Improper Integral

May 3, 2025

Introduction In this chapter we will extend the notion of integration to include cases where
the function is unbounded or the interval is unbounded or both.

1 Definitions and properties

1.1 Definition
Let I be an non empty interval ;
A function f is said locally integrable on I if and only if f is integrable on each
bounded and closed subinterval of I.
The set of locally integrable functions on I is denoted RLoc (I).

Example

1. f1 (x) = ex beeing continuous on R then f1 ∈ RLoc (R).

1
2. f2 (x) = beeing continuous on R∗ then f2 ∈ RLoc (R+? ) et f2 ∈ RLoc (R−? )
x

1.2 Definition
1. Let f ∈ RLoc [a, b[ with b ∈ R ∪ {+∞}
Z y 
• If ∃l ∈ R such that lim f (t) dt = l ,then we say that the improper integral
y→b a
<
Z b Z b
f (t) dt converges and we write: f (t) dt = l
a a
Z y 
• If lim f (t) dt is infinite or doesn’t exist , we say that he improper integral
y→b a
<
Z b
f (t) dt diverges .
a
• In the two
Z b
previous cases, we say that b is the singularity of the improper
integral f (t) dt.
a

2. Let f ∈ RLoc (]a, b]) with a ∈ R ∪ {−∞}

1
 Z b !
• If ∃l ∈ R such that lim f (t) dt = l ,then we say that the improper integral
y→a y
>
Z b Z b
f (t) dt converges and we write f (t) dt = l .
a a
Z b !
• If lim f (t) dt is infinite or doesn’t exist , we say that the improper integral
y→a y
>
Z b
f (t) dt diverges.
a
• In the two
Z b
previous cases, we say that a is the singularity of the improper
integral f (t) dt.
a

Example Detemine whether the following intergals converge or diverge:

Z +∞
1
1. dt.
0 1 + t2
1
Let f (t) = , f is continuous on R so f ∈ RLoc [0, +∞[
1 + t2
Z +∞ Z y
π

f (t) dt = lim f (t) dt = lim (arctgy − arctg0) = ∈ R.
0 y→+∞ 0 y→+∞ 2
Z +∞ Z +∞
π
Conclusion: f (t) dt is convergente and f (t) dt = .
0 0 2
Z 1
1
2. dt.
0 t
1
Let f (t) = , f is continuous on ]0, 1] so f ∈ RLoc ]0, 1]
t
Z 1 Z 1 
f (t) dt = lim f (t) dt = lim (Log(1) − Log(y)) = +∞.
0 y→0 y y→0
> >

Z 1
1
Conclusion: dt is divergente.
0 t

Proposition
Let f ∈ RLoc ([a, b[) with b ∈ R.
Z b
If ∃l ∈ R such that lim f (t) = l then f (t) dt converges, moreover
t→b a
<

Z b Z b
f (t) dt = f˜(t) dt
a a

where f˜ is the extension by continuity of f on [a, b].

Z b
We say that b is a removable singularity of f (t) dt.
a

2
proof 1.2.1
 :
f (t) si t ∈ [a, b[
f˜(t) =
l si t=b
1. Let’s prove that f˜ is Riemann integrable on [a, b]: we have
lim f˜(t) = l ∈ R =⇒ f˜ is bounded in a left neighborhood of b
t→b
<

=⇒ ∃η > 0 such that f˜ is bounded on [b − η, b]

f ∈ RLoc ([a, b[) =⇒ f is R.I on [a, c] ∀c < b

=⇒ f is bounded on [a, c] ∀c < b
=⇒ f is bounded on [a, b − η]
=⇒ f˜ is bounded on [a, b − η] since f˜(x) = f (x) ∀x ∈ [a, b − η]
Conclusion 1: f˜ is bounded on [a, b − η] and bounded on [b − η, b] so f˜ is bounded on [a, b].
Conclusion 2: we proved that
• f˜ is defined and bounded on [a, b].
• f˜ is integrable on each interval [a, d] d < b since f ∈ RLoc ([a, b[) and f = f˜ on [a, b[.
Therfore f˜ is integrable on [a, b].(??)
Z b
2. Let’s prove that f (t) dt converges. We have
a
Z y  Z y 
lim f (t) dt = lim f˜(t) dt since f (t) = f˜(t) on [a, b[
y→b a y→b a
< <
Z b
= f˜(t) dt since f˜ is R.I on [a, b]
a
Z b Z b Z b
Conclusion: f (t) dt converges moreover f (t) dt = f˜(t) dt.
a a a

Remark
1. Be prudent, the previous proposition isn’t true in the case b = +∞ orb = −∞

2. We have the same result if f ∈ RLoc (]a, b]) with a ∈ R .

Example Detemine whether the following integrals converge or diverge:

Z 1
sin t
1. dt
0 t
sin t
Let f (t) = ; f is continuous on ]0, 1] so f ∈ RLoc (]0, 1])
t Z 1
sin t
lim f (t) = 1 ∈ R =⇒ dt converges.
t→0
>
0 t
We say that 0 is a removable singularity of the improper integral.
Z 1
Logt
2. dt
1
2
t−1
Logt h h h h
Let f (t) = ; f is continuous on 12 , 1 so f ∈ RLoc ( 12 , 1 )
t−1 Z 1
Logt
lim f (t) = 1 ∈ R =⇒ 1 dt converges. (1 is a removable singularity. )
t→1
< 2
t − 1

3
Remark

1. If lim f (t) doesn’t exist or is infinite , we can’t say any thing about the the convergence
t→b
<
Z b
of f (t) dt.
a

2. The infinity is always a singularity:

1.3 Proposition
Let f ∈ RLoc ([a, b[) with the b ∈ R ∪ {+∞} .
∀c ∈ [a, b[, the improper integrals :
Z b Z b
f (t) dt and f (t) dt either both converge or both diverge.
a c Z b Z c Z b
And when they converge, we obtain: f (t) dt = f (t) dt + f (t) dt
a a c

proof 1.3.1 : The theoremZ means that

Z b b
f (t) dt converges ⇐⇒ f (t) dt converges .
a Z b c Z y
By definition f (t) dt converges ⇐⇒ lim f (t) dt ∈ R
a y→b a
<
By the Chasles’ relation Z y Z c Z y
f (t) dt = f (t) dt + f (t) dt
a a c
Z c
Since f (t) dt is a Riemann integral , then
a
Z y Z y
lim f (t) dt exists ⇐⇒ lim f (t) dt exists
y→b a y→b c
< <

Z b Z b
therefore f (t) dt and f (t) dt have the same behavior at b, and when they converge we
c a
obtain:
Z b Z y Z c Z y 
f (t) dt = lim f (t) dt = lim f (t) dt + f (t) dt
a y→b a y→b a a
< <
Z c Z y
= f (t) dt + lim f (t) dt
a y→b c
<
Z c Z b
= f (t) dt + f (t) dt
a c

Z 1000
sin t Z 0,00001
sin t
Example dt and dt either both converge or both diverge.
0 t 0 t

Remark We have the same result when f ∈ RLoc (]a, b]) with a ∈ R ∪ {−∞}.

4
 1.4 Properties
∀f, g ∈ RLoc ([a, b[) we have:
Z b Z b Z b
1. If f (t) dt and g(t) dt converge then ∀β ∈ R (f (t) + βg(t)) dt is convergent
a a a
and Z b Z b Z b
(f (t) + βg(t)) dt = f (t) dt + β g(t) dt.
a a a
Z b Z b
2. If f (t) ≥ 0 ∀t ∈ [a, b[ and f (t) dt converges then f (t) dt ≥ 0.
a a
Z b Z b
3. If f (t) dt and g(t) dt converge and f (t) ≥ g(t) ∀t ∈ [a, b[
a Z b Za b
then f (t) dt ≥ g(t) dt.
a a

proof 1.4.1 1. f, g ∈ RLoc ([a, b[) so (f (t) + β g(t)) ∈ RLoc ([a, b[)
Assume that both integrals are convergent so
Z y Z y Z y 
lim (f (t) + βg(t)) dt = lim f (t) dt + βg(t) dt
y→b a y→b
< < | a {z a }
Riemannintegration
Z y Z y
= lim f (t) dt + lim βg(t) dt property of finite limits
y→b a y→b a
< <
Z b Z b
= f (t) dt + β g(t) dt. since both integrals converge.
a a

2. Result of limit of functions.

3. Result of limit of functions.

Remark

1. We have the same results if f, g ∈ RLoc (]a, b]).

Z b Z b Z b
2. If f (t) dt converges and g(t) dt diverges then (f (t) + g(t)) dt diverges.
a a a

1.5 First Basic improper integrals

1. Let α ∈ R ,and ∀c > 0

Z +∞
1
dt converges ⇐⇒ α > 1.
c tα

We call it the Improper Riemann integral in infity.

5
 2. Let α ∈ R , ∀c > 0 Z c
1
dt converges ⇐⇒ α < 1.
0 tα
We call it the Improper Riemann integral in 0.

3. Let α ∈ R , ∀c ∈ R
Z +∞
eαt dt converges ⇐⇒ α < 0.
c

We call it the improper exponential integral .

Examples Determine whether the integrals converge or diverge.

Z +∞
1
1. dt converges since I.R.I at +∞ with 2 > 1.
1 t2
Z 5
1
2. dt diverges since I.R.I at 0 with 2 ≥ 1.
0 t2
Z +∞
3. et dt diverges since I.I.exp with 1 ≥ 0.
0
Z +∞
4. e−t dt converges since I.R.I exp with −1 < 0.
5

1.6 Definition: Improper integral with an interior singularity

Let f be a function such that f ∈ RLoc ([a, x1 [) and f ∈ RLoc (]x1 , b])
Z b Z x1 Z b
f (t) dt converges ⇐⇒ f (t) dt converges and f (t) dt converges.
a a x1
Z b Z x1 Z b
In the case of convergence, we write: f (t) dt = f (t) dt + f (t) dt
a a x1

1.7 Definition: Improper integral with more than one singularity

Let f ∈ RLoc (]a, b[).
Z b Z c Z b
f (t) dt converges ⇐⇒ ∃c ∈ ]a, b[ such that f (t) dt converges and f (t) dt converges.
a Z b a Z c Z b c

In the case of convergence , we write: f (t) dt = f (t) dt + f (t) dt.

a a c

Remark
1. Let f ∈ RLoc (]a, b[),
Z b Z c Z b
f (t) dt diverges ⇐⇒ ∀c ∈ ]a, b[ , f (t) dt diverges or f (t) dt diverges
a Zac Zc b
⇐⇒ ∃c ∈ ]a, b[ , f (t) dt diverges or f (t) dt diverges
a c

6
 2. Let x1 , x2 , ..., xn ∈ ]a, b[ such that x1 < x2 < x3 < ... < xn−1 < xn and f a function such
that:
f ∈ RLoc ([a, x1 [) , f ∈ RLoc (]x1 , x2 [) , ...., f ∈ RLoc (]xn−1 , xn [) , f ∈ RLoc (]xn , b]) ;
By definition:
Z b Z x1 Z x2 Z b
f (t) dt converges ⇐⇒ f (t) dt , f (t) dt ,... f (t) dt are convergente.
a a x1 xn

1 Z +∞ Z 1
1 Z +∞
1
Example 2
dt converges ⇐⇒ 2
dt converges and dt converges.
0 t 0 t Z +∞ 1 t2
Z 1
1 1
Since dt diverges I.R.I at 0 with 2 ≥ 1 consequently dt diverges.
0 t2 0 t2

2 Improper integrals of nonnegative functions

2.1 Comparison test for integrals

Proposition Let f, g ∈ RLoc ([a, b[) such that:
0 ≤ f (t) ≤ g(t) ∀t ∈ [a, b[
Z b Z b
F If g(t) dt converges then so does f (t) dt .
a a
Z b Z b
F If f (t) dt diverges then so does g(t) dt .
a a

Example :
Z +∞
sin2 (t)
1. Determine whether dt converges or diverges.
1 t3
Z +∞ √
t
2. Determine whether e dt converges or diverges.
1

Remark
1. In the comparison test, we can replace the assumption: 0 ≤ f (t) ≤ g(t) ∀t ∈ [a, b[ by
0 ≤ f (t) ≤ g(t) in a left neighborhood of b
2. We have the same result as the previous one when f ∈ RLoc (]a, b]).

2.2 Equivalence test

Proposition Let f, g ∈ RLoc ([a, b[) such that: f ≥ 0 on [a, b[.
Z b Z b
If f ∼− g then f (t) dt and g(t) dt either both converge or both diverge.
b a a

7
Example
Z +∞
1 1
1. Determine whether e t dt is cv or dv
1 t2
Z +∞
Log(1 + t4 e−t )
2. Study the convergence of dt
0 t2 + t4
Log(1 + t4 e−t )
f (t) = ∈ RLoc (]0, +∞[
t2 + t4
The improper integral has two singularities at 0 and at +∞. By definition,
Z +∞ Z 1 Z +∞ 
f (t) dt converges ⇐⇒ f (t) dt converges and f (t) dt converges
0 0 1
Z 1
F Study of the convergence of f (t) dt :
0

Log(1 + t4 e−t ) ∼ t4 e−t since lim t4 e−t = 0


Log(1 + t4 e−t ) t4 e−t


 0 t→0
and =⇒ ∼ = t2 e−t
 t2 + t4 ∼ t2

 t2 + t4 0 t2
0

lim t2 e−t = 0 =⇒ lim f (t) = 0 ∈ R

t→0 t→0
=⇒ 0 is a removable singularity f
so Z 1
f (t) dt converges.
0
Z +∞
F Study of the convergence of f (t) dt:
1
Log(1 + t4 e−t ) ∼ t4 e−t since lim t4 e−t = 0

Log(1 + t4 e−t ) t4 e−t


 +∞ t→+∞
and =⇒ ∼ =
 t2 + t4 ∼ t4

 t2 + t4 +∞ t4
+∞
e−t
e−t ≥ 0 donc f (t) ≥ 0 ; By the equivalent test :
Z +∞ Z +∞
f (t) dt et e−t dt have the same behavior at +∞
1 1

By the first basic examples

Z +∞
e−t dt converges since Improper expo integral with − 1 < 0
1

thus Z +∞
f (t) dt converges.
1

Conclusion:
Z 1 Z +∞ Z +∞
f (t) dt et f (t) dt converges then f (t) dt converges.
0 1 0

8
Remark
1. The previous proposition is true even if f is a non negative in a left neighborhood of b.
Z b
2. The equivalence test remains valid in the case f ≤ 0 since the integrals f (t) dt ,
Z b a

(−f (t)) dt either both converge or both diverge

a

3. The equivalence test remains valid when f, g ∈ RLoc (]a, b]) .

2.3 The absolute convergence

Definition Let f ∈ RLoc ([a, b[) ,

Z b Z b
f (t) dt converges absolutely if and only if |f (t)| dt converges .
a a

Z +∞
sin(t)
Example Determine if the following integral converges absolutely: 3 dt.
1 t2
|sin(t)|
|f (t)| = 3 ∈ RLoc ([1, +∞[) ; only One singularity at +∞.
t2
1
0 ≤ |f (t)| ≤ 3
t2
We know that Z +∞
1 3
3 dt converges car I.R à +∞, with > 1
1 t
2 2
thus by the comparison test Z +∞
|f (t)| dt converges
1
Z +∞
so f (t) dt is absolument convergente.
1

2.3.1 The absolute convergence test

Let f ∈ RLoc ([a, b[); We have
Z b Z b
If f (t) dt converges absolutely then f (t) dt converges.
a a

Example :
Z +∞
sin(t)
We have already proved that 3 dt is absolument convergente ; thus , by the previous
1 t2
Z +∞
sin(t)
proposition 3 dt is convergente.
1 t2

Remark :
1. We say that the absolute convergence implies the convergence.
2. Le converse of the previous proposition is false.
3. We have the same result if f ∈ RLoc (]a, b]).

9
2.3.2 The conditional convergence

Definition : Let f ∈ RLoc ([a, b[),

Z b Z b Z b
f (t) dt is conditionally convergente if and only if f (t) dt converges and |f (t)| dt is
a a a
divergente.

2.4 The Abel 1 test

Theorem
Let f, g ∈ RLoc ([a, b[) :
If

F f is monotone on [a, b[, and limf (t) = 0

t→b
<
Z c
F ∃M ∈ R+ such that ∀c ∈ [a, b[ g(t) dt ≤ M.
a
Z b
then f (t)g(t) dt converges.
a

Example
Z +∞
sin(t)
1. Detemine whether dt converges or diverges.
2 log t
Z +∞
sin(t)
2. Prove that dt is conditionally convergent.
2 log t

2.5 3rd Basic Examples

Z +∞
sin(βt)
For c > 0 and β 6= 0, Let dt be an improper integral at +∞;
c tα
Z +∞
sin(βt)
1. If α > 1 then dt is absolutely convergent.
c tα
Z +∞
sin(βt)
2. If 0 < α ≤ 1 then dt is conditionnaly convergent.
c tα
Z +∞
sin(βt)
3. If α ≤ 0 then dt is divergente.
c tα

Z +∞
cos(βt)
Remark We have the same result for dt with c > 0 and β 6= 0.
c tα

10
2.6 The Abel 2 test :

Theorem Let f, g ∈ RLoc ([a, b[):

If

F f is monotone, and bounded on [a, b[, and

Z b
F g(t) dt converges
a
Z b
then f (t)g(t) dt converges.
a

!
Z +∞
log t
Example Study the convergence of (Arctgt) dt
1 t2
log t
Let’s set : f (t) = Arctgt and g(t) = .
t2
f, g ∈ RLoc ([1, +∞[), By the Abel 2 test
1
f 0 (t) = ≥ 0 donc f is monotone;
1 + t2
π
|f (t)| = |arctgt| ≤ donc f is bounded.
2
Z +∞
log t Z +∞
log t
2
dt and dt have the same behavior at +∞.
Z1+∞ t 2 t2
log t
dt is convergent since it’s a Bertrand integral at +∞ with α = 2  1.
2 t2 Z +∞
log t
Conclusion: (Arctgt) 2 dt is convergente.
1 t

2.7 Integration by parts and chang of variable in an improper inte-

gral
2.7.1 Chang of variable

Theorem Let

• a ∈ R and b ∈ R ∪ {+∞}

• f a continuous function on I

• φ is a strictly monotone function of class C 1 on J = [a, b[ such that φ (J) ⊂ I.

• L = lim φ(x). (the limit exsist since φ is monotone)

x→b
<

Z b Z L
0
f (φ(t)) φ (t) dt and f (u) du are either both convergent or both divergent.
a φ(a)
If they converge, they will be equal.

Remark : We have the same result in ]a, b] or ]a, b[.

11
 Z +∞
Example Study the convergence of sin(t2 ) dt :
1

f (t) = sin(t2 ) ∈ RLoc ([1, +∞[ only one singularity at + ∞.

Let y = t2 , so
dy 1
= 2t =⇒ dt = √ dy
dt 2 y
Z +∞ Z +∞
sin(y)
sin(t2 )dt has the same behavior at +∞ as √ dy.
1 1 2 y
or Z +∞
sin(y) 1
√ dy converges because >0
1 2 y 2
Therefore Z +∞
sin(t2 )dt conveges.
1

2.7.2 Integration par parts

Theorem :
Let f, g ∈ RLoc ([a, b[) with f, g differentiable on [a, b[ , such that f g 0 , f 0 g ∈ RLoc ([a, b[).
If
lim f (x) g(x) ∈ R
x→b
<

then
Z b Z b
0
f (t)g (t) dt and f 0 (t)g(t) dt either both converge or both diverge.
a a

Whent they converge, we get

 
Z b Z b
f (t) g 0 (t) dt = lim f (x) g(x) − f (a) g(a) − f 0 (t) g(t) dt.
a x→b a
<

Z +∞
sin(t)
Example Study the convergence of √ dt.
1 t

2.8 Limied developpement technic

Example Use the Limited developpemnt to determine if the integral converges or diverges;
!
Z +∞
sin(t)
log 1 + √ dt.
0 t

sin(t)
f (t) = log(1 + √ ) ∈ RLoc (]0, +∞[)
t

; the function has two singularities one at 0+ the other at +∞.

at 0+ :

12
sin(t) t √ sin(t) sin(t)
√ v √ = t =⇒ lim √ = 0 =⇒ lim log(1 + √ ) = 0.
t 0 +
t t→0+
t t→0+
t
so 0 is a removable singularity .

at +∞:

sin(t) sin(t) sin(t) 1 sin2 (t) sin2 (t)

lim √ = 0 =⇒ log(1 + √ ) = √ − + ε(t).
t→+∞ t t t 2 t t
Z +∞
sin(t) 1
√ dt converges because > 0.
A>0 t 2
1 sin2 (t) sin2 (t) 1 sin2 (t)
− + ε(t) v − <0
2 t t +∞ 2 t
By the equivalent test the integrals have the same behavior at +∞.

sin2 (t) 1 − cos(2t) 1 cos(2t)

= = − .
t 2t 2t 2t
Z +∞
cos(2t) Z +∞
1
dt converges because 1 > 0 and dt diverges
A>0 2t A>0 2t
Therfore
Z +∞
sin t
log(1+ √ )dt diverges because it is a sum of convergent integral and divergent integral.
A>0 t

13