Proof. Fix > 0. By the convergence of

an it follows that there

n=1

exists n N such that

an+1 + + an+p < , n n , p 1.

Taking account of 1) we obtain:
|fn+1 (x) + + fn+p (x)| |fn+1 (x)| + + |fn+p (x)|
an+1 + + an+p
for all n max{n , n0 }, p 1 and x A.
Therefore the sequence of partial sum of the
damental. The series
criterion.

fn is uniformly fun-

n=1

fn is uniformly convergent, in view of Cauchys

n=1


cos nx
Example 3.2.1. Prove that the series
is uniformly and
n2 + x2
n=1
absolutely convergent on R.

cos nx
, x R, n N . We have
Solution. Take fn (x) = 2
n + x2


 cos nx 
1
1


 n2 + x2  n2 + x2 n2 , x R, n 1

Sequences and series of functions

73


1
is convergent, the conclusion follows according to Weiern2
n=1
strass test.


and since

The properties of uniformly convergent sequences of functions are

valid for series of functions as follows:
- the sum of a uniformly convergent series of continuous function is a
continuous function
- a uniformly convergent series of continuous functions on the interval
[a, b] can be integrated termwise
- a convergent series of functions with the property that the series of
the derivatives of the terms is uniformly convergent can be dierentiated
termwise.

3.3. Power series

Power series are particular cases of functional series, their sequence of
partial sums is polynomial. They play an important role in the denition
of elementary functions.
Definition 3.3.1. Let (an )n0 be a sequence of complex numbers and
z0 C. A series of functions of the form

an (z z0 )n = a0 + a1 (z z0 ) + a2 (z z0 )2 + . . .

n=0

is called a power series with center z0 .

Denoting z z0 = t we obtain a power series with center 0:

an tn = a0 + a1 t + a2 t2 + . . . ,

n=0

therefore in what follows we consider only power series with center zero.

74


n=0

Chapter 3
Theorem 3.3.1. (First Abels theorem) For every power series
an z n there exists R [0, ) {} such that:
1) The series converges absolutely for z C, |z| < R;
2) The series diverges for z C, |z| > R;
3) The series is uniformly convergent for z C, |z| r < R.
Proof. If the series

an z n converges only for z = 0 the conclusions

n=0

of the theorem are true. We prove that if the series converges for z0 = 0,
then it converges for z C, |z| < |z0 |. By the convergence of the series


an z0n it follows that there exists M > 0 such that
n=0

|an z0n | M, n N.

(1)

Let z C, |z| < |z0 |. We have:


 n
 n
  
 n z n
 
z
n
n  z 


  , n 0.
=
|a
z
|

M
(2)
|an z | = an z0
n
0
 z0 
 z0 
z0 
 n


z
By the convergence of the geoemetric series
M   and (2) folz0
n=0


|an z n |, according to the rst comparison test.
lows the convergence of
n=0

Take now
(3)


R = sup |z0 | :


an z0n is convergent .

n=0

1) For z C, |z| < R, there exists z0 C, |z| < |z0 | < R, and since


n
an z0 is absolutely convergent
an z n is absolutely convergent.

n=0

n=0

2) Let z C, |z| > R, and suppose that

R |z|, contradiction.


n=0

an z n converges. Then

Sequences and series of functions

75

3) For x C, |z| r < R, since the series of numbers

an r n is

n=0

absolutely convergent and

|an z n | |an r n |, n 0,
the series

an z n is uniformly and absolutely convergent, in view of

n=0

Weierstrass criterion.
Definition 3.3.2. Let

an z n be a power series. The unique ele-

n=0

ment R [0, ) {}, dened by Theorem 3.3.1 is called radius of

convergence of the series.
The set
D(0, R) = {z C | |z| < R}
is called the disk of convergence of the series.

Theorem 3.3.2. (Cauchy-Hadamard) Let

an z n be a power se n=0
ries of radius of convergence R such that lim n |an | exists. Then:
n

R=

n

lim

1

n

|an |

|an |, [0, ) {} and let z C be


a xed number. We apply Cauchys root test to the series
|an z n |. We
Proof. Denote = lim

n=0

have:
lim


n

|an z n | = |z| lim


n

|an | = |z|.

1
For |z| < 1 |z| < , the series is convergent.

1
For |z| > 1 |z| > , the series is divergent.

76

Chapter 3
1
It follows R = , in view of Abels rst theorem.

Remark 3.3.1. One prove a stronger result (see [1]). For every power


an z n the radius of convergence is given by the relation
series
n=0

R=

lim

1

n

|a|


an z n be a power series with the property that
Remark 3.3.2. Let
n=0


 an 
 exists. Then
lim 
n  an+1 


 an 

.
R = lim 
n an+1 


 an+1 

n

, in view of a consequence of StolzProof. lim |an | = lim 
n
an 
Cesaro lemma.

For the convergence of a power series for |z| = R we have the following
result.
Theorem 3.3.3. (Second Abels theorem) Let

an xn be a se-

n=0

ries of powers, where (an )n0 is a sequence of real numbers. If the series
converges for x = R or x = R, then its sum is continuous at R respectively R.
For the proof of the theorem see [13].
The properties of powers series follows from the properties of series
of functions and are given bellow.

Sequences and series of functions

77

Theorem 3.3.4. Let (an )n0 be a sequence of real numbers and suppose that

x (R, R).

an xn = S(x),

n=0

Then:
1) S is dierentiable on (R, R) and





n
an x
=
nan xn1 = S  (x) on (R, R).
n=0

n=1

2) For every t (R, R) the following relation holds:


t
t 

t

n
an x dx =
an xn dx =
S(x)dx.
0

n=0

n=0

Example 3.3.1. Find the set of convergence and the sum of the power
series

(1)n n2 xn .

n=0



n 2


n
(1)
 = 1. For x = 1 the series
Solution. R = lim 
n+1
2
n (1)
(n + 1) 
diverges, therefore the set of convergence is (1, 1). We have:
(1)

(1)n xn =

n=0

1
,
1+x

x (1, 1).

By dierentiation of relation (1) one gets

(2)

(1)n nxn1 =

n=0

1
,
(1 + x)2

x (1, 1).

Now multiplying (2) by x, a new dierentiation gives


n=0

(1)n n2 xn1 =

x1
,
(1 + x)2

x (1, 1)

78

Chapter 3

which leads to

(1)n n2 xn =

n=0

x(x 1)
,
(1 + x)2

x (1, 1).

Example 3.3.2. Find the sum of the series of numbers


(1)n1
.
n(n
+
1)
n=1
Solution. The series converges in view of Leibniz test for alternating
series. Now consider the power series


(1)n1 n+1
x .
n(n
+
1)
n=1

We obtain R = 1 and denote by S its sum on (1, 1)

(1)


(1)n1 n+1
x ,
S(x) =
n(n + 1)
n=1

x (1, 1).

By dierentiation one obtains

(2)

S  (x) =


(1)n1
n=1

(3)

S  (x) =

x (1, 1)

xn ,

(1)n1 xn1 =

n=1

1
,
1+x

x (1, 1)

Integrating the previous relation it follows

(4)

S  (x) = ln(1 + x) + C,

x (1, 1).

For x = 0 in (2) and (4) one gets C = 0.

Forward by (4) follows


(5)
S(x) = ln(1 + x)dx = x ln(1 + x)
= (1 + x) ln(1 + x) x + C.

x
dx
1+x

Sequences and series of functions

79

Putting x = 0 in (1) and (5) one gets C = 0.

Since the power series (1) is convergent for x = 1 we have


(1)n1
= lim S(x) = 2 ln 2 1
n(n + 1) x1
n=1

in view of Second Abel Theorem.

3.4. Taylor series

Let I be an open interval in R, f C (I) and x0 I.
Definition 3.4.1. The power series given by
(1)


f (n) (x0 )
n=0

n!

(xx0 )n = f (x0 )+

f  (x0 )
f (x0 )
(xx0 )+
(xx0 )2 +. . .
1!
2!

is called Taylors series of f at the point x0 (or in powers of x x0 ).

For x0 = 0 the corresponding series in called sometimes Maclaurins
series.
The case when Taylors series of the function f at x0 is convergent in
a neighborhood of x0 to that very function is of particular importance.
In this case
(2)

f (x) =


f (n) (x0 )
n=0

n!

(x x0 )n ,

x (x0 , x0 + ).

where (x0 , x0 + ) I.
The function f is said to be expanded into Taylors series in powers
of x x0 .
Generally the equality (2) does not holds. Consider the function

1
e x , x > 0
f : R R, f (x) =
0,
x0

80

Chapter 3
One prove that f (n) (0) = 0 for every n N, therefore the Taylors

series of f at x0 = 0 is given by


0 n
x = 0, x R.
n!
n=0

Hence there is no neighborhood of x0 = 0 such that the equality (2)

holds.
Theorem 3.4.1. The relation (2) holds if and only if the sequence
(Rn (x))n1 of the remainders in Taylors formula converges to zero for
all x (x0 , x0 + ).
Proof. For every n 1 and every x (x0 , x0 + ) we have:
(3)

f (x) =

n

f (k) (x0 )
k=0

k!

(x x0 )k + Rn (x)

in view of Taylors formula. It follows by (3) that the equality (2) holds
if and only if lim Rn (x) = 0 for all x (x0 , x0 + ).
n

Corollary 3.4.1. Suppose that there exists M > 0 such that

(4)

|f (n) (x)| M, n 1, x (x0 , x0 + ).

Then f can be expanded on Taylors series in powers of x x0 on the

interval (x0 , x0 + ).
Proof. For every x (x0 , x0 + ) and every n N we have
Rn (x) =

f (n+1) ()
(x x0 )n+1
(n + 1)!

where = x0 + (x x0 ), 0 < < 1. Then:

|Rn (x)| M

|x x0 |n+1
(n + 1)!

Sequences and series of functions

81

|x x0 |n+1
= 0 it follows lim Rn (x) = 0.
n (n + 1)!
n

and since lim

Finally we give the expansion in Taylors series of some elementary

functions.
Theorem 3.4.2. The following relations hold:
x
x2
+
+ . . . , x R;
1! 2!
x3 x5
sin x = x
+
..., x R
3!
5!
x2 x4
cos x = 1
+
. . . , x R;
2!
4!
x2 x3
+
. . . , |x| < 1;
ln(1 + x) = x
2
3
( 1) 2

x + . . . , |x| < 1, R.
(1 + x) = 1 + x +
1!
2!

1) ex = 1 +
2)
3)
4)
5)

Proof. 1) Let a > 0 and f : (a, a) R, f (x) = ex . Since f (n) (x) =

ex and |f (n) (x)| ea , n N, x (a, a) it follows
f (x) =


f (n) (0)

n!

n=0

x =


xn
n=0

n!

x (a, a)

and since a is arbitrary the relation holds on R.

The relation 2), 3) follows analogously.
4) For |t| < 1 we have the expansion in geometric series
1
= 1 t + t2 . . .
1+t
By integration on [0, x], |x| < 1, the previous relation leads to
ln(1 + x) = x

x2 x3
+
...
2
3

5) The radius of convergence of the binomial series

1+

( 1) 2
x+
x + ...
1!
2!

82

Chapter 3

is



 ( 1) . . . ( n) ( 1) . . . ( n + 1) 


R = lim 
:

n
(n + 1)!
n!


 n
 = 1.
= lim 
n n + 1 
Denote by f (x) its sum for x (1, 1). We have:
( 1) . . . ( n + 1) n1
( 1)
+
x ++
x
+ ...
1!
2!
(n 1)!

f  (x) =
and
xf  (x) =

( 1) 2

( 1) . . . ( n + 1) n
x+
x ++
x + ...
1!
2!
(n 1)!

Summing the last equalities it follows



( 1)

(1 + x)f (x) = +
+ x
1!


( 1)( 2) ( 1)
+
x2 + . . .
+
2!
1!



( 1) 2
= 1+ x+
x + ...
1!
2!
= f (x).
The relation (1 + x)f  (x) = f (x) is equivalent with
(1 + x) f  (x) (1 + x)1 f (x) = 0,
or

f (x)
(1 + x)

x (1, 1)


= 0,

x (1, 1).

Then there exists c R such that

f (x) = c(1 + x) ,

x (1, 1)

Sequences and series of functions

83

By the last relation one gets c = f (0) = 1, hence



f (x) = (1 + x) .

Using power series one can dene some elementary complex functions. For every z C dene the functions ez , sin z, cos z by the relations:
z2
z
+
+ ...
1! 2!
z3 z5
sin z = z
+
...
3!
5!
z2 z4
cos z = 1
+
...
2!
4!

ez = 1 +
(5)

The previous denition is correct, since the radius of convergence of

the power series from the right hand is R = .
A nice relation, called Eulers formula, follows from (5):
(6)

eiz = cos z + i sin z,

z C.

Now, replacing z by z in (6) one obtains:

(7)

eiz = cos z i sin z,

z C.

From (6) and (7) we get:

eiz + eiz
2
,
iz
e eiz
sin z =
2i

cos z =
(8)

z C.

Example 3.4.1. Expand in power series of x the functions:

5 2x
;
1) f : R \ {2, 3} R, f (x) = 2
x 5x + 6
2) f : (1, 1) R, f (x) = arcsin x;
ez
3) f : C \ {1} C, f (z) =
.
1+z

84

Chapter 3
Solution. 1) f (x) =

We have:

5 2x
1
1
=
+
, x R\{2, 3}.
(x 2)(x 3)
2x 3x

and


x  x 2
1+ +
+ ... ,
2
2

|x| < 2,



1
1
1
x  x 2
1
=
+ ... ,
=
1+ +
x3
3 1 x
3
3
3
3

|x| < 3

1
1
1
1
=
=
x
2x
2 1
2
2





1
xn
xn
1
f (x) =
+
=
+ n+1 xn ,
n+1
n+1
n+1
2
3
2
3
n=0
n=0
n=0

|x| < 2.

2) First we expand in power series of x the derivative of f using the

binomial series.

 

1
1
1

2
2

2 12
2
2
f (x) = (1 x ) = 1 +
(x ) +
(x2 )2 + . . .
1!
2!
 
 

1
1
1
1 ... n+ 1

2
2
2
(x2 )n + . . .
+
n!
2n 1
1 3

...

2 2
2 x2n
=1+
n!
n=1
=1+


1 3 . . . (2n 1)
n=1


n=0

2n n!

x2n

(2n)! 2n
x , |x| < 1.
4n (n!)2

By integration on [0, x], |x| < 1, it follows


(2n)!
x2n+1
arcsin x =

,
4n (n!)2 2n + 1
n=0

|x| < 1.

Sequences and series of functions

85

3) The Cauchy product of the expansions

z
z2 z3
+

+ ...,
1! 2!
3!

zC

1
= 1 z + z2 z3 + . . . ,
1+z

|z| < 1

ez = 1

leads to



z
z2 z3
1 +

+ . . . (1 z + z 2 z 3 + . . . )
1! 2!
3!






1
1
1
1
1
1
2
z+ 1+ +
z + 1+ + +
z3 + . . .
= 1 1 +
1!
1! 2!
1! 2! 3!




1
1
n
=
(1) 1 + + +

z n , |z| < 1.
1!
n!
n=0
f (z) =

3.5. Trigonometric series

The harmonic oscillatory motion is described by the equation
(1)

y = A sin(t + ).

2
.

Consider in what follows only functions of period 2 and denote by x

The function from the right hand of (1) is periodic of period T =

their independent variable. Then (1) becomes:

(2)

y = A sin(x + ).
Other functions with the same period are
y = Ak sin(kx + k ),

and their sum

y=

n

k=0

Ak sin(kx + k ),

k N,

n N,

86

Chapter 3

called trigonometric polynomial of order n. Therefore it is naturally to

consider the problem of representation of a periodic function f , with period T = 2, as a trigonometric polynomial, or the problem of expansion
in a trigonometric series, i.e.
(3)

f (x) =

An sin(nx + n ).

n=0

The general term of the series from relation (3), called the nth order
harmonic of f , can be written as follows
An sin(nx + n ) = an cos nx + bn sin nx,
where
an = An sin n ,

bn = An cos n ,

n N.

Definition 3.5.1. A series of functions of the following form:

a0 
+
(an cos nx + bn sin nx)
2
n=1

(4)

where (an )n0 , (bn )n1 are sequences of real numbers is called trigonometric series.
The sequence of partial sums of a trigonometric series is called
trigonometric polynomial. The trigonometric polynomials are functions of period T = 2, hence it suces to study trigonometric series on
an interval of length 2, for example [, ].
Remark 3.5.1. The trigonometric series (4) is convergent if the series
of numbers
|a0 | 
(|an | + |bn |)
+
2
n=1

(5)

Sequences and series of functions

87

is convergent, in view of Weierstrass test. But series (4) can converge

without series (5) being convergent. Abel-Dirichlet test (see chapter 1)
is suitable to investigate the convergence in this case.
We are looking now for a relation between the sum and the coecients
of a trigonometric series.
Lemma 3.5.1. The following relations hold:

sin mx sin nxdx = 0, m, n N, m = n;

cos mx cos nxdx = 0, m, n N, m = n;

(6)

sin mx cos nxdx = 0, m, n N;

2
cos mxdx =
sin2 mxdx = , m N .

Proof. We have


1
sin mx sin nxdx =
(cos(m n)x cos(m + n)x)dx
2


1

sin(m n)x

2(m n)


1


sin(m + n)x = 0.
2(m + n)

The proof of the next two relations is analogous.





sin 2mx 
1 + cos 2mx
1
2
x+
cos mxdx =
dx =
 =
2
2
2m

sin mxdx =

1 cos 2mx
1
dx =
2
2

sin 2mx
x
2m



 =

88

Chapter 3
Theorem 3.5.1. If
a0 
+
(an cos nx + bn sin nx) = f (x)
2
n=1

(7)

uniformly on the interval [, ] then:

1
cos nxdx, n 0,
an =

(8)

1
f (x) sin nxdx, n 1.
bn =

Proof. We integrate the relation (7), taking account of the uniform
convergence, and obtain




cos nxdx + bn
an
a0 +

n=1

Since


sin nxdx =

f (x)dx.

cos nxdx =

sin nxdx = 0

it follows
1
f (x)dx.

Now multiplying relation (7) by cos mx, uniformly convergence is

a0 =

kept, and integrating the relation on [, ], one gets:






a0
cos mxdx +
cos nx cos mxdx+ bn
sin nx cos mxdx
an
2

n=1

=
f (x) cos mxdx.

Taking account of the Lemma 3.5.1 it follows

am
cos2 mxdx =
f (x) cos mxdx

am =

f (x) cos mxdx.

Sequences and series of functions

89

The relation for bm follows analogously multiplying (7) by sin mx and

integrating on [, ].

3.6. Fourier series

Let f : R R be a periodic function of period T = 2, integrable on
the interval [, ].
Definition 3.6.1. The trigonometric series
a0 
(an cos nx + bn sin nx)
+
2
n=1

where the coecients are given by the relations

1
f (x) cos nxdx, n 0,
an =


1
bn =
f (x) sin nxdx, n 1

is called the Fourier series of f and an , bn are called Fourier coecients of f .
Remark 3.6.1. 1) A Fourier series may be convergent or not. If it is
convergent its sum may diers from the function f .
2) If f is an even function then the Fourier coecients of f are given
by the relations
2
an =

f (x) cos nxdx,

bn = 0,

n1

and if f is an odd function then

an = 0,

n0

n0

90

Chapter 3
bn =

f (x) sin nxdx,

n 1.

Proof. The relation for Fourier coecients follows taking account

that if f : [a, a] R is an odd function then
a
f (x)dx = 0
a

and if f is an even function

a
f (x)dx = 2
a

f (x)dx.

The following estimations for the Fourier coecients hold.

Theorem 3.6.1. Suppose that f has a continuous derivative of order
p satisfying on the real axis the inequality
|f (p) (x)| M.
Then the Fourier coecients of f satisfy the inequalities
|an |

2M
,
np

|bn |

2M
,
np

n 1.

Proof. Integrating by parts and taking into account that f () =

f () we have
1
an =

f (x) cos nxdx




1
sin nx
sin nx 

=
f (x)
f (x)
dx


n
n

1
1

f (x) sin nxdx =
f  (x)(cos nx) dx
=
n
n2




1



f (x) cos nxdx
f (x) cos nx
=
n2

1
= 2
f  (x) cos nxdx = . . .
n

Sequences and series of functions

91

Generally an is given by the relation

f (p) (x) cos nxdx, if p is even

np
an =

f (p) (x) sin nxdx, if p is odd
np
Therefore

1
2M
(p)
|f
(x)|dx

.
np
np
The relation for bn follows analogously.
|an |

Theorem 3.6.2. (Bessels inequality) Let f : [, ] R be an

integrable function and (an )n0 , (bn )n1 the Fourier coecients of f .
The following inequality holds:
a20  2
1
+
(ak + b2k )
2

k=1
n

f 2 (x)dx, n 1.

Proof. Let (Tn )n1 be given by

a0 
+
(ak cos kx + bk sin kx),
2
k=1
n

Tn (x) =

We have:

(f (x) Tn (x)) dx =

f (x) 2

x [, ], n 1.

f (x)Tn (x)dx +

Forward


a0
f (x)Tn (x)dx =
f (x)dx
2

n 

f (x) cos kx + bk
+
ak
k=1

n

a20
=
(a2k + b2k )
+
2
k=1

Tn2 (x)dx.


f (x) sin kxdx

92

Chapter 3

and

Tn2 (x)dx

 n


n

a20
(ak cos kx + bk sin kx)
+ a0
4
k=1

2
(ak cos kx + bk sin kx)

k=1

a2
= 0+
2

a20 
+
ak
2
k=1
n

2
(ak cos kx + bk sin kx)

cos2 kxdx +

ak aj

k<j

bk bj

n


bk

sin2 kxdx

cos kx cos jxdx

sin kx sin jxdx +

n


ak bj

k,j=1

k=1

k<j

dx

k=1

n


n


cos kx sin jxdx

(a2k + b2k ).

k=1

Hence


(f (x) Tn (x))2 dx =


2

f (x)dx

a20  2
+
(ak + b2k )
2
k=1
n

which leads to Bessels inequality.

Corollary 3.6.1. Suppose that f : [, ] R is integrable on
[, ] and let (an )n0 , (bn )n0 be the Fourier coecients of f . Then:


(a2n + b2n ) converges;
1) The series
n=1

2) lim an = 0, lim bn = 0;
n

Sequences and series of functions

93

3) If the Fourier series of f is uniformly convergent to f , then Parsevals equality holds, i.e.:
a20  2
1
(an + b2n ) =
+
2

n=1

Proof. 1) The convergence of

f 2 (x)dx.

(a2n + b2n ) follows from Bessels

n=1

inequality.


(a2n + b2n ) < lim an = lim bn = 0.
2)
n=1

3) Letting n in the relation

(f (x) Tn (x))2 dx =

f 2 (x)dx


n
a20  2
+
(an + b2n ) ,
2
k=1

obtained in the proof of Theorem 3.6.2, one gets Parseval equality.

Remark 3.6.2. The following series


cos nx

n
n=1

is an example of a trigonometric series, which is not a Fourier series.


cos nx

Proof. Suppose that

is a Fourier series. Then there exists
n
n=1
a function f : [, ] R such that the Fourier coecients of f , a0 = 0,
1
an = , bn = 0, n 1, satisfy Bessels inequality, i.e.:
n

1
1
1 2
f (x)dx, n 1,
1+ ++
2
n

which leads to contradiction, since



1
1
= .
lim 1 + + +
n
2
n

94

Chapter 3
One of the main problems in the theory of Fourier series is to nd

conditions for a given function f , such that its Fourier series converges
to f . Many mathematicians tried to give an answer to this problem. Here
we shall give some tests for convergence of Fourier series without proving
them.
Definition 3.6.2. A function f : R R is called piecewise
smooth on R if for every compact [a, b] there exists a partition a =
x0 < x1 < < xn = b of [a, b] such that f continuously dierentiable
on (xk1 , xk ), 1 k n, and at every point xk , 0 k n, there exist
the left and the right limits of f and f  and are nite.
Theorem 3.6.3. If f : R R is periodic of period 2, is continuous
throughout the real axis and it has a piecewise smooth derivative, then its
Fourier series if uniformly convergent to f , i.e.
a0 
+
(an cos nx + bn sin nx),
f (x) =
2
n=1

x R.

Theorem 3.6.4. (Dirichlet) Let f : R R be a function of period

2 with the property that there exists a partition of the interval [, ],
= x0 < x1 < < xn = such that on every interval (xk1 , xk ),
1 k n, f is bounded and monotone. Then the Fourier series of f is
convergent on R and
a0 
f (x + 0) + f (x 0)
+
,
(an cos nx + bn sin nx) =
2
2
n=1

x R.

Remark 3.6.3. If f : [0, ] R satises the conditions of Dirichlets theorem, then it can be expanded in a Fourier series of cosinus,
respectively sinus, if we extend the function f to an even function f1 ,

Sequences and series of functions

95

respectively to an odd function f2 on the interval [, ]:


f (x), x [, 0)
f1 (x) =
f (x),
x [0, ]

f (x), x [, 0)
f2 (x) =
0,
x=0

f (x),
x (0, ]
Remark 3.6.4. If f : [a, b] R is a function that satises the conditions of Dirichlets theorem on [a, b] it can be expanded in a Fourier
series as follows.
First we introduce the function f : [a, b] R dened by

f (x), x (a, b]
f(x) =
f (b), x = a
and we extend f on R to a function with period 2T ,
T =

ba
.
2

Now we consider the function g : [, ] R given by

 
T
g(t) = f
t , t [, ].

The following relation holds

a0 
g(t + 0) + g(t 0)
+
(an cos nt + bn sin nt) =
2
2
n=1

where
1
an =

bn =

g(t) cos ntdt,

n0

g(t) sin ntdt,

n 1.

96

Chapter 3

Finally for the function f we get the expansion



a0 
f (x + 0) + f (x 0)
2n
2n
an cos
+
x+bn sin
x =
, x [a, b]
2 n=1
ba
ba
2
and
an =

2
ba

2
bn =
ba

f (x) cos

2n
xdx,
ba

n0

f (x) sin

2n
xdx,
ba

n 1.

a
b

Example 3.6.1. Expand in Fourier series f : [, ] R given by

f (x) = x,

x [, ].

Solution. We have an = 0 for every n N, since f is an odd

function. We have


2
2
bn =
x sin nxdx =
x(cos nx) dx
0
n 0



2
2

x cos nx
=
cos nxdx = (1)n+1 .
0
n
n
0
By Dirichlet theorem one obtains:

2

sin x sin 2x sin 3x

+
...
1
2
3


=

x, x (, )
0, x =

Example 3.6.2. Expand into a Fourier series the function f : R R,

of period 2, given by the relation
f (x) = x2 ,

x [, ],

and prove the relation


1
2
=
.
n2
6
n=1

Sequences and series of functions

97

Solution. Since f is an even function, bn = 0 for every n N .

2
2 2
f (x)dx =
a0 =
0
3





2
2

2
2
x sin nx 2
x cos nxdx =
x sin nxdx
an =
0
0
n
0




4
4
4


x(cos nx) =
cos nxdx = (1)n 2 .
=
x cos nx
2
2
0
n 0
n
n
0
By Dirichlets theorem follows the expansion


(1)n
2
x =
cos nx,
+4
3
n2
n=1
2

x [, ].

For x = 0 in the previous relation we get


1
2
.
=
2
n
6
n=1

Example 3.6.3. Expand f : [0, ] R, f (x) = x, in a Fourier series

of cosinus.
Solution. We extend f to an even function f : [, ] R,
f (x) = |x|,

x [, ].

Then bn = 0 for every n N and

2
a0 =
xdx =
0

2
2
an =
x cos nxdx =
((1)n 1),
0
n2
One gets the expansion

4  cos(2n 1)x
,

2 n=1 (2n 1)2

n 1.

x=

x [0, ].

The complex form of Fourier series can be obtained on the basis

of Eulers formulas.

98

Chapter 3
Let f : [, ] R be an integrable function and an , bn its Fourier

coecients. We have:
an cos nx + bn sin nx = an

einx + einx
einx einx
+ bn
2
2

= cn einx + cn einx
where b0 = 0 and
cn =

an ibn
,
2

cn =

an + ibn
,
2

n 0.

Hence


1
1
(cos nx i sin nx)f (x)dx =
f (x)einx dx
2
2


1
1
=
(cos nx + i sin nx)f (x)dx =
f (x)einx dx
2
2

cn =
cn

n0

The previous two equalities can be written as a unique formula

1
cn =
2

f (x)einx dx,

nZ

and the Fourier series of f takes the form

n=+


cn einx .

n=

Remark 3.6.5. For a complex function of a real variable z : [a, b]

C, z(t) = f (t) + ig(t), where f , g are integrable on [a, b], one dene

z(t)dt =
a

f (t)dt + i
a

g(t)dt.
a

Sequences and series of functions

99

3.7. Problems
1. Prove that the sequence of functions (fn )n1 , fn : R R,
fn (x) =

1
arctan xn ,
n

n N , x R,

is uniformly convergent on R but



lim fn (x) (1) = lim fn (1).
n

2. Let (fn )n1 , fn : [0, 1] R, be a sequence of functions given by

fn (x) = nx(1 x2 )n .
Prove that

1
0

lim fn (x) dx = lim

fn (x)dx.

3. Is it possible termwise integration of the series

2x(n2 en

2 x2

2 x2

(n 1)2 e(n1)

n=1

on the interval [0, 1]?

4. Prove that the series

en sin nx

n=1

is pointwise convergent on R and its sum is continuously dierentiable

on R.
5. Find the set of convergence and the sum of the following power
series:
a)
b)


(1)n1 2n
x ;
n(2n + 1)
n=1


xn
n=1

n(n + 1)(n + 2)

100

Chapter 3
c)
d)


n=1


n=0

n3 xn ;
x2n
;
(2n)!


x3n
;
e)
(3n)!
n=0


x4n+1
;
f)
(4n + 1)!
n=0


n2 + 1 n
x ;
g)
n n!
2
n=0


enx
h)
.
n!
n=0
6. Compute the sum of the following series:


(1)n
;
a)
3n + 1
n=0


(1)n1
b)
;
n(n + 1)(2n + 1)
n=1


(1)n n
;
c)
(2n + 1)!
n=0


(1)n n3
;
d)
(n + 1)!
n=0


(1)n1
;
e)
n(4n
+
1)
n=1


n4
.
f)
n!
n=0
7. Expand in power series of x the functions:

a) f (x) = arctan x, x R;

b) f (x) = ln(x + x2 + 1), x R;

c) f (x) = (arctan x)2 , x R;

d) f (x) = x arctan x ln x2 + 1, x R;

Sequences and series of functions

e) f (x) = ln

101


1 4x +

3x2 ,


1
,
;
3

1
, x R \ {1};
(x 1)3
8 2x
g) f (x) = 2
, x R \ {3, 5};
x 8x + 15
x
h) f (x) = 2
, x R;
(x + 4)2
i) f (x) = sin3 x, x R;
ln(1 + x)
j) f (x) =
, x (1, ).
1+x
8. Expand in powers of x a, a R the following functions:
1
a) f (x) = , x R \ {0};
x
b) f (x) = ln(1 + x), x (1, );

c) f (x) = x, x [0, );
ex
d) f (x) = .
x
9. Expand in Fourier series the functions f : R R, of period 2,
f) f (x) =

given by the relations:


0, x (, 0]
a) f (x) =
x, x (0, ]

ax, x (, 0)
b) f (x) =
, a, b R.
bx, x [0, ]
c) f (x) = eax , x (, ], a R;
d) f (x) = cos ax, x [, ], a R \ Z;
e) f (x) = sin ax, x (, ], a R \ Z
and calculate the sum of the series of numbers


n=1


n=1

n2

1
,
+ a2

1
,
n2 a2

a R;

a R \ Z.

102

Chapter 3
10. Expand in Fourier series the following functions on the given

intervals:
a) f (x) =
b) f (x) =

x
, x (0, 2];
2
sin x, x (0, )

x [, 2]


(1)n1  (1)n1
and nd the sum of the series
,
.
2n 1 n=1 4n2 1
n=1
11. Expand the function f : [0, a] R, a > 0, f (x) = x, in a series
0,

of sinus on [0, a].

12. Expand in a Fourier series of sinus the function f : [0, ] R,
f (x) = x2 , and prove the following relations


sin(2n 1)x
n=1

(2n 1)3

( x)
=
,
8


(1)n1
3
=
.
(2n 1)3
32
n=1

13. Expand in a Fourier series the function f (x) = 10 x on the

interval (5, 15).