Fubini's Theorem
Fubini's Theorem :-
In the world of mathematics integration plays an important role . So we have discussed the advance form of integration which results the famous theorem Fubini's Theorem.
Statement:
If a double integral , I = ∫∫ f dx dy
R
exists over a rectangle R = [a,b;c,d] , and if
d
∫ f dy also exists , for each fixed x in [a,b] ,
c b d
then the iterated integral ∫ dx ∫ f dy exists
a c
and is equal to double integral I.
Proof :
Let ε be any positive number.
Since the upper integral , Iᵘ is the infimum of the upper sums , there exists a partition P of R such that
Σ Σ Mᵢⱼ ΔRᵢⱼ < Iᵘ + ε
i j
or Σ Σ Mᵢⱼ Δxᵢ Δyⱼ < Iᵘ + ε ...........(1)
i j
Again , since for any fixed value of x ∈ Δxᵢ
, Mᵢⱼ is upper bound (not necessarily the supremum) of f in Δyⱼ , therefore
_d
∫ f dy ≤ Σ Mᵢⱼ Δyⱼ , when x∈ Δxᵢ
c j ..........(2)
, Mᵢⱼ is upper bound (not necessarily the supremum) of f in Δyⱼ , therefore
_d
∫ f dy ≤ Σ Mᵢⱼ Δyⱼ , when x∈ Δxᵢ
c j ..........(2)
Again , from equation(2) , Σ Μᵢⱼ Δyⱼ is upper bound of the function
d
Φ(x) = ∫ f dy , x∈ Δxᵢ
c
therefore by the same reasoning as above , we get
b
∫̅ Φ(x) dx ≤ Σ(Σ Μᵢⱼ Δyⱼ)Δxᵢ
a i j
or
_b _d
∫ dx ∫ f dy ≤ Σ Σ Mᵢⱼ Δyⱼ Δxᵢ < Iᵘ +ε
a c i j [by(1)] ........(3)
Also , since by hypothesis ,
d _d d
∫ f dy = ∫ f dy = ∫ f dy
̅ c c c
and ε is an arbitrary positive number , we get from equation(3),
_b _d _b d
∫ dx ∫ f dy = ∫ dx ∫ f dy ≤ Iᵘ ........(4)
a c a c
By considering the lower integral Iₗ, we can similarly show that
b _d
∫ dx ∫ f dy ≥ Iₗ ..........(5)
̅ a a
Again , since Iₗ = Iᵘ = I as the double integral exists , from (4) and (5) , we get
b d _d
I ≤ ∫ dx ∫ f dy ≤ ∫ f dy ≤ I
̅a c a
b d _b d
⇒ ∫ dx ∫ f dy = ∫ dx ∫ f dy = ∫∫ f dx dy
̅ a c a c R
b d
Thus ∫ [∫ f dy ] dx exists and equals
a c
∫∫ f dx dy . (Proved)
R
Concise Introduction To Measure Theory for undergraduate students .
d
Φ(x) = ∫ f dy , x∈ Δxᵢ
c
therefore by the same reasoning as above , we get
b
∫̅ Φ(x) dx ≤ Σ(Σ Μᵢⱼ Δyⱼ)Δxᵢ
a i j
or
_b _d
∫ dx ∫ f dy ≤ Σ Σ Mᵢⱼ Δyⱼ Δxᵢ < Iᵘ +ε
a c i j [by(1)] ........(3)
Also , since by hypothesis ,
d _d d
∫ f dy = ∫ f dy = ∫ f dy
̅ c c c
and ε is an arbitrary positive number , we get from equation(3),
_b _d _b d
∫ dx ∫ f dy = ∫ dx ∫ f dy ≤ Iᵘ ........(4)
a c a c
By considering the lower integral Iₗ, we can similarly show that
b _d
∫ dx ∫ f dy ≥ Iₗ ..........(5)
̅ a a
Again , since Iₗ = Iᵘ = I as the double integral exists , from (4) and (5) , we get
b d _d
I ≤ ∫ dx ∫ f dy ≤ ∫ f dy ≤ I
̅a c a
b d _b d
⇒ ∫ dx ∫ f dy = ∫ dx ∫ f dy = ∫∫ f dx dy
̅ a c a c R
b d
Thus ∫ [∫ f dy ] dx exists and equals
a c
∫∫ f dx dy . (Proved)
R
Note :
- Similarly , if ∫∫ f dx dy and ∫ f dx both exist
R a
then the double integral can be expressed as
d b
∫∫ f dx dy = ∫ dy ∫ f dx
R c a
2. Since ε is an arbitrary positive number , inequality (3) shows that
_b _d __
∫ dx ∫ f dy ≤ ∫∫ f dx dy
a c R
and similarly by considering the lower integrals ,
b d
∫ dx ∫ f dy ≥ ∫∫ f dx dy
̅ a ̅ c ̅ ̅ ̅ R
Remarks :
1. The theorem holds even if f has a finite number of discontinuities , or an infinite of discontinuities lying on a finite number of lines x = Cᵢ , i= 1,2,...m parallel to y-axis .
d
For , the function Φ(x) = ∫ f dy will be
c
discontinuous at a finite number of points C₁,C₂,.........,Cₘ only and will therefore be still integrable . A similar remark holds for discontinuities parallel to x - axis .
2. If a double integral exists , then the two repeated integrals cannot exist without being equal . However , if the double integral does not exist , nothing can be said about the repeated integrals ; they may or may not exist .
Also one of the repeated integrals may exist or even that both may exist and be equal and yet the double integral may not exist , i.e. the existence of one or of both of the repeated integrals is no guarantee for the existence of the double integrals .
However , if the two repeated integrals exist both are unequal , the double integral cannot exist .
Example :
A function is defined on a rectangle [0,1;0,1] as
{ 1/2 when y is rational
f(x,y) = {
{ x , when y is irrational
Prove that the iterated integral
1 1
∫ dy ∫ f dx = 1/2 , but the other iterated
0 0
integral does not exist .
Proof :
1 1
For y rational ∫ f dx = ∫ 1/2 dx = 1/2
0 0
For y irrational
1 1
∫ f dx = ∫ x dx = 1/2
0 0
1 1 1
∴ ∫ dy ∫ f dx = ∫ 1/2 dx = 1/2
0 0 0
Proceeding as for a function of a single
variable , we may show that the integral
1 1 1
∫ f dy does not exist and so ∫ dx ∫ f dy
0 0 0
also does not exist .
It was shown earlier that the double integral ∫∫ f dx dy does not exist .
R
(Proved)
About Scientist :-
Guido Fubini (19 January 1879 – 6 June 1943) was an Italian mathematician, known for Fubini's theorem and the Fubini–Study metric. Born in Venice, he was steered towards mathematics at an early age by his teachers and his father, who was himself a teacher of mathematics.For more
Recommended Books :
Measure And Integration by Springer for undergraduate studies.
Comments
Post a Comment
If Any Doubt Ask Me