Calculus on Rn

Seminar 10 2020

Multiple Integrals

1 Multiple Integrals Over Compact Intervals

Let a1 < b1 , a2 < b2 ,...,an < bn be real numbers. Then, the product of intervals

[a1 , b1 ] × [a2 , b2 ] × ... × [an , bn ] ⊂ Rn

is a compact interval in the vector space Rn .

For example, in R2 , such a compact interval is a rectangle, while in R3 , such an interval
is a rectangular parallelepiped.
Continuous functions defined on compact intervals from Rn can be Riemann integ-
rated, with a theory similar to the one in the real case. Thus, the Riemann integral is
the limit of a sequence of Riemann sums, defined in terms of the values of the functions
at certain intermediate points, and the measure of some partitions. (For details see the
lecture notes)
In practice, when computing multiple integrals over compact sets (much like when
computing partial derivatives), we integrate with respect to one variable (at the same
time consider all the others as constant), and continue the procedure, until we are done
integrating with respect to all variables.
The order in which we integrate is random, so we are free to start with whichever
variable we like. We just have to make sure that we do not forget the domain it is
considered on. Moreover, as we will see in some exercises, sometimes the exercise is easier
to solve when starting with one of the variables, and harder if when starting with the
other one. (This depends on the particular expression of the function we have to integrate)

Example 1:
Determine the following double integral
Z 2Z 5 
1
x+ dxdy.
0 1 y

Solution:
First we identify the elements of the problem. We have the set

A = [0, 2] × [1, 5] ⊂ R2

which is a rectangle. The function is

1
f : A → R, f (x, y) = x + , ∀(x, y) ∈ A.
y

1
Since f is a composition of elementary functions, it is continuous on A, and this A is a
compact set. Hence f is Riemann integrable on A. We start computing the integral. As
we mentioned earlier, the order of integration is random, but we have to be careful at
where we apply the values. Let us consider the case when we start integrating first with
respect to y.
Z 2Z 5  Z x=2 Z y=5   
1 1
I= x+ dxdy = x+ dy dx.
0 1 y x=0 y=1 y
In practice, sometimes we can see better if we solve separately the integral. Recall that
in this approach the variable x is considered a constant. Then
Z y=5   Z y=5 Z y=5 Z y=5 Z y=5
1 1 1
J= x+ dy = xdy + dy = x dy + dy =
y=1 y y=1 y=1 y y=1 y=1 y

= x · y|y=5 y=5
y=1 + ln y|y=1 = x(5 − 1) + ln 5 − ln 1 = 4x + ln 5.

Now we are done with y, so we may pass to x, thus completing I.

Z x=2 Z x=2 Z 2 Z 2
I= Jdx = (4x + ln 5) dx = 4 xdx + ln 5 · dx =
x=0 x=0 0 0

x2 2
=4· | + ln 5 · x|20 = 8 + 2 ln 5.
2 0

Remark: If the function to integrate may be written as a product of function with

separable variable, so if, for

A ⊆ R2 , f : A → R, ∃g, h : R → R, f (x, y) = g(x) · h(y), ∀(x, y) ∈ A,

then Z Z Z Z
f (x, y)dxdy = g(x)dx · h(y)dy.

The same happens for more than two variables. This situation is encountered in the
following example.

Example 2:

Determine the following double integral

Z 2Z 5 
1
x· dxdy.
0 1 y

Solution:
First we identify the elements of the problem. We have the set

A = [0, 2] × [1, 5] ⊂ R2

2
which is a rectangle. The function to be integrated is
1
f : A → R, f (x, y) = x · , ∀(x, y) ∈ A.
y
Since f is a composition of elementary functions, it is continuous on A, and this A is
a compact set. Hence f is Riemann integrable on A. Moreover, in this case f can be
written as
f (x, y) = g(x) · h(y), with
1
g(x) = x ∀x ∈ [0, 2] and h(y) = ∀y ∈ [1, 5].
y
Then Z 2 Z 5
1
I= x · dxdy = I1 · I2
0 1 y
with
2
x2 2
Z
I1 = xdx = | =4
0 2 0
and Z 5
1
I2 = dy = ln y|51 = ln 5 − ln 1 = ln 5.
1 y
Then
I = I1 · I2 = 4 · ln 5.

Remark: Notice that in Example 1, the function does not have separable variables,
so the procedure adopted for Example 2 in terms of the multiplication of simple integrable
does not apply in that case.
Remark: The procedure when dealing with triple integrals is quite the same.

Example 3:

Determine the following double integral, where a, b, c > 0

Z aZ bZ c
2z
2
dxdydz.
0 0 0 (x + y + 1)

Solution:
First we identify the elements of the problem. We have the set

A = [0, a] × [0, b] × [0, c] ⊂ R3 .

which is a rectangular parallelepiped. Moreover,

1
f (x, y, z) = 2z · , ∀(x, y, z) ∈ A.
(x + y + 1)2

3
This function does not have all the variables separable, however, z can be separated, so,
there exist the functions

g : [0, a] × [0, b] → R h : [0, c] → R

1
g(x, y) = and h(z) = 2z
(x + y + 1)2
so that
f (x, y, z) = g(x, y) · h(z).
Thus Z Z Z Z a Z b Z c
I= f (x, y, z)dxdydz = g(x, y)dxdy · h(z)dz.
A 0 0 0
Let us start with
Z aZ b Z a Z b Z a Z b
1
I1 = g(x, y)dxdy = g(x, y)dxdy = dxdy.
0 0 0 0 0 0 (x + y+)2

We see that the functions seems equally easy to integrate in both ways, thus we start
with y.
Z a Z b  Z a Z b  Z a
1 −2 −1 b


I1 = 2
dy dx = (x + y + 1) dy dx = −(x + y + 1) |0 dx =
0 0 (x + y + 1) 0 0 0
Z a 
1 1
=− − dx = − ln(x + b + 1)|a0 + ln(x + 1)|a0 =
0 x + b + 1 x + 1 + 0
(a + 1)(b + 1)
−(ln(a + b + 1) − ln(b + 1)) + (ln(a + 1) − ln 1) = ln .
a+b+1
Then we continue with c
z2 c
Z
I2 = 2zdz = 2 · | = c2 .
0 2 0
By combining them we get

(a + 1)(b + 1) 2 (a + 1)(b + 1)
I = I1 · I2 = ln · c = c2 · ln .
a+b+1 a+b+1

Remark: Be careful not to forget to integrate with respect to all variables, even in
the cases when in the explicit expression of f , not all the variables appear. This is the
case of the following example

Example 4:

Determine the following double integral, where a, b, c > 0

Z πZ πZ π
2 4
(sin x + cos y)dxdydz.
0 0 0

4
Solution:
In this case we start directly
Z πZ πZ π Z π Z π Z π 
2 4 2 4
I= (sin x + cos y)dxdydz = (sin x + cos y)dz dxdy
0 0 0 0 0 0

We notice that
Z π Z π
(sin x + cos y)dz = (sin x + cos y) dz = (sin x + cos y) · z|π0 = (sin x + cos y) · π
0 0

Hence
Z π Z π Z π Z π
!
2 4 2 4
I= (π · (sin x + cos y)) = π · (sin x + cos y)dy dx =
0 0 0 0

Z π Z π √ !
2
 π  2 π 2
=π· (y sin x + sin y) |0 4
dx = π · sin x + dx =
0 0 4 2
" √ # √ !
π  π 2 π π 2π
=π· · − cos x|0 +
2
· x|02 = π · (−0 + 1) + =
4 2 4 4
√
1+ 2 2
= ·π .
4

Homework 1: Solve the exercises 1 and 2 from Seminar9-ttrif

2 Integration over Nonempty Bounded Sets

In Rn there are many bounded sets, which are not generalized compact intervals in Rn .
ON such set, integration is possible. In R2 we are able to deliver the value of the Riemann
integral, when the set is a normal domain with respect to one of the axes (ox or oy - we
talked such details during the live seminar on Friday 08.05.2020)
A set A ⊂ R2 is said to be normal w.r.t to the 0y axis, if x ∈ [a, b], thus x is in a
compact interval [a, b] and y can be expressed as bounded by two functions depending on
x, hence, there exist the functions ψ1 and ψ2 : [a, b] → R such that

ψ1 (x) ≤ y ≤ ψ2 (x).

In this case Z Z Z x=b Z y=ψ2 (x)

f (x, y)dxdy = f (x, y)dxdy
A x=a y=ψ1 (x)

and in order to solve if, we have to start integrating first with respect to the dependent
variable, which in this case is variable y. Therefore
!
Z Z Z Z x=b y=ψ2 (x)
f (x, y)dxdy = f (x, y)dy dx
A x=a y=ψ1 (x)

5
.
A set B ⊂ R2 is said to be normal w.r.t to the 0z axis, if y ∈ [c, d], thus y is in a
compact interval [c, d] and x can be expressed as bounded by two functions depending
on y, hence, there exist the functions φ1 and φ2 : [c, d] → R such that

φ1 (x) ≤ y ≤ φ2 (x).

In this case Z Z Z x=φ2 (y) Z y=d

f (x, y)dxdy = f (x, y)dxdy
B x=φ1 (y) y=c

and in order to solve if, we have to start integrating first with respect to the dependent
variable, which in this case is variable x. Therefore
!
Z Z Z Zx=d x=φ2 (x)
f (x, y)dxdy = f (x, y)dx dy
B y=c x=φ1 (x)

Remark:In practice the same set A my be normal with respect to both axes or just
one. The picture attached to the set is extremely useful in finding a good approach. In
the case when A is normal w.r.t. to both axes we are free to fix any of the variables.
In this case the choice depends on the structure of A and on the explicit formula of the
function f .

Example 5:

Determine the following double integral

Z Z
f (x, y)dxdy
A

where

A = {(x, y) ∈ R: (x, y) is bounded by y = x2 andx = 3},

and
2x
f (x, y) = , ∀(x, y) ∈ A.
(1 + x2 + y)2
Solution:
If you plot the image you can see that the set A is situated below the graph of the
function y = x2 , is bounded by the ox axes and by the line with the equation x = 3.
Now, judging by the things explained at the beginning of the section, we see that our
function is a normal domain with respect to both axes. We are going to solve it in two
ways, so that you can notice easily the difference.

6
 Approach 1: We consider the set A normal with respect to the 0y axis. This means
that x gets to have exact bounds, which can be read from the graph. In this case

x ∈ [0, 3].

Then, y has to be bounded by functions depending on x. Because the set is below the
graph of y = x2 , we immediately get the condition

y ≤ x2 .

Now we need a lower bound for y. By looking at the graph... we are not given that bound
explicitly, however, we notice that the only lower bound in this case is the line having
the equation y = 0. Because we are situated above it, we get the lower bound for y, thus

0 ≤ y.

Notice that in this case ψ1 (x) = 0 is a the constant function, while ψ2 (x) = x2 . Therefore
Z Z Z x=3 Z y=x2
2x
I= f (x, y)dxdy = dxdy =
A x=0 y=0 (1 + x2 + y)2
Z x=3 Z y=x2 !
2x
= dy dx.
x=0 y=0 (1 + x2 + y)2
Now we are forced to start integrating with respect to y.
Z y=x2 Z y=x2
2x 1 1 y=x2
I1 = dy = 2x dy = −2x | =
y=0 (1 + x2 + y)2 y=0 (1 + x2 + y)2 1 + x2 + y y=0
 
1 1 2x 2x
= −2x 2
− 2
= 2
− .
1 + 2x 1+x 1+x 1 + 2x2
Going back to the original problem, we get
Z x=3 Z x=3  
2x 2x 1
I= I1 dx = 2
− 2
dx = ln(1 + x2 )|30 − ln(1 + 2x2 )|30 =
x=0 x=0 1+x 1 + 2x 2
r
1 100
= ln 10 − ln 19 = ln .
2 19

Approach 2: This approach is considered just for the sake of a good exemplification
(Approach 1 is more than enough for this exercise). In this case we consider the set A
normal with respect to the 0x axis. This means that y gets to have exact bounds, which
can be read from the graph. In this case

y ∈ [0, 9].

Then, x has to be bounded by functions depending on y. Because the set is below the
graph of y = x2 , and all the values are positive we immediately get the condition
√
y < x.

7
Now we need an uppr bound for x. By looking at the graph... we are not given that bound
explicitly, however, we notice that the only upper bound in this case is the line having
the equation x = 3. Because we are situated on its left, we get the condition

x ≤ 3.

Passing on to the integration we get

Z Z Z x=3 Z y=9
2x
I= f (x, y)dxdy = dxdy =
A
√
x= y y=0 (1 + x2 + y)2

We are forced to start integrating with respect to x, since its limits depend on y. Thus,
we have Z Z Z y=9 Z x=3 !
2x
I= f (x, y)dxdy = √ 2 2
dx dy =
A y=0 x= y (1 + x + y)

In order to simplify the notation, we compute

Z x=3
2x 1
J= 2
√ (1 + x + y) 2
dx == − |x=3√ =
2 + y x= y
x= y 1 + x
 
1 1 1 1
=− − = − .
10 + y 1 + 2y 1 + 2y 10 + y
Going back to the original problem, we get
Z y=9 Z y=9  
1 1 1
I= Jdy = − dy = ln(1 + 2y)|90 − ln(10 + y)|90 =
y=0 y=0 1 + 2y 10 + y 2
r
1 1 100
= (ln 19 − ln 1) − ln 19 + ln 10 = − ln 19 + ln 10 = ln .
2 2 19
Not at all surprisingly :), the results of both approached are the same.

Homework 2: Solve the exercises 3,4 and 5 from Seminar9-ttrif

Homework 3: Take a look at the remaining exercises from
Seminar9-ttrif. We will tackle them with change of variables in
the following seminar.