Chapter 3

Series of functions

1 De…nitions
In all this chapter, we let I be a real interval and we denote by F (I; R) the set of all real
functions de…ned on I: For any f 2 F (I; R), the symbol kf k denotes the supremum of the
function jf j on the interval I; that is
kf k = sup jf (x)j :
x2I

De…nition 1.1 Let (fn )n2N be a sequence of functions in F (I; R). We call the series of gen-
P P
fn the in…nite sum +1
eral term P n=1 fn (x) (or n 1 fn (x)). The sequence (Sm (x))m 1 , where
Sm (x)P = m f
n=1 n (x) for all m 2 N; is called
P+1 the sequence of partial sums associated with the
+1
series n=1 fn (x) : The series Rm (x) = n=m+1 fn (x) is called the rest of order m:

Example 1.2
P+1 n n
1. n=0 x , here fn (x) = x ; x 2 R (geometric series):

P+1 1 1
2. n=0 , here fn (x) = ; x 2 R:
n2 + jxj n2 + x
De…nition 1.3 Let (fn )n2N F (I; R) be a sequence of functions and let (Sm (x))m 1 be
P+1 P+1
the sequence of partial sums associated with the series n=1 fn (x): The series n=1 fn (x)
is said to be convergent at x0 2 I if the sequence (Sm (x0 ))m 1 is. In such a situation, if
P
lim SmP(x0 ) = S (x0 ) we say that the series +1
n=1 fn (x0 ) has the sum equals to S (x0 ) and we
+1
write n=1 fn (x0 ) = S (x0 ). The set
( )
X
+1
D= x2I: fn (x) converge
n=1
P+1
is called the domain
P+1 of convergence of the series n=1 fn (x0 ) : P
The series n=1 fn (x) is said to be absolutely convergent at x0 2 I if the series +1
n=1 jfn (x)j
is convergent.

Example 1.4
P+1
1. The domain of convergence of the series n=0 xn is D = ( 1; 1) :
P+1 xn xn
2. Consider the series n=0 and set u n (x) = : We have
n! n!
un+1 (x) xn+1 n! x
= n
= ! 0 as n ! 1:
un (x) (n + 1)! x n+1
Therefore the domain of convergence of this series is D = R:

1
2 Uniform convergence and normal convergence
In all this section, we let (fn )n2N be a sequence in F (I; R) and (Sm (x))m 1 be the sequence of
P
partial sums associtated with the series +1 n=1 fn (x):
P
De…nition 2.1 The series of functions +1 n=1 fn (x) is said to be uniformely convergent on I if
the sequence (Sm (x))m 1 converge uniformely on I:

The following theorem is a consequence of Theorem 2.7 in Chapter 2.

P
Theorem 2.2 (Cauchy) The series of functions +1 n=1 fn (x) converge uniformely on I if and
only if for all > 0 there exists n 2 N such that for all m; n 2 N,
m+p m+p
X X
m; p n =) fn = sup fn (x) :
x2I
n=m+1 n=m+1

Example 2.3
P P+1 n
1. Consider the geometric series +1 n=0 x n
; x 2 ( 1; 1) : We know that n=0 x = S(x) =
1
for all x 2 ( 1; 1) : For any a 2 (0; 1) we have then
1 x
X
+1
kS Sm k = sup jS(x) Sm (x)j = sup xn
x2[ a;a] x2[ a;a] n=m+1
m
x
= sup am ! 0 as m ! 1:
x2[ a;a] 1 x
P+1
This shows that the geometric series n=0 xn converges uniformely on each interval
[ a; a] with a 2 (0; 1) :
P xn
2. Consider the series +1 n=0 ; x 2 R: Using d’Alembert criterion we see that this series
n!
converge for all x 2 R: and we have For any a > 0
X
+1
xn
kS Sm k = sup jS(x) Sm (x)j = sup
x2[ a;a] x2[ a;a] n=m+1
n!
X
+1
jxjn X
+1
jajn
sup ! 0 as m ! 1:
x2[ a;a] n=m+1 n! n=m+1
n!
P+1 n
This shows that the geometric series n=0 x converges uniformely on each interval
[ a; a] ; with a > 0:
P+1
De…nition 2.4
P+1 The series of functions n=1 fn (x) is said to be normally convergent on I if
the series n=1 kfn k is convergent.
P
Proposition 2.5 If a series +1 n=1 fn (x) is normally convergent, then it converges uniformely.

2
Proof. Observe that for any m; p 2 N we have
m+p m+p
X X
fn kfn k :
n=m+1 n=m+1

P+1
Hence, taking in consideration the convergence of the series n=1 kfn k, we conclude that for
any > 0 there exists n 2 N such that for all m; p 2 N,
m+p m+p
X X
m; p n =) fn kfn k :
n=m+1 n=m+1

P+1
Thus, Theorem 2.2 ensures the uniforme convergence of the series n=1 fn (x) on I:

Example 2.6
P+1 1 P P+1 1
1. The series n=1 converges normally since +1
n=1 kfn k = n=1 2 is convergente.
n2 +x 2 n
P+1 1
2. Consider the Riemann series n=1 x ; x 2 R: It is well known that its domain of con-
n
1
vergence is (1; +1) : Put fn (x) = x : For any a > 1 we have
n
X
+1 X
+1
1 X 1 +1
kfn k = sup x = converges:
n=1 n=1
x a n
n=1
na

P+1 1
This shows that the series n=1 x converges uniformely on each interval [a; +1) with
n
a > 1:

Remark 2.7 The inverse implication P of the that in Proposition 2.5 does not hold. Indeed,
considering the series of functions +1
n=1 fn (x) on [0; 1] where

0, if x 6= 1=n;
fn (x) =
1=n, if x = 1=n:

P P+1 1
We have +1 n=1 kfn k = n=1 is divergent, however the series is pointwise and uniformely
n
convergent to
0, if x 6= 1=n;
S(x) =
1=n, if x = 1=n;
since
X
m X
+1
1
S fn = sup fn (x) = ! 0 as m ! +1:
n=1 x2[0;1] n=m+1 m

3
3 Abel’s creterion of uniform convergence
Theorem 3.1 Let (fn )n2N ; (gn )n2N F (I; R) be two sequences of functions satisfying the
following conditions.
P
i) There exists M > 0 such that kFm k M , where Fn (x) = m n=1 fn (x):
P
ii) The series +1n=1 kgn gn+1 k converges and limn!+1 kgn k = 0:
P+1
Then the series n=1 fn (x)gn (x) converge uniformely on I:

Proof. For all m; p 2 N we have

m+p m+p
X X
fn (x) gn (x) = (Fn (x) Fn 1 (x)) gn (x)
n=m+1 n=m+1
m+p m+p
X X
= Fn (x) gn (x) Fn 1 (x) gn (x)
n=m+1 n=m+1
m+p m+p 1
X X
= Fn (x) gn (x) Fn (x) gn+1 (x)
n=m+1 n=m
m+p 1
X
jFn (x)j jgn (x) gn+1 (x)j + jFm (x)j jgm+1 (x)j + jFm+p (x)j jgm+p (x)j
n=m+1
m+p 1
X
M kgn gn+1 k + M kgm+1 k + M kgm+p k ;
n=m+1

leading to
m+p m+p 1
X X
fn gn M kgn gn+1 k + M kgm+1 k + M kgm+p k :
n=m+1 n=m+1

Therefore, from Condition ii) we conclude that any > 0 there is n 2 N such that
m+p 1
X
kgn gn+1 k ; kgm+1 k and kgm+1 k for all m n:
n=m+1
3M 3M 3M

At the end, we obtain for all m; p 2 N with m n

m+p m+p 1
X X
fn gn M kgn gn+1 k + M kgm+1 k + M kgm+p k
n=m+1 n=m+1

M + +
= :
3M 3M3M
P
By Theorem 2.2 we conclude that the series +1
n=1 fn (x)gn (x) converge uniformely on I:

4
 P+1 sin(nx) P+1 cos(nx)
Example 3.2 Consider the series n=1 and n=1 with > 0 and x 2
n n
[a; a] with a 2 (0; =6) : Set gn (x) = 1=n : Clearly, we have kgn k = 1=n ! 0 as n ! +1
and !
1 1 1 1
kgn gn+1 k = = 1 1+ v +1 at + 1:
n (n + 1) n n n
This shows that Condition ii) of Theorem 3.1 is satis…ed.
Now, taking in account that if z = x + iy, then

max (jxj ; jyj) jzj

and
X
m X
m X
m
n 1 ei(n+1)x
cos(nx) + i sin(nx) = eix = :
n=0 n=0 n=1
1 eix
Hence, we have
! s
X
m X
m
1 ei(m+1)x 1 cos ((m + 1) x)
max cos(nx) ; sin(nx) =
n=0 n=0
j1 eix j 1 cos (x)
1 1
p =p :
1 cos (x) 2 jsin (x=2)j
This leads to
!
X
m X
m
1
max sup cos(nx) ; sup sin(nx) p :
x2[a; a] n=0 x2[a; a] n=0 2 jsin (a=2)j

P+1 sin(nx) P cos(nx)

By Abel’s crirerion each of the series n=1 and +1
n=1 converges uniformely
n n
on [a; a] :
( 1)n P
Example 3.3 Consider the series n 1 . It is well known that this series has its domain
nx
of converges equals to (0; +1) : We prove below that it converges uniformely on any interval
having
Pm the form Pm [a; +1) with a > 0. Put fn (x) = ( 1)n and gn (x) = 1=nx . First, we have
k n=1 fn k = j n=1 ( 1)n j 1 for all m 2 N and kgn k = 1=na ! 0 as n ! 1: Then
1 1 1 1
kgn gn+1 k = sup = sup 1 exp x ln 1 +
x a nx (n + 1)x x a nx n
( 1
, if a > 1;
na
max supx2[a;b] hn (x); supx b hn (x) ; if a 1<b
8
< 1 , if a > 1;
>
= na
> b 1
: + b ; if a 1 < b:
n a+1 n
P
This shows that n 1 kgn gn+1 k converges.

5
4 Properties of the uniform convergence
In all this section, we let (fn )n2N F (I; R) be a sequence of functions:
P+1
Theorem 4.1 SupposeP+1 that n=1 fn converge uniformely on I and fn is continuous on I for
all n 2 N. Then n=1 fn is continuous I: That is
X
+1 X
+1 X
+1
lim fn (x) = lim fn (x) = fn (a) for all a 2 I:
x!a x!a
n=1 n=1 n=1

Proof. Applying Corollary 3.5 of Chapter sequences of functions, we obtain for all a 2 I
! !
X
+1 Xm Xm
lim fn (x) = lim lim fn (x) = lim lim fn (x)
x!a x!a m!+1 m!+1 x!a
n=1 n=1 n=1
! !
X
m X
m
= lim lim fn (x) = lim fn (a)
m!+1 x!a m!+1
n=1 n=1
X
+1
= fn (a) :
n=1

The theorem is proved.

P+1
Theorem 4.2 Suppose that P n=1 fn converge uniformely on [a; b] I and fn is integrable on
[a; b] for all n 2 N. Then +1
n=1 fn is integrable on [a; b] and we have
Z b X ! +1 Z b
+1 X
fn (t) dt = fn (t) dt :
a n=1 n=1 a
P+1 R x R x P+1
Moreover, the series of functions n=1 a
fn (t) dt converge uniformely to a n=1 fn (t) dt:
Proof. Applying Theorem 3.2 of Chapter sequences of functions, we obtain
Z b Z b ! Z b X !
Xm Xm +1
lim fn (t) dt = lim fn (t) dt = fn (t) dt;
m!+1 a a m!+1 a
n=1 n=1 n=1

proving that !
+1 Z
X b Z b X
+1
fn (t) dt = fn (t) dt:
n=1 a a n=1
P+1
Now,
P set S(x) = n=1 fn (x) and let > 0 be arbitrary. There exists n 2 N such that
k m f
n=1 n Sk = (b a) for all m n : Hence, for all x 2 [a; b] and all m n we have
Z x Z x Z x Z x X !
Xm m
S (t) dt fn (t) dt = S (t) dt fn (t) dt
a n=1 a a a n=1
Z x X
m
S (t) fn (t) dt
a n=1
Z b X
m
fn S dt :
a n=1

6
The theorem is proved.

Example 4.3 We know that

1 X n
= t2 for all t 2 ( 1; 1) :
1 + t2 n 0

Since this series converges uniformely on any interval [a; b] ( 1; 1) ; for all x 2 ( 1; 1) we
have
Z x Z x X !
dt n
X Z x X ( 1)n
2 2 n
arctan (x) = 2
= t dt = t dt = x2n+1 :
0 1+t 0 n 0 n 0 0 n 0
2n + 1

Letting x ! 1 we get
X ( 1)n
= :
n 0
2n + 1 4

Theorem 4.4 Suppose that the following conditions are satis…ed.

i) fn is continuously di¤erentiable on I for all n 2 N:

P+1
ii) n=1 fn (x0 ) converges for some x0 2 I:
P+1 0
iii) n=1 fn (x) is uniformely convergent on I:
P
Then +1 n=1 fn (t) converges uniformely on I and for all x 2 I we have
!0 +1
X
+1 X
fn (x) = fn0 (x) :
n=1 n=1

Proof. We have Z !
X
m X
m x X
m
fn (x) = fn (x0 ) + fn0 (t) dt:
n=1 n=1 x0 n=1

Passing to the limit when m ! +1, we obtain from Theorem 4.2

Z ! !
X
+1 X
m x X
m
fn (x) = lim fn (x0 ) + fn0 (t) dt
m!+1 x0
n=1 n=1 n=1
Z !
X
+1 x X
+1
= fn (x0 ) + fn0 (t) dt:
n=1 x0 n=1

This ends the proof.

The following corollary is a consequence of the above theorem.

Corollary 4.5 Suppose that the following conditions are satis…ed.

i) fn is of class C k on I for all n 2 N:

7
 P+1 (i)
ii) n=1 fn (x0 ) converges for some x0 2 I and for i 2 f0; 1; : : : ; k 1g :
P+1 (i)
iii) n=1 fn (x) is uniformely convergent on I for i 2 f0; 1; : : : ; k 1g :
P+1
Then n=1 fn (t) converges uniformely on I and for all i 2 f0; 1; : : : ; kgall x 2 I we have
!(i)
X
+1 X
+1
fn (x) = fn(i) (x) :
n=1 n=1

Example 4.6 In this example, we will show that

X
+1 n
x
x
e = for all x 2 R:
n=0
n!

We have seen that

P Dxc n = R and it is clear that the conditions of Theorem 4.2 are satis…ed.
Setting S(x) = +1
n=0 n! , we have

X
+1
nxn 1 X
+1
xn 1 X
+1 n
x
0
S (x) = = = = S(x):
n=1
n! n=1
(n 1)! n=0 n!

Hence, S satis…es the di¤erential equation y 0 = y with the intial condition y(0) = 1: We conclude
that S(x) = ex :

Example 4.7 Here, we will show that

X
+1
( 1)n 2n X
+1
( 1)n 2n+1
cos x = x and sin x = x for all x 2 R:
n=0
(2n)! n=0
(2n + 1)!

Appying d’Alembert criterion, we obtain that the two series converge for allPx 2 R: Moreover,
+1 ( 1)n 2n
they satisfy the conditions of Corollary 4.4 for k = 2: Setting U (x) = n=0 (2n)! x and
P+1 ( 1)n 2n+1
V (x) = n=0 (2n+1)! x we have

X
+1
( 1)n 2n 2n X
+1
( 1)n 2n X
+1
( 1)n 2n+1
0 1 1
U (x) = x = x = x = V (x)
n=1
(2n)! n=1
(2n 1)! n=0
(2n + 1)!

and
X
+1
( 1)n (2n + 1) 2n X ( 1)n 2n
+1
V 0 (x) = x = x 1
= U (x):
n=0
(2n + 1)! n=1
(2n)!
These show that U and V are solutions of the di¤erential equation y 00 + y = 0 with the intial
conditions U (0) = 1; U 0 (0) = 0 and V (0) = 0; V 0 (0) = 1: We conclude then U (x) = cos x and
V (x) = sin x: