Chapter

Integers series 4

4.1. Integers series

Integer series are series of functions of a particular form. They are well suited to the derivation

operation, and therefore to solving differential equations.

X
Definition 1. We call an integer series a series of functions f n such that for all n ∈ N, f n
n
is defined as follows,

f n (z) = an z n ,

the variable z can be real or complex.

4.1.1. Radius of convergence

The convergence radius of an integer series approximately characterises the convergence modes of
X
the series of functions an z n and the analytical properties of the sum.
n
X
Lemma 4.1.1 ( Abel’s lemma). Let an z n be an integer series. Assume that the sequence (an z0n )n
n
is bounded for some z0 ∈ C. Then for all z ∈ C, if |z| < |z0 |, the series with general term an z n is

absolutely convergent.

X
Theorem 1. Let an x n be an integer series. The set of positive reals r such that the

sequence (an r n )n is bounded is an interval of R+ containing 0.

Proof. If (an r n )n is bounded and if 0 6 s < r, then the series with general term an s n is absolutely

convergent according to the previous lemma, so the sequence (an s n )n tends to 0 and is therefore

bounded.

X
Definition 2 ( Convergence radius). The radius of convergence of the integer series an z n

is the element

sup{r > 0 : (an r n )n is bounded} ∈ [0, +∞]

X
If R is the radius of convergence of the integer series an z n , the set of r > 0 such that the

sequence (an r n )n is bounded is [0, R].

Theorem 2.

1. If R = +∞, then for all z ∈ C, the series with general term an z n is absolutely conver-

gent.

2. If R = 0, for all z ∈ C \ {0}, the sequence (an z n )n is not bounded, in particular, the

series diverges.

3. If R 6= 0 and R 6= +∞ :

• For all z ∈ C such that |z| < R, the series with general term an z n is absolutely

convergent.

• If |z| > R, the sequence (an z n )n is not bounded, so the series diverges grossly.

• If |z| = R, we can say nothing in general. We thus have a partition of the complex

plane into three parts (in the last case).

X
Example 1. The series z n is absolutely convergent if |z| < 1 and diverges if |z| > 1, so

we conclude that the radius of convergence is R = 1.

4.1.2. Radius of convergence of a sum and a product
X
Theorem 3. Let R a be the radius of convergence of an integer series an z n , R b that of an
X
integer series bn z n . Then the Radius of Convergence of the sum and product series are
X
greater than min(R a , R b ). And for sum: If R a 6= R b , the radius of convergence of (an + bn )z n

is equal to min(R a , R b ).

Example 2.

1. ∀n ∈ N, an = −bn = 1.

Then R a = R b = 1 but the radius of convergence of the sum series equals +∞.

2. We take (an )n such that,

+∞
1−z X
= an z n , and bn = (−1)n an .
1 + z n=0
Thus, 
+∞ n 1 if n = 0,

X 1+z X
bn z n = and cn = ak bn−k =
1−z 0
n=0 otherwise.
k=0


We then have R a = R b = 1, but the radius of convergence of the product series equals

+∞.

With
+∞ +∞
X 1 − 2z X 1 + 2z
an z =
n
, R a = 1, bn z = . n

n=0
1+z n=0
1−z
R b = 2 and the radius of convergence of the product series equals +∞.

+∞
X
Definition 3. If R is the radius of convergence of an integer series an z n , the open disk
n=0
D◦ (0, R) = {z ∈ C | |z| < R}

is called the disk of convergence of the integer series.

 +∞
X
Remark 1. If R is finite, we don’t know a priori whether an z n will converge for |z| = R.
n=0

X X
Theorem 4. Let an z n and bn z n be two integer series of radii of convergence R a and

R b respectively. Then we have

∃n0 ∈ N, ∀n > n0 : |an | 6 |bn | ⇒ R a > R b ,

and more generally,

∃α ∈ R, ∃k > 0, ∃n0 ∈ N, ∀n > n0 : (|an | 6 k|bn |nα ) ⇒ (R a > R b ),

an
 ‹
•∃α ∈ R, ⇒ (R a = R b ),
bn nα
in particular,
an
 ‹
• ∼ 1 ⇒ (R a = R b ).
bn

4.1.3. Methods for calculating the convergence radius

Theorem 5 ( d’Alembert’s rule). Let (an )n be a sequence of complexes such that,

1. There exists n0 ∈ N such that ∀n > n0 , an 6= 0.

an+1
2. The sequence tends to l ∈ [0, +∞].
an
+∞
X 1
Then the radius of convergence of the integer series an z n is R = ∈ [0, +∞].
n=0
l

Example 3.
+∞
X (−1)n+1
1. Let z n , we calculate the radius of convergence. So, we have,
n=0
(n + 1)n
an+1 (n + 1)n n+1 n 1
 ‹
= = e × 0 = 0.
an (n + 2)n+1 n + 2 n + 2 +∞
+∞
X (−1)n+1 n
So the radius of convergence R = +∞, so z converges in C.
n=0
(n + 1) n
 +∞
X n!
2. z n , radius of convergence, study in ±R? Then, we have,
n=0
(n + 1) · · · (2n + 1)
an+1 (n + 1)n! (n + 1) · · · (2n + 1) (n + 1)2
= × = → .
an (n + 2) · · · (2n + 3) n! (2n + 2)(2n + 3) +∞
So the radius of convergence R = 4.
+∞
X n!
• Study in ±4. For 4n
n=0
(n + 1) · · · (2n + 1)
f n+1 (4) an+1 (n + 1)2 (n + 1)
= 4 =4 =2 < 1,
f n (4) an (2n + 2)(2n + 3) (2n + 3)
so the sequence (4n an )n decreases. We are looking for an equivalent of 4n an in +∞.

We will look for α and k > 0 such that f n ∼ knα . We are looking for α such that the
+∞
fn
 ‹
sequence of general term ln ◦ converges.
n
fn f n−1 fn n−1
 ‹  ‹  ‹  ‹
ln α − ln = ln + α ln
n (n − 1)α f n−1 n
2n n−1
 ‹  ‹
= ln + α ln
2n + 1 n
1 n−1
 ‹  ‹
= − ln 1 + + α ln
2n n
1 1 1
 ‹  ‹
=− α+ +0 2 .
2 n n

1
Thus, if we take α = − , the series converges absolutely, so the sequence with general term
2
fn
 ‹
α
ln converges to λ ∈ R and f n ∼ eλ . i.e., f n ∼ eλn , we have therefore also found k such
an
that
k
fn ∼ p .
n
k
Thus, at R = 4, there is divergence because 4n an ∼ p . And at R = −4, there is convergence
n
by the Leibniz criterion.
4.2. Functional properties of an integer series

+∞
X
Theorem 6. Let an z n be an integer series of radius of convergence R. The series converges
n=0
normally on any compact included in the open disk of convergence (in the case of a complex

variable) or the open interval of convergence (in the case of a real variable). In the real case,

there is in particular normal convergence of the integer series on any segment of type [a, b]

or [−a, a] included in ]−R, R[ .

4.2.1. Continuity of the sum of an integer series

+∞
X
Theorem 7 ( Continuity of the sum of an integer series of real variables). Let an x n
n=0
be an integer series with a real variable, a convergence radius R and a sum S. The function

S is continuous on the open interval of convergence. ]−R, R[ .

Proof. The continuity of the functions. ∀n ∈ N, x 7→ an x n on any interval [a, b] ⊂] − R, R[ and the

normal convergence on [a, b] of the series of these functions (??), means that the sum of this series

(i.e. the sum of the integer series) is continuous on any interval [a, b] ⊂] − R, R[, and therefore on

] − R, R[ itself.

Theorem 8 ( Continuity of the sum of an integer series of complex variables).

+∞
X
Let an z n be an integer series of complex variable, radius of convergence R and sum S.
n=0
The function S is continuous on the open disc D(0, R).

4.2.2. Primitives of the sum of an integer series of real variables

Theorem 9.
+∞
X
Let an x n be an integer series of real variables, of convergence radius R and of sum
n=0
+∞
X
S(x) = an x n .
n=0
 We can integrate S term by term on any segment contained in ] − R, R[. In particular, S has

primitives on ] − R, R[ which are equal to

+∞
X x n+1
c+ an , where c ∈ C.
n=0
n+1
+∞
X
These primitives have the same convergence radius R as an x n .
n=0

Proof. Since S is continuous on ] − R, R[, it has primitives there. Moreover, for 0 6 a < R, the
+∞
X
integer series an x n converges normally sur[−a, a] so we can calculate the primitive term by
n=0
term.

Finally the primitives of S on [−a, a] (which are all equal up to an additive constant) are:
Z +∞ Z +∞
X X x n+1
S(x)d x = c + an x n d x = c + an
n=0 n=0
n+1
where c is a real or complex constant. These new integer series have a convergence radius R p .
+∞ +∞
Xx n+1 X xn
• For x non-zero, the convergence of an is equivalent to that of an , since
n=0
n+1 n=0
n+1
they are equal up to a multiplicative constant. But
an
∀n ∈ N, 6 |an | ,
n+1
and we deduce that R p > R.

• Then ∀z ∈ C : |z| < R p , ∃ρ ∈ R∗ , |z| < ρ < R p . We can then write

ρn |z n |
• ˜
∀n ∈ N : |an z | = |an |
n
(n + 1) n ,
n+1 ρ
n ‹
|z |

and as the sequence (n + 1) n tends to 0, due to the theorem of comparative growths,
ρ n
we deduce that
|z n | ρn
∃n0 ∈ N, ∀n > n0 : (n + 1) 6 1and |a n z n
| 6 |a n | ,
ρn n+1
+∞ +∞
X ρn X
or the series |an | converges (since 0 < ρ < R p ), and therefore the series an x n
n=0
n + 1 n=0
 
converges absolutely. We deduce that |z| 6 R, therefore 0, R p ⊂ [0, R] . and finally R p = R,

and the primitive series have the same radius of convergence as the initial series.
 Example 4. Determine the radius of convergence and the sum of the real integer series,
+∞ +∞ +∞
X X x n+1 X
nx n−1 , , n2 x n−1 .
n=1 n=0
n + 1 n=1

+∞
X +∞
X
n−1
1. nx is the derivative series of the integer series x n with radius of convergence
n=1 n=0
R = 1 and sum
1
S(x) = on ]−1, 1[ ,
1− x
and
+∞
X 1
S (x) =
0
nx n−1 = , ∀x ∈ ]−1, 1[
n=1
(1 − x)2
+∞ +∞
X x n+1 X
2. is the primitive of the integer series x n with radius of convergence R = 1
n=0
n + 1 n=0
and
+∞ Z X+∞ Z
X x n+1 1
= an x d x =
n
d x = − ln(1 − x), ∀x ∈ ]−1, [ .
n=0
n+1 n=0
1− x

3.
+∞
X +∞
X +∞ +∞

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

+∞
X +∞
X
n−2
n(n − 1)x est la série dérivée d’ordre 2 de la série entière x n , donc
n=1 n=0
+∞
X 2x 1 1+ x
n2 x n−1 = xS 00 (x) + S 0 (x) = + = , ∀x ∈ ]−1, 1[ .
n=1
(1 − x)2 (1 − x)2 (1 − x)3

4.2.3. Derivability and C ∞ nature of the sum of an integer series

Theorem 10.
+∞
X
Let an x n be an integer series of real variable, of radius of convergence R and of sum
n=0
+∞
X
S(x) = an x n .
n=0
On ]R; R[, the function S is of class C ∞ and we obtain its successive derivatives by term-by-

term derivation of the function S.

 +∞
X
All the integer series derived from S have the same radius of convergence R as an x n .
n=0
Moreover
+∞
X +∞
X
• ∀x ∈ ]−R, R[ , S (x) =
0
nan x n−1
= (n + 1)an+1 x n
n=1 n=0

+∞ +∞
X (n + p)!
(p)
X n!
• ∀p ∈ N, ∀x ∈ ]−R, R[ , S (x) = an x n−p = an+p x n .
n=p
(n − p)! n=0
n!

+∞
X
The coefficients of the integer series an x n then check
n=0

S (n) (0)
∀n ∈ N, an = .
n!

Proof. First, let’s show that S ∈ C in f t y (] − R, R[). We have

+∞
X
∀x ∈ ]−R, R[ , S(x) = an x n .
n=0
+∞
X
an x n converges normally and therefore uniformly on any interval [−r, r] such that 0 < r < R.
n=0
On the other hand, the function f n : x 7→ f n (x) is of class C ∞ on R and therefore in particular

on [−r, r].

Let us show by recurrence that,


 0 if p > n
f n(p) (x) = n!
an x n−p if p 6 n.
(n − p)!


This expression is true for p = 1, in fact

f n0 (x) = nan x n−1 , ∀n > 1,

and

f n0 (x) = 0forn = 0.
Assume that the hypothesis is true for (p) and prove it for (p + 1). We have



 0 if p > n,
0
 n!
f n(p+1) (x) = f n(p) (x) = (n − p)an x n−p−1 if p < n,



 (n − p)!
p = n,

 0



 0 if p > n − 1,

 n!
an x n−p−1

 if p 6 n − 1,
(n − p − 1)!

=


 0 if p + 1 > n,

 n!
an x n−(p+1) if p + 1 6 n,


(n − (p + 1))!

+∞
X n!
so the relation is true for (p + 1). It remains to show that an x n−p converges uniformly
n=0
(n − p)!
on [−r, r], 0 6 r 6 R.
+∞
X
We have the series an x n and the derivative series having the same radius of convergence R,
n=0
+∞
X
from which f n(p) (x) converges uniformly on [−r, r] ⊂ ]−R, R[ . Therefore, given the derivation
n=0
theorem for series of functions (Theorem ??), we conclude that,

S ∈ C p ([−r, r]) for all p > 1,

S (n) (0)
so S ∈ C ∞
([−R, R[). Let’s show that ∀n ∈ N : an = . We have,
n!

 0 if p > k,
(p)
f k (x) = k!
 ak x k−p if p 6 k,
(k − p)!
‚+∞ Œ(n) +∞ +∞
(n)
X X
(n)
X k!
S (x) = f k (x) = f k (x) = ak x k−n
n=0 n=0 k>n
(k − n)!
+∞
n! X k!
= an x +
0
ak x k−n
0! k>n+1
(k − n)!
+∞
X k!
= an n! + x ak x k−n−1 ,
k>n+1
(k − n)!
and consequently
S (n) (0)
S (n) (0) = an n! ⇒ an = .
n!
 +∞
X
Theorem 11. If the series an x n converges for x = R (or x = −R), then the sum S(x) =
n=0
+∞
X
an x n is continuous on ]−R, R[ (or [−R, R]).
n=0

4.3. Applications of integer series

4.3.1. Functions that can be developed into integer series

Definition 4.

Let I be an interval of R containing 0 and let f be a function of I in R. We say that f is

+∞
X
developable as an integer series at 0 if and only if there exists an integer series an x n of
n=0
non-zero radius of convergence R, and 0 < r 6 R such that
+∞
X
∀x ∈ ]−r, r[ ∩ I, f (x) = an x n .
n=0

4.3.2. Necessary existence conditions for the development of inte-

ger series

Theorem 12.

Let r > 0, and let f be a function of ]−r, r[ in R, developable as an integer series in 0 such

that
+∞
X
∀x ∈ ]−r, r[ , f (x) = an x n .
n=0
(n)
f (0)
Then f is of class C ∞ on ]−r, r[ and ∀n ∈ N : an = .
n!

Proof. f coincides with the sum of an integer series on ]−r, r[. The result follows from theorem

??.
4.3.3. Sufficient condition for integer series development

Theorem 13.

Let f :] − r, r[→ R be a function of class C ∞ in a neighborhood of 0. We assume that there

exists M > 0 such that

for alln ∈ N, and for all x ∈] − r, r[, | f (n) (x)| 6 M . (4.3.1)

+∞ (n)
X f (0) n
Then the series x is simply convergent in ] − r, r[ and we have
n=0
n!
+∞ (n)
X f (0) n
f (x) = x , ∀x ∈] − r, r[.
n=0
n!

Proof. By hypothesis, there exists M > 0 such that for any n ∈ N, and for any x ∈] − r, r[, we have

| f (n) (x)| 6 M .

The Taylor expansion of f in the neighbourhood of 0 to order n gives,

n
X f (k) (0) k f (n+1) (θ x) n+1
f (x) = x + x , with0 < θ < 1.
k=0
k! (n + 1)!
f (n+1) (θ x) n+1
To prove the theorem, it suffices to prove that lim x = 0. Indeed, x ∈] − r, r[
n→+∞ (n + 1)!
(n+1)
⇒ |x| < r ⇒ |θ x| < r ⇒ | f (θ x)| < r, and so,
f (n+1) (θ x) n+1 M
x 6 r n+1 .
(n + 1)! (n + 1)!
M
Now, the series with general term un = r n+1 is convergent because,
(n + 1)!

un+1 r
lim = lim = 0 < 1,
n→+∞ u n→+∞ (n + 1)
n
and as a result
f (n+1) (θ x) n+1
lim x = 0,
n→+∞ (n + 1)!

which yields
+∞ (n)
X f (0) n
f (x) = x .
n=0
n!
 Theorem 14.

Let f :] − r, r[→ R be a function of class C ∞ in a neighborhood of 0. We assume that there

exists M > 0 such that

∀n ∈ N, and ∀x ∈] − r, r[, | f (n) (x)| 6 M . (4.3.2)

+∞ (n)
X f (0) n
Then the series x is simply convergent in ] − r, r[ and we have,
n=0
n!
+∞ (n)
X f (0) n
f (x) = x , ∀x ∈] − r, r[.
n=0
n!

Proof. By hypothesis, there exists M > 0 such that for all n ∈ N, and for all x ∈] − r, r[, we have

| f (n) (x)| 6 M .

The Taylor development of f near 0 to order n gives

n
X f (k) (0) k f (n+1) (θ x) n+1
f (x) = x + x , with 0 < θ < 1.
k=0
k! (n + 1)!
f (n+1) (θ x) n+1
To prove the theorem, it suffices to prove that lim x = 0.
n→+∞ (n + 1)!

In fact, x ∈] − r, r[ ⇒ |x| < r ⇒ |θ x| < r ⇒ | f (n+1) (θ x)| < r, and so,

f (n+1) (θ x) n+1 M
x 6 r n+1 .
(n + 1)! (n + 1)!
M
Or the series with general term un = r n+1 is convergent because
(n + 1)!
un+1 r
lim = lim = 0 < 1,
n→+∞ u n→+∞ (n + 1)
n

and consequently
f (n+1) (θ x) n+1
lim x = 0,
n→+∞ (n + 1)!
which gives
+∞ (n)
X f (0) n
f (x) = x .
n=0
n!

The condition (??) is not necessary, as shown in the following example

 +∞ n
X 2 2n
Example 5. Let f (x) = x n . We put an = . We calculate R.
n=0
n! n!
an+1 2n+1 n!
= → 0 ⇒ R = +∞.
an (n + 1)! 2n n→+∞

Then f is of class C ∞ on R, and we have

f (n) (0)
an =
n!
2n
So, f (n) (0) = an n! = n! = 2n → +∞. f (n) (0) is not bounded, so we conclude that the
n! n→+∞

derivatives of f are not bounded.

4.3.4. Necessary and sufficient condition for integer series develop-

ment

Theorem 15.

Let f be a function of class C ∞ on ] − r, r[. f is developable as an integer series if and only

f (n+1) (θ x) n+1
if the Mac-Laurin remainder x , with 0 < θ < 1 holds,
(n + 1)!
f (n+1) (θ x) n+1
∀x ∈] − r, r[: lim x =0
n→+∞ (n + 1)!

Example 6.

1. The exponential function: x 7→ f (x) = e x .

This function is infinitely differentiable in R, and we have

∀n ∈ N : f (n) (x) = e x .
e(0x)
The Mac-Laurin remainder is x n+1 . We check as before (proof of Theorem ??)
(n + 1)!
that this limit tends to 0 when n tends to +∞, and this whatever x in R. Finally,
+∞ n
x x2 X x
∀x ∈ R : e x = 1 + + + ··· = .
1! 2! n=0
n!

2. The hyperbolic functions.

The cosine-hyperbolic and sine-hyperbolic functions have the same radius of conver-
 gence as the exponential function, in other words, R = +∞.
+∞
e x + e−x x2 x4 X x 2n
cosh x = =1+ + + ··· = ,
2 2! 4! n=0
(2n)!
+∞
e x − e−x x3 x5 X x 2n+1
sinh x = =x+ + + ··· = .
2 3! 5! n=0
(2n + 1)!

3. The circular functions.

(a) The sine function: x 7→ f (x) = sin x. f ∈ C ∞ (R), and we have,

f (x) = sin x, f 0 (x) = cos x, f 00 (x) = − sin x, f 000 (x) = − cos x,

which allows us to deduce that for all p ∈ N :

f (4p) (x) = sin x ⇒ f (4p) (0) = 0,

f (4p+1) (x) = cos x ⇒ f (4p+1) (0) = 1,

f (4p+2) (x) = − sin x ⇒ f (4p+2) (0) = 0,

f (4p+3) (x) = − cos x ⇒ f (4p+3) (0) = −1.

Derivatives of any order are bounded above(upper bounded) by 1, and this

whatever x in R. We then have,

+∞
X x 2n+1
sin x = (−1)n , and R = +∞.
n=0
(2n + 1)!
(b) The cosine function: x 7→ f (x) = cos x.
+∞
X x 2n
f (x) = cos x = (sin x)0 = (−1)n , and R = +∞.
n=0
(2n)!

4. The binomial series (Binomial family)

Considérons la fonction

x 7→ f (x) = y = (1 + x)α , α ∈ R.

Its domain of definition is ] − 1, +∞[. We have a simple relation between the function

and its derivative

y = (1 + x)α ⇒ y 0 = α(1 + x)α−1 ,

hence the differential equation,

y 0 (1 + x) = α y. (4.3.3)
All the solutions to this equation are of the form y = c(1 + x)α , where c is an arbitrary

constant. Now let’s see if there is a solution f that can be developed into an integer

series in the neighbourhood of 0.

X
Suppose that f (x) = an x n is a solution of (??). For such a function to exist it is
n>0
necessary to have the relations,
X X X
(1 + x) f 0 (x) − α f (x) = 0 ⇔ (1 + x) nan x n−1 − α an x n = [(n + 1)an+1 − (α − n)an ] x n
n>1 n>0 n>0

= 0,
then we deduce that, for all n ∈ N,

(n + 1)an+1 − (α − n)an = 0,

and therefore, for all n ∈ N,

α−n
an+1 = an ,
n+1
because an integer series is a zero series if and only if all its coefficients are zero,


 a1 = αa0 ,
α−1



a2 = a1 ,


2


..

.
α−n+2


an−1 = an−2 ,


n−1


 a = α − n + 1a ,



n n−1
n
which finally gives
α(α − 1)(α − 2) · · · (α − n + 1)
an = a0 .
n!
X α(α − 1)(α − 2) · · · (α − n + 1)
Let the series a0 x n , the radius of convergence R is
n>0
n!
given by the relation
1 α(α − 1)(α − 2) · · · (α − n) n! α−n
= lim = lim = 1,
R n→+∞ (n + 1)! α(α − 1)(α − 2) · · · (α − n + 1) n→+∞ n + 1

X α(α − 1)(α − 2) · · · (α − n + 1)
by construction, the series f (x) = a0 x n is a solution
n>0
n!
of the differential equation (??), so it is of the form f (x) = c(1 + x)α .
 Since c = a0 = f (0), we can deduce that for all x ∈ ]−1, 1[ :
X α(α − 1)(α − 2) . . . (α − n + 1)
(1 + x)α = 1 + x n , R = 1.
n>1
n!

This series is known as the binomial series. For example

p X 1 · 1.3 · 5 . . . (2n − 3) n 1 1 2
1+ x =1+ (−1)n−1 n n!
x = 1 + x − x + ···
n>1
2 2 8
1 X 1.3.5 . . . (2n − 1) n 1 3 2
p = 1 + (−1)n x = 1 − x + x − ···
1+ x n>1
2 n n! 2 8

1
5. The function: x 7→ .
1− x
We note on the one hand that for all |x| < 1, lim |x|n = 0 and on the other hand
n→+∞
n+1
1 x
= 1 + x + x2 + · · · + xn + for all |x| < 1.
1− x 1− x
hence,
1 X 1 X
= xn with R = 1 and = (−1)n x n , R = 1.
1− x n>0
1+ x n>0

6. La fonction: x 7→ ln(1 + x).

Some integer series developments can be obtained using the theorems on the integra-

tion and derivation of integer series, so from the series development of the function
1
we deduce by integration that
1+ x
X (−1)n
ln(1 + x) = x n+1 , R = 1.
n>0
n + 1
Similarly, we have
X 1
ln(1 − x) = − x n+1 , R = 1.
n>0
n+1
So,
X xn
∀x ∈ [−1, 1] : ln(1 − x) = − ,
n>1
n
1 X 1 · 3 · 5 · · · (2n − 1) 1 2 3 4
(arcsin x)0 = p =1+ n n!
x 2n
= 1 + x + x + ···
1 − x2 n>1
2 2 8
Knowing that arcsin 0 = 0, we obtain
X 1 · 3 · 5 · · · (2n − 1) 1 3 5
arcsin x = x + x 2n+1 = x + x 3 + x + ···
n>1
2 · 4 · · · · · (2n)(2n + 1) 6 40
We use the same process to develop new functions x 7→ arccos x, x 7→ arg sinh x,

x 7→ arctan x, x 7→ arg th x.
 Example 7. Determine the integer series expansions in the neighbourhood of 0 of the

functions defined by,

ex 5
f (x) = , g(x) = .
1− x x4 − 13x 2 + 36
1
a) We have f (x) = f1 (x) × f2 (x) with f1 : x 7→ e x , f2 : x 7→ .
1− x
The function f1 is developable as an integer series on R, f2 is developable as an integer

series on ] − 1, 1[. We deduce from this,

‚+∞ Œ ‚+∞ Œ +∞ n
X X xn X X 1
∀x ∈] − 1, 1[: f (x) = xn × = an x n with an = .
n=0 n=0
n! n=0 p=0
p!
hence,

f (x) = a0 + a1 x + a2 x 2 + · · ·
n
1 1 1 1 1 1 n
 ‹  ‹ X
= 1 + 2x + 1 + + x + 1+ + +
2
x + ··· +
3
x + ···
1! 2! 1! 2! 3! p=0
p!

b. We gave x 4 − 13x 2 + 36 = (x 2 − 9)(x 2 − 4).

The function g is therefore not defined at the points −3, −2, 2, 3. It follows that the

largest interval with centre 0 on which g can be developed as an integer series is the

interval ] − 2, 2[. Let x be a point in this interval. We have

5 1 1 1
 ‹
= − 2 =− .
(x − 9)(x − 4)
2 2 (x − 9) (x − 4)
2 2
9 · 1 − 9 + 14 − x4
x 2

+∞
1 X x2 x2
For all y such that | y| < 1, we have = n
y . We have < 1 and < 1.
1− y n=0
9 4
As a result
+∞ 
1 1
X ‹
∀x ∈] − 2, 2[, g(x) = x 2n − .
n=0
4n + 1 9n + 1

4.3.5. Solving differential equations in the form of an integer series

Example 8. In each of the following cases, find the functions that are solutions of the

differential equations and that can be developed into an integer series

1. Consider the differential equation

 x 2 (1 − x) y 00 − x(1 + x) y 0 + y = 0 (4.3.4)
X
If an x n is the integer series development of a function f which is a solution to the
n>0
differential equation, we have

X
0
(x) = nan x n−1 ,



 y
n>1
X
00
(x) = n(n − 1)an x n−2 .



 y
n>2
0 00
Substituting y and y into (??), we obtain,
X X X
(??) ⇔ x (1 − x)
2
n(n − 1)an x n−2
− x(1 + x) nan x n−1
+ an x n = 0
n>2 n>1 n>0
X X X
⇔ 2a2 x 2 + (n − 1) [nan − (n − 2)an−1 ] x n − a1 x − [nan + (n + 1)an−1 ] x n + an x n
n>3 n>2 n>0

=0
X
⇔ a0 + a2 x 2 − a1 x 2 + [(n(n − 1) − n + 1)an − ((n − 1)(n − 2) + (n + 1))an−1 ] x n = 0
n>3



 a0 = 0

⇔ a2 − a1 = 0


 (n − 1)2 a − (n − 1)2 a
n−1 = 0, ∀n > 3.

n



 a0 = 0,

⇔ a2 = a1 ,


 a = a , ∀n > 3.

n n−1

On the other hand, by making x zero in the expression of f , we have f (0) = a0 = 0. So

X
the series an x n has radius of convergence R = 1. Consequently, any solution of (??)
n>1
that can be developed as an integer series is proportional to the particular solution,
X x
f1 (x) = xn = .
n>1
1− x

2. Consider the equation

x 2 y 00 + x(1 + x) y 0 − y = 0 (4.3.5)
 X
If an x n is the integer series expansion of a function f which is the solution to the
n>0
differential equation (??), we have
X X X
(??) ⇔ x 2 n(n − 1)an x n−2 + x(1 + x) nan x n−1 − an x n = 0
n>2 n>1 n>0
X X X X
⇔ n(n − 1)an x n + nan x n + nan x n+1 − an x n = 0
n>2 n>1 n>1 n>0
X
⇔ [n(n − 1)an + nan + (n − 1)an−1 − an ] x n + a1 x − a0 − a1 x = 0
n>2
X
⇔ −a0 + (n − 1) [(n + 1)an + an−1 ] x n = 0
n>2

 a0 = 0
⇔
 (n − 1) [(n + 1)a + a ] , ∀n > 2.
n n−1

 a0 = 0
⇔ 1
 an = − an−1 , ∀n > 2
(n + 1)

or else, a0 = 0 and for all n ∈ N∗

1
an+1 = − an
(n + 2)
−1 −1
= × an−1
(n + 2) (n + 1)
−1 −1 −1
= × × an−2
(n + 2) (n + 1) n
(−1)n
=2 a1 .
(n + 2)!
In addition
an+1
lim = 0,
n→+∞ an
hence,

R = +∞,

so the solutions of (??) that can be developed as an integer series are the functions

defined on R by
X (−1)n X (−1)n
f (x) = a0 + a1 x + 2a1 x = 2a1
n
x n+1 ,
n>2
(n + 2)! n>0
(n + 2)!
 and as a result we have
X (−1)n+2

x f (x) = 2a1 x n+2 = 2a1 e−x + x − 1 ,
n>0
(n + 2)!
then any function which can be developed into an integer series and is a solution of

(??) is therefore proportional to the particular solution f0 defined by,

e−x + x − 1
f (0) = 0 and ∀x ∈ R∗ : f (x) = .
x

4.3.6. Definitions of functions of complex variable in integer series

The complex exponential

We have seen that the only function which is equal to its derivative (over an interval) is the

exponential function, and this is why it is used to solve differential equations of order 2. We have
X xn
seen that the radius of convergence of the real integer series is +∞ and that for all x ∈ R,
n>0
n!
X xn
the radius of convergence of the real integer series is +∞
n>0
n!
X xn
ex = .
n>0
n!
X zn
We generalize this expression to all z ∈ C and we set ez = .
n>0
n!

Propriété.

1. For all z1 , z2 ∈ C : ez1 ez2 = ez1 +z2 .

X (−1)n X (−1)n
2. For all x ∈ R : cos x = x 2n , sin x = x 2n+1 .
n>0
(2n)! n>0
(2n + 1)!

3. For all x ∈ R : e i x = cos x + i sin x.

4. For all z ∈ C, For all n ∈ N∗ : (ez )n = e nz .

Proof.
 1. For all z1 , z2 ∈ C we have
‚ Œ ‚ Œ n
X zn X zn X X X z1k z2n−k
1 2
e e =
z1 z2
× cn = cn where cn = .
n>0
n! n>0
n! n>0 n>0 k=0
k!(n − k)!
X X zn X zn
1 2
( cn is the product series of × which are absolutely convergent). Therefore
n>0 n>0
n! n>0
n!
‚ n k n−k
Πn
X 1 X z z
1 2
n! XX z1k z2n−k n!
e z1 e z2 = =
n>0
n! k=0 k!(n − k)! n>0 k=0
n!k!(n − k)!
X 1
= (z1 + z2 )n = ez1 +z2 .
n>0
n!

2. According to the Taylor-Lagrange formula for the neighbourhood of 0 of order 2n + 1:

n  
X
(2k) x 2k (2k+1) x 2k+1 (2n+2) x 2k+2
cos x = cos (0) + cos (0) + cos (ck ) .
k=0
(2k)! (2k + 1)! (2n + 2)!
or for all n ∈ N :

cos(2k) (0) = (−1)n cos x0 = (−1)n et cos(2k+1) (0) = (−1)n+1 sin 0 = 0,

so,
n
X x 2k |x|2n+2
lim cos x − (−1)k 6 lim = 0.
n→+∞
k=0
(2k)! n→+∞ (2n + 2)!

The same procedure for sin x.

3.
X (i x)n X (i x)2p X (i x)2p+1 X (−1) p x 2p X (−1) p x 2p+1
ei x = = + = +i
n>0
n! p>0
(2p)! p>0
(2p + 1)! p>0 (2p)! p>0
(2p + 1)!

= cos x + i sin x,

because (i)2 p = (−1)n , (i)2p+1 = i(−1)n .

Remark 2. The function z 7→ ez is periodic with period 2πi i.e.,

ez+2πi = ez e2πi = ez (cos 2πi + i sin 2πi) = ez .

Hyperbolic functions with complex values

For all z ∈ C:
 ez + e−z X z 2n X z 2n+1 ez − e−z
cosh z = = , sinh z = , th z = .
2 n>0
(2n)! n>0
(2n + 1)! ez + e−z
All these integer series therefore have an infinite radius of convergence.

The functions z 7→ cos z, z 7→ sin z, z ∈ C

e iz + e−iz e iz − e−iz 1 − e2iz

e iz = cos z + i sin z 7→ cos z = , sin z = , tan z = i ,
2 2i 1 + e2iz
sin iz = i sinh z, cos iz = cosh z, tan iz = i th z.

Definition of the function z 7→ ln z

Theorem 16. Let Z be a non-zero complex number, for any complex numberz :

ez = Z ⇔ ∃k ∈ Z/z = ln(|Z|) + i(arg(Z) + 2πk).

In particular, the function exp : C → C∗ , z 7→ ez is surjective.

Proof. Let Z ∈ C, z ∈ C, put z = x + i y where x and y are two real numbers

ez = Z ⇔ e x e i y = Z = |Z|e i arg(Z)

 e x = |Z|,
⇔
 ∃k ∈ Z : y = arg(Z) + 2πk,

⇔ ∃k ∈ Z : z = ln(|Z|) + i(arg(Z) + 2πk).

Example 9. Find ln(−3) and solve the equation cos z = 2. in C.

− 3 = 3e iπ = 3(cos π + i sin π) ⇒ ln(−3) = ln 3 + i(π + 2πk)

cos z = 2
e iz + e−iz
⇔ = 2 ⇔ e2iz + 1 − 4e iz = 0
2
p p
⇔ e iz1 = 2 − 3 ∨ e iz2 = 2 + 3
p p
⇔ iz1 = ln(2 − 3) + i2πk ∨ iz2 = ln(2 + 3) + i2πk
p p
⇔ z1 = −i ln(2 − 3) + 2πk ∨ z2 = −i ln(2 + 3) + 2πk, k ∈ Z.